The flexibility of the LeaLex Tumor Associated Antigen central fragment studied by systematic and stochastic searches as well as dynamic simulations

Article abstract of DOI:10.1016/j.bmc.2009.01.020

The solution conformational behavior of the Tumor-Associated Carbohydrate Antigen Le^aLe^x central fragment: methyl α-l-fucopyranosyl-(1→4)-2-acetamido-2-deoxy-β-d-glucopyranosyl-(1→3)-β-d-galactopyranoside was studied using three comp

Full text of DOI:10.1016/j.bmc.2009.01.020

Bioorganic & Medicinal Chemistry 17 (2009) 1514–1526
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry
journal homepage: www.elsevier.com/locate/bmc
The flexibility of the LeaLex Tumor Associated Antigen central fragment
studied by systematic and stochastic searches as well as dynamic simulations
Trudy A. Jackson a, Valerie Robertson a, Anne Imberty b, France-Isabelle Auzanneau a,
*
a Department of Chemistry, University of Guelph, Guelph, Ontario, Canada N1G 2W1
b CERMAV-CNRS (Affiliated with Université Joseph Fourier and Member of the Institut de Chimie Moléculaire de Grenoble), BP 53, F-38041, Grenoble Cedex 9, France
a r t i c l e i n f o
a b s t r a c t
The solution conformational behavior of the Tumor-Associated Carbohydrate Antigen LeaLex central frag-
Article history:
Received 15 October 2008
Revised 2 January 2009
Accepted 8 January 2009
Available online 15 January 2009
ment: methyl
a-L-fucopyranosyl-(1?4)-2-acetamido-2-deoxy-b-D-glucopyranosyl-(1?3)-b-D-galacto-
pyranoside was studied using three computational methods: a rigid systematic search as implemented
in Sybyl, a stochastic search as implemented in MOE2004, and dynamics simulations using the SANDER
module of AMBER9. Our results illustrate the complementarity of these methods to identify energetically
relevant conformations and flexible linkages. In particular, the b-GlcNAc-(1?3)-Gal linkage was shown
to be extremely flexible adopting a wide range of orientations around two energy minima. The modeling
results were validated by comparison of theoretical distances, derived from the simulations, with exper-
imental measurements obtained from 1D selective ROESY buildup curves on the synthetic fragment.
Ó 2009 Elsevier Ltd. All rights reserved.
Keywords:
Carbohydrate-based anti-cancer vaccines
LeaLex
Conformational analysis
NMR spectroscopy
Stochastic search
Systematic search
Molecular dynamics
Synthesis
1. Introduction
fragment: methyl
a-L-fucopyranosyl-(1?4)-2-acetamido-2-deoxy-
b- -glucopyranosyl-(1?3)-b-
D
D
-galactopyranoside than that
2
Tumor-Associated Carbohydrate Antigens (TACAs) are inter-
esting targets for the development of anti-cancer vaccines. One
such TACA, the LeaLex hexasaccharide 1, has been shown to be
over-expressed at the surface of squamous lung carcinoma
cells.1–3 However, to date, this hexasaccharide has not been
extensively investigated for efficacy in anti-cancer vaccines be-
cause its terminal non-reducing Lea trisaccharide is also com-
monly expressed on normal non-tumor cells.4 Thus, polyclonal
antibodies generated from a vaccine based on the hexasaccha-
ride will likely cross-react with Lea leading to autoimmune reac-
tions. However, a monoclonal antibody 43-9F that was raised
against LeaLex, has shown high selectivity toward internal epi-
topes of the hexasaccharide, while it did not bind efficiently to
the terminal trisaccharide residue.5–7 With the ultimate goal to
develop a selective anti-tumor vaccine, we have embarked on
the identification of such internal Tumor-Associated Carbohy-
drate Epitopes (TACE) and have undertaken detailed studies of
internal fragments of the hexasaccharide 1. Here we describe
an improved synthesis of the central LeaLex trisaccharide
previously reported.8 The conformational properties of the trisac-
charide were investigated by both rigid systematic and relaxed
stochastic molecular mechanics searches and by molecular
dynamics simulations in explicit solvent. The modeling results
were validated by comparison of theoretical distances derived
from the simulations with experimental measurements obtained
from ROESY spectra at several mixing times.
HO
OH
OH
O
OH
O
HO
O
OH
O
OH
O
HO
HO
OH
O
O
O
O
O
O
OH
NHAc
NHAc
OH
O
OH
OH
HO
1
HO
OH
OH
O
OH
O
HO
OH
O
O
O
NHAc
OMe
HO
OH
* Corresponding author. Tel.: +1 519 824 4120; fax: +1 519 766 1499.
E-mail addresses: Anne.Imberty@cermav.cnrs.fr (A. Imberty), fauzan-
ne@uouelph.ca (F.-I. Auzanneau).
2
0968-0896/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmc.2009.01.020

T. A. Jackson et al. / Bioorg. Med. Chem. 17 (2009) 1514–1526
1515
ꢀ30 °C with up to 2.0 equiv of TMSOTf and continuous stirring
overnight at ꢀ30 °C successfully afforded the desired disaccharide
7 in 88% yield. Disaccharide 7 was subsequently converted to the
acceptor 11 in four synthetic steps (Scheme 2). Selective deacety-
lation (2% HCl in MeOH) gave the triol 8 in 77% yield and that
was converted to the benzylidene acetal 9 [PhCH(OMe)2/TsOH].
Acetylation of the crude alcohol 9 gave disaccharide 10 (94%)
which was reductively opened yielding the desired acceptor 11
(86%). The regioselectivity of the benzylidene ring opening was
confirmed by acetylation of an analytical sample of acceptor 11
2. Results and discussion
2.1. Synthesis of trisaccharide 2
For the synthesis of the trisaccharide fragment, we envisaged a
‘1 + 1 + 1’ linear addition of the monosaccharide precursors. We
first attempted the regioselective glycosylation of the known9 diol
4 (Scheme 1) with the known10 donor 3. The reaction was carried
out in CH2Cl2 at ꢀ78 °C, using an excess of acceptor 4 (1.5 equiv)
and under promotion by TMSOTf (0.16 equiv). However under
these conditions, neither the desired 3-O-glycosylated disaccha-
ride nor its regioisomer glycosylated at O-4 were observed and
the only product isolated in 76% yield was the di-O-glycosylated
trisaccharide 5. Thus, in contrast to previous reports of such selec-
tive glycosylation at O-3 of galactosyl residues in lactoside accep-
tors,11 there was no significant difference in reactivities between
O-3 and O-4 in the monosaccharide diol acceptor 4 that would al-
low its selective glycosylation. Abandoning this regioselective
strategy, we prepared acceptor 6 selectively free at O-3 in 81%
yield by treating diol 4 with trimethyl orthobenzoate under acid
catalysis and opening the resulting orthoester by treatment with
aq AcOH (Scheme 2).
1
yielding acetate 12. Indeed, H NMR showed that signal H-40 ap-
peared at 4.84–4.95 ppm in compound 12, while this signal was
found upfield at 3.66 ppm in acceptor 11, supporting that O-40
was indeed acetylated in 12 and free in acceptor 11. The disaccha-
ride acceptor 11 was then glycosylated with the known13 b- or
a-
fucose thioethyl donors 13 or 14 under promotion with MeOTf to
yield trisaccharide 15 in 85% and 77% yield, respectively (Scheme
3). The trisaccharide 15 was deprotected in three steps (72% overall
SEt
SEt
This constitutes an alternative synthesis of tribenzoate 6, which
O
O
OBn
had previously been obtained12 via the palladium catalyzed deben-
or
OBn
OBn
OBn
BnO
BnO
zylation of methyl 2,4,6-tri-O-benzoyl-3-O-benzyl-b-D-galactopy-
14
ranoside. Coupling of the alcohol 6 with the donor 3 starting at
ꢀ78 °C, adding up to 0.5 equiv of TMSOTf, and allowing the reac-
tion to reach room temperature over 18 h resulted only in the sily-
lation of the acceptor. In contrast, carrying out the glycosylation at
13
MeOTf
Et2O, rt
11
85% (from 13)
77% (from 14)
BzO
OBz
BnO
O
O
O
OMe
OAc
O
AcO
OBz
O
TrocHN
3
HO
HO
BzO
RHN
AcO
AcO
O
O
+
OBn
OMe
OBn
CCl3
O
BnO
BzO
4
15 R = Troc
16 R = Ac
NH
Zn, Ac2O, 79%
TMSOTf
CH2Cl2
NaOMe MeOH
-78 oC to rt
HO
OH
OAc
O
RO
HO
O
O
O
AcO
AcO
OMe
O
O
OBz
HO
TrocHN
O
AcHN
O
O
AcO
AcO
AcO
OR
OMe
O
OR
RO
BzO
TrocHN
17 R = Bn
2 R = H
H2, Pd/C 92%
5
Scheme 1.
Scheme 3.
OBz
BzO
O
OBz
BzO
BzO
HO
RO
RO
RO
O
O
(i)
O
3
4
OMe
OBz
OMe
TMSOTf
CH2Cl2
-30 oC
88%
BzO
TrocHN
(ii)
6
7 R = Ac
8 R = H
BzO
O
O
OBz
BzO
(iii)
Ph
BnO
R2O
R1O
O
O
O
O
(v)
O
OMe
OMe
O
RO
BzO
TrocHN
BzO
TrocHN
(iv)
11 R1 = Ac, R2 = H
9 R = H
(vi)
10 R = Ac
12 R1, R2 = Ac
Scheme 2. Reagents and conditions: (i) (a) PhC(OMe)3, TsOH, DMF; (b) 80% AcOH; 81%. (ii) HCl/MeOH, 77%. (iii) PhCH(OMe)2, TsOH, DMF. (iv) Ac2O, C6H5N, 94%. (v) NaCNBH3,
HCl–Et2O, THF, 86%. (vi) Ac2O, C6H5N, 68%.

