Synthesis and conformational analysis of a conformationally constrained trisaccharide, and complexation properties with concanavalin A

Article abstract of DOI:10.1002/(SICI)1521-3765(19990802)5:8<2281::AID-CHEM2281>3.0.CO;2-D

The trisaccharide β-D-GlcNAc(1→2)α-D-Man(1→3)D-Man is a fragment of a biantennary glycan that is recognized by α-D-Man-specific lectins such as concanavalin A (ConA), Lathyrus ochrus lectin, lentil lectin, and can adopt several conformations upon binding.

Full text of DOI:10.1002/(SICI)1521-3765(19990802)5:8<2281::AID-CHEM2281>3.0.CO;2-D

FULL PAPER
Synthesis and Conformational Analysis of a Conformationally Constrained
Trisaccharide, and Complexation Properties with Concanavalin A
Nathalie Navarre,[a] Nicolas Amiot,[a, f] Anita van Oijen,[a] Anne Imberty,[b] Ana Poveda,[d]
Jesus JimeÂnez-Barbero,[c] Alan Cooper,[e] Margaret A. Nutley,[e] and Geert-Jan Boons*[a, f]
Abstract: The
trisaccharide
b-d-
ally constrained compound has a more
favorable entropy term but this term is
offset by a smaller enthalpy term. NMR
spectroscopic studies have shown that
the cyclic compound is indeed consid-
erably less flexible than the linear com-
pound and both compounds adopt main-
ly one conformation. SYBYL software
together with energy parameters appro-
priate for carbohydrates was used for a
systematic conformational search. The
linear compound is very flexible. A
clustering method determined seven
main conformational families. Six possi-
ble conformational families were iden-
tified for the cyclic compound when
considering the orientations of the
GlcNAc(1 !2)a-d-Man(1 !3)d-Man is
a fragment of a biantennary glycan that
is recognized by a-d-Man-specific lec-
tins such as concanavalin A (ConA),
Lathyrus ochrus lectin, lentil lectin, and
can adopt several conformations upon
binding. To probe the importance of loss
of flexibility of a saccharide during
binding with ConA this trisaccharide
has been synthesized in a conformation-
ally constrained form where a methyl-
ene acetal bridge mimics a GlcNAc ± O-
6'' ´´´ Man ± O-4 intramolecular hydro-
gen bond. Microcalorimetry measure-
ments revealed that the conformation-
bGlcNAc(1 !2)Man
and
abMan-
(1 !3)Man glycosidic bonds. The cen-
tral mannose residue was docked in the
binding site of ConA and the complex
was refined. The results are compared
with crystal structures of legume lectin ±
oligosaccharide complexes and with the
NMR and thermodynamic data.
Keywords: calorimetry
´ carbohy-
drates ´ lectins ´ molecular model-
ing ´ oligosaccharides
These interactions are also important in health science, and
are involved in the invasion and attachment of pathogens,
inflammation, metastasis, blood group immunology, and
xenotransplantation. Among protein ± carbohydrate com-
plexes those involving lectins are of considerable interest
because of the high specificities of their interactions.
Introduction
Protein ± carbohydrate interactions are implicated in many
essential biological processes. Examples include embryogen-
esis, fertilization, neuronal development, hormonal activities,
and cell proliferation and organization into specific tissues.
The molecular basis of lectin ± carbohydrate interactions
has been widely studied[1] but the thermodynamics of these
interactions are complex and not well understood.[2] Conse-
quently, the prediction of binding constants is still a difficult
and unreliable process. While counterexamples exist, pro-
tein ± carbohydrate association is typified by a favorable
enthalpy term, which is offset by an unfavorable entropy
contribution. It is widely accepted that the enthalpy term
arises from a large number of hydrogen bonds and extensive
van der Waals interactions. Furthermore, it has been pro-
posed that the dynamic rearrangement of water may also
contribute to the enthalpy of complexation. The entropy term
has been considered to arise either from solvation effects[3] or
from the loss of conformational flexibility of the carbohydrate
ligand.[4]
[a] Prof. G.-J. Boons, N. Navarre, N. Amiot, Dr. A. van Oijem
The School of Chemistry, University of Birmingham
Edgbaston, Birmingham B152TT (UK)
[b] Dr. A. Imberty[‡]
CERMAV-CNRS, BP 53
F-38041 Grenoble Cedex 9 (France)
Â
[c] Dr. J. Jimenez-Barbero
Instituto de Quimica Organica, C.S.I.C.
Juan de la Cierva 3, E-28006 Madrid (Spain)
[d] Dr. A. Poveda
Servicio Interdepartamental de Investigacion
Â
Universidad Autonoma de Madrid, Cantoblanco
E-28049 Madrid (Spain)
[e] Prof. A. Cooper, M. A. Nutley
Department of Chemistry, University of Glasgow
Glasgow G128QQ (UK)
[f] Present address:
The majority of oligosaccharides are flexible and in many
cases a lectin binds a saccharide in a secondary rather than a
global minimum conformation. For example, the bGlcNAc-
Complex Carbohydrate Research Center, 220 Riverbend Road
Athens, Georgia, GA 30602-4712 (USA)
[‡] Associated with University Joseph Fourier, Grenoble
Chem. Eur. J. 1999, 5, No. 8
ꢀ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 1999
0947-6539/99/0508-2281 $ 17.50+.50/0
2281

G.-J. Boons et al.
FULL PAPER
(1 !2)aMan(1 !3)Man trisaccharide is known to adopt
different conformations when crystallized in different crystal
forms of Lathyrus ochrus lectin.[5] Another different con-
formation is observed when it is complexed with animal
S-lectin.[6] It has been shown that many of these conforma-
tions do not correspond to the global lowest energy con-
formation of the uncomplexed trisaccharide but to secondary
minima.[7] This trisaccharide has not been cocrystallized with
ConA but has been the subject of a systematic conformational
search when docked in the lectin-binding site.[8] It has been
predicted that several conformations can be complexed in the
lectin-binding site.
with loss of flexibility of the cyclic oligosaccharide was
expected to be smaller. Indeed, the measured thermodynamic
data revealed a more favorable entropy of binding of
compound III, which surprisingly was offset by a less
favorable enthalpy term.
Results and Discussion
Synthesis of the cyclic trisaccharide:
Macrocyclization is a very challenging aspect in the
preparation of the conformationally restricted trisaccharide
III. A successful approach is outlined in Scheme 1. The
cyclization was achieved by an intramolecular glycosylation
utilizing the precursor 10, which already possessed a methyl-
ene acetal linker. This approach was more fruitful than
macrocyclization by methylene
In order to study both conformational and thermodynamic
aspects of protein ± carbohydrate interactions, two cyclic
trisaccharides have been designed (Figure 1). The design of
acetal formation.
The precursor 10 for the
macrocyclization was assem-
bled from the monomeric
building blocks 1,[10] 2,[11] and
6.[12] These compounds were
readily available by standard
carbohydrate protecting-group
interconversion
strategies.
Thus, coupling of glycosyl do-
nor 1 with glycosyl acceptor 2 in
dichloromethane/ether in the
presence of N-iodosuccinimide
Figure 1. Schematic representation of the linear trisaccharide and the cyclic analogues.
trisaccharide II[9] is based on
BnO
OBn
O
BnO
Ph
OAc
O
OAc
O
O
O
mimicking an intramolecular
hydrogen bond O2 ´´´ O6''
which is present in the crystal-
line lectin ± oligosaccharide
BnO
BnO
BnO
BnO
NIS/TMSOTf
OR
O
HO
OBn
+
R'O
OMe
O
SEt
1
2
3. R, R' = PhCH
NaCNBH3
CH3SCH3
4. R = Bn, R' = H
OMe
5. R = Bn, R' = MTM
complex[5] by a methylene ace-
tal. The O2 ´´´ O6'' hydrogen
bond is also present in a pre-
dicted docking mode of the
trisaccharide to ConA.[8] Trisac-
charide II has been recently
synthesized[9] and its binding
properties are currently under
investigation.
OBn
OBn
PhtN
RO
O
HO
BnO
O
O
TMSOTf
BnO
NIS/TfOH
OR'
OMP
BnO
O
BnO
BnO
OBn
OBn
5
NPht
O
6
O
O
7. R = MP, R' = Ac
8. R = MP, R' = H
9. R = R' = H
OMe
KOtBu/MeOH
CAN
10. R = OC(NH)CCl3, R' = H
Cl3CCN, DBU
OBn
OBn
OH
OH
PhtN
O
O
AcHN
O
OBn
OH
O
O
O
Here, we present the synthe-
sis of trisaccharide III, which
BnO
BnO
HO
O
O
1. H2NNH2
HO
HO
OBn
OH
BnO
O
O
2. Ac2O
O
O
contains
a methylene acetal
3. H2, Pd(OAc)2
O
O
that mimics an O-4 ´´´ O-6'' in-
tramolecular bond as was pre-
dicted to occur in another pos-
sible docking mode of bGlcN-
Ac(1 !2)aMan(1 !3)Man with ConA.[8] Molecular model-
ing and NMR studies have demonstrated that the introduction
of the methylene tether does not distort the conformation of
the cyclic compound, although its flexibility is significantly
reduced. On the basis of these results, it was anticipated that
the conformationally constrained trisaccharide III would lose
less conformational flexibility than the corresponding linear
trisaccharide and therefore the entropic barrier associated
11
III
OMe
OMe
Scheme 1. Synthesis of trisaccharide III.
(NIS)/catalytic trimethylsilyl triflate (TMSOTf)[13] afforded
the a-linked disaccharide 3 in an excellent yield (96%). The
benzylidene group of 3 was selectively opened by treatment
with sodium cyanoborohydride, followed by HCl in ether[14] to
give compound 4 (93%). The methylthiomethyl protecting
group was introduced[15] with dimethyl sulfide and benzoyl
peroxide in acetonitrile to yield compound 5 (87%). The
methylene acetal functionality was introduced by coupling of
2282
ꢀ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 1999
0947-6539/99/0508-2282 $ 17.50+.50/0
Chem. Eur. J. 1999, 5, No. 8

Cyclic Trisaccharide
2281±2294
Table 1. Low-energy conformations of linear trisaccharide I and structural
characteristics (torsion angle values in 8 and relative energy value in
kcalmol 1).
the compounds 5 and 6 in the presence of NIS/triflic acid
(TfOH), and the required saccharide 7 was obtained in a yield
of 76%. Next, methylene-linked compound 7 had to be
converted into a glycosyl donor suitable for an intramolecular
glycosylation. To this end, the protecting group at the
anomeric center of the 2-deoxy-2-phthalimidoglycosyl moiety
of compound 7 had to be converted into a suitable leaving
group and the 2-OH had to be deprotected. We selected an
anomeric trichloroacetamido functionality as the anomeric
leaving group.[16] Thus, cleavage of the acetyl group of 7 with
potassium tert-butoxide in methanol gave 8 and the p-
methoxyphenyl group (MP) of 8 was removed by reaction
with cerium ammonium nitrate (CAN)[17] to afford 9. Treat-
ment of 9 with trichloroacetonitrile and DBU resulted in the
formation of the trichloroacetimidate 10. The regioselectivity
of the latter reaction was achieved by virtue of the higher
acidity of the anomeric hydroxyl group; however, care had to
be taken to avoid bis(trichloroacetimidate) formation. In the
following step, TMSOTf-mediated intramolecular glycosyla-
tion of 10 gave the macrocyclic
Family
F1±2
Y1±2
F1±3
Y1±3
DE
Linear_AA
Linear_AB
Linear_BC
Linear_BA
Linear_BB
Linear_BD
Linear_DC
50.7
171.7
68.7
99.4
0.0
0.8
1.6
2.0
2.5
3.3
3.6
51.8
44.4
44.1
46.4
56.7
104.0
174.5
143.7
146.7
149.5
159.1
71.4
81.8
164.6
68.9
92.5
133.6
152.9
44.6
174.5
104.3
39.9
66.2
160.1
on the energy maps of both linkages (Figure 2) which were
calculated recently[19] with the MM3 program.[20] For both
disaccharides, the main low-energy region has been labeled A,
whereas the second low-energy region has been given the
label B. Regions C and D correspond to remote parts of the
main low-energy region that could be considered as secondary
minima. The conformations in Table 1 have been identified by
the orientation of each linkage: Linear_AB corresponds to a
compound 11 in a very good
yield of 82%. Interestingly, a
similar cyclization for the prep-
aration of II resulted in a low
yield of cyclic compound. Com-
pound 11 was deprotected as
follows: conversion of the
phthalimido functionality into
an NHAc moiety was accom-
plished by treatment with hy-
drazine monohydrate[18] fol-
lowed by acetylation with acetic
anhydride. The benzyl ether
protecting groups were re-
moved by catalytic hydrogena-
tion over palladium acetate to
afford the requisite saccharide
III.
Figure 2. Superimposition of the conformations predicted for the isolated and ConA-bond trisaccharide I and III
on the energy maps of each linkage, as calculated with the MM3 program.[19] The conformations observed in some
crystal structures of lectin ± oligosaccharide complexes are also depicted.
The structural integrity of
compound III was confirmed
by NMR spectroscopy and FAB
mass
spectrometry
(594,
[M‡Na] ). The 1H and 13C NMR signals were unambiguously
assigned by two-dimensional homonuclear correlation spec-
troscopy (COSY, TOCSY). The assignments were aided by
heteronuclear, proton ± carbon chemical shift correlation
experiments (d (13C) ˆ 103.1 (JC1, H1 ˆ 173.5 Hz, C-1), 102.0
(JC1', H1' ˆ 173.3 Hz, C-1'), 99.6 (JC1'', H1'' ˆ 166.4 Hz, C-1).
conformation with the bGlcNAc(1 !2)Man linkage in con-
formation A and the aMan(1 !3)Man linkage in conforma-
tion B of their respective energy map.
‡
The lowest energy conformation, Linear_AA, corresponds
to the one predicted by NMR and molecular modeling
studies.[21] However, some of the other conformations are not
much higher in energy. More particularly, the conformations
Linear_BA and Linear_BB correspond to those observed in
N-glycan fragments when cocrystallized with Lathyrus ochrus
lectin.[5, 6] The occurrence of a conformation (B on each map)
that greatly differs from the absolute minimum is a character-
istic of oligosaccharide flexibility. The occurrence of a small
percentage of these anti conformers has also been demon-
strated by high-resolution nuclear magnetic resonance spec-
troscopy (NMR). For the a-linkage, the secondary minimum
corresponds to the gauche orientation of the Y angle, which
has been proven to exist in solution in mannobiose[22] and
Molecular modeling of the uncomplexed trisaccharides:
A four-dimensional systematic conformational search per-
formed on the F and Y torsion angles of the bGlcNAc-
(1 !2)aMan(1 !3)Man trisaccharide I yielded 26712 con-
1
formations in a 10 kcalmol energy window. A cluster
analysis[8] gave seven families, and the lowest energy con-
formation of each of those families was fully optimized. The
resulting geometries are described in Table 1. All seven
conformations retained present very different geometries.
The F and Y values of these conformers have been reported
Chem. Eur. J. 1999, 5, No. 8
ꢀ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 1999
0947-6539/99/0508-2283 $ 17.50+.50/0
2283

