Reconciling solvent effects on rotamer populations in carbohydrates - A joint MD and NMR analysis

Article abstract of DOI:10.1139/V06-036

The rotational preferences of the hydroxymethyl group in pyranosides is known to depend on the local environment, whether in solid, solution, or gas phase. By combining molecular dynamics (MD) simulations with NMR spectroscopy the rotational preferences for the ω angle in methyl 2,3-di-O-methyl- α-D-glucopyranoside (3) and methyl 2,3-di-O-methyl-α-D- galactopyranoside (6) in a variety of solvents, with polarities ranging from 80 to 2.3 D have been determined. The effects of solvent polarity on intramolecular hydrogen bonding have been identified and quantified. In water, the internal hydrogen bonding networks are disrupted by competition with hydrogen bonds to the solvent. When the internal hydrogen bonds are differentially disrupted, the rotamer populations associated with the ω angle may be altered. In the case of 3 in water, the preferential disruption of the interaction between HO6 and O4 destabilizes the tg rotamer, leading to the observed preference for gauche rotamers. Without the hydrogen bond enhancement offered by a low polarity environment, both 3 and 6 display rotamer populations that are consistent with expectations based on the minimization of repulsive intramolecular oxygen-oxygen interactions. In a low polarity environment, HO6 prefers to interact with O4, however, in water these interactions are markedly weakened, indicating that HO6 acts as a hydrogen bond donor to water.

Full text of DOI:10.1139/V06-036

569
Reconciling solvent effects on rotamer
populations in carbohydrates — A joint MD and
NMR analysis1
Jorge Gonzalez-Outeiriño, Karl N. Kirschner, Smita Thobhani, and
Robert J. Woods
Abstract: The rotational preferences of the hydroxymethyl group in pyranosides is known to depend on the local envi-
ronment, whether in solid, solution, or gas phase. By combining molecular dynamics (MD) simulations with NMR
spectroscopy the rotational preferences for the ω angle in methyl 2,3-di-O-methyl-α-D-glucopyranoside (3) and methyl
2,3-di-O-methyl-α-D-galactopyranoside (6) in a variety of solvents, with polarities ranging from 80 to 2.3 D have been
determined. The effects of solvent polarity on intramolecular hydrogen bonding have been identified and quantified. In
water, the internal hydrogen bonding networks are disrupted by competition with hydrogen bonds to the solvent. When
the internal hydrogen bonds are differentially disrupted, the rotamer populations associated with the ω angle may be al-
tered. In the case of 3 in water, the preferential disruption of the interaction between HO6 and O4 destabilizes the tg
rotamer, leading to the observed preference for gauche rotamers. Without the hydrogen bond enhancement offered by a
low polarity environment, both 3 and 6 display rotamer populations that are consistent with expectations based on the
minimization of repulsive intramolecular oxygen–oxygen interactions. In a low polarity environment, HO6 prefers to in-
teract with O4, however, in water these interactions are markedly weakened, indicating that HO6 acts as a hydrogen
bond donor to water.
Key words: carbohydrate, rotamer, molecular dynamics simulation, MD, NMR.
Résumé : Il est bien connu que les préférences rotationnelles des groupes hydroxyméthyles des pyranosides dépend de
l’environnement local que ce soit à l’état solide, en solution ou en phase gazeuse. En combinant des simulations de dy-
namique moléculaire (DM) avec la spectroscopie RMN, on a pu déterminer les préférences rotationnelles pour l’angle
ω du 2,3-di-O-méthyl-α-D-glucopyranoside de méthyle (3) 2,3-di-O-méthyl-α-D-galactopyranoside de méthyle (6), dans
une variété de solvants dont les polarités s’étalent de 80 à 2,3 D. On a identifié et quantifié les effets de polarité du
solvant sur la liaison hydrogène intramoléculaire. Dans l’eau, la liaison hydrogène interne est perturbée par la compéti-
tion avec les liaisons hydrogènes vers le solvant. Lorsque les liaisons hydrogènes internes sont perturbées de façons
différentes, les populations de rotamères associées à l’angle ω peuvent être modifiées. Dans le cas du composé 3, l’eau
provoque une perturbation préférentielle de l’interaction entre le OH en position 6 et l’oxygène en 4 qui déstabilise le
rotamère tg, ce qui conduit à une observation préférentielle des rotamères gauches. Lorsqu’ils ne présentent pas
d’augmentation de la liaison hydrogène en raison d’un environnement de faible polarité, les populations de rotamères
des composés 3 et 6 sont en accord avec ce qu’on pourrait attendre sur la base d’une minimisation des interactions ré-
pulsives oxygène–oxygène intramoléculaires. Dans une environnement de faible polarité, le OH en position 6 préfère
l’interaction avec l’oxygène en 4; toutefois, dans l’eau, ces interactions sont de beaucoup affaiblies ce qui indique que
le OH en position 6 agit comme donneur de liaison hydrogène pour l’eau.
Mots clés : hydrate de carbone, rotamère, simulation de dynamique moléculaire, DM, RMN.
[Traduit par la Rédaction] Gonzalez-Outeiriño et al. 579
Introduction
oligo- and poly-saccharides are increasingly implicated in
both the healthy growth and functioning of organisms, as
well as in a diverse range of disease states (1–3). Further,
the covalent attachment of carbohydrates to proteins may af-
fect both protein stability and structure (4–7), as well as,
The importance of understanding the relationships be-
tween conformation and biological and chemical function in
carbohydrates has come to the forefront of glycobiology as
Received 18 August 2005. Published on the NRC Research Press Web site at http://canjchem.nrc.ca on 5 May 2006.
This manuscript is dedicated to Professor Walter A. Szarek.
J. Gonzalez-Outeiriño, K.N. Kirschner, S. Thobhani, and R.J. Woods.2 Complex Carbohydrate Research Center, University of
Georgia, 315 Riverbend Road, Athens, GA 30602, USA.
1This article is part of a Special Issue dedicated to Professor Walter A. Szarek.
2Corresponding author (e-mail: rwoods@ccrc.uga.edu).
Can. J. Chem. 84: 569–579 (2006)
doi:10.1139/V06-036
© 2006 NRC Canada