1516
T. A. Jackson et al. / Bioorg. Med. Chem. 17 (2009) 1514–1526
yield): treatment with Zn dust in acetic anhydride gave the N-ace-
tate 16 which was submitted to Zemplén deacylation followed by
catalytic hydrogenation. From the protected monosaccharides, the
overall yield of our synthesis was 34% a considerable improvement
over the previously reported synthesis of 2 in which the trisaccha-
ride was obtained in 11% yield from protected monosaccharides.8
Having this fragment of LeaLex in hand we proceeded to study its
conformational behavior using a combination of NMR experiments
and computational methods.
2.2. NMR measurements
3
The 1H and 13C chemical shift assignments and JH,H coupling
constants for trisaccharide 2 were determined by a combination
of 1D and 2D NMR experiments. The 13C chemical shift values
are in close agreement with values reported previously.8 The vici-
nal coupling constants measured for the three sugar units sup-
ported an average 4C1 conformation for the galactose and N-
acetylglucosamine rings and a 1C4 conformation of the fucose ring.
Interresidue NOE interactions were evaluated using 2D NOESY
experiments at 300 and 310 K as well as 1D selective ROESY exper-
iments at 300 K. At 300 K, the molecular tumbling rate of trisaccha-
ride 2 was such that at 600 MHz NOESY signals were not of
sufficient intensity for quantification even when increasing the
number of scans eight fold or increasing sample concentration.
Dilution of the NMR sample and increasing of the temperature to
310 K to reduce sample viscosity and decrease sc allowed the col-
lection of usable NOE data. Figure 1 shows the significant region of
the NOESY spectrum obtained in these conditions at 800 ms mix-
ing time. The molecular tumbling rate of the trisaccharide was suf-
Figure 2. (a) 1H NMR spectrum (600 MHz, 300 K) of trisaccharide 2; (b) 1H, 1H
ROESY spectrum (mixing time 100 ms) upon selective excitation of proton H-100; (c)
1H, 1H ROESY spectrum (mixing time 100 ms) upon selective excitation of proton H-
10.
ficiently fast such that
xsc was less than 1.12, and cross-peaks in
the NOESY spectra were positive when the diagonal was phased
as negative. All expected intra-residue cross-peaks were evident:
medium to strong cross-peaks were observed for the glycosidic
protons H-100/H-40 and H-10/H-3, weak to medium cross-peaks
were observed for H-100/H-60a and H-100/H-60b as well as H-10/H-4
(Fig. 1). However, the %NOEs were still low (ꢁ1.2%), and the sig-
nal-to-noise ratio at this temperature required that the buildup
curves be fitted only to points at high mixing times (400 ms–1 s,
see Supplementary data). Thus, we acquired 1D ROESY experi-
ments at 300 K, selectively irradiating H-10 and H-100. As can be
seen in Figure 2, these experiments gave significantly better signal
intensities and cross relaxation buildup curves could be obtained
over mixing times ranging from 20 ms to 400 ms. Distances were
evaluated using the initial slopes of the normalized cross relaxa-
tion buildup curves (Fig. 3) fitted to a double exponential equation.
The R2 values for these fits were at least 0.999 in every instance.
These results are compared below to the outcome of our computa-
tional studies of the conformational behavior of trisaccharide 2.
1
Figure 3. 1H, H ROESY cross-relaxation build up curves for proton pairs H-10/H-3
(s), H-10/H-4 (.), H-10/H-30 ), H-100/H-40 (d).
(D
2.3. Computational studies
The different orientations around the glycosidic linkages of oli-
gosaccharides have the most substantial impact on the overall
molecular shape,14 and are therefore often the main underlying
determinant of binding properties to receptors such as antibod-
ies.15 We therefore focused our attention on the conformational
features associated with these linkages in the trisaccharide. The
orientations adopted around the glycosidic linkages are described
by two dihedral angles:
U = O5–C1–O1–Cx and W = C1–O1–Cx–
Cx + 1. The signs of the torsions are in agreement with the recom-
mendations of the IUPAC-IUB Commission of Biochemical
Nomenclature.16
2.3.1. Rigid search: potential energy surface of trisaccharide 2
Rigid-residue searches provide computationally efficient means
of studying potential energy surfaces and identifying key minima
of molecules.14 Rigid isoenergy contour maps were calculated for
trisaccharide 2 using the Tripos17 force field with the inclusion of
the PIM18 parameters for carbohydrates. For each trisaccharide
Figure 1. 600 MHz NOESY spectrum recorded at 310 K (800 ms mixing time).
linkage the resulting U–W map has been superimposed over the

T. A. Jackson et al. / Bioorg. Med. Chem. 17 (2009) 1514–1526
1517
previously reported relaxed energy maps of the corresponding
constituent disaccharide (Fig. 4a and b).19 As can be seen in Figure
4a and b, the overall conformational space as well as the locations
of minima agree closely with those obtained for the disaccharides.
For the b-GlcNAc-(1?3)-Gal linkage (Fig. 4a), the low-energy areas
identified for the rigid search (colored surfaces) are almost identi-
cal to those obtained for the disaccharide MM3 grid search19 (black
contours).
The potential energy surface displays four principal minima A–
D. The global minimum A, as well as local minima B and C adopt
linkage conformations corresponding to the preferred exo-ano-
meric U1 orientations20 while the local minimum D adopts a syn
U1 orientations. For the W1 linkage, minima A and D correspond
to (+) gauche orientation whereas minimum B fits an anti orienta-
tion and C a (ꢀ) gauche orientation. This potential energy map
(Fig. 4a) suggests that the conformationally accessible region sur-
rounding the D minimum is more extended for the trisaccharide
than for the corresponding disaccharide.
searches of corresponding disaccharide19 (black contours Fig. 4b).
However, it is difficult to say whether this indicates any reduced
flexibility of this linkage in the trisaccharide relative to the disac-
charide since this discrepancy can also be explained by the various
approximations inherent to the rigid-residue approach.14 For this
linkage, the potential energy map shows two main minima A0
and B0 (Fig. 4b). The U2 torsions of these minima both correspond
to the preferred orientation directed by the exo-anomeric effect.20
Minimum A0, which is the global minimum for the disaccharide,
corresponds to a (ꢀ) gauche orientation of the
minimum B0 corresponds to a (+) gauche orientation. Overall, the
fewer minima found for the -Fuc-(1?4)-GlcNAc linkage suggests
W
2 linkage, whereas
a
less potential flexibility for this linkage than for the b-GlcNAc-
(1?3)-Gal linkage.
2.3.2. Stochastic search: conformational families of
trisaccharide 2
Mapping the potential energy surface of the trisaccharide, and
identifying the main accessible low-energy plateaus, provides a
good starting point for predicting its solution conformational
behavior. However, it is possible that subtle structural variations
around the main minima significantly impact the overall shape
In contrast to the b-GlcNAc-(1?3)-Gal linkage ( Fig. 4a), the
diameters of the energy plateaus surrounding the minima for the
a-Fuc-(1?4)-GlcNAc linkage of the trisaccharide (colored surface
Fig. 4b) are marginally more restricted than for the MM3 grid
Figure 4. Potential energy maps of the glycosidic linkages of trisaccharide 2 superimposed on the energy maps obtained from previous MM3 grid searches for the
corresponding disaccharide linkages (black contours).19 (a) b-GlcNAc-(1?3)-Gal linkage and (b)
-Fuc-(1?4)-GlcNAc linkage: PIM rigid systematic search (colored); (c) b-
GlcNAc-(1?3)-Gal linkage and (d) -Fuc-(1?4)-GlcNAc linkage: MOE stochastic search (colored).
a
a