G.-J. Boons et al.
FULL PAPER
maltose.[23] For the b-linkages, the secondary minimum is the
gauche conformation about the y angle that has been referred
to as the gauche ± gauche or alternate exo-anomeric effect.
Independent NMR studies on the b(1 !3) linkage,[24] the
b(1 !2) linkage,[25] and the C-analogue of the b(1 !4)
linkage[26] in solution demonstrated the existence of the
alternate conformation in addition to the usual syn con-
former.
A systematic conformational search was run on nine of the
torsion angles of the 13-membered-ring cyclic trisaccharide
III. The F and Y torsion angles of both glycosidic linkages as
well as five torsion angles of the methylene acetal bridge were
rotated, resulting in 1203 conformations. After partial opti-
mization and rejection of conformers with incorrect stereo-
chemistry or van der Waals conflicts, 792 conformations were
1
considered in an energy window of 10 kcalmol . As a result
of convergence during the last step of geometry optimization,
only 23 conformations presented different geometries in
terms of the 13-membered-ring shape and were stored. Since
the acetal bridge appears to be the most flexible part of this
macrocycle, the 23 conformations can be classified in six
families, based on the conformation of the two glycosidic
linkages (Table 2).
The conformations, grouped in six families, are displayed in
Figure 3. Despite the cyclization constraints, both linkages can
adopt several very different conformations in an energy
Figure 3. Graphical representation of the 23 low-energy conformers of
compound III distributed in six families, based on the conformations of the
glycosidic linkages. The lowest energy conformation of each family is
represented by sticks. Color-coding is as follows: red for the bGlcNAc,
green for the central aMan, and yellow for the aManOMe; the tether is
colored violet.
1
window of 10 kcalmol . For each family, the methylene
acetal bridge moves freely. In order to evaluate the influence
of cyclization on the conformational behavior of the glyco-
sidic linkages, the six conformations have been reported on
the energy maps of both linkages (Figure 2). From this
comparison it can be seen that the cyclization does not force
the linkages into high-energy regions: all the calculated
conformations are encountered by the most external iso-
1
energy contour (8 kcalmol ). The trisaccharide is still rather
flexible since many areas of the energy maps are spanned by
the calculated conformations. However, there is a significant
reduction of flexibility when compared with the linear
trisaccharide, and, for example, the AB conformation cannot
be adopted by the constrained
Table 2. Low-energy conformations of cyclic trisaccharide III, and their classification in conformational families
based on the conformation at the two glycosidic linkages (torsion angle values in 8 and relative energy value in
kcalmol 1).
compound.
When comparing the confor-
mational behavior of the linear
and cyclic trisaccharides, the
global minimum conformation
of each oligosaccharide appears
to be almost identical. Howev-
er, the secondary minima do
not appear in the same order.
Family
F1±2
Y1±2
F1±3
Y1±3
w1
w2
w3
w4
w5
DE
Cyclic_AA
44.4
169.4
75.9
89.3
76.1
56.9
42.3
61.0
73.2
42.2
47.3
42.6
44.5
74.5
80.7
69.1
94.5
44.1
76.2
97.5
112.2
107.2
128.6
117.8
161.9
162.6
176.6
24.2
142.5
50.3
88.5
159.6
85.9
166.3
72.4
179.9
155.0
51.8
175.5
173.7
118.7 0.0
95.8 1.3
57.7 2.6
122.5 3.9
154.9 4.0
103.8 4.2
40.2 4.8
37.1 8.2
40.1 8.4
43.3
49.4
38.4
37.6
29.1
47.7
38.7
38.6
168.8
163.7
170.7
175.7
172.1
163.3 123.2
166.1 122.8
156.6 143.4
62.9
77.8
77.5
67.6
97.1
84.5
117.5
146.1
138.2
86.3
101.8
81.7
60.6
Conformational analysis by
NMR spectroscopy:
Cyclic_BA
43.0
47.5
66.3
45.8
45.7
150.4
150.2
172.3
142.1
147.8
74.0
79.2
57.6
70.1
68.0
95.3
113.4
75.7
134.9
93.9
71.7
59.7
77.2
60.6
60.1
70.6
107.0
85.8
68.2
144.0
57.8
145.7
126.3
151.1
78.6
156.0
177.7
168.5
176.9
125.2
129.0 0.3
56.5 2.2
64.4 3.4
134.3 3.6
89.5 3.8
1
The H NMR and 13C NMR
spectra were completely as-
signed by a combination of
homonuclear COSY, TOCSY,
and heteronuclear HMQC,
HMBC, and HMQC-TOCSY
techniques. The latter two tech-
niques were crucial to resolve
the final ambiguities. The cor-
responding 1H and 13C NMR
chemical shifts of compounds I
and III are listed in Table 3.
Cyclic_BD
Cyclic_CA
52.0
148.1 140.5
55.0
65.6
73.4
95.4
144.7
66.0 2.7
59.6
57.0
51.6
54.7
86.5
83.0
77.2
84.5
54.2
52.8
51.9
52.0
53.1
46.4
63.7
48.0
69.7
56.7
45.3
45.4
131.5
76.6
120.3
139.5
57.6
40.9
132.8
58.2
142.5
165.7
166.7
134.7
52.7 3.3
102.7 4.4
57.7 6.4
80.0 7.3
Cyclic_CB
Cyclic_BB
68.0
96.2
74.0
1.5
49.3
75.4
452.8
23.9
98.1 5.6
69.0
68.6
65.1
167.0
179.2 124.5
161.1 141.5
87.4
16.3
30.8
26.7
63.7
66.0
58.5
112.9
101.5
101.1
148.7
160.9
153.7
2.3
38.9
32.4
99.2 5.8
118.9 6.2
107.8 6.4
2284
ꢀ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 1999
0947-6539/99/0508-2284 $ 17.50+.50/0
Chem. Eur. J. 1999, 5, No. 8

Cyclic Trisaccharide
2281±2294
Table 3. 500 MHz 1H and 13C NMR chemical shifts for trisaccharides I and
III at 303 K.
(two small consecutive JH1,H2 and JH2,H3), although all cross-
peaks H-2 !H-6(S, R) can be deduced from H-2 Mana(1 !3),
and H-2 !H-5 connectivities from H-2 ManaOMe, as ex-
pected for the Man residues adopting the 4C1 conformation. In
addition, medium-strong H-1 ± H-2 cross-peaks are observed
for both Man anomeric protons. The NOESY and ROESY
experiments were used to estimate proton ± proton interresi-
due distances qualitatively.[31] NOESY cross-peaks are pos-
itive at 299 K and 500 MHz. (Figures 4 and 5) 1H NMR cross-
relaxation rates[31, 32] (sROESY and sNOESY) were obtained from
the 2D-NOESY and 2D-ROESY experiments. sROESY/sNOESY
ratios[33] are independent of interproton distances and allow us
to estimate specific correlation times and therefore interpro-
ton distances (see Experimental Section). For both com-
pounds, overall correlation times around 0.2 ± 0.3 ns were
obtained. This fact indicates that both molecules tumble
almost isotropically in solution and that the isolated-spin-pair
approximation (ISPA) can be safely applied to deduce
proton ± proton distances. In fact, the results for the distances
given in Tables 4 ± 6 are very similar to those estimated by the
ISPA method. The corresponding H-1 ± H-2 intraresidue
signals were used as reference (2.4 Š) for the a-linkages,
and the H-1 ± H-3 and H-1 ± H-5 intensities for the unique b-
linkage. The distances calculated by molecular mechanics are
also shown in Tables 4 ± 6.
Proton
Trisaccharide I
Trisaccharide III
GlcNAc H-1
GlcNAc H-2
GlcNAc H-3
GlcNAc H-4
4.52/101.7
3.68/57.8
3.53/75.8
3.49/72.2
3.47/78.3
3.92/63.0
3.79/63.0
5.12/100.3
4.16/78.7
3.89/71.9
3.52/69.6
3.69/76.0
3.91/63.6
3.64/63.6
4.72/102.2
4.06/72.0
3.85/80.5
3.75/68.2
3.65/75.0
3.95/63.0
3.78/63.0
±
4.92/99.6
3.95/56.6
3.61/76.4
3.53/71.6
3.71/78.7
4.10/70.4
4.02/70.4
5.68/102.0
4.38/75.0
4.01/71.8
3.71/69.3
3.72/75.7
3.97/63.4
3.86/63.4
4.82/103.1
4.10/72.5
4.11/77.7
3.84/76.9
3.77/74.4
3.95/62.7
3.84/62.7
5.17/100.4
5.03/100.4
GlcNAc H-5
GlcNAc H-6A
GlcNAc H-6B
Mana(1 !3) H-1
Mana(1 !3) H-2
Mana(1 !3) H-3
Mana(1 !3) H-4
Mana(1 !3) H-5
Mana(1 !3) H-6A
Mana(1 !3) H-6B
ManaOMe H-1
ManaOMe H-2
ManaOMe H-3
ManaOMe H-4
ManaOMe H-5
ManaOMe H-6A
ManaOMe H-6B
pro-R CH2
pro-S CH2
±
The existence of molecular motion around the glycosidic
linkages of oligosaccharides has now been firmly establish-
ed.[27±29] In addition, recent investigations have revealed that
the rates of overall and internal motions of small and medium-
size oligosaccharides may occur on similar time scales.[30]
Since NMR parameters are es-
For the acyclic trisaccharide I, the experimentally deduced
NMR distances arise from different conformational isomers
(Table 4). Indeed, the observed NOEs for the Man-
a(1 !3)Man linkage (weak NOE for Mana(1 !3) H-1 ±
sentially time-averaged, the in-
formation that can be deduced
from NOE experiments corre-
sponds to a time-averaged con-
formation in solution.
For both trisaccharides, the
pyranoid rings can be described
as essentially monoconforma-
4
tional C1, as deduced from the
proton ± proton coupling pat-
terns and intraresidue NOE
data. Small couplings are ob-
served for both Man anomeric
protons (<2 Hz) and H-2 Man-
a(1 !3) (<4 Hz), while an
8.4 Hz coupling is measured
for the GlcNAc analogue. In
fact, for the GlcNAc residue all
proton chemical shifts (H-
1 !H-6(S, R)) can be obtained
through a TOCSY experiment
as expected for an all-axial
conformation of the ring pro-
tons. In addition, medium-
strong H-1 ± H-3 and H-1 ± H-5
cross-peaks support the usual
chair conformation. For both
Man residues, the TOCSY
Figure 4. NOE spectrum of compound I.
transfer from H-1 stops at H-2
Chem. Eur. J. 1999, 5, No. 8
ꢀ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 1999
0947-6539/99/0508-2285 $ 17.50+.50/0
2285