570
Can. J. Chem. Vol. 84, 2006
Fig. 1. (A) Schematic diagram indicating glycosidic torsion angles in carbohydrates. (B) Nomenclature for the ω torsion angle (O6-C6-
C5-O5). Qualitatively, the three staggered rotamers are commonly defined with respect to the O6-C6-C5-O5 and O6-C6-C5-C4 torsion
angles as gauche-gauche (gg), gauche-trans (gt), and trans-gauche (tg).
OH
(A)
(B)
gt
tg
gg
H
H
O
H
HS
O6
R
HO
HO
H
O5
O6
O5
O5
HS
C4
HS
C4
O6
C4
H
OH
H
O
H
HR
ω
H
HR
H
H5
H5
H5
H
O
H
HO
HO
O
OH
O
HO
ω ~ 60o
ω ~ 180o
ω ~ 300o
H
OH
OH
Ψ
ΦH
H
H
OH
H
conversely, function (8–10). The ability to correlate biologi-
cal function with glycan structure can provide not only
mechanistic insight (11), but is also of benefit in optimizing
carbohydrate-based therapeutic agents (12).
choice of classical mechanical force field. Force field
development remains an active research area; however, MD
simulations are generally able to accurately identify the
conformational states present in solution and can often pre-
dict their populations with reasonably accuracy (20–24, 19).
In solution, 3-D structures of oligosaccharides have been
shown to exist as an equilibrium among several conforma-
tions, owing to the flexibility introduced in the structure by
the glycosidic linkages, Φ, Ψ, and ω (see Fig. 1). The con-
formations of the first two torsion angles (Φ, Ψ) are domi-
nated mainly by stereolectronic (anomeric effect) and steric
effects (25, 26). As would be expected for an ether-type
linkage, the ω torsion angle may adopt any of the three stag-
gered rotamers, with populations depending heavily on the
configuration at C4 and on the polarity of the local environ-
ment (27, 23).
In the solid state, an early report based on a statistical
analysis of X-ray structures (28) revealed two remarkably
different conformational behaviors for the ω torsion angle in
carbohydrates. The glucopyranosides and mannopyranosides
showed clear preferences for only the gg and gt conforma-
tions, with ratios of approximately 40:0:60 (gt:tg:gg), while
galactopyranosides showed a significant population of the tg
rotamer, with a ratio of approximately 58:34:8 (gt:tg:gg).
Similar trends have been inferred from NMR data in aque-
ous solution for both the glucopyranosyl (50:0:53 (29),
38:4:58 (30), 44.7:3.5:51.8 (31), and 52:7:41 (32)) and