1518
T. A. Jackson et al. / Bioorg. Med. Chem. 17 (2009) 1514–1526
of the molecule. In addition, the energetically accessible combina-
tions of torsions around the glycosidic linkages are not readily
apparent from the energy maps of each linkage considered in iso-
lation. Thus, we proceeded to attempt the identification of such
combination via stochastic searches using the AMBER94 force
field21 as it is implemented in the Molecular Operating Environ-
ment (MOE2004) software package.22 In these stochastic searches,
minima in the potential energy surface are randomly sampled
through the rotation of all bonds (including ring bonds) to random
dihedral angles as well as through a random 0.4 Å Cartesian pertur-
bation of atom positions. Two searches of 150,000 iterations were
carried out to generate new conformers that were each minimized
with implicit solvation by the Generalized Born/Surface Area (GB/
SA) continuum solvation model.23 In the first search an RMS toler-
ance of 0.01 Å on heavy atoms was used to determine whether
generated conformations were considered as new. In the second
search an RMS tolerance of 0.1 Å was applied on all atoms includ-
ing all hydrogen atoms to account for possible hydrogen bonding.
The databases were combined and only those conformations found
within 10 kcal/mol from the global minimum were retained,
resulting in 1450 conformers. The potential energy maps obtained
for the two glycosidic linkages are shown as colored surfaces in
Figure 4c and d superimposed over the maps of the constituent
disaccharides shown as black contours. The minima located for
the individual linkages are in good agreement with the low-energy
areas previously identified for the corresponding disaccharides as
well as those obtained for the trisaccharide from the Tripos PIM ri-
gid systematic search. However, as can been seen when comparing
Figure 4a and c for the b-GlcNAc-(1?3)-Gal linkage as well as Fig-
optimization using AM1 semi-empirical calculations with implicit
COSMO solvation led to an identical conformation with an inter-
mediate
U
2 torsion of ꢀ110°. Thus, the orientations obtained from
the two force fields are both likely to be close in energy and closely
associated with the true global minimum orientation. Globally,
three conformational families (I, V, IX) were found to present ori-
entations for the two glycosidic bonds belonging to the energy pla-
teaus surrounding a combined AA0 global minimum identified by
the systematic search. Similarly, two conformational families (II,
III) corresponded to orientations close to a combined BA0 local
minimum and two families (IV, VI) were found that corresponded
to a DA0 combined local minimum. The remaining families of high-
er energy (VII, VIII, X) corresponded, respectively to combined lo-
cal minima AB0, BB0 and CA0. Overall, the stochastic search and the
rigid search gave conformational profiles that were in good agree-
ment and the stochastic search complemented well the rigid
search by showing three combinations of glycosidic torsions that
seem to be energetically favored: the global minimum AA0 and
two local minima BA0 and DA0 (Fig. 4). As shown below these early
conclusions were in full agreement with the molecular dynamics
simulations and NMR experiments.
2.3.3. Molecular dynamics simulations: flexibility of
trisaccharide 2
Molecular dynamics simulations at 300 and 310 K were carried
out using AMBER924 with the inclusion of Glycam04 parameters
for carbohydrates.25 We ran trajectories of 8 ns in explicit water
starting from the global minimum (AA0) obtained in the Tripos-
PIM systematic search. The average values of the
U and W dihedral
ure 4b and d for the a-Fuc-(1?4)-GlcNAc linkage, the low energy
angles for both glycosidic linkages as well as the associated calcu-
lated standard deviations are given in Table 1, entries 1 and 2.
areas identified in the stochastic search appear to be much more
restricted than in those found by the PIM rigid search. This results
from the fact that such stochastic searches are unlikely sampling
the entire conformational space. In fact, our searches did not meet
the ‘number of failures to find new conformations’ termination cri-
teria (10,000 iterations in a row), and terminated at the end on the
150,000 iterations. The conformations obtained within 3 kcal/mol
of the global minimum were clustered into families based upon
The superimposition of the
U–W trajectories for each linkage
over the adiabatic maps for the corresponding disaccharides19
are shown in Figure 5 and the variations of the individual
glycosidic torsions for the length of the simulation are shown in
Figure 6.
The data obtained when running the dynamic simulation at
300 K and seen in Figures5a, b and 6a, b show that at this temper-
ature both the b-GlcNAc-(1?3)-Gal and a-Fuc-(1?4)-GlcNAc link-
orientations around the two sets of
U and W linkages. A conforma-
tion was considered as belonging to a different family if at least one
of its glycosidic torsions differed by more than 30° from another
conformation. Ten different conformational families (I–X) were
identified and the lowest energy representative conformations of
each family are identified in Figure 4c and d.
ages remained in the same energy plateau. With respect to the b-
GlcNAc-(1?3)-Gal linkage, the trisaccharide remained in the
low-energy region surrounding the global minimum A, and local
minimum B throughout most of the simulation. Standard devia-
tions values of 22o and 24o, for the
U W
1 and 1 torsions, respectively
For the b-GlcNAc-(1?3)-Gal linkage, 7 out of the 10 families
adopt orientations that belong to the same energy plateau as min-
ima A and B identified in the systematic searches, of which 3 fam-
ilies correspond to minimum A and 3 to minimum B. Two families
were found to belong to the plateau in which the rigid search ori-
entation D was found and one conformational family was identi-
fied in the energy plateau that contained orientation C found by
(Table 1, entry 1) showed that there were significant fluctuations
around these torsions such that the entire low-energy region was
explored and the multiple distinct conformational families associ-
ated with the minima A and B energy wells were visited.
Table 1
the rigid search. For the
a-Fuc-(1?4)-GlcNAc linkage, the energy
Calculated torsion anglesa
plateau that includes the minimum orientation A0 identified by
the systematic search dominates the trisaccharide families, and
conformations close to this minimum are adopted in 8 of the 10
families identified. The plateau represented by the orientation B0
found in the systematic search is identified in the remaining two
families.
Entry
U1
W1
U2
W2
Molecular dynamics simulationsb
T (K)
300
1
2
ꢀ90
22c
94
ꢀ85
16c
ꢀ130
24c
24c
310
ꢀ74
101
ꢀ94
ꢀ132
52c
30c
26c
23c
When comparing the global minimum found by the stochastic
Rigid search conformationsd
3
4
5
search (I, Fig. 4c and d) to that found by the rigid search (AA0,
AA0
BA0
DA0
ꢀ79
ꢀ73
62
58
148
104
ꢀ76
ꢀ76
ꢀ76
ꢀ147
ꢀ146
ꢀ142
Fig. 4a and b), we noticed that the U2 torsions of the
a-Fuc-
(1?4)-GlcNAc linkage differed by 70° (compare Fig. 4b and d)
while the remaining torsions were found to be in excellent agree-
ment with one another. This difference is likely resulting from the
different parameterizations of the force fields involved. Indeed, for
both the MOE AMBER94 and TRIPOS PIM global minima, further
a
b
c
U1,
W = a-Fuc-(1?4)-GlcNAc.
1 = b-GlcNAc-(1?3)-Gal; U2, W2
Average torsions calculated for the MD simulations.
Standard deviation.
Selected from the rigid search and minimized with the Tripos-PIM force field.
d

T. A. Jackson et al. / Bioorg. Med. Chem. 17 (2009) 1514–1526
1519
Figure 5. Trajectories of the glycosidic torsion angles
energy maps obtained for the disaccharide linkages (contours).19 The main minima identified in the PIM rigid systematic search are labeled A–D and A0, B0. (a) and (c) U1
for the b-GlcNAc-(1?3)-Gal linkage; (b) and (d) U2 W2 for the
-Fuc-(1?4)-GlcNAc linkage.
U, W during the 8 ns dynamics simulations at 300 K (a and b) and 310 K (c and d) superimposed on the MM3 grid search
,
W1
,
a
The
a
-Fuc-(1?4)-GlcNAc linkage appeared to be less flexible
drastic conformational change around the GlcNAc-(1?3)-Gal link-
age, the -Fuc-(1?4)-GlcNAc linkage remained in the energy well
than the b-GlcNAc-(1?3)-Gal linkage. It remained in the energy
well surrounding the A0 global minimum identified by the TRIPOS
PIM force field throughout the entire simulation with standard
a
surrounding the A0 minimum throughout the entire simulation
(Fig. 5d). The average U2 and W2 values calculated for the 310 K
simulation were similar to that calculated at 300 K and only a
small increase in flexibility around U2 was shown by the greater
standard deviation obtained for this torsion. While for both the
deviations of 16o and 24o for U2 and
W
2, respectively. The results
of the 8 ns simulation at 310 K are shown in Figures 5c, d and 6c,
d and support an expected increased flexibility of both glycosidic
bonds at this higher temperature. Indeed at 310 K a marked in-
creased flexibility of the b-GlcNAc-(1?3)-Gal linkage was ob-
served. While the average values and standard deviations
calculated for W1 were similar to that obtained at 300 K, the U1
torsion and its associated standard deviation were significantly dif-
ferent from that obtained at 300 K (Table 1, entry 2). Indeed the
significantly increased flexibility around that torsion was further
U
2 and 2 torsions significant fluctuations in the torsions adopted
W
were observed such that the entire low-energy region associated
with the minimum A0 energy well was explored, no transitions
were observed to the B0 minimum of this linkage. However, this
linkage in trisaccharide 2 still shows considerable flexibility when
compared to branched trisaccharides containing the same linkage
as well as to the corresponding disaccharide.26
Overall, the trisaccharide displays significant flexibility in solu-
tion. As could be anticipated from the results of the combined rigid
and stochastic searches, it readily adopts conformations around
the AA0 and BA0 minima at 300 K but also visits the second local
minimum DA0 at 310 K. Indeed, 8 ns molecular dynamics simula-
tions at 300 and 310 K starting from this local minimum (DA0) con-
firmed that while visited, the trisaccharide did not maintain
extensively the D orientation around the b-GlcNAc-(1?3)-Gal link-
age (Fig. 7). Indeed, at both temperatures, this glycosidic torsion
quickly returned to the minimum energy plateau containing the
A and B orientations, albeit more slowly at 310 K confirming that
seen when looking at the U1
–
W1 trajectories for this linkage (Fig-
ure 5c) and the variations of the individual U1 torsion (Fig. 6c).
While the trisaccharide remained in the low-energy region sur-
rounding the A and B minima throughout most of the simulation,
transitions occurred to the region surrounding the D orientation
(Fig. 5c) corresponding to U1 orientations between 0° and 60°.
Careful examination of the trajectory followed by the
at 310 K (Fig. 6c), showed that once the conformation had reached
the local minimum D with a
remained in that energy well for some time before returning to the
U
1 torsion
U
1 value of ꢁ60° (Table 1, entry 5), it
energy well surrounding the minima A and B. In contrast to the

1520
T. A. Jackson et al. / Bioorg. Med. Chem. 17 (2009) 1514–1526
Figure 6. Glycosidic dihedral trajectories during the 8 ns dynamics simulations at 300 K (a and b) and 310 K (c and d). (a) and (c) b-GlcNAc-(1?3)-Gal linkage. (b) and (d)
Fuc-(1?4)-GlcNAc linkage.
a-
this conformation is more likely to be populated at higher temper-
atures. As expected, the -Fuc-(1?4)-GlcNAc linkage in these later
simulations remained throughout the simulations in the energy
well surrounding the A0 minimum (data shown in Supplementary
data).
iors. The H-10/H-3 and H-100/H-4 theoretical and experimental dis-
a
tances agree completely while the H-10/H-4 distance is larger than
that measured for the global minimum AA0 and well within exper-
imental range of the calculated distances at 300 and 310 K. Thus
while the
a-Fuc-(1?4)-GlcNAc linkage adopts predominantly the
Finally, Figure 8 shows the overlay of 200 conformations ex-
tracted from the 8 ns MD simulation at 310 K starting from the glo-
bal minimum AA0. The flexibility of the molecule and the particular
flexibility of the b-GlcNAc-(1?3)-Gal linkage are clearly evident.
A0 type orientation with very little contribution from the B0 orien-
tation the GlcNAc-(1?3)-Gal linkage is likely sampling the A and B
orientation as well as possibly the D as observed in the 310 K
molecular dynamics simulation (Fig. 4c). Overall, in terms of the
conformations of the trisaccharide, these favored orientations of
the glycosidic linkages support the presence of significant popula-
tions of trisaccharide conformational families AA0, BA0 and DA0 in
solution. The extensive flexibility of both glycosidic linkages sug-
gests that the distinct conformational families of the trisaccharide
can readily interconvert in solution.
2.4. Comparison of experimental and theoretical data
Assuming isotropic motion, theoretical hrꢀ6i averaged distances
were calculated for H-100/H-40, H-10/H-3 and H-10/H-4 over all of the
conformations generated during the MD simulations started from
the global minimum AA0 at 300 and 310 K (Table 2, entries 1 and
2). Table 2 also gives the distances measured for the global mini-
mum AA0 and local minima BA0 and DA0 selected from the rigid
search and minimized with the Tripos-PIM force field (entries 3–
5). It is interesting to notice that the H-10/H-4 distance is substan-
tially longer (>4 Å) for both the B and D orientations around the
GlcNAc-(1?3)-Gal linkage (Table 2, entries 4 and 5) than for the
A orientation (entry 3). Indeed, the calculated average distance be-
tween these two hydrogens is 2.8 Å in both molecular dynamics
runs at 300 and 310 K and there is no significant difference in this
calculated distance whether the molecule visits the energy plateau
surrounding the D orientation, or whether it remains in the energy
plateau that encompasses the A and B minima. Therefore, experi-
mental measurement of this distance does not allow confirmation
of whether or not the transition to the DA0 local minimum effec-
tively occurs at 310 K. Indeed, the distances measured experimen-
tally (ROE) at 300 K are in good agreement (Table 2, entries 1, 2 and
6) with the distances calculated at 300 or 310 K suggesting that the
true solution behavior is compatible with both simulation behav-
3. Conclusion
We have outlined a more efficient synthesis and the major con-
formational features of a trisaccharide fragment of the LeaLex hexa-
saccharide tumor antigen as part of our ongoing efforts to identify
internal epitopes that may be useful in the development of anti-
cancer vaccines. Explorations of the conformational space occupied
by the trisaccharide were carried out using the unprecedented
combination of a rigid systematic search as implemented in Sybyl
and using the Tripos force field along with the PIM parameters and
a stochastic search as implemented in MOE2004 and using the Am-
ber94 force field. Indeed, we have shown that the two methods
complement each other well and, in this case, allow the identifica-
tion of three conformations that are energetically relevant. Molec-
ular dynamics simulations in explicit water using the Amber9 force
field with the inclusion of the Glycam parameters confirmed that
this linear trisaccharide was highly flexible. Dynamic simulations
starting from the global minimum at 300 K showed that two of