G.-J. Boons et al.
FULL PAPER
Mana H-2 NOE contacts are
fairly strong. Since no GlcNAcb
H-2 ± Mana H-2 NOE is ob-
served, conformer B is less than
5% present in solution at this
linkage, within experimental er-
ror. In addition, no conformer
can simultaneously satisfy the
corresponding two close distan-
ces, and therefore the presence
of conformational averaging
between conformers A and D
is firmly confirmed.
The observed NOEs for cy-
clic compound III (Tables 5 and
6) are fairly distinct for the
GlcNAcb(1 !2)Man linkage
and now the interresidue
GlcNAcb H-1 ± Mana H-2 dis-
tance is appreciably larger than
the corresponding GlcNAcb
H-1 ± Mana H-1, as can be de-
duced from the relative inten-
sities of the NOEs. Again, no
GlcNAcb H-2 ± Mana H-2 is
observed; this indicates that
conformer B does not partici-
pate in the conformational
Figure 5. NOE spectrum of compound III.
Table 4. Experimental and calculated proton ± proton distances for trisacchar-
ide I. The experimental distances have been calculated from the NOE and ROE
cross-relaxation rates as described in the experimental part. The calculated
distances correspond to the representative structure for every local minimum.
Estimated errors in the experimental distances are smaller than 5%.
Table 5. Experimental and calculated proton ± proton distances for trisac-
charide III.
H ± H Pair GlcNAcb(1 !2)Mana linkage Mana(1 !3)Mana linkage
Conformer
Conformer
Exp
AA
BA
CA, CB
Exp
AA
BB
BD
H ± H Pair GlcNAcb(1 !2)Mana linkage Mana(1 !3)ManaOMe linkage
1 ± 1
1 ± 2
1 ± 3
1 ± 4
2 ± 1
2 ± 2
5 ± 2
2.2
3.2
nd
nd
nd
nd
nd
2.1
3.3
4.8
3.9
4.7
4.1
4.2
3.6
3.6
5.3
4.2
4.2
2.2
4.6
3.3
3.6
3.6
2.0
4.6
4.5
5.4
nd[a]
ov[b]
2.2
3.3
nd
6.0
4.3
2.4
3.7
7.1
5.2
2.4
4.7
3.5
3.6
2.3
6.0
3.9
5.1
5.9
4.2
2.4
3.5
6.2
5.0
3.9
Conformer
Conformer
Exp
A
B
C
D
Exp
A
B
C
D
1 ± 1
1 ± 2
1 ± 3
1 ± 4
2 ± 1
2 ± 2
5 ± 2
2.4
2.6
nd
nd
2.1
3.3
4.8
3.9
3.6
3.6
5.3
4.2
4.2
2.2
4.6
4.1
2.3
3.8
3.8
4.3
4.2
4.5
nd[a]
3.3
2.2
3.5
nd
6.0
4.3
2.4
3.7
7.1
5.2
2.4
4.7
3.5
3.6
2.3
6.0
3.9
5.1
4.5
2.1
2.9
4.3
5.3
3.3
5.0
5.9
4.2
2.4
3.5
6.2
5.0
3.9
nd
ov
ov[b] 4.7
[a] nd ˆ not determined. [b] ov ˆ overlapping signal.
nd
nd
4.1
4.2
nd
2.7
Table 6. Experimental and calculated proton ± proton distances for the
bridge methylene protons of trisaccharide III.
[a] nd ˆ not determined. [b] ov ˆ overlapping signal.
Pair
Exp AA BA BB BD CA CB
ManaOMe H-2, strong NOE for Mana(1 !3) H-1 ±
ManaOMe H-3, and medium NOE for Mana(1 !3) H-5 ±
ManaOMe H-2) are basically identical to those observed for
the disaccharide indicating that the GlcNAc residue does not
have a significant effect on the conformational behavior of
this linkage. The conformation around this glycosidic linkage
may be described as a conformational equilibrium around
conformer A, with minor excursions towards the regions
defined by conformers C and D, and in all cases the values of
the F and Y angles are positive. The NOE data obtained for
the GlcNAcb(1 !2)Man linkage unequivocally indicate that
the conformation of this linkage can be described by a
conformational equilibrium between conformers A and D,
since both GlcNAcb H-1 ± Mana H-1 and GlcNAcb H-1 ±
pro-R CH2 ± Mana(1 !3) H-1
pro-R CH2 ± Mana(1 !3) H-2
pro-R CH2 ± GlcNAcb H-6S
pro-R CH2 ± GlcNAcb H-6R
pro-R CH2 ± ManaOMe H-4
pro-S CH2 ± GlcNAcb H-6S
pro-S CH2 ± GlcNAcb H-6R
2.8
2.6
3.2
2.9
2.7
2.7
2.9
2.6 2.6 4.2 3.4 4.2 4.2
2.5 2.3 6.6 4.5 6.6 6.6
3.6 3.1 3.4 4.0 3.1 2.7
2.5 3.8 2.8 3.3 2.7 3.7
2.7 2.8 3.9 3.6 3.8 3.2
3.8 3.7 3.4 3.5 3.9 2.6
3.1 4.1 2.2 2.2 3.6 3.6
2.6 3.0 3.8 4.1 3.4 3.7
2.3 2.1 3.3 3.9 3.6 3.5
pro-S CH2 ± ManaOMe H-6AB 3.3
pro-S CH2 ± ManaOMe H-4 2.8
equilibrium. According to these results, the cyclization
rigidifies the GlcNAcb(1 !2)Man linkage and conformer A
is basically the only one in solution. On the other hand, the
NOEs for the Mana(1 !3)Man linkage are identical to the
linear compound. Although there is overlapping between
2286
ꢀ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 1999
0947-6539/99/0508-2286 $ 17.50+.50/0
Chem. Eur. J. 1999, 5, No. 8

Cyclic Trisaccharide
2281±2294
ManaOMe H-3 and H-2 protons, the NOE cross-peak
intensity between these resonances and Mana(1 !3)Man
H-1 can be ascribed exclusively to the ManaOMe H-3 ±
Mana(1 !3)Man H-1 cross-peak. In fact, according to the
modeling results, and despite the extensive search performed
linkage, while an equilibrium between conformers A and D
was predicted. The calculations for compound III predict a
major conformer AA with variation around the angles
defining the methylene orientation. These predictions are
consistent with the experimental data, and the only discrep-
ancy is the absence of conformer BA, which according to the
calculations should be present to some extent.
(see below), no conformer with
a ManaOMe H-2 ±
Mana(1 !3)Man H-1 distance below 3.82 Š could be detect-
ed. This is due to the fact that conformation C around this
glycosidic linkage cannot be attained because of the cycliza-
tion. Therefore, the presence of the cycle poses a major
constraint around the GlcNAcb(1 !2)Mana linkage and now
the experimental NOEs of this compound can be described by
conformer AA. In the absence of J coupling data, the
assignment of diastereotopic protons heavily relies on the
presence of a major conformation.[31] Thus, by considering the
geometries (not the relative energies) deduced from the
modeling studies, the diastereotopic methylene bridge pro-
tons could also be assigned. As mentioned above, cyclization
of the trisaccharide highly restricts the GlcNAcb(1 !2)Mana
linkage. Several NOEs were observed for the methylene
bridge protons. Particularly important were those between the
low-field proton with both Mana(1 !3) H-1 and H-2. Again,
and according to the modeling results, for all the found
conformers, only the pro-R proton may show short distances
with these two protons at the Man residue, thus providing the
key for diastereotopic assignment. Additional NOEs are also
observed between both methylene bridge protons with
ManaOMe H-4, both GlcNAc and ManaOMe H-6s, and
ManaOMe H-5. The pattern of the observed NOEs between
both methylene bridge protons and the GlcNAc H-6s (strong
pro-R-high field, weak pro-R-low field, strong pro-S-low field,
weak pro-S-high field) seems to indicate that the low-field
GlcNAc H-6 proton is indeed the pro-S proton. This assign-
ment is also supported by observed intraresidue NOEs
between the methylene GlcNAc H-6s with H-4 and H-5, that
is medium/strong high field GlcNAc H-6 ± H-4, medium/
strong low-field GlcNAc H-6 ± H-5, medium/weak high field
GlcNAc H-6 ± H-5, weak low-field GlcNAc H-6 ± H-4. This is
the expected NOE pattern for the usual gg/gt equilibrium
around the GlcNAc C-5 ± C-6 bond.[34] The JH5, H6S and JH5, H6R
of this residue are 2.0 and 4.9 Hz and therefore the exper-
imental equilibrium is approximately gg:gt 60:40.[34]
Molecular modeling of the complexes between ConA and the
trisaccharides:
All conformational families determined for the linear
compound I were docked in the ConA binding site. All of
the seven tested conformations could be accommodated in the
binding site with no major changes in their geometries
(Table 7), except for the Linear_BD conformation, which
Table 7. Low-energy conformations of linear trisaccharide I and cyclic trisac-
charide III when docked in the binding site of ConA. Hb indicates the number
of hydrogen bonds between the ligand and the protein. DEtot (kcalmol 1) is the
difference between the energy of the complex and the energy of the
noninteracting protein and trisaccharide, whereas DEint is the interaction
energy between protein and ligand.
Conformation
F1±2
Y1±2
F1±3
Y1±3
Hb DEtot
DEint
ConA ± Linear_BA
ConA ± Linear_BC
ConA ± Linear_BB
ConA ± Linear_AA
ConA ± Linear_AB
ConA ± Linear_DC
52.2
151.7
150.5 161.7
146.5 139.3
166.6
158.2
99.5 152.6
71.0
102.3
166.1 11
8
37.9
45.9
51.5
54.3
55.8
52.6
82.8
35.4
35.0
31.5
31.1
26.4
43.7
43.2
38.2
39.2
42.3
69.8
87.2
68.6
145.7 11
88.9
55.7 10
90.1
20.2 10
61.3 11
71.6
8
6
6
89.9
83.9
ConA ± Cyclic_BA
ConA ± Cyclic_BD
ConA ± Cyclic_AA
ConA ± Cyclic_BB
ConA ± Cyclic_CA
ConA ± Cyclic_CD
48.8
52.9
55.9
57.3
62.1
42.6
148.9 70.7
9
39.2
36.3
33.4
31.0
27.5
24.2
45.6
42.3
41.2
40.9
47.6
38.3
149.1 141.3
154.7 100.5
147.1 156.6
83.1
156.8 128.4
7
60.3
8
gave a much higher energy complex than the others. The
complex with the lowest energy is ConA ± Linear_BA. The
preference of conformation B (the one that does not
correspond to the exo-anomeric effect) for the
bGlcNAc(1 !2)Man linkage is marked, since this orientation
is predicted to occur in the three lowest energy docking
modes. This result is in agreement with previous calculations
performed on the same complex by a different approach.[8]
The conformation of this linkage also affects the orientation
of the central mannose in the binding site. When conforma-
tion B is adopted, the mannose is very deep in the binding site
and all the hydrogen bonds have the characteristics observed
in the ConA ± aMan complex.[35] When conformation A is
taken, the mannose is not so deeply docked, probably because
of steric interactions of the GlcNAc with the protein surface,
and some hydrogen bonds are weaker. The GlcNAc residue
also establishes hydrogen bonds with the peptide, but only if
linkage b(1 !2) is not in conformation A. The third predicted
conformation for the bGlcNAc(1 !2)Man is termed D. In this
intermediate case, the mannose is not very deep in the binding
site, but many hydrogen bonds still exist, and furthermore, the
GlcNAc presents some additional hydrogen bonds with the
Thr226 side chain and Gly224 backbone. It has to be noted that
this peculiar conformation is the one observed for the
The simultaneous existence of all the aforementioned
NOEs can only be explained by the presence of mobility in
this region of the molecule. Therefore, a major conformer AA
may explain the conformational behavior of this trisaccharide,
with local flexibility around the C-6-O-CH2-O region of the
macrocycle. Although only qualitative, the geometry of this
conformer may explain the unusual chemical shift observed
for H-1 and H-2 of the Mana(1 !3) residue. These two atoms
are in close proximity to O-4 ManaOMe (2.3 Š) and O-5
GlcNAc (2.5 Š). In addition, the relative deshielding of the
pro-R methylene bridge proton with respect to the pro-S
analogue could also be explained by its proximity to O-5
GlcNAc (2.4 Š) in the AA conformation.
The modeling results of the uncomplexed molecules are in
good agreement with the experimental results, because the
calculations (see above) predict an equilibrium in which
conformer A is the major one for the GlcNAcb(1 !2)Man
Chem. Eur. J. 1999, 5, No. 8
ꢀ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 1999
0947-6539/99/0508-2287 $ 17.50+.50/0
2287