T. A. Jackson et al. / Bioorg. Med. Chem. 17 (2009) 1514–1526
1521
Figure 7. Trajectories of the glycosidic torsion angles U1 1 for the b-GlcNAc-(1?3)-Gal linkage during the 8 ns dynamics simulations starting from the local minimum D.
, W
Simulations (a) and (b) at 300 K; (c) and (d) at 310 K. For (a) and (c) the trajectories are superimposed on the MM3 grid search energy maps obtained for the disaccharide
linkages (contours).19 The main minima identified in the PIM rigid systematic search are labeled A–D.
Table 2
Calculated and measured inter-proton distances
Entry
H-10/H-3 (Å)
H-10/H-4 (Å)
H-100/ H-40 (Å)
Molecular dynamics simulationsa
T (K)
1
2
300
2.2
0.2b
2.3
0.5b
2.8
2.2
0.8b
2.8
0.2b
2.2
310
1.0b
0.2b
Rigid search conformationsc
3
4
5
AA0
BA0
DA0
2.4
2.6
3.6
2.6
4.4
4.1
2.2
2.2
2.2
NMR measurements
6
2.3
3.2
2.4
a
Average distances calculated for the MD simulations.
Standard deviation.
Selected from the rigid search and minimized with the Tripos-PIM force field.
b
c
the three conformers identified by the combined systematic and
stochastic searches were visited while simulations at 310 K
showed that the third conformation was also accessible. The theo-
Figure 8. Overlay of 200 snapshots from the 8 ns molecular dynamics simulation at
310 K.

1522
T. A. Jackson et al. / Bioorg. Med. Chem. 17 (2009) 1514–1526
retical distances averaged from these dynamic simulations were
matched to the NOE derived distances confirming that the trisac-
charide was highly flexible in solution. However, because the
experimental distances measured from the NOE buildup curves
were in good agreement with the distances calculated from both
sets of dynamic simulations starting from the global minimum at
300 and 310 K, it was impossible to confirm whether or not both
identified local minima were indeed populated in solution. The
b-GlcNAc-(1?3)-Gal linkage was shown to be particularly flexible
adopting a wide range of orientations around these conformations.
Additional studies to shed light on the conformational behavior of
this linkage in the LeaLex hexasaccharide are ongoing in our labora-
tory and will be reported in due time. However, our preliminary re-
sults do indicate that this linkage is, indeed, quite flexible even in
the larger hexasaccharide. The flexibility around this central link-
age suggests that there may be multiple internal epitopes of the
hexasaccharide that can be targeted for anti-cancer vaccine
development.
J = 10.0 Hz, H-2), 5.23 (d, 4H, J = 8.6 Hz, H-1), 5.19–4.95 (m, 4H,
H-300, H-500, OCH2CCl3), 4.91 (d, 1H, J = 8.7 Hz, NH00), 4.72 (d, 1H,
J = 12.2 Hz, OCHHCCl3), 4.68–4.62 (m, 2H, H-600a, OCHHCCl3),
4.59 (d, 1H, J = 8.2 Hz, H-100), 4.53–4.42 (m, 3H, H-10, H-40, H-
600b), 4.40–4.32 (m, 2H, H-50, H-6a), 4.21 (dd, 1H, J = 3.8,
12.2 Hz, H-60a), 4.10 (br d, 1H, J = 10.5 Hz, H-6b), 4.00 (br d, 1H,
J = 11.3 Hz, H-60b), 3.96–3.82 (m, 3H, H-3, H-4, H-200), 3.72–3.63
(m, 2H, H-5, H-400), 3.59 (t, 1H, J = 9.1 Hz, H-200), 3.44 (s, 3H,
CH3O), 2.18–1.92 (6s, 6 ꢂ 3H, 6 ꢂ CH3CO). 13C NMR (100 MHz,
CDCl3): d 170.0 (CO), 133.8, 133.6, 130.0, 129.9, 129.0, 128.8
(Ar), 101.8 (C-100, 101.2 (C-10), 99.8 (C-1), 79.9 (C-3 or C-4 or C-
20), 74.7 (OCH2CCl3), 72.6, 72.2 (C-3 or C-4 or C-20, C-2, C-5, C-
30, C-40, C-50, C-300, C-400, C-500), 70.7 (OCH2CCl3), 69.2 (C-600), 68.8
(C-6), 62.3 (C-60), 62.1 (C-2), 56.9 (C-3 or C-4 or C-20, C-200), 56.4
(CH3O), 21.0, 20.8 (CH3CO). HRMS calcd for C51H58Cl6N2O26
[M+K]+ 1363.1046, found 1363.1057.
4.1.3. Methyl 2,4,6-tri-O-benzoyl-b-D-galactopyranoside (6)
Diol 4 (1.2 g, 3.0 mmol) was dissolved in CH2Cl2 (42 mL),
trimethylorthobenzoate (1.8 mL, 10 mmol) and TsOH (42 mg,
0.25 mmol) were added and the mixture was stirred 1 h under
4. Experimental
N2 at room temperature. Et3N (40 lL) was added to the reaction
4.1. Synthesis of fragment 2
mixture, and solvents were evaporated. The residue was dis-
solved in 80% AcOH–H2O (55 mL), and the solution was stirred
at room temperature for 2-h. Solvents were removed by co-
evaporation with toluene, and chromatography of the residue
(7:3 hexanes–EtOAc) gave acceptor 6 (1.22 g, 81%) pure as a col-
orless glass. 1H NMR (400 MHz, CDCl3): d 7.39–8.17 (m, 15 H,
Ar), 5.78 (d, 1H, J = 3.4 Hz, H-4), 5.38 (dd, 1H, J = 7.9, 10.0 Hz,
H-2), 4.63 (d, 1H, J = 7.9 Hz, H-1), 4.60 (dd, 1H, J = 6.8, 11.3 Hz,
H-6a), 4.42 (dd, 1H, J = 6.2, 11.4 Hz, H-6b), 4.18–4.05 (m, 2H,
H-3, H-5), 3.57 (s, 3 H, CH3O). 13C NMR (100 MHz, CDCl3): d
166.8, 166.4, 166.2 (CO), 133.7, 133.5, 133.4, 130.2, 130.0,
129.9, 129.6, 129.2, 128.7, 128.6, 128.5 (Ar), 102.2 (C-1), 73.6
(C-2), 71.9 (C-3 or C-5), 71.5 (C-3 or C-5), 70.6 (C-4), 62.5 (C-
6), 57.2 (CH3O). The NMR data agreed with that previously
reported.12
4.1.1. Materials and methods
1H (600, 400 or 300 MHz) and 13C NMR (150, 100 or 75.5 MHz)
spectra were recorded for compounds solubilized in CDCl3 or D2O.
1H chemical shits were referenced to TMS (d 0.0, CDCl3) or sodium
2,2-dimethyl-2-silapentane-5-sulfonate (external standard DSS, d
0.0). Chemical shifts and coupling constants were obtained from
a first-order analysis of one-dimensional spectra. Data are reported
as s = singlet, d = doublet, t = triplet, q = quartet, m = multiplet,
b = broad and coupling constant(s) are in Hertz. 13C chemical shifts
were referenced to internal CDCl3 (d 77.0) or external DSS (d 0.0).
Assignments of proton and carbon resonances were based on COSY
and 13C–1H heteronuclear correlated experiments. Mass spectra
were obtained under electron spray ionisation (ESI) on a high res-
olution mass spectrometer. TLC were performed on precoated alu-
minum plates with Silica Gel 60 F254 and detected with UV light
and/or charred with a solution of 10% H2SO4 in EtOH. Compounds
were purified by flash chromatography with Silica Gel 60 (230–400
mesh) unless otherwise stated. Solvents were distilled and dried
according to standard procedures,27 and organic solutions were
dried over Na2SO4 and concentrated under reduced pressure below
40 °C. Centrifugal chromatography was performed on a silica gel
plate (2 mm thickness). Optical rotation measurements were
uncorrected.
4.1.4. Methyl 3-O-[3,4,6-tri-O-acetyl-2-deoxy-2-(2,2,2-
trichlorethoxycarbonylamino)-b-
benzoyl-b- -galactopyranoside (7)
The acceptor 6 (175 mg, 0.35 mmol), the donor 3 (374 mg,
0.60 mmol) and activated powdered molecular sieves
D-glucopyranosyl]-2,4,6-tri-O-
D
4
Å
(480 mg) were stirred in CH2Cl2 (6.4 mL) at room temperature
under N2 for 1 h. The mixture was cooled to ꢀ60 °C, TMSOTf
(64 lL, 0.35 mmol) was added, and the mixture was stirred at
ꢀ30 °C overnight. Et3N (45
lL) was added to the reaction mix-
ture; the solids were filtered off, washed with CH2Cl2 and the
filtrate and washings were combined and concentrated. Chroma-
tography (8:2 to 1:1 hexanes–EtOAc) of the residue gave disac-
4.1.2. Methyl 3,4-di-O-[3,4,6-tri-O-acetyl-2-deoxy-2-(2,2,2-
trichlorethoxycarbonylamino)-2,6-di-O-benzoyl-b-
D-
glucopyranosyl]-b- -galactopyranoside (5)
D
charide 7 (296 mg, 88%) pure as colorless glass. [a]D = +38 (c 2.3,
The donor 3 (22.2 mg, 0.04 mmol), the acceptor 4 (21.2 mg,
0.05 mmol) and activated powdered 4 Å MS (200 mg) were stir-
red in CH2Cl2 (2 mL) at room temperature under N2 for 1 h. The
mixture was cooled to ꢀ78 °C, TMSOTf (0.28 M in CH2Cl2,
CHCl3). 1H NMR (400 MHz, CDCl3): d 8.17–7.37 (m, 15H, Ar),
5.85 (d, 1H, J = 3.2 Hz, H-4), 5.62 (dd, 1H, J = 7.9, 9.9 Hz, H-2),
5.37 (t, 1H, J = 9.9 Hz, H-30), 5.01 (d, 1H, J = 8.0 Hz, H-10), 4.94
(t, 1H, J = 9.6 Hz, H-40), 4.89 (d, 1H, J = 7.7 Hz, NH), 4.57 (d, 1H,
J = 7.9 Hz, H-1), 4.53 (dd, 1H, J = 5.3, 11.7 Hz, H-6a), 4.48 (dd,
1H, J = 7.1, 11.7 Hz, H-6b), 4.26–4.06 (m, 5H, H-3, H-5, H-60a,
H-60b, OCHHCCl3), 3.79–3. 62 (m, 2H, H-50, OCHHCCl3), 3.50 (s,
22.3 lL, 6.2 lmol) was added, and the mixture was allowed to
warm to room temperature overnight. The mixture was cooled
to ꢀ78 °C, additional TMSOTf (0.28 M in CH2Cl2, 45.0 L, 13 mol)
was added and the mixture was again allowed to warm to room
temperature overnight. Et3N (5 L) was added, the mixture was
l
l
3
H, CH3O), 3.20 (q, 1
H, J = 8.8 Hz, H-20), 1.97, 1.89 (3s,
l
3 ꢂ 3H, 3 ꢂ CH3CO). 13C NMR (100 MHz, CDCl3): d 171.1, 166.5,
165.3 (CO), 133.7, 133.5, 130.5, 130.2, 130.0, 129.8, 128.9,
128.8, 128.7 (Ar), 102.4 (C-1, C-10), 95.6 (CCl3), 78.1 (C-3 or C-
5), 73.8 (OCH2CCl3), 72.1 (C-3 or C-5), 71.9 (C-2), 71.5 (C-30),
70.3 (C-4), 68.9 (C-40), 63.3 (C-6), 61.8 (C-60), 57.2 (CH3O), 56.8
(C-20), 20.9, 20.9, 20.7 (CH3CO). HRMS calcd for C43H44Cl3NO18
[M+NH4]+ 985.1968, found 985.1971.
filtered, the solids were washed with CH2Cl2 and the filtrate
and washings were combined and concentrated. Chromatography
(6:4 to 1:1 hexanes–EtOAc) gave trisaccharide 5 (18.0 mg, 76%)
pure as an amorphous solid. [
(400 MHz, CDCl3): 7.38–8.15 (m, 10H, Ar), 6.06 (d, 1H,
J = 8.8 Hz, NH0), 5.42 (t, 1H, J = 9.1 Hz, H-30), 5.32 (t, 1H,
a]
D = +6 (c 0.7, CHCl3). 1H NMR
d

T. A. Jackson et al. / Bioorg. Med. Chem. 17 (2009) 1514–1526
1523
4.1.5. Methyl 2,4,6-tri-O-benzoyl-3-O-[2-deoxy-2-(2,2,2-
trichlorethoxycarbonylamino)-b-D-glucopyranosyl]-b-D-
galactopyranoside (8)
layers were washed with water (30 mL), combined, dried and con-
centrated. Chromatography (6:4 hexanes–EtOAc) of the residue
gave the disaccharide acceptor 11 (148 mg, 86%) pure as a white
AcCl (400
lL) was added to a solution of disaccharide 7
amorphous powder. [a]
D = +16 (c 1.2, CHCl3). 1H NMR (300 MHz,
(206 mg, 0.21 mmol) in MeOH (5 mL) and the reaction mixture
was stirred overnight under N2 at room temperature. Solid NaHCO3
was added to the reaction mixture, solids were filtered off and
washed with MeOH (10 mL). The filtrate and washings were com-
bined, concentrated and chromatography of the crude residue
(49:1 CHCl3–MeOH) gave triol 8 (138 mg 77%) pure as a colorless
CDCl3): d 8.15–7.28 (m, 20H, Ar), 5.84 (d, 1H, J = 3.0 Hz, H-4),
5.54 (dd, 1H, J = 7.9, 9.9 Hz, H-2), 4.98 (t, 1H, J = 9.8 Hz, H-30),
4.81 (d, 1H, J = 8.1 Hz, H-10), 4.74 (d, 1H, J = 8.1 Hz, NH), 4.59–4.
45 (m, 5H, H-1, H-6a, H-6b, CH2Ph), 4.21 (dd, 1H, J = 10.0, 3.1 Hz,
H-3), 4.13 (t, 1H, J = 7.1 Hz, H-5), 4.03 (dd, 1H, J = 11.9 Hz,
OCHHCCl3), 3.93 (dd, 1H, J = 12.1 Hz, OCHHCCl3), 3.74 (dd, 1H,
J = 4.06, 10.4 Hz, H-60a), 3.79 (dd, 1H, J = 10.4 Hz, H-60b), 3.66 (t,
1H, J = J = 9.1 Hz, H-40), 3.50 (br s, 4H, H-50,CH3O), 3.29 (q, 1H,
J = 9.0 Hz, H-20), 3.07 (br s, 1H, OH), 1.93 (s, 3H, CH3CO). 13C NMR
(75 MHz, CDCl3): d 172.3, 166.6, 165.7 (CO), 137.5-128.3 (Ar),
108.7 (C-1), 101.7 (C-10), 95.6 (CCl3), 77.9 (C-3), 74.6 (C-30, C-50),
74.2 (OCH2CCl3, CH2Ph), 72.0 (C-5), 71.8 (C-2), 70.7 (C-40), 70.3
(C-4), 69.6 (C-60), 57.4 (CH3O, C-20), 21.0 (CH3CO). HRMS calcd for
C46H46Cl3NO16 [M+Na]+ 996.1780, found 996.1779.
glass. [a]
D = +10 (c 0.7, MeOH). 1H NMR (400 MHz, CDCl3): d
8.18–7.38 (m, 15H, Ar), 5.99 (d, 1H, J = 3.2 Hz, H-4), 5.56 (dd, 1H,
J = 8.1, 9.6 Hz, H-2), 5.31 (d, 1H, J = 6.7 Hz, NH), 4.81 (br d, 1H,
J = 6.4 Hz, H-10), 4.62–4. 50 (m, 2H, H-1, H-6a), 4.38 (dd, 1H,
J = 5.8, 11.5 Hz, H-6b), 4.22–4. 01 (m, 3H, H-3, H-5, OCHHCCl3),
3.82–3.52 (m, 6H, H-30, H-60a, H-60b, OCHHCCl3, 2 ꢂ OH), 3.51 (s,
1H, CH3O), 3.27 (br s, 1H, H-50), 3.28 (br s, 1H, H-40), 2.85 (br s,
1H, H-20), 2.00 (br s, 1H, OH). 13C NMR (100 MHz, CDCl3): d
167.0, 166.6, 165.7, 154.6 (CO), 134.2, 133.8, 130.7, 130.4, 130.1,
129.9, 129.4, 129.0, 128.9 (Ar), 102.6 (C-1, C-10), 76.2, 71.7, 70.9
(C-2, C-3, C-4, C-5, C-30, C-40, C-50, OCH2CCl3), 62.7 (C-6), 61.6 (C-
60), 58.8 (C-20), 57.4 (CH3O). HRMS calcd for C37H38Cl3NO15
[M+Na]+ 864.1205, found 864.1160.
4.1.8. Methyl 3-O-[3,4-di-O-acetyl-6-O-benzyl-2-deoxy-2-(2,2,2-
trichlorethoxycarbonylamino)-b-
D-glucopyranosyl]-2,4,6-tri-O-
benzoyl-b- -galactopyranoside (12)
D
A solution of the alcohol 11 (16.0 mg, 0.02 mmol) in Ac2O
(0.2 mL) and pyridine (0.4 mL) was stirred under N2 at room tem-
perature for 2 h. Solvents were co-evaporated with toluene, and
the crude product was purified by chromatography (6:4 hex-
anes–EtOAc) to yield the pure diacetate 12 (11.3 mg, 68%).
4.1.6. Methyl 3-O-[3-O-acetyl-4,6-O-benzylidene-2-deoxy-2-
(2,2,2-trichlorethoxycarbonylamino)-b-
2,4,6-tri-O-benzoyl-b- -galactopyranoside (10)
Benzaldehyde dimethyl acetal (480 L, 3.20 mmol) and TsOH
D-glucopyranosyl]-
D
l
[a]
D = +36 (c 0.9, CHCl3). 1H NMR (400 MHz, CDCl3): d 8.18–7.21
(92 mg, 0.53 mmol) were added to a solution of triol 8 (1.28 g,
1.52 mmol) in DMF (29 mL) and the solution was stirred at 80 °C
for 1 h. Et3N (80 lL) was added to the reaction mixture, solvents
(m, 20H, Ar), 5.88 (d, 1H, J = 3.2 Hz, H-4), 5.60 (dd, 1H, J = 8.1,
9.7 Hz, H-2), 5.30 (t, 1H, J = 9.9 Hz, H-3’), 4.95–4.84 (m, 2H, H-10,
H-40), 4.77 (d, 1H, J = 7.9 Hz, NH), 4.55–4.42 (m, 5H, H-1, H-6a, H-
6b, CH2Ph), 4.20–4.10 (m, 2H, H-3, OCHHCCl3), 4.05 (t, 1H,
J = 6.3 Hz, H-5), 3.78 (d, 1H, J = 12.1 Hz, OCHHCCl3), 3.70–3.61 (m,
1H, H-50), 3.60–3.50 (m, 2H, H-60a, H-60b), 3.50 (s, 1H, CH3O),
3.21 (q, 1H, J = 8.9 Hz, H-20), 1.87 (2s, 2 ꢂ 3H, 2 ꢂ CH3CO). 13C
NMR (75 MHz, CDCl3): d 171.8, 166.1, 154.3, 153.2 (CO), 144.1,
138.0, 133.4, 133.3, 133.1, 130.2, 130.0, 130.0, 128.6, 128.4, 127.7
(Ar), 117.1 (CCl3) 102.2 (C-1, C-10), 73.6 (OCH2CCl3, CH2Ph), 73.4,
72.3, 69.9 (C-2, C-4, C-5, C-30, C-40), 69.7 (C-60), 62.9 (C-6), 57.1
(CH3O), 56.5 (C-20), 20.6, 20.4 (CH3CO). HRMS calcd for
C48H48Cl3NO17 [M+NH4]+ 1033.2332, found 1033.2343.
were evaporated and the crude residue was dissolved in pyridine
(10 mL) and Ac2O (5 mL). The solution was stirred under N2 at room
temperature for 1.5 h diluted with water (ꢁ20 mL) and extracted
with CH2Cl2 (3 ꢂ ꢁ20 mL). The combined organic phases were
washed successively with 1 M HCl, satd aq NaHCO3 and water and
theaqueouslayerswerere-extractedwithCH2Cl2. Theorganiclayers
were combined, dried, concentrated, and chromatography (3:1 hex-
anes–EtOAc) of the residue gave disaccharide 10 (1.38 g, 94%) pure
as an amorphous powder. [a]
D = +13 (c 3.1, CHCl3). 1H NMR
(400 MHz, CDCl3): d 8.18–7.31 (m, 20H, Ar), 5.81 (d, 1H, J = 2.9 Hz,
H-4), 5.62 (dd, 1H, J = 8.1, 9.7 Hz, H-2), 5.42 (s, 1 H, CHPh), 5.25 (t,
1H, J = 9.8 Hz, H-30), 4.82 (d, 1H, J = 8.2 Hz, H-10), 4.73 (d, 1H,
J = 8.6 Hz, NH), 4.57 (d, 1H, J = 7.9 Hz, H-1), 4.54–4. 46 (m, 2H, H-
6a, H-6b), 4.31 (dd, 1H, J = 4.8, 10.7 Hz, H-60a), 4.20–4.10 (m, 3H,
H-3, H-5, OCHHCCl3), 4.03 (d, 1H, J = 12.2 Hz, OCHHCCl3), 3.67 (t,
1H, J = 10.2 Hz, H-60b), 3.59–3.32 (m, 6H, H-20, H-40, H-50, CH3O),
1.91 (s, 3H, CH3CO). 13C NMR (100 MHz, CDCl3): d 166.6, 165.4
(CO), 137.2, 133.9, 133.7, 130.3, 130.1, 129.5, 129.1, 128.9, 128.6,
126.5 (Ar), 102.6, 101.7 (C-1, C-10, CHPh), 95.8 (CCl3), 78.8 (C-3, C-
40), 74.2 (OCH2CCl3), 72.1, 71.7, 70.4 (C-2, C-4, C-5, C-30), 68.7 (C-
60), 66.7 (C-50), 63.3 (C-6), 57.3 (C-20, CH3O), 21.0 (CH3CO). HRMS
calcd for C46H44Cl3NO16 [M+H]+ 972.1804, found 972.1762.
4.1.9. Methyl 3-O-[3-O-acetyl-6-O-benzyl-4-O-(2,3,4-tri-O-
benzyl-
trichlorethoxycarbonylamino)-b-
benzoyl-b- -galactopyranoside (15)
a
-
L
-fucopyranosyl)-2-deoxy-2-(2,2,2-
D-glucopyranosyl]-2,4,6-tri-O-
D
Method A. A solution of acceptor 11 (93.1 mg, 0.10 mmol), thio-
glycoside donor 13 (162 mg, 0.34 mmol) in Et2O (5 mL) containing
activated powdered 4 Å molecular sieves (173 mg) was stirred un-
der N2 at room temperature for 1 h. MeOTf (53
added and the mixture was stirred at room temperature for an
additional 1.5 h. Et3N (63 L) was added to the mixture; the solids
lL, 0.47 mmol) was
l
were filtered off and washed with Et2O (ꢁ10 mL). The filtrate and
washings were combined, concentrated and centrifugal purifica-
tion (17:3 to 7:3 hexanes–EtOAc) of the crude residue gave trisac-
charide 15 (114 mg, 85%) pure as an amorphous powder.
4.1.7. Methyl 3-O-[3-O-acetyl-6-O-benzyl-2-deoxy-2-(2,2,2-
trichlorethoxycarbonylamino)-b-
D-glucopyranosyl]-2,4,6-tri-O-
benzoyl-b- -galactopyranoside (11)
D
Method B. Coupling of the acceptor 11 (29.6 mg, 0.03 mmol),
NaCNBH3 (165 mg, 2.63 mmol) and activated 3 Å molecular
sieves (0.66 g) were added to a solution of benzylidene acetal 10
(172 mg, 0.18 mmol) stirred under N2 at room temperature in an-
hyd THF (3.8 mL). A solution of HCl in Et2O (2.0 M, 1.7 mL,
3.4 mmol) was then added dropwise until the evolution of gas
had ceased and the mixture remained acidic (litmus paper). The
mixture was stirred at room temperature for 45 min, decanted into
a separatory funnel and satd aq NaHCO3 (20 mL) was added. The
mixture was extracted with CH2Cl2 (2 ꢂ ꢁ20 mL) and the organic
and thioglycoside 14 (51.5 mg, 0.11 mmol) under MeOTf (20 lL,
0.18 mmol) activation was carried out in Et2O (1.5 mL) as de-
scribed above for the coupling of acceptor 11 with donor 13. Work
up, as described above, followed by column chromatography (7:3
hexanes–EtOAc) of the crude residue gave trisaccharide 15
(32.5 mg, 77%) pure.
Analytical data for 15.
(300 MHz, CDCl3):
[
a
]
D = ꢀ3 (c 1.8, CHCl3). 1H NMR
d
8.18–7.15 (m, 35H, Ar), 5.84 (d, 1H,
J = 2.6 Hz, H-4), 5.58 (br t, 1H, J = 8.8 Hz, H-2), 5.02 (br t, 1H,