G.-J. Boons et al.
FULL PAPER
bGlcNAc(1 !2)Man disacchar-
ide in the crystal structure of
ConA ± pentasaccharide com-
plex.[36]
All of the 23 low-energy
conformations of the cyclic tri-
saccharide III were considered
for the docking study with
ConA. After the central man-
nose residue was fitted in the
binding site and the energy of
the complexes was optimized
with appropriate energy param-
eters,[37] 20 different conformers
were stored in an energy win-
1
dow of 15 kcalmol . The other
conformers were rejected ei-
ther because of their high en-
ergy or because their geometry
converged close to one already
considered. The conformers
have been clustered in six fam-
ilies based on the conformation
at the bGlcNAc(1 !2)Man and
aMan(1 !3)Man linkages (Ta-
ble 7). All conformations are
displayed in Figure 6. When
docked in the ConA binding
site, the trisaccharide can in
principle still adopt several dif-
ferent conformations. Howev-
er, as for the linear saccharide,
the
B conformation of the
bGlcNAc(1 !2)Man linkage is
energetically favored in the
ConA complex state.
Figure 6. Graphical representation of the 20 conformers of cyclic trisaccharide III when interacting with ConA,
distributed in six docking modes, based on the glycosidic linkage conformation. The protein is depicted by a
ribbon-type representation.
Complexation studies of the
saccharides with ConA:
Thermodynamic data ob-
tained by isothermal microca-
lorimetry for complexation of
Table 8. Thermodynamic data for the binding of trisaccharide I and III with ConA compared with literature data.
Compound
DG
[kcalmol
DH
[kcalmol
TDS
[kcalmol ]
Ref.
1
1
1
]
]
Man
aManOMe
4.4
5.3
5.3
5.3
5.3
5.2
5.7
5.7
6.2
6.0
5.9
6.1
5.8
5.7
6.6
6.8
8.2
6.8
5.3
10.2
7.8
10.7
7.4
9.5
4.6
2.7
1.3
1.3
1.6
2.9
1.5
0.1
4.5
2.0
4.5
1.4
3.6
1.5
3.1
[45]
[46]
[45]
[47]
[48]
[46]
[47]
[46]
[47]
[47]
Present work
Present work
Present work
the
disaccharide
aMan-
(1 !3)aManOMe, the trisac-
charide I and III with ConA
are listed in Table 8 together
with literature data for related
compounds. Although the en-
tropy and enthalpy contribu-
tions of particular interactions
vary between reporting authors,
some clear tendencies can be
identified. First, there is a lim-
bGlcNAc(1 !2)Man
aMan(1 !3)Man
aMan(1 !3)aManOMe
bGlcNAc(1 !2)aMan(1 !3)aManOMe (I)
4 !6, cyclic trisaccharide (III)
ited variation of free energy of binding between mono-, di-,
and trisaccharides, a phenomenon that is attributed to
entropy/enthalpy compensation. Second, binding of
aMan(1 !3)Man-containing fragments has a more favorable
enthalpy but a less favorable entropy contribution than its
bGlcNAc(1 !2)Man counterpart. The linear trisaccharide I
has a slightly more favorable energy of binding than both
disaccharide constituents do. Nevertheless, the thermody-
namic characteristics of the linear oligosaccharide are close to
those of the bGlcNAc(1!2)Man disaccharide.
It was anticipated that the conformationally constrained
trisaccharide III can adopt a conformation required for
2288
ꢀ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 1999
0947-6539/99/0508-2288 $ 17.50+.50/0
Chem. Eur. J. 1999, 5, No. 8

Cyclic Trisaccharide
2281±2294
binding with the lectin ConA but will lose less conformational
flexibility than the corresponding linear trisaccharide. There-
fore, the entropic barrier associated with loss of flexibility of
the cyclic oligosaccharide was expected to be smaller and
should thus result in more favored binding. Indeed, the
measured thermodynamic data revealed more favorable
entropy of binding of compound III, which surprisingly is
offset by a less favorable enthalpy term.
design high-affinity carbohydrate ligands.[40] The NMR studies
demonstrated that in solution both compounds populate the
BA binding conformer less than 5%. Thus, upon binding a
considerable unfavorable enthalpy term is introduced for this
conformational change and this term may well be larger for
the conformationally constrained compound explaining the
less favorable enthalpy of binding. We attempted to confirm
the predicted BA bound conformation of the saccharides by
measuring TR-NOEꢁs. However, despite major efforts, no
TR-NOEꢁs could be measured and the failure of these
experiments is probably due to the time scale of complex
dissociation and/or to the presence of a paramagnetic Mn2‡
ion close to the sugar-binding site.
The conformational studies have shown that the introduc-
tion of the methylene acetal tether does not distort the
conformation of compound III, but only reduces its conforma-
tional space as intended. Furthermore, the docking studies
indicate that both compounds complex in similar secondary
minimum energy conformations (BA) and have similar
interactions with the protein surface (Figure 7).
It is important to note that the calculations do not take into
account enthalpy effects arising from water rearrangement
and the assumption was made that the bulk solvent should
have the same enthalpic effect on the complexes involving the
linear and cyclic trisaccharides.
Recently, Bundle et al.[38] reported complexation studies of
preorganized branched trisaccharides with a monoclonal
However, we may speculate
that the desolvation enthalpy
of the cyclic compound is sig-
nificantly higher than for the
linear compound.
Conclusion
The synthesis of a cyclic con-
formationally constrained tri-
saccharide is desribed, the de-
sign of which is based on mim-
icking an internal hydrogen
bond which is present in a
docking mode of a ConA ± sac-
charide complex.
NMR spectroscopic and mo-
lecular modeling studies have
shown that the cyclic compound
III is indeed considerably less
flexible than the linear com-
pound I; however, both com-
pounds adopt mainly the con-
formation AA. It was anticipat-
ed that the entropy barrier of
complexation of compound III
Figure 7. Comparison of the lowest energy complexes of ConA and both the linear trisaccharide I and the cyclic
trisaccharide III. The solvent-accessible surface of the protein has been represented with the MOLCAD option in
the SYBYL package. The linear trisaccharide I is colored by atom type whereas the cyclic trisaccharide III is
colored by residue.
antibody. They found that the introduction of conformational
constraints had little effect on the thermodynamic parameters
of complexation and their data suggested that interresidue
flexibility is not a major contributor to the weak association
that characterizes oligosacharide ± protein interactions. They
also proposed that the absence of large impacts on DH and DS
implies that oligosaccharides display a restricted range of
conformations.
The data presented in this paper show that loss of flexibility
contributes significantly to the entropy of binding. Recently,
Quiocho and co-workers[39] made a similar observation for the
binding of linear and cyclic oligosaccharides to the maltodex-
trin-binding protein of Escherichia coli.
with ConA should be smaller because less conformational
flexibility is lost during binding. Indeed, the complexation has
a more favorable entropy term but surprisingly this term is
offset by a smaller enthalpy contribution, and the linear and
cyclic compounds have similar binding constants. Molecular
modeling studies have been performed to explain the loss of
enthalpy of binding, and it was predicted that both com-
pounds are complexed in the local minimum energy con-
formation BA. Figure 7 displays a comparison of the lowest
energy complexes of the two compounds; both have very
similar interactions with the protein, and thus the small
enthalpy term of complexation of the cyclic compound cannot
be explained by fewer interactions with the protein.
It is important to speculate about the origin of loss of
enthalpy of binding since it may provide opportunities to
The NMR studies have shown that in solution the BA
conformation for both compounds is less than 5% populated.
Chem. Eur. J. 1999, 5, No. 8
ꢀ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 1999
0947-6539/99/0508-2289 $ 17.50+.50/0
2289

G.-J. Boons et al.
FULL PAPER
dissolved in CH2Cl2 (100 mL), and the solution was washed with Na2S2O3
(1m, 2 Â 100 mL) and water (100 mL). The organic layer was dried
(MgSO4), filtered, and concentrated to dryness. Purification of the crude
product by column chromatography on silica gel (CH2Cl2/acetone, 99:1 v/v)
Thus, a considerable amount of energy will be lost for this
conformational change and it may be possible that this energy
term is considerably more unfavorable for the cyclic com-
pound. On the other hand, the observed differences of
enthalpy for III and I may arise from differences in solvent
reorganization.
The data presented in this paper suggest that loss of
flexibility during complexation accounts for a significant
entropic penalty. Further thermodynamic measurements and
NMR experiments will be conducted to validate the models
proposed in the present study, and these results will be
important for the future design of putative high-affinity
carbohydrate ligands. Cocrystallization experiments with
ConA and the ligand are under way, and these experiments
will confirm whether the ligands are bound in the predicted
BA conformation. Other conformationally constrained sac-
charides that adopt the BA conformation will be prepared to
study the general applicability of the observed binding data.
afforded the desired disaccharide (3) as a colorless oil (8.0 g, 96% yield); Rf
22
(acetone/CH2Cl2, 3:97 v/v) ˆ 0.70; [a]
ˆ ‡12.1 (CH2Cl2, c ˆ
ꢀkap†Dꢀ=kap†
13.13 mgmL 1); 1H NMR (300 MHz, CDCl3): d ˆ 7.52 ± 7.15 (m, 25H,
arom H), 5.63 (s, 1H, PhCH-), 5.62 (m, 1H, H-2'), 5.32 (d, 1H, J1', 2'
2.0 Hz, H-1'), 4.89 (d, 1H, Jgem ˆ 11.0 Hz, -CH2- benzyl), 4.67 (d, 1H, J1, 2
ˆ
ˆ
1.5 Hz, H-1), 4.77 ± 4.60 (4d, 4H, Jgem ˆ 12.0 Hz, -CH2- benzyl), 4.51 ± 4.41
(3d, 3H, -CH2- benzyl), 4.25 (m, 2H, H-3, H-5'), 3.98 (dd, 1H, J3', 2'
ˆ
3.5 Hz, J3', 4' ˆ 9.0 Hz, H-3'), 3.92 ± 3.65 (m, 8H, H-4, H-4', H-5, H-2, H-6,
H-6'), 3.30 (s, 3H, -OCH3), 2.10 (s, 3H, -CH3 acetyl); 13C NMR (75 MHz,
ˆ
CDCl3): d ˆ 170.1, (C O), 138.7, 138.4, 137.9, 137.4 (Cq benzyl ‡ Cq
benzylidene), 128.6 ± 126.1 (arom C), 101.1 (PhCH-), 100.6, 98.9 (C-1,
C-1'), 79.1, 78.0, 77.4, 74.3, 73.2, 72.2, 68.2, 63.9 (C-2, C-3, C-4, C-5, C-2',
C-3', C-4', C-5'), 75.1, 73.7, 73.5, 71.6 (-CH2- benzyl), 69.1, 68.8 (C-6, C-6'),
‡
54.9 (-OCH3), 21.1 (-CH3 acetyl); MS (FAB): m/z (%): 885 (4) [M‡K] ,
‡
‡
869 (100) [M‡Na] ; HRMS (FAB): calcd C50H54O12Na [M‡Na] 869.35;
found 869.35.
Methyl (2-O-acetyl-3,4,6-tri-O-benzyl-a-d-mannopyranosyl)-(1 !3)-2,6-
di-O-benzyl-a-d-mannopyranoside (4): 4 Š molecular sieves (7 g) and
sodium cyanoborohydride (4.8 g, 76.47 mmol) was added to a solution of
compound 3 (6.47 g, 7.65 mmol) in dry THF (100 mL). Subsequently, a
solution of HCl in ether (1m, 99 mL) was slowly added over a period of
20 min to the mixture, which was stirred for 3 h under argon. The reaction
mixture was neutralized by the addition of triethylamine, and washed with
saturated NaHCO3 (2 Â 100 mL) and water (100 mL). The organic layer
was dried (MgSO4), filtered, and concentrated under reduced pressure. The
crude product was purified by column chromatography with acetone/
CH2Cl2 (3:97 v/v) as the eluent, and the desired disaccharide (4) was
Experimental Section
Nomenclature: Schematic representations of the cyclic trisaccharide I and
III are given in Figure 1. The three carbohydrate residues are labeled from
unprimed to primed and double-primed from the ManOMe residue, by
analogy to the linear compound. The relative orientation of a pair of
contiguous residues about each glycosidic linkage is described by a set of
two torsional angles: F1±3 ˆ O-5'-C-1'-O-1'-C-3 and Y1±3 ˆ C-1'-O-1'-C-3-C-
4 for the aMan(1 !3)Man linkage and F1±2 ˆ O-5''-C-1''-O-1''-C-2' and
Y1±2 ˆ C-1''-O-1''-C-2'-C-3' for the bGlcNAc (1 !2)Man linkage. The
orientation of the methylene acetal bridge is given by a set of five torsion
angles: w1 ˆ O-5''-C-5''-C-6''-O-6'', w2 ˆ C-5''-C-6''-O-6''-C-7'', w3 ˆ C-6''-
O-6''-C-7''-O-4, w4 ˆ O-6''-C-7''-O-4-C-4 and w5 ˆ C-7''-O-4-C-4-C-5 (Fig-
ure 3). The sign of the torsion angles is given as proposed by the
Commission on Nomenclature.[41]
isolated as a colorless oil (6.03 g, 93% yield); Rf (acetone/CH2Cl2, 2:98
22
v/v) ˆ 0.10; [a]
ˆ ‡27.5 (CH2Cl2, c ˆ 29.26 mgmL 1); 1H NMR
ꢀkap†Dꢀ=kap†
(300 MHz, CDCl3): d ˆ 7.40 ± 7.12 (m, 25H, arom H), 5.50 (m, 1H, H-2'),
5.37 (s, 1H, H-1'), 4.89 (d, 1H, Jgem ˆ 11.0 Hz, -CH2- benzyl), 4.75 ± 4.45 (m,
11H, Jgem ˆ 11.4 Hz, -CH2- benzyl, H-1), 4.11 (dd, 1H, J4, 3 ˆ 9.2 Hz, H-4),
3.98 (m, 3H, H-3', H-5, H-5'), 3.74 (m, 7H, H-2, H-3, H-6, H-4', H-6'), 3.33
(s, 3H, -OCH3), 2.10 (s, 3H, -CH3 acetyl); 13C NMR (75 MHz, CDCl3): d ˆ
1
Equipment: H and 13C NMR spectra were recorded on a Bruker AC300
ˆ
170.4, (C O), 138.6, 138.0, 137.9 (Cq benzyl), 128.4-127.6 (arom C), 98.8,
98.7 (C-1, C-1'), 77.9, 77.5, 74.5, 71.9, 71.0, 68.8, 68.6 (C-2, C-3, C-4, C-5, C-2',
C-3', C-4', C-5'), 74.9, 73.7, 72.6, 71.8, 70.8, 69.4 (-CH2- benzyl, C-6, C-6'),
spectrometer equipped with a B-ACS60 autochanger, and an Aspect 3000
off-line editing computer. The 1H 2D-COSY[45] spectra were recorded on a
Bruker AMX400 spectrometer and an Aspect station I off-line editing
computer. Chemical shifts (d) were measured with tetramethylsilane as
internal standard. Fast atom bombardment (FAB) mass spectra were
recorded by means of a VG Zabspec spectrometer with m-nitrobenzyl
alcohol as matrix.
‡
54.9 (-OCH3), 21.1 (-CH3 acetyl); MS (FAB): m/z (%): 887 (4) [M‡K] ,
‡
‡
871 (100) [M‡Na] ; HRMS (FAB): calcd C50H56O12Na [M‡Na] 871.37;
found 871.37.
Methyl (2-O-acetyl-3,4,6-tri-O-benzyl-a-d-mannopyranosyl)-(1 !3)-2,6-
di-O-benzyl-4-O-methylthiomethyl-a-d-mannopyranoside (5): Dimethyl
sulfide (5.2 mL, 70.3 mmol) was added to a cooled solution (08C) of 4
(5.96 g, 7.03 mmol) in dry acetonitrile (80 mL). Benzoyl peroxide (6.8 g,
28.12 mmol) was added in portions over a period of 45 min under the
exclusion of light, and the mixture was stirred for 18 h and gradually
allowed to reach room temperature. TLC analysis (acetone/CH2Cl2, 3:97
v/v) showed complete conversion of the starting material into the product.
The solution was diluted with ethyl acetate, and washed with NaOH (1m,
2 Â 100 mL) and brine (100 mL). The organic layer was dried (MgSO4),
filtered, and concentrated in vacuo. Purification by column chromatog-
raphy on silica gel (petroleum ether 40 ± 60/CH2Cl2, 1:1 v/v, followed by
CH2Cl2 and acetone/CH2Cl2, 1:99 v/v) gave the compound 5 as a pale
All calculations were performed on Silicon Graphics workstations.
General methods and materials: All chemicals were purchased from
Aldrich, Fluka, and Lancaster. 4 Š Molecular sieves were purchased from
Avocado, activated at 3008C for 5 h, and stored at 1808C. Hydrogen and
nitrogen (White Spot) were supplied by British Oxygen Company.
All reaction solvents were distilled prior to use: dichloromethane and 1,2-
dichloroethane were distilled from P2O5, acetonitrile, pyridine, and diethyl
ether from CaH2, tetrahydrofuran (THF) from LiAlH4, and methanol from
magnesium activated with iodine.
Chromatography was performed with Merck 7734 silica gel 60, and flash
chromatography with Crosfield ES70X microspheroidal silica gel. TLC
analyses were carried out on silica gel plates (Merck 1.05554 Kieselgel
60F254). Compounds on TLC were visualized by UV light and/or by dipping
in H2SO4/MeOH (1:10 v/v/v) followed by subsequent charring at 1408C.
yellow foam (5.55 g, 87% yield); Rf (acetone/CH2Cl2, 3:97 v/v) ˆ 0.62;
22
[a]
ˆ ‡37.8 (CH2Cl2, c ˆ 25.73 mgmL 1); 1H NMR (300 MHz,
ꢀkap†Dꢀ=kap†
CDCl3): d ˆ 7.39 ± 7.15 (m, 25H, arom H), 5.35 (dd, 1H, J2', 1' ˆ 1.8 Hz,
Methyl (2-O-acetyl-3,4,6-tri-O-benzyl-a-d-mannopyranosyl)-(1 !3)-2-O-
benzyl-4,6-di-O-benzylidene-a-d-mannopyranoside (3): A mixture of ethyl
2-O-acetyl-3,4,6-tri-O-benzyl-1-thio-a-d-mannopyranoside (1) (6.06 g,
11.31 mmol), methyl 2-O-benzyl-4,6-di-O-benzylidene-a-d-mannopyrano-
side (2) (3.66 g, 9.84 mmol), and 4 Š powdered molecular sieves (10 g) was
J2', 3' ˆ 3.3 Hz, H-2'), 5.10 (d, 1H, H-1'), 4.89 (d, 1H, Jgem ˆ 11.0 Hz, -CH2-
benzyl), 4.78 (d, 1H, Jgem ˆ 11.4 Hz, -CH2- benzyl), 4.72 (d, 1H, J1, 2
ˆ
1.8 Hz, H-1), 4.67 (d, 1H, -CH2- benzyl), 4.64 ± 4.43 (m, 9H, -CH2SCH3,
-CH2- benzyl), 4.06 (dd, 1H, J3, 2 ˆ 3.3 Hz, J3, 4 ˆ 9.2 Hz, H-3), 3.97 (dd, 1H,
J3', 4' ˆ 9.2 Hz, H-3'), 3.88 (m, 2H, H-4, H-5'), 3.77 (m, 4H, H-2, H-4', H-6a,
H-6a'), 3.68 (m, 3H, H-5, H-6b, H-6b'), 3.30 (s, 3H, -OCH3), 2.22 (s, 3H,
-CH3 acetyl), 2.02 (s, 3H, -SCH3); 13C NMR (75 MHz, CDCl3): d ˆ 170.3,
stirred for 1 h in CH2Cl2 (120 mL). Subsequently,
a solution of N-
iodosuccinimide (120 mL, 0.1m in ether/CH2Cl2, 1:1 v/v) and TMSOTf
(0.232 mL, 1.2 mmol) were added. The mixture was stirred at room
temperature for 1 h, and then neutralized with triethylamine. The solution
was filtered through Celiteꢂ, washed with MeOH/CH2Cl2, 5:95 v/v/v, and
the combined filtrates were concentrated to dryness. The residue was
ˆ
(C O), 138.7, 138.4, 138.1, 137.9 (Cq benzyl), 128.5 ± 127.6 (arom C), 99.8,
99.3 (C-1, C-1'), 77.9, 77.3, 74.5, 74.4, 72.2, 71.4, 69.1 (C-2, C-3, C-4, C-5, C-2',
C-3', C-4', C-5'), 75.0, 73.5, 73.4, 73.0, 72.1, 72.0, 69.5, 69.3 (-CH2- benzyl,
-CH2SCH3, C-6, C-6'), 54.9 (-OCH3), 21.2 (-CH3 acetyl), 14.8 (-SCH3); MS
2290
ꢀ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 1999
0947-6539/99/0508-2290 $ 17.50+.50/0
Chem. Eur. J. 1999, 5, No. 8