1524
T. A. Jackson et al. / Bioorg. Med. Chem. 17 (2009) 1514–1526
J = 8.5 Hz, H-30), 4.90 (d, 1H, J = 11.5 Hz, CHPh), 4.85 (d, 1H,
J = 2.4 Hz, H-100), 4.78–4. 23 (m, 12H, H-1, H-10, H-6a, H-6b, NH,
7 ꢂ CHPh), 4.20–3.88 (m, 5H, H-3, H-5, H-60a, OCH2CCl3), 3.93
(dd, 1H, J = 3.4, 10.2 Hz, H-200), 3.79–3. 67 (m, 2H, H-300, H-500),
3.65–3.41 (m, 7H, H-40, H-50, H-60b, H-400, CH3O), 3.24 (q, 1H,
J = 8.9 Hz, H-20), 1.85 (s, 3H, CH3O), 0.97 (d, 3H, J = 6.4 Hz, H-600).
13C NMR (75 MHz, CDCl3): d 170.8, 166.1, 165.0, 153.8 (CO),
138.7, 138.4, 130.2, 130.0, 129.7, 128.6, 128.4, 128.3, 128.2,
128.2, 127.7, 127.7, 127.6, 127.4, 127.3 (Ar), 102.2 (C-1), 101.0
(C-10), 100.1 (C-100), 95.4 (CCl3), 79.0, 77.9, 77.2, 76.5, 75.2, 73.0,
71.8 (C-3, C-5, C-30, C-40, C-50, C-200, C-400, C-500), 74.9 (CH2), 73.9
(CH2), 73.8 (CH2), 73.4 (CH2), 72.8 (CH2), 71.5 (C-2), 70.6 (C-4),
69.5 (C-60), 67.7 (C-300), 62.9 (C-6), 5.68, 56.6 (CH3O, C-20), 21.0
(CH3CO), 16.3 (C-600). HRMS calcd for C73H74Cl3NO20 [M+K]+
1428.3507, found 1428.3438.
73.1 (CH2Ph), 72.9 (CH2Ph), 72.2 (C-300), 70.0 (C-30), 69.2 (C-6),
68.4 (C-50), 68.1 (C-200), 62.5 (C-60), 58.2 (C-200), 57.1 (CH3O), 23.6
(CH3CON), 16.7 (C-600). HRMS calcd for C49H61NO15 [M+NH4]+
926.3939, found 926.3922.
4.1.12. Methyl 3-O-[2-acetamido-2-deoxy-4-O-(
a-L-
fucopyranosyl)-b- -glucopyranosyl]-b- -galactopyranoside (2)
D
D
Trisaccharide 16 (44.2 mg, 0.04 mmol) was dissolved in metha-
nolic sodium methoxide (1 M, 2.5 mL), and the mixture was stirred
at room temperature for 1.5 h. Work up of the reaction was carried
out as described above and the crude trisaccharide 17 was filtered
through a bed of silica gel eluted with a mixture of CHCl3–MeOH
(19:1). After concentration, the semi crude trisaccharide 17 was
dissolved in MeOH (4 mL), 10% Pd/C (71.2 mg) was added and
the mixture was stirred overnight at room temperature under H2
atmosphere at 200 psi. The solids were filtered off on Celite,
washed with MeOH (8 mL) and the combined filtrate and washings
were concentrated. Lyophilization of the residue from water gave
trisaccharide 2 (17.6 mg, 92%) as a white amorphous powder:
4.1.10. Methyl 3-O-[3-O-acetyl-6-O-benzyl-4-O-(2,3,4-tri-O-
benzyl-a-
L
-fucopyranosyl)-2-deoxy-2-acetamido-b-
D-
glucopyranosyl]-2,4,6-tri-O-benzoyl-b-
D
-galactopyranoside (16)
Activated Zn dust (65 mg) was added to a solution of trisaccha-
ride 15 (20.6 mg, 0.02 mmol) in Ac2O (1 mL) and the mixture was
stirred under N2 at room temperature overnight. The solids were
filtered off on Celite, washed with AcOH (2 mL) and the combined
filtrate and washings were co-concentrated with toluene. Column
chromatography (7:1 to 1:1 hexanes–EtOAc) of the residue gave
[a]
D = ꢀ93 (c 0.2, H2O). 1H NMR (600 MHz, D2O, 300 K): d 4.95
(d, 1H, J = 3.8 Hz, H-100), 4.69 (d, 1H, J = 8.4 Hz, H-10), 4.34 (q, 1H,
J = 6.6 Hz, H-500), 4.29 (d, 1H, J = 8.0 Hz, H-1), 4.13 (d, 1H,
J = 3.2 Hz, H-4), 3.94 (dd, 1H, J = 2.0, 12.3 Hz, H-60a), 3.85 (dd, 1H,
J = 4.2, 12.3 Hz, H-60b), 3.82 (dd, 1H, J = 3.2, 10.5 Hz, H-300), 3.75–
3.66 (m, 7H, H-6a, H-6b, H-20, H-200, H-400), 3.66–3.58 (m, 3H, H-3,
H-5, H-30), 3.56–3.48 (m, 6H, H-2, H-4, H-50, CH3O), 2.03 (s, 1H,
NHCO), 1.15 (d, 1H, J = 6.7 Hz, H-600). 13C NMR (150 MHz, CDCl3):
d 175.3 (CO), 104.2 (C-1), 102.9 (C-10), 99.9 (C-100), 82.6 (C-3),
77.4 (C-40), 75.4 (C-50), 75.1 (C-5), 72.7 (C-30), 72.3 (C-400), 70.1
(C-300), 69.8 (C-2), 68.7 (C-200), 68.4 (C-4), 67.4 (C-500), 61.3 (C-6,
C-600), 57.6 (CH3O), 56.6 (C-20), 22.5 (CH3CO), 15.6 (C-600). HRMS
calcd for C49H61NO15 [M+Na]+ 926.3939, found: 926.3922. The
NMR data agreed with that previously reported.8
the amide 16 (14.8 mg, 79%) as an amorphous powder. [
a]
D = ꢀ6
(c 0.6, CHCl3). 1H NMR (300 MHz, CDCl3): d 8.19–7.12 (m, 35H,
Ar), 5.84 (d, 1H, J = 3.1 Hz, H-4), 5.57 (dd, 1H, J = 8.1, 9.8 Hz, H-2),
5.04–4.88 (m, 3H, H-30, NH, CHPh), 4.85 (d, 1H, J = 3.5 Hz, H-100),
4.80–4.25 (m, 11H, H-10, H-1, H-6a, H-6b, 7 ꢂ CHPh), 4.12 (dd,
1H, J = 9.9, 3.4 Hz, H-3), 4.05–3.93 (m, 2H, H-5, H-60a), 3.92 (dd,
1H, J = 3.4, 10.2 Hz, H-200), 3.79–3.68 (m, 2H, H-300, H-500), 3.63–3.
37 (m, 8H, H-20, H-40, H-50, H-60b, H-400, CH3O), 1.86 (s, 3H,
CH3COO), 1.28 (s, 3H, CH3CON), 0.97 (d, 3H, J = 6.4 Hz, H-600). 13C
NMR (75 MHz, CDCl3): d 173.9, 171.1, 170.0, 166.1, 163.7 (CO)
138.4, 135.2, 133.5, 133.1, 131.4, 130.2, 129.9, 129.7, 129.1,
128.8, 128.7, 128.4, 128.3, 128.2, 128.1, 127.9, 127.7, 127.7,
127.5, 127.4, 127.3, 127.1 (Ar), 102.1 (C-1), 101.5 (C-10), 99.9 (C-
100), 74.9 (CH2Ph), 73.8 (CH2Ph), 73.3 (CH2Ph), 72.8 (CH2Ph), 70.0,
71.0, 71.5, 73.8, 73.9, 75.1 (C-3, C-5, C-20, C-30, C-40, C-50, C-200, C-
300, C-400, C-500), 72.8 (CH2Ph), 68.5 (C-60), 67.2 (C-4, C-2), 62.2 (C-
6), 56.8 (C-20, CH3O), 22.7 (NHCO), 21.1 (CH3CO), 14.1 (C-600). HRMS
calcd for C72H75NO19 [M+NH4]+ 1275.5277, found 1275.5240.
4.2. Nuclear magnetic resonance spectroscopy
4.2.1. Sample preparation
Trisaccharide 2 (12.4 mg) was lyophilized three times from D2O
(95%) and then dissolved in 0.7 mL of D2O (99.96%) to give a final
0.033 M concentration.
4.2.2. NMR experiments
NMR experiments were recorded in 5 mm NMR tubes at
400 MHz (HSQC) and 600 MHz (1H, 13C, COSY, HMBC, NOESY and
ROESY). All spectra were recorded at 300 K with the exception of
NOESY experiments, which were also recorded at 310 K.
Phase-sensitive two-dimensional NOESY spectra were acquired
with a spectral width of 5000 Hz, 2048 complex data points and 64
scans each for 128 increments. A 1.5 s relaxation delay was imple-
mented between scans. NOESY spectra were recorded at seven
mixing times: 100 ms increments from 400 ms to 1 s. Before Fou-
4.1.11. Methyl 3-O-[2-acetamido-6-O-benzyl-4-O-(2,3,4-tri-O-
benzyl-a-L-fucopyranosyl)-2-deoxy-b-D-glucopyranosyl]-b-D-
galactopyranoside (17)
Trisaccharide 16 (14.7 mg, 0.01 mmol) was dissolved in metha-
nolic sodium methoxide (0.25 M, 1.5 mL), and the mixture was
stirred under N2 at room temperature for 3 h. The mixture was
deionized with Dowex 50 (H+) resin, filtered, and solids were
washed with MeOH (3 mL). The filtrate and washings were com-
bined and concentrated. Chromatography (19:1 CHCl3–MeOH) of
rier transformation a p/2 shifted sine squared window was applied
in both dimensions and the FID was zero-filled to 2048 points in
the F2 direction and 512 points in the F1 direction. Further han-
dling of the NOESY spectra and integrations of the cross-peaks to
obtain buildup curves is described in Supplementary data.
the residue gave trisaccharide 17 (8.3 mg, 79%) pure. [
a]
D = ꢀ25
(c 0.5, CHCl3). 1H NMR (400 MHz, CDCl3): d 7.43–7. 17 (m, 20H,
Ar), 6.72 (br s, 1H, NH), 5.22 (d, 1H, J = 8.4 Hz, H-10), 4.97 (d, 1H,
J = 11.4 Hz, CHPh), 4.92 (d, 1H, J = 3.7 Hz, H-100), 4.85–4.58 (m, 5H,
One-dimensional ROESY spectra were acquired with selective
excitations of anomeric protons effected using a 60 ms Gaussian-
shaped pulse and a spin-lock field of 2.0 kHz. A 5 s relaxation delay
was implemented between scans. ROESY spectra were recorded at
eleven mixing times from 20 ms to 400 ms. Before Fourier transfor-
mation, the FIDs were zero-filled once and multiplied with a 1 Hz
line broadening factor. Spectra were phase and baseline corrected
and integrated. The integrals measured for the irradiated H-10 and
H-100 signals were plotted against mixing time and the obtained
curves were fitted to a double exponential decaying function:
5 ꢂ CHPh), 4.27 (d, 1, J = 12.0 Hz, CHHPh), 4.20 (d,
1 H,
J = 11.9 Hz, CHHPh), 4.16 (d, 1H, J = 7.8 Hz, H-1), 4.12–3.96 (m,
3H, H-30, H-500, H-200), 3.93–3.52 (m, 11H, H-2, H-3, H-5, H-6a, H-
6b, H-50, H-60a, H06b, H-300, H-400, CH3O,), 3.31 (t, 1H, J = 8.7 Hz, H-
40), 3.17 (q, 1H, J = 8.3 Hz, H-20), 1.98 (s, 3H, CH3CON), 1.14 (d,
3H, J = 6.4 Hz, H-600). 13C NMR (75 MHz, CDCl3): d 172.1 (CO),
138.5 138.4 138.1 128.5 128.4 128.3 127.9 127.8 127.7 127.5
(Ar), 104.0 (C-1), 100.1 (C-10), 99.9 (C-100), 83.0 (C-3), 81.0 (C-40),
79.0 (C-5), 75.9 (CH2Ph), 75.0 (C-400), 74.2 (CH2Ph), 74.0 (C-4),