Cyclic Trisaccharide
2281±2294
‡
‡
‡
‡
‡
(FAB): m/z (%): 948 (8) [M‡K] , 932 (100) [M‡Na] ; HRMS(FAB): calcd
[M‡K] , 1330 (100) [M‡Na] ; MS (FAB): calcd C77H81NO18Na [M‡Na]
‡
C52H60O12SNa [M‡Na] 931.37; found 931.37.
1330.54; found 1330.53.
Methyl 2,6-di-O-benzyl-3-O-(3,4,6-tri-O-benzyl-a-d-mannopyranosyl)-a-
d-mannopyranoside-trichloroacetimidate 3'',4''-di-O-benzyl-2''-deoxy-2''-
phthalimido-b-d-glucopyranoside-4,6''-methylidene acetal (10): - Trichlor-
oacetonitrile (0.782 mL, 7.80 mmol) and 1,8-diazabicyclo[5.4.0]undec-7-ene
(19.5 mL, 0.13 mmol) were added to a solution of 9 (1.7 g, 1.3 mmol) in
CH2Cl2 (30 mL) and the reaction mixture was stirred under argon. After
20 min the TLC analysis (CH2Cl2/acetone, 95:5 v/v) showed almost
complete conversion of the starting material into the product. The mixture
was concentrated in vacuo and the oily residue was applied onto a column
of silica gel and eluted with acetone/CH2Cl2 (1:99 v/v). Concentration of the
appropriate fractions gave 10 as a white foam (1.22 g, 65% yield); Rf
Methyl
2,6-di-O-benzyl-3-O-(2-O-acetyl-3,4,6-tri-O-benzyl-a-d-manno-
pyranosyl)-a-d-mannopyranoside-p-methoxyphenyl 3'',4''-di-O-benzyl-2''-
deoxy-2''-phthalimido-b-d-glucopyranoside-4,6''-methylidene acetal (7): A
mixture of 5 (5.1 g, 5.62 mmol), p-methoxyphenyl 3,4-di-O-benzyl-2-deoxy-
2-phthalimido-b-d-glucopyranoside (6) (2.90 g, 4.88 mmol) and powdered
molecular sieves (8 g) was stirred for 1 h in a mixture of THF and 1,2-
dichloroethane (1:1 v/v, 100 mL) under argon. The solution was cooled to
08C. N-iodosuccinimide (1.35 g, 6 mmol) was dissolved in a mixture of THF
and 1,2-dichloroethane (60 mL, 1:1 v/v). Trifluoromethanesulfonic acid
(54 mL, 0.6 mmol) was then added and the solution stirred for further 30 s.
The resulting mixture (0.1m, 58.7 mL) was added quickly to the initial
solution which was then stirred for 30 min at 08C. TLC analysis indicated
the conversion of the starting material into a major product. The reaction
mixture was neutralized by the addition of triethylamine and filtered
through Celite. The resulting solution was diluted with CH2Cl2
(200 mL)and washed with Na2S2O3 (2 Â 100 mL), NaHCO3 (100 mL) and
brine (100 mL). The organic layer was dried (MgSO4), filtered, and
concentrated to dryness. Purification of the crude product by column
chromatography on silica gel (CH2Cl2/acetone, 98:2 v/v) afforded the
(acetone/CH2Cl2, 5:95 v/v) ˆ 0.50; [a]2D2 ˆ ‡53.3 (CH2Cl2,
c
ˆ1
1
9.6 mgmL 1); H NMR (300 MHz, CDCl3): d ˆ 8.41 (s, 1H, C NH), 7.65
ˆ
(brs, 4H, arom H -Pht), 7.40 ± 6.80 (m, 35H, arom H), 6.37 (d, 1H, J1'', 2''
ˆ
8.5 Hz, H-1''), 5.15 (d, 1H, J1', 2' ˆ 1.1 Hz, H-1'), 4.87 (d, 1H, Jgem ˆ 11.4 Hz,
-CH2- benzyl), 4.85 (d, 1H, Jgem ˆ 11.0 Hz, -CH2- benzyl), 4.82 ± 4.35 (m,
17H, -OCH2O-, -CH2- benzyl, H-1, H-2'', H-3''), 4.15 (brs, 1H, H-2'), 4.03
(dd, 1H, J3, 2 ˆ 3.3 Hz, J3, 4 ˆ 9.2 Hz, H-3), 3.98 ± 3.52 (m, 14H, H-2, H-4,
H-5, H-6, H-3', H-4', H-5', H-6', H-4'', H-5'', H-6''), 3.31 (s, 3H, -OCH3),
13
ˆ
3.10 (brs, 1H, -OH); C NMR (75 MHz, CDCl3): d ˆ 168.0 (C O), 160.9
trisaccharide 7 as a colorless oil (5.39 g, 76% yield); Rf (acetone/CH2Cl2,
22
4:96 v/v) ˆ 0.66; [a]
ˆ ‡44.4 (CH2Cl2, c ˆ 26.8 mgmL 1); 1H NMR
(Cq imidate), 138.7, 137.9 (Cq benzyl), 133.8, 123.3 (arom C -Pht), 131.5 (Cq
-Pht), 128.5 ± 127.4 (arom C), 102.1, 98.3, 94.0 (C-1, C-1', C-1''), 97.6
(-OCH2O-), 80.2, 79.5, 78.9, 77.6, 75.7, 74.8, 74.6, 71.3, 68.4 (C-2, C-3, C-4,
C-5, C-2', C-3', C-4', C-5', C-3'', C-4'', C-5''), 74.9, 74.7, 73.5, 73.2, 72.1, 69.7,
69.5, 67.1 (-CH2- benzyl, C-6, C-6', C-6''), 54.8 (-OCH3, C-2''); MS (FAB):
ꢀkap†Dꢀ=kap†
(300 MHz, CDCl3): d ˆ 7.65 (brs, 4H, arom H -Pht), 7.38 ± 6.85 (m, 35H,
arom H), 6.80, 6.65 (2 m, 4H, arom H -MP), 5.59 (d, 1H, J1'', 2'' ˆ 8.1 Hz,
H-1''), 5.56 (dd, 1H, J2', 1' ˆ 1.8 Hz, J2', 3' ˆ 3.3 Hz, H-2'), 5.21 (d, 1H, H-1'),
5.02 (d, 1H, Jgem ˆ 6.3 Hz, -OCH2O-), 4.94 ± 4.42 (m, 16H, H-1, -OCH2O-,
-CH2- benzyl), 4.40 (m, 1H, H-2''), 4.02 (dd, 1H, J3, 2 ˆ 2.6 Hz, J3, 4 ˆ 9.6 Hz,
H-3), 4.38 ± 3.53 (m, 15H, H-2, H-4, H-5, H-6, H-3', H-4', H-5', H-6', H-3'',
H-4'', H-5'', H-6''), 3.62 (s, 3H, -OCH3 -MP), 3.25 (s, 3H, -OCH3), 2.10 (s,
m/z (%) 1473 (73) [M‡Na‡3  35Cl] , 1476 (100), [M‡Na‡2  35Cl‡1 Â
‡
37Cl] , 1479 (32) [M‡Na‡1  35Cl‡2  37Cl] ; HRMS (FAB): calcd
‡
‡
C79H81N2O1835Cl3Na
[M‡Na]
1473.44;
found
1473.44;
calcd
‡
C79H81N2O1835Cl237ClNa [M‡Na]
1475.44; found 1475.44; calcd
‡
3H, -CH3 acetyl); 13C NMR (75 MHz, CDCl3): d ˆ 170.0 (C O), 151.0 (Cq
ˆ
C79H81N2O1835Cl37Cl2Na [M‡Na] 1477.44; found 1477.44.
‡
-MP), 138.0 (Cq benzyl), 133.7, 123.3 (arom C -Pht), 131.6 (Cq -Pht), 128.5 ±
127.4 (arom C), 118.6, 114.4 (arom C -MP), 100.1, 98.8, 97.7 (C-1, C-1',
C-1''), 97.8 (-OCH2O-), 79.6, 79.1, 77.8, 77.5, 74.3, 74.1, 72.2 (C-2, C-3, C-4,
C-5, C-2', C-3', C-4', C-5', C-3'', C-4'', C-5''), 74.8, 73.4, 73.0, 72.1, 71.9, 69.8,
69.2, 67.5 (-CH2- benzyl, C-6, C-6'), 64.1 (C-6''), 55.9, 55.7, 54.8 (-OCH3 -MP,
-OCH3, C-2''), 21.1 (-CH3 acetyl); MS (FAB): m/z (%): 1478 (100)
Methyl
(1 !2)-(3,4,6-tri-O-benzyl-a-d-mannopyranosyl)-(1 !3)-2,6-di-O-benzyl-
a-d-mannopyranoside-4,6''-methylidene acetal (11): mixture of 10
(3,4-di-O-benzyl-2-deoxy-2-phthalimido-b-d-glucopyranosyl)-
A
(1.22 g, 0.84 mmol) and 4 Š powdered molecular sieves (1 g) was stirred
for 30 min in CH2Cl2 (50 mL) under argon. The solution was cooled to
208C and trimethylsilyl trifluoromethanesulfonate (8 mL, 0.042 mmol)
was added and the reaction mixture was stirred for 10 min. The reaction
mixture was quenched by the addition of triethylamine and filtered through
Celite; the filtrate was diluted with CH2Cl2, and washed with NaHCO3 (2 Â
50 mL) and water (50 mL). The organic layer was dried (MgSO4), filtered,
and concentrated to dryness. Purification of the residue by column
chromatography on silica gel (CH2Cl2/acetone, 98:2 v/v) afforded the
cyclic trisaccharide 11 as a colorless oil (891.8 mg, 82% yield); Rf (CH2Cl2/
acetone, 96:4 v/v) ˆ 0.78; [a]2D2 ˆ ‡42.1 (CH2Cl2, c ˆ 20.53 mgmL 1);
1H NMR (300 MHz, CDCl3): d ˆ 7.52 (brs, 4H, arom H-Pht), 7.41 ± 6.80
(m, 35H, arom H), 5.75 (2d, 2H, J1'', 2'' ˆ 7.7 Hz, H-1'', H-1'), 5.06 (d, 1H,
Jgem ˆ 4.0 Hz, -OCH2O-), 4.88 (d, 1H, Jgem ˆ 11.0 Hz, -CH2- benzyl), 4.67 (d,
1H, J1, 2 ˆ 1.5 Hz, H-1), 4.83 ± 4.23 (m, 15H, -OCH2O-, -CH2- benzyl, H-2'',
H-2'), 4.15 (d, 1H, Jgem ˆ 10.7 Hz, -CH2- benzyl), 4.11 (dd, 1H, J3, 2 ˆ 2.9 Hz,
H-3), 3.99 (2dd, 2H, J4, 3 ˆ 9.9 Hz, H-4, H-4''), 3.82 ± 3.44 (m, 13H, H-2,
‡
‡
[M‡Na] ; HRMS (FAB): calcd C86H89NO20Na [M‡Na] 1478.59; found
1478.59.
Methyl 2,6-di-O-benzyl-3-O-(3,4,6-tri-O-benzyl-a-d-mannopyranosyl)-a-
d-mannopyranoside-3'',4''-di-O-benzyl-2''-deoxy-2''-phthalimido-b-d-glu-
co-pyranose-4,6''-methylidene acetal (9): Potassium tert-butoxide (0.412 g,
3.67 mmol) was added to a solution of trisaccharide 7 (5.34 g, 3.67 mmol) in
methanol (50 mL). The resulting mixture was stirred at room temperature
under argon. After 4 h, TLC analysis (acetone/CH2Cl2, 3:97 v/v) showed
the complete conversion of the starting material. The reaction mixture was
‡
neutralized with Dowex-50WX8-[H ] and filtered, and the filtrate con-
centrated to dryness in vacuo. The crude compound 8 was used without
further purification. Compound 8 was dissolved in a mixture of toluene/
acetonitrile/water (60 mL, 1:4:1 v/v/v), and cerium ammonium nitrate
(6.04 g, 11.01 mmol) was added. The mixture was stirred for 4 h at room
temperature under the exclusion of light, after which time the solution was
diluted with ethyl acetate. The organic layer was washed with saturated
NaHCO3 (2 Â 75 mL) and brine (75 mL), dried (MgSO4), and filtered, and
the filtrate concentrated under reduced pressure. The crude product was
purified by column chromatography on silica gel with acetone/CH2Cl2 (4:96
v/v) as the eluent. The trisaccharide 9 was obtained as a pale yellow oil
(2.16 g, 45% yield); Rf (acetone/CH2Cl2, 5:95 v/v) ˆ 0.20; [a]2D2 ˆ ‡29.8
(CH2Cl2, c ˆ 20.8 mgmL 1); 1H NMR (300 MHz, CDCl3): d ˆ 7.80 ± 7.60
H-5, H-6, H-3', H-4', H-5', H-6', H-3'', H-5'', H-6''), 3.25 (s, 3H, -OCH3);
13
ˆ
C NMR (75 MHz, CDCl3): d ˆ 167.9 (C O), 139.1, 138.4, 138.1, 137.9,
137.7 (Cq benzyl), 133.4, 123.1 (arom C -Pht), 131.6 (Cq -Pht), 128.5 ± 126.8
(arom C), 100.0, 99.0, 94.5 (C-1, C-1', C-1''), 96.9 (-OCH2O-), 79.2, 78.9,
77.2, 76.8, 76.0, 75.3, 74.5, 72.0, 71.8, 71.3 (C-2, C-3, C-4, C-5, C-2', C-3', C-4',
C-5', C-3'', C-4'', C-5''), 74.9, 74.3, 73.5, 73.2, 72.5, 72.3, 69.1, 68.8, 66.2
(-CH2- benzyl, C-6, C-6', C-6''), 54.7, 53.8 (C-2'', (-OCH3); MS (FAB): m/z
‡
‡
(%): 1328 (8) [M‡K] , 1312 (100) [M‡Na] ; HRMS (FAB): calcd
(4H, m, arom H -Pht), 7.41 ± 6.88 (m, 35H, arom H), 5.32 (d, 1H, J1', 2'
1.1 Hz, H-1'), 5.25 (d, 1H, J1'', 2'' ˆ 8.5 Hz, H-1''), 4.85 (2d, 2H, Jgem
ˆ
ˆ
‡
C77H79NO17Na [M‡Na] 1312.52; found 1312.52.
11.0 Hz, -CH2- benzyl), 4.70 (d, 1H, J1, 2 ˆ 1.4 Hz, H-1), 4.81 ± 4.39 (m,
15H, -OCH2O-, -CH2- benzyl, H-3''), 4.27 (brs, 1H, H-2'), 4.13 (dd, 1H,
J2'', 3'' ˆ 10.7 Hz, H-2''), 4.09 (m, 1H, H-3), 4.01 ± 3.54 (m, 14H, H-2, H-4,
Methyl
(3,4-di-O-benzyl-2-acetimido-2-deoxy-b-d-glucopyranosyl)-
(1 !2)-(3,4,6-tri-O-benzyl-a-d-mannopyranosyl)-(1 !3)-2,6-di-O-benzyl-
a-d-mannopyranoside-4,6''-methylidene acetal (12): Hydrazine monohy-
drate (1.78 mL, 36.69 mmol) was added to a solution of 11 (945.8 mg,
0.733 mmol) in EtOH (10 mL) and the reaction mixture was heated under
reflux for 18 h. TLC analysis with ninhydrin showed complete conversion
of the starting material into an amino-containing intermediate (Rf
(acetone/CH2Cl2, 4:96 v/v) ˆ 0.10). The solvent was removed in vacuo
and the residue coevaporated with toluene. The amine thus obtained was
subsequently dissolved in a mixture of pyridine (10 mL) and acetic
H-5, H-6, H-3', H-4', H-5', H-6', H-4'', H-5'', H-6''), 3.31 (s, 3H, -OCH3);
13
ˆ
C NMR (75 MHz, CDCl3): d ˆ 168.2 (C O), 138.7, 138.4, 138.3, 138.2,
138.0 (Cq benzyl), 133.7, 123.2 (arom C -Pht), 131.8 (Cq -Pht), 128.5 ± 126.6
(arom C), 100.7, 98.6, 92.9 (C-1, C-1', C-1''), 97.6 (-OCH2O-), 79.7, 79.3, 79.2,
78.0, 76.3, 75.7, 74.6, 74.5, 71.7, 71.5, 68.3 (C-2, C-3, C-4, C-5, C-2', C-3', C-4',
C-5', C-3'', C-4'', C-5''), 74.8, 74.6, 73.4, 71.9, 69.5, 69.3 (-CH2- benzyl, C-6,
C-6', C-6''), 57.7, 54.9 (-OCH3, C-2''); MS (FAB): m/z (%): 1346 (3)
Chem. Eur. J. 1999, 5, No. 8
ꢀ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 1999
0947-6539/99/0508-2291 $ 17.50+.50/0
2291