T. A. Jackson et al. / Bioorg. Med. Chem. 17 (2009) 1514–1526
1525
fðsmÞ ¼ ꢀA½expðBsmÞ ꢀ expðCsmÞꢃ
attempt to generate
a
new conformer. In these stochastic
searches, new conformations are generated by the rotation of
all bonds (including ring bonds) to random dihedral angles. In
our searches a bias of 30° was used. This means that dihedral
angles were selected with a sum-of-Gaussians distribution with
peaks at multiples of 30°. In addition to bond rotations, atom
positions were randomly perturbed by 0.4 Å. These random per-
turbations were repeated 150,000 times per conformational
search and each generated conformer was subjected to 500 steps
of minimization with full degrees of freedom with implicit solva-
tion by the Generalized Born/Surface Area (GB/SA) continuum
solvation model23 using a dielectric constant of 78. Two separate
searches were performed and the databases were combined for
analysis. Initially each new conformation was checked against
the previously found conformations using a root-mean-square
(RMS) tolerance of 0.1 Å on all atoms, including the hydrogen
atoms to account for possible hydrogen bonding. The second
where sm is the mixing time and A, B and C are adjustable parame-
ters. The values of these integrals were extrapolated to 0 ms mixing
time, and the integrals from cross-relaxation peaks were normal-
ized through division by these extrapolated values. The normalized
cross-relaxation integrals were plotted against the mixing times
and the buildup curves were fitted to a double exponential equation
of the form
fðsmÞ ¼ A½expðBsmÞ ꢀ expðCsmÞꢃ
and the initial slopes at 0 ms mixing times were determined from
the calculated first derivatives,f0(0) = A(B ꢀ C).28
Interproton distances were calculated based on the isolated
spin pair approximation (ISPA):
1=6
riJ ¼ rref ðSref =SijÞ
where S is the initial slope at sm = 0, and r is the proton-proton dis-
search was carried out using
a more stringent RMS value
tance. The H-10/H-30 intra-residue cross-peak was used as the refer-
ence for the distance determinations. The reference distance used
was 2.62 Å.
(0.01 Å) applied to the heavy atoms only. In both searches the
calculations terminated at the end of the 150,000 iterations as
the number of failures criteria to find new conformations
(10,000 iterations in a row) was never met. Structures that did
4.3. Computational methods
not have the 4C1 conformation for
D-glucosamine or
D-galactose
or those that did not maintain a 1C4 conformation for
L
-fucose
4.3.1. Sybyl systematic search
were eliminated. The results of the two searches were combined
for analysis. And conformations with energies over 10 kcal/mol
above the global minima were rejected resulting in 1450
structures.
A rigid search was carried out in Sybyl7.117 using the Tripos
force field along with PIM parameters18 developed for carbohy-
drates. The starting molecule was built in Sybyl. To eliminate
excessive biasing energies from steric interactions associated with
rigid positions of ring substituents normally expected to be highly
flexible, hydroxyl hydrogen atoms were removed and the
hydroxymethyl groups were converted to methyl groups. During
the search, the individual monosaccharide rings were held fixed,
4.3.3. Molecular dynamics simulations
The molecular dynamics simulations were carried out using
the SANDER module of AMBER924 with the inclusion of Gly-
cam04 parameters25 for carbohydrates. The initial structure
was built using the XLEAP module of AMBER. The molecule
was solvated with 10 shells of TIPS3P water molecules30 giving
a total of 1070 solvent molecules in a 30 Å cubic box. Dynamics
trajectories of 8 ns were calculated. Each simulation was carried
out in five stages using the SANDER module of AMBER. First the
water molecules were minimized at constant volume (NVT)
whereas the trisaccharide was held fixed. A 1000 cycle limit
was employed with 50 cycles of steepest descent followed by
conjugate gradient minimization. A 1 ꢂ 10ꢀ3 kcal molꢀ1 Å gradi-
ent convergence criteria was used. A 10 Å cutoff was set, and
non-bonded updates were made every 10 steps. The dielectric
constant was set to 1. Then, the entire system was minimized
without any restraints. A 2500 cycle limit was used with 500 cy-
cles of steepest descent followed by conjugate gradient minimi-
zation. Afterward, the system was heated for 20 ps from 0 to
300 K or 310 K with the saccharide weakly restrained. Time steps
of 2 fs were used and translational momentum was removed
every 1000 steps. Bonds involving hydrogens were constrained
using the SHAKE algorithm with 0.00001 Å tolerance. Coordi-
nates were output every 250 steps. The system was then equil-
ibrated at a constant pressure (NPT) of 1 atm for 100 ps with no
restraints. Finally, production runs of 8 ns were carried out at
300 or 310 K. The PTRAJ module of AMBER was used to analyze
the results.
and the glycosidic
U and W torsions were varied in 10° incre-
ments throughout the entire possible conformational space, and
energies were evaluated after each variation. The van der Waals
radius scale factors were reduced to very low values (0.3). Elec-
trostatic contributions were not taken into account. Conforma-
tions within 40 kcal molꢀ1 of the global minimum were saved
and the minima within 3 kcal molꢀ1 of the global minimum were
identified. The glycosidic torsions obtained for these minima were
then applied to the starting molecule containing all hydrogen
atoms and hydroxymethyl groups, and for each minimum, struc-
tures containing the nine different combinations of staggard ori-
entations of the hydroxymethyl groups were generated. Each
structure was optimized using conjugate gradient minimization
with
a a
0.05 kcal molꢀ1 gradient termination criteria and
10,00,000 iteration limit. A cutoff of 8 Å was used for nonbonding
interactions. For the optimization, electrostatic contributions
were taken into account, and a dielectric constant of 78.5 was
employed. The lowest energy structure found for each minimum
was used as the representative molecule for distance determina-
tions (Table 1).
4.3.2. MOE stochastic search
The trisaccharide was subjected to all atom and heavy atom
stochastic conformational searches using the AMBER94 force
field21 in the Molecular Operating Environment (MOE2004)22 pro-
gram suite. In this implementation of AMBER94, any parameters
missing for a class of compound are approximated by MOE
based on parameter values for similar molecular fragments.
The starting structure was built using the carbohydrate builder
in MOE2004 and submitted to the stochastic search. This search
is similar to the random incremental pulse search (RIPS) meth-
od29 in which the coordinates of each atom are randomly per-
turbed, after which the entire molecule is minimized to
Acknowledgments
The authors thank the National Sciences and Engineering Re-
search Council of Canada, the University of Guelph, the Canada
Foundation for Innovation and the Ontario Innovation Trust for
financial support. This work was also made possible by the facili-
ties of the Shared Hierarchical Academic Research Computing Net-
work (SHARCNET: www.sharcnet.ca).