G.-J. Boons et al.
FULL PAPER
anhydride (10 mL) and the solution was stirred at room temperature for
4 h. The mixture was then concentrated and coevaporated with toluene
(3 Â 20 mL), ethanol (2 Â 20 mL) and CH2Cl2 (2 Â 20 mL). Purification of
the crude product by column chromatography on silica gel (CH2Cl2/
acetone, 97:3 v/v) gave the title compound 12 (563.4 mg, 64% yield) as a
colorless oil; Rf (acetone/CH2Cl2, 4:96 v/v) ˆ 0.23; [a]2D4 ˆ ‡31.2 (CH2Cl2,
c ˆ 16.93 mgmL 1); 1H NMR (300 MHz, CDCl3): d ˆ 7.45 ± 7.11 (m, 35H,
arom H), 5.43 (d, 1H, JNH, 2'' ˆ 6.3 Hz, NH), 5.39 (s, 1H, H-1'), 5.22 (d, 1H,
J1'', 2'' ˆ 8.8 Hz, H-1''), 4.91, 4.90, 4.85 (3d, 3H, Jgem ˆ 11.0 Hz, -CH2- benzyl),
4.79 ± 4.48 (m, 15H, -OCH2O-, -CH2- benzyl, H-1, H-3''), 4.22 (s, 1H, H-2'),
4.08 ± 3.97 (m, 2H, H-3, H-4), 3.88 ± 3.48 (m, 13H, H-2, H-5, H-6, H-3', H-4',
atom O-4 and O-6''. The solution with the best energy was taken as a
starting point for the cyclic compound. The coordinates of concanavalin A
were taken from the 2.0 Š resolution crystal structure of the ConA ±
mannose complex.[35] Four conserved water molecules, which are involved
in the coordination of the structural cations, were incorporated into the
model. All hydrogen atoms were added and their position optimized with
the Tripos force field.[44]
Systematic conformational search for the trisaccharides: For the linear
trisaccharide, a four-dimensional systematic search was performed by
rotating the F and Y angles of the two glycosidic linkages by 108 steps. The
SEARCH procedure of the SYBYL software was used for this purpose
together with energy parameters appropriate for carbohydrates.[37] In order
to avoid limitation of conformational space due to steric conflicts in the
rigid-residue approach, the hydroxyl hydrogens were omitted and the
hydroxymethyl groups at C-6 were replaced by methyl groups. In order to
limit the computation time needed, all conformations with penetration of
van der Waals spheres larger than 40% were rejected prior to any energy
calculations.
H-5', H-6', H-4'', H-5'', H-6''), 3.27 (s, 3H, -OCH3), 2.91 ± 2.80 (m, 1H,
13
ˆ
H-2''); C NMR (75 MHz, CDCl3): d ˆ 171.0 (C O), 139.3, 138.5, 138.3,
138.1, 137.8 (Cq benzyl), 129.0 ± 127.5 (arom C), 102.9, 98.6, 96.3 (C-1, C-1',
C-1''), 97.2 (-OCH2O-), 79.0, 78.2, 78.0, 77.5, 76.8, 76.3, 75.9, 74.3, 72.6, 72.1,
70.5 (C-2, C-3, C-4, C-5, C-2', C-3', C-4', C-5', C-3'', C-4'', C-5''), 75.1, 74.8,
74.5, 73.6, 72.6, 69.3, 69.1, 66.1 (-CH2- benzyl, C-6, C-6', C-6''), 56.2, 54.8 (C-
‡
2'', -OCH3), 23.5 (-CH3 acetyl); MS (FAB): m/z (%): 1240 (5) [M‡K] ,
‡
‡
1224 (100) [M‡Na] ; HRMS (FAB): calcd C71H79NO16Na [M‡Na]
A systematic search of the possible conformations of the 13-membered ring
has been performed on the cyclic trisaccharide. In the 13-membered ring of
this cyclic trisaccharide, nine torsion angles could be rotated, one of which
1224.53; found 1224.53.
Methyl
(2-acetimido-2-deoxy-b-d-glucopyranosyl)-(1 !2)-(a-d-manno-
pyranosyl)-(1 !3)-a-d-mannopyranoside-4,6''-methylidene acetal (III):
Compound 12 (536.6 mg, 0.447 mmol) in ethanol (10 mL) was hydro-
genated (H2, 54 mg Pd(OAc)2) for 18 h. The mixture was filtered though
Celite and concentrated. The cyclic trisaccharide III was obtained as a
white solid (230 mg, 90%yield); Rf (MeOH/CH2Cl2, 20:80 v/v) ˆ 0.05;
[a]2D2 ˆ ‡34.7 (MeOH, c ˆ 3.4 mgmL 1); 1H NMR (400 MHz, D2O): d ˆ
5.68 (d, 1H, J1', 2' ˆ 1.3 Hz, H-1'), 5.15, 5.02 (2d, 2H, Jgem ˆ 5.1 Hz, -OCH2O-
), 4.95 (d, 1H, J1'', 2'' ˆ 8.4 Hz, H-1''), 4.83 (d, 1H, J1, 2 ˆ 1.5 Hz, H-1), 4.37
(dd, 1H, J2', 3' ˆ 3.9 Hz, H-2'), 4.20 ± 3.92 (m, 8H, H-2, H-3, H-6a, H-3',
H-6a', H-2'', H-6a'', H-6b''), 3.90 ± 3.56 (m, 9H, H-4, H-5, H-6b, H-4', H-5',
was used as
a ring-closure bond. An eight-dimensional systematic
conformational search was performed by rotating the F and Y torsion
angles of the two glycosidic linkages as well as the torsion angles of the
methylene acetal bridge by 108 steps, except for w3, which corresponds to
the ring-closure bond.
In order to establish the correct stereochemistry at the ring-closure point
and to relieve small steric conflicts, each of the resulting conformations of
the cyclic trisaccharide was submitted to several steps of energy minimi-
zation. The hydroxyl atoms were restored, and the charges and atom types
specially defined for carbohydrate parameterization[37] were used. The first
cycles included constraints on the torsion angles. The final optimization was
run with a termination gradient of 0.05 rms with no constraints. The
structures obtained after the systematic search were checked for the
puckering of the pyranose rings and chirality of carbon atoms.
A family analysis method[8] was used to identify groups of conformations. In
this approach, a family algorithm was used which concludes that an object
belongs to a group if there is only a single-step change to at least one of the
objects in the group. For this purpose, the torsion angles of both the
bGlcNAc(1 !2)Man and the aMan(1 !3)Man glycosidic linkage were
used as selection criteria and the step limit was fixed at 308.
H-6b', H-3'', H-4'', H-5''), 3.49 (s, 3H, -OCH3), 2.13 (s, 3H, -CH3 acetyl);
13
ˆ
C NMR (100 MHz, D2O): d ˆ 177.5 (C O), 103.1 (JC1, H1 ˆ 173.5 Hz, C-1),
102.0 (JC1', H1' ˆ 173.3 Hz, C-1'), 100.4 (-OCH2O-), 99.6 (JC1'', H1'' ˆ 166.4 Hz,
C-1''), 78.7, 77.7, 76.9, 76.4, 75.7, 75.0, 74.4, 72.5, 71.8, 71.6, 69.3 (C-2, C-3,
C-4, C-5, C-2', C-3', C-4', C-5', C-3'', C-4'', C-5''), 70.4 (C-6''), 63.4, 62.7 (C-6,
C-6'), 57.3 (-OCH3), 56.6 (C-2''), 24.8 (-CH3 acetyl); MS (FAB): m/z (%):
‡
‡
616 (7) [M‡2  Na] , 594 (100) [M‡Na] ; MS (FAB): calcd C22H37NO16Na
‡
[M‡Na] 594.20; found 594.20.
Microcalorimetry: Isothermal titration calorimetry (ITC) experiments to
measure the binding of saccharides to concanavalin A were done at 258C
by means of a Microcal MCS titration microcalorimeter following standard
instrumental procedures[42] with a 250 mL injection syringe and 400 rpm
stirring. Concanavalin A (Sigma) was used without further purification and
was dissolved in a buffer (0.1m Tris, 0.5m NaCl, 1mm MnCl2, 1mm CaCl2,
pH 7.0) and degassed gently immediately before use. Saccharide ligands
were dissolved in the same buffer. Protein concentrations in the ITC cell
were determined from UV absorbance measurements at 280 nm with the
molar extinction coefficient e280 ˆ 33000 (ConA). A typical binding experi-
ment involved 25 Â 10 mL injections of ligand solution (typically around
10mm concentration) into the ITC cell (ca. 1.3 mL active volume)
containing protein (0.1 ± 0.5mm). Control experiments were performed
under identical conditions by injection of ligand into buffer alone (to
correct for heats of ligand dilution) and injection of buffer into the protein
mix (to correct for heats of dilution of the protein). Integrated heat effects,
after correction for heats of dilution, were analyzed by nonlinear regression
in terms of a simple single-site binding model using the standard Microcal
ORIGIN software package. For each thermal titration curve this yields
estimates of the apparent number of binding sites (N) on the protein, the
binding constant (K, m 1) and the enthalpy of binding (DH, kcalmol 1). In
cases of weak ligand binding the titration curve is too gradual to allow
unambiguous estimation of N, and in such cases the stoichiometry was fixed
at N ˆ 1 for regression fits. Other thermodynamic quantities were
calculated with the standard expression DG8 ˆ RTlnK ˆ DH8 TDS8.
Docking in the binding site of ConA: The lowest energy conformation of
each family of compound I and each of the low-energy conformations of
compound III, in an energy window of 10 kcalmol 1, were used as starting
structures to be docked in the binding site of ConA. This was performed by
superimposing the central mannose unit of the cyclic trisaccharide onto the
sugar ring in the ConA ± mannose complex.[35] The hydroxyl groups were
oriented in order to create the appropriate hydrogen network between the
mannose residue and the amino acids of the binding site. Each of these
ConA ± trisaccharide complexes was optimized by means of the appro-
priate energy parameters.[37] Two shells of amino acids were considered for
the optimization cycle. A 10 Š shell around the binding site (33 amino acids
as well as the water molecules and cations) was taken into account for the
energy calculations. A 4 Š shell containing the 15 amino acids closest to the
carbohydrate (Tyr12, Asn14, Thr97, Gly98, Leu99, Tyr100, Ser168, Ala207, Asp208
,
Gly224, Ser225, Thr226, Gly227, Arg228, and Leu229) was defined as the hot region
to be optimized. In the first optimization cycles, only the cyclic trisacchar-
ide and the side chains of these 15 amino acids were allowed to optimize. In
the second step, the backbone of these amino acids was also optimized.
Conformational analysis by NMR spectroscopy: NMR experiments were
recorded on a Bruker DRX500 spectrometer, with an approximately
1
5 mgmL solution of the trisaccharides in D2O at 299 K. Chemical shifts
are reported in ppm, with the residual HDO signal (d ˆ 4.72) and external
TMS (d ˆ 0) as references. The double quantum filtered COSY spectrum
was recorded with a data matrix of 256 Â 1 k to digitize a spectral width of
2000 Hz. 16 scans were used with a relaxation delay of 1 s. The 2D-TOCSY
experiment was performed with a data matrix of 256 Â 2 k to digitize a
spectral width of 2000 Hz; 16 scans were used per increment with a
relaxation delay of 2 s, and MLEV 17 with 100 ms isotropic mixing time.
The one-bond proton ± carbon correlation experiment was collected by
means of the gradient-enhanced HMQC sequence. A data matrix of 256 Â
Molecular modeling:
Starting models for the oligosaccharides and for ConA: The trisaccharide
bGlcNAc(1 !2)aMan(1 !3)Man was constructed with the monosacchar-
ides obtained from a database of three-dimensional structures.[43] All
subsequent calculations were performed by SYBYL software. Cyclization
was performed through a systematic search around the four torsion angles
of the glycosidic linkage, while a short distance was imposed between the
2292
ꢀ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 1999
0947-6539/99/0508-2292 $ 17.50+.50/0
Chem. Eur. J. 1999, 5, No. 8

Cyclic Trisaccharide
2281±2294
1 k was used to digitize a spectral width of 2000 Hz in F2 and 10000 Hz in
F1. 4 scans were used per increment with a relaxation delay of 1 s and a
delay corresponding to a J value of 145 Hz. 13C decoupling was achieved by
the WALTZ scheme.
[12] T. Nakano, Y. Ito, T. Ogawa, Carbohydr. Res. 1993, 243, 43 ± 69.
[13] G. H. Veeneman, S. H. van Leeuwen, J. H. van Boom, Tetrahedron
Lett. 1990, 31, 1331 ± 1334.
[14] P. J. Garegg, H. Hultberg, S. Wallim, Carbohydr. Res. 1982, 108, 97 ±
101.
The 2D-HMQC-TOCSY experiment was conducted with 80 ms mixing
time (MLEV 17). The same conditions as for the HMQC were employed.
HMBC experiments were performed with the gradient-enhanced sequence
with a data matrix of 256 Â 2 k to digitize a spectral width of 2000 Â
15000 Hz. Eight scans were acquired per increment with a delay of 65 ms
for evolution of long-range couplings. NOESY experiments were recorded
with mixing times of 100, 200, 300, and 400 ms. ROESY experiments used
mixing times of 100, 200, 300, and 400 ms. The rf carrier frequency was set
at d ˆ 6.0, and the spin locking field was 3.0 kHz. Good linearity of the
build-up curves was observed up to 250 ms (NOESY) and 300 ms
(ROESY). Estimated errors are smaller than 10%. Assuming that the
motion of two interacting protons can be described by a monoexponential
autocorrelation function, the corresponding cross-relaxation rates are
given in Equations (1) and (2).
[15] P. J. L. M. Quaedflieg, C. M. Timmers, V. E. Kal, G. A. van der Marel,
E. Kuyl-Yeheskiely, J. H. van Boom, Tetrahedron Lett. 1992, 33,
3081 ± 3084.
[16] R. R. Schmidt, Angew. Chem. 1986, 98, 213 ± 236; Angew. Chem. Int.
Ed. Engl. 1986, 25, 212 ± 235.
[17] T. Fukuyama, A. A. Laird, L. M. Hotchkiss, Tetrahedron Lett. 1985,
26, 6291 ± 6292.
[18] J. Alais, A. Veyrieres, Carbohydr. Res. 1990, 207, 11 ± 31.
Â
[19] T. Sokolowski, T. Peters, S. Perez, A. Imberty, J. Mol. Graph. 1997, 15,
37 ± 42.
[20] N. L. Allinger, Y. H. Yuh, J. H. Lii, J. Am. Chem. Soc. 1989, 111, 8551 ±
8566.
[21] a) J. R. Brisson, J. P. Carver, Biochemistry 1983, 22, 3671 ± 3689; b) H.
Paulsen, T. Peters, V. Sinnwell, R. Lebuhn, B. Meyer, Liebigs Ann.
Chem. 1985, 489 ± 509; c) S. W. Homans, R. A. Dwek, T. W. Rade-
macher, Biochemistry 1987, 26, 6553 ± 6560.
sNOESY ˆ (k2/10r6)[6J(2w) J(0)]
sROESY ˆ (k2/10r6)[2J(0) ‡ 3J(w)]
(1)
(2)
Â
[22] A. Imberty, V. Tran, S. Perez, J. Comput. Chem. 1989, 11, 205 ±
216.
Cross-relaxation rates were estimated from the build-up courses by
extrapolation at zero mixing time.[31, 32] Effective correlation times and,
thus, interproton distances, for selected proton pairs may be obtained from
sNOESY/sROESY ratios, since they only depend on tc. Developing the spectral
density functions, J(nw), as a function of the correlation time, tc, and of the
spectrometer frequency, w0, we obtain Equation (3), which is a quartic
Â
[23] S. Perez, F. R. Taravel, C. Vergelati, Nouv. J. Chim. 1985, 9, 561 ± 564.
[24] J. Dabrowski, T. Kozar, H. Grosskurth, N. E. Nifantꢁev, J. Am. Chem.
Soc. 1995, 117, 5534 ± 5539.
[25] C. Landersjö, R. Stenutz, G. Widmalm, J. Am. Chem. Soc. 1997, 119,
8695 ± 9698.
[26] J. F. Espinosa, M. Martin-Pastor, J. L. Asensio, H. Dietrich, M.
Â
Martin-Lomas, R. R. Schmidt, J. Jimenez-Barbero, Tetrahedron Lett.
1995, 36, 6329 ± 6332.
sROESY/sNOESY ˆ (5 ‡ 22w02 tc2 ‡ 8w04 tc4)/(5 ‡ w02 tc2 4w20 tc4)
(3)
[27] a) M. Hricovini, R. N. Shah, J. P. Carver, Biochemistry 1992, 3,
10018 ± 10023; b) T. J. Rutherford, J. Partridge, C. T. Weller, S. W.
Homans, Biochemistry 1993, 32, 12715 ± 12724; c) C. A. Bush, Curr.
Opin. Struct. Biol. 1992, 2, 655 ± 663; d) K. G. Rice, P. Wu, L. Brand,
Y. C. Lee, Curr. Opin. Struct. Biol. 1993, 3, 669 ± 674.
[28] a) C. Mukhopadhyay, K. E. Miller, C. A. Bush, Biopolymers 1994, 34,
21 ± 29; b) T. J. Rutherford, D. G. Spackman, P. J. Simpson, S. W.
Homans, Glycobiology 1994, 4, 59 ± 68; c) R. U. Lemieux, Chem. Soc.
Rev. 1989, 18, 347 ± 374; d) K. Bock, Pure Appl. Chem. 1983, 55, 605 ±
622.
[29] a) J. P. Carver, Pure Appl. Chem. 1993, 65, 763 ± 770; b) A. Ejchart, J.
Dabrowski, C. W. von der Lieth, Magn. Reson. Chem. 1992, 30, S105 ±
S114; c) T. Peters, T. Weimar, J. Biomol. NMR 1994, 4, 97 ± 116.
[30] a) L. Poppe, H. van Halbeek, J. Am. Chem. Soc. 1992, 114, 1092 ±
1094; b) M. Hricovini, G. Torri, Carbohydr. Res. 1995, 268, 159 ± 175;
c) B. J. Hardy, W. Egan, G. Widmalm, Int. J. Biol. Macromol. 1995, 17/
equation in tc, which can be easily solved. It should be stressed that the
assumption made here does not require equal mobility between different
proton pairs, as in a rigid molecule. Therefore, it is possible to obtain
internuclear distances r (geometrical parameter) and local correlation
times (dynamic parameter) for any pair of protons.[33] This approach
reduces the intrinsic error resulting from the use of an internal reference (as
in the isolated-spin-pair approximation).
Acknowledgment
The research is supported by TMR Marie Curie Research Grant Financial
(N.N.), by DGICYT (project PB96-0833) (J.J.B.), and by Pfizer (N.A.).
SIdI-UAM provided facilities throughout this work (A.P.).
Â
Â
18, 149 ± 160; d) C. Meyer, S. Perez, C. Herve du Penhoat, V. Michon,
J. Am. Chem. Soc. 1993, 115, 10300 ± 10310; e) I. Braccini, V. Michon,
[1] a) J. M. Rini, Annu. Rev. Biophys. Biomol. Struct. 1995, 24, 551 ± 577;
b) N. K. Vyas, Curr. Opin. Struct. Biol. 1991, 1, 732 ± 740; c) W. I. Weis,
K. Drickamer, Annu. Rev. Biochem. 1996, 65, 441 ± 473; d) K.
Drickamer, Structure 1997, 5, 465 ± 472; e) H. Lis, N. Sharon, Chem.
Rev. 1998, 98, 637 ± 674.
[2] E. J. Toone, Curr. Opin. Struct. Biol. 1994, 4, 719 ± 728.
[3] R. U. Lemieux, L. T. J. Delbaere, H. Beierbeck, U. Spohr, Ciba Found.
Symp. 1991, 158, 231 ± 238.
Â
Â
C. Herve du Penhoat, A. Imberty, S. Perez, Int. J. Biol. Macromol.
1993, 15, 52 ± 55; f) A. Poveda, J. L. Asensio, M. Martin-Pastor, J.
Jimenez-Barbero, Chem. Commun. 1996, 421 ± 422; g) S. Bagley, H.
Kovacs, J. Kowalewski, G. Widmalm, Magn. Reson. Chem. 1992, 30,
733 ± 739; h) L. Maler, J. Lang, G. Widmalm, J. Kowalewski, Magn.
Reson. Chem. 1995, 33, 541 ± 548; i) J. F. Espinosa, J. L. Asensio, M.
Â
Â
Bruix, J. Jimenez-Barbero, An. Quim. Int. Ed. 1996, 96, 320 ± 324.
[31] D. Neuhaus, M. P. Williamson, The Nuclear Overhauser Effect in
Structural and Conformational Analysis, VCH, New York, 1989.
[32] G. Esposito, A. Pastore, J. Magn. Reson. 1988, 76, 331 ± 336, and
references therein.
[4] J. P. Carver, Pure Appl. Chem. 1993, 65, 763 ± 770.
Â
[5] Y. Bourne, P. Rouge, C. Cambillau, J. Biol. Chem. 1992, 267, 197 ± 204;
Â
b) Y. Bourne, J. Mazurier, D. Legrand, P. Rouge, J. Montreuil, G. Spik,
C. Cambillau, Structure 1994, 2, 209 ± 219.
[33] a) H. Desvaux, P. Berthault, N. Bilirakis, M. Goldman, J. Magn.
Reson. 1994, A108, 219 ± 229; b) G. Lippens, J. M. Wieruzeski, P.
Talaga, J. P. Bohin, H. Desvaux, J. Am. Chem. Soc. 1996, 118, 7227 ±
7228; c) D. G. Davis, J. Am. Chem. Soc. 1987, 109, 3471 ± 3472; d) A.
[6] Y. Bourne, B. Bolgiano, D. I. Liao, G. Strecker, P. Cantau, O.
Herzberg, T. Feizi, C. Cambillau, Nat. Struct. Biol. 1994, 1, 863 ± 870.
Â
Â
[7] a) A. Imberty, Y. Bourne, C. Cambillau, P. Rouge, S. Perez, Adv.
Biophys. Chem. 1993, 3, 71 ± 117; b) A. Imberty, Curr. Opin. Struct.
Biol. 1997, 7, 617 ± 623.
Â
Poveda, J. L. Asensio, M. Martin-Pastor, J. Jimenez-Barbero, J.
Â
Biomol. NMR 1997, 10, 29 ± 43.
[8] A. Imberty, S. Perez, Glycobiology 1994, 4, 351 ± 366.
[34] a) K. Bock, J. Duus, J. Carbohydr. Chem. 1994, 13, 513 ± 543; b) G. D.
Rockwell, T. B. Grindley, J. Am Chem. Soc. 1998, 120, 10953 ± 10963.
[35] J. H. Naismith, C. Emmerich, J. Habash, S. J. Harrop, J. R. Helliwell,
W. N. Hunter, J. Raftery, A. J. (Gilboa) Kalb, J. Yariv, Acta Crystallogr.
1994, D50, 847 ± 848.
[9] N. Navarre, A. H. van Oijen, G. J. Boons, Tetrahedron Lett. 1997, 38,
2023 ± 2026.
[10] T. Peters, Liebigs Ann. Chem. 1991, 2, 135 ± 141.
[11] a) Y. Kondo, K. Noumi, S. Kitagawa, S. Hirano, Carbohydr. Res. 1983,
Â
123, 157 ± 159; b) H. B. Boren, P. J. Garegg, N. H. Wallim, Acta. Chem.
[36] D. N. Moothoo, J. H. Naismith, Glycobiology 1998, 8, 173 ± 181.
Scand. 1972, 26, 1082 ± 1086.
Chem. Eur. J. 1999, 5, No. 8
ꢀ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 1999
0947-6539/99/0508-2293 $ 17.50+.50/0
2293

G.-J. Boons et al.
FULL PAPER
Â
[37] S. Perez, C. Meyer, A. Imberty in: Modelling of Biomolecular
[42] a) T. Wiseman, S. Williston, J. F. Brandts, L.-N. Lin, Analyt. Biochem.
1989, 179, 131 ± 137; b) A. Cooper, C. M. Johnson, in: Microscopy,
Optical Spectroscopy, and Macroscopic Techniques, Methods in
Molecular Biology, Vol. 22 (Eds.: C. Jones, B. Mulloy, A. H. Thomas),
Humana, Totowa, NJ, 1994, pp. 137 ± 150.
Structures and Mechanisms (Eds.: A. Pullman, J. Jortner, B. Pullman),
Kluwer Academic, Dordrecht, 1995, pp. 425 ± 454.
[38] D. R. Bundle, R. Alibes, S. Nilar, A. Otter, M. Warwas, P. Zhang, J.
Â
Am. Chem. Soc. 1998, 120, 5317 ± 5318.
Â
[39] J. Thomson, Y. F. Liu, J. M. Sturtevant, F. A. Quiocho, Biophys. Chem.
1998, 70, 101 ± 108.
[43] S. Perez, M. M. Delage, Carbohydr. Res. 1991, 212, 523 ± 529.
[44] M. Clark, R. D. Cramer IV, N. van den Opdenbosch, J. Comp. Chem.
1989, 10, 982 ± 1012.
[45] F. P. Schwarz, K. D. Puri, R. G. Bhat, A. Surolia, J. Biol. Chem. 1995,
268, 7668 ± 7677.
[46] B. A. Williams, M. C. Chervenak, E. J. Toone, J. Biol. Chem. 1992, 267,
22907 ± 22911.
[47] D. K. Mandal, N. Kishore, C. F. Brewer, Biochemistry 1994, 33, 1149 ±
1156.
[40] Several other attempts have been reported to bias different structures
towards the bioactive conformation; see: a) R. U. Lemieux, T. C.
Wong, H. Thongersen, Can. J. Chem. 1982, 60, 81 ± 86; b) K. Bock,
J. O. Duus, O. Hindsgaul, I. Lindl, Carbohydr. Res. 1992, 228, 1 ± 20;
c) I. Lindh, O. Hindsgaul, J. Am. Chem. Soc. 1991, 113, 216 ± 223;
d) H. C. Kolb, B. Ernst, Chem. Eur. J. 1997, 3, 1571 ± 1578; e) M.
Wilstermann, J. Balogh, G. Magnusson, J. Org. Chem. 1997, 62, 3659 ±
3665.
[48] M. C. Chervenak, E. J. Toone, Biochemistry 1995, 34, 5685 ± 5695.
[41] IUPAC-IUB Commission on Nomenclature: http://www.chem.qm-
w.ac.uk/iupac/
Received: October 5, 1998
Revised version: February 22, 1999 [F1370]
2294
ꢀ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 1999
0947-6539/99/0508-2294 $ 17.50+.50/0
Chem. Eur. J. 1999, 5, No. 8