1526
T. A. Jackson et al. / Bioorg. Med. Chem. 17 (2009) 1514–1526
14. Imberty, A.; Pérez, S. Chem. Rev. 2000, 100, 4567.
Supplementary data
15. Bundle, D. R. Recognition of Carbohydrate Antigens by Antibody Binding sites.
In Bioorganic Chemistry: Carbohydrates;; Hecht, S. M., Ed.; Oxford University
Press: New York, 1999; pp 370–440.
16. IUPAC-I.U.B. Commission of Biochemical Nomenclature. Eur. J. Biochem. 1997,
243, 9–9.
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.bmc.2009.01.020.
17. SYBYL 7.1, Tripos Associates, 1699 S. Hanley Road, Suite 303, St Louis, MO
63144, USA.
References and notes
18. Imberty, A.; Hardman, K. D.; Carver, J. P.; Pérez, S. Glycobiology 1991, 1, 631.
19. Imberty, A.; Mikros, E.; Koca, J.; Mollicone, R.; Oriol, R.; Pérez, S. Glycoconjugate
J. 1995, 12, 331; Imberty, A.; Mazeau, K.; Pérez, S.; Rivet, A.; Sabin, C.
Disaccharide Database. http://www.cermav.cnrs.fr/cgi-bin/di/di.cgi (accessed
August 19, 2008).
1. Pettijohn, D. E.; Pfenninger, O.; Brown, J.; Duke, R.; Olsson, L. Proc. Natl. Acad.
Sci. U.S.A. 1988, 85, 802.
2. Stranahan, P. L.; Howard, R. B.; Pfenninger, O.; Cowen, M. E.; Johnston, M. R.;
Pettijohn, D. E. Cancer Res. 1992, 52, 2923.
20. Lemieux, R. U.; Koto, S.; Voisin, D.. In ACS Symposium Series; American Chemical
Society: Washington, DC, 1979; Vol. 87, pp 17–29.
3. Stranahan, P. L.; LaRoe, J.; McCombs, R.; Goldsmith, A.; Rahim, I.; Overland, M.;
Pettijohn, D. E. Glycoconjugate J. 1996, 13, 741.
21. Cornell, W. D.; Cieplak, P.; Bayly, C. I.; Gould, I. R.; Merz, K. M.; Ferguson, D. M.;
Spellmeyer, D. C.; Fox, T.; Caldwell, J. W.; Kollman, P. A. J. Am. Chem. Soc. 1995,
117, 5179.
22. Chemical Computing Group, Montreal, QC, Canada.
23. Still, W. C.; Tempczyk, A.; Hawley, R. C.; Hendrickson, T. J. Am. Chem. Soc.
1990, 112, 6127; Qiu, D.; Shenkin, P. S.; Hollinger, F. P.; Still, W. C. J. Phys.
Chem. A 1997, 101, 3005; Schaefer, M.; Karplus, M. A. J. Phys. Chem. 1996,
100, 1578.
4. Lemieux, R. U.; Baker, D. R.; Weinstein, W. M.; Switzer, C. M. Biochemistry 1981,
20, 199.
5. Pettijohn, D. E.; Stranahan, P. L.; Due, C.; Rønne, E.; Sørenson, H. R.; Olsson, L.
Cancer Res. 1987, 47, 1161.
6. Martensson, S.; Due, C.; Pahlsson, P.; Nilsson, B.; Eriksson, H.; Zopf, D.; Olsson,
L.; Lundblad, A. Cancer Res. 1988, 48, 2125.
7. Battifora, H.; Sørenson, H. R.; Mehta, P.; Ahn, C.; Niland, J.; Hage, E.; Pettijohn,
D. E.; Olsson, L. Cancer 1992, 70, 1867.
24. Case, D. A.; Darden, T. A.; Cheatham et al., T. E. I. AMBER 9, University of
California, San Francisco, 2006.
25. Woods, R. J. Phys. Chem. 1995, 99, 3832.
26. Mukhopadhyay, C.; Bush, C. A. Biopolymers 1991, 31, 1731.
27. Armarego, W. L. F.; Perrin, D. D. Purification of Laboratory Chemicals, 4th ed.;
Elsevier Science Ltd, 1996.
28. Maaheimo, H.; Kosma, P.; Brade, L.; Brade, H.; Peters, T. Biochemistry 2000, 39,
12778.
8. Jain, R. K.; Kohata, K.; Abbas, S. A.; Matta, K. L. Carbohydr. Res. 1988, 172, 27.
9. Ogawa, T.; Nukada, T.; Matsui, M. Carbohydr. Res. 1982, 101, 263.
10. Dullenkopf, W.; Castro-Palomino, J. C.; Manzoni, L.; Schmidt, R. R. Carbohydr.
Res. 1996, 296, 135.
11. Sato, S.; Ito, Y.; Nukada, T.; Nakahara, Y.; Ogawa, T. Carbohydr. Res. 1987, 167,
197; Windmüller, R.; Schmidt, R. R. Tetrahedron Lett. 1994, 43, 7927; Cao, S.;
Gan, Z.; Roy, R. Carbohydr. Res. 1999, 318, 75.
12. Kovac, P.; Glaudemans, C. P. J.; Taylor, R. B. Carbohydr. Res. 1985, 142, 158.
13. Lonn, H. Carbohydr. Res. 1985, 139, 105; Murphy, P. V.; Hubbard, R. E.;
Manallack, D. T.; Wills, R. E.; Montana, J. G.; Taylor, R. J. K. Bioorg. Med. Chem.
1998, 6, 2421.
29. Chang, G.; Guida, W. C.; Still, W. C. J. Am. Chem. Soc. 1989, 111, 4379.
30. Jorgensen, W. L.; Chandrasekhar, J.; Madura, J.; Klein, M. L. J. Chem. Phys. 1983,
79, 926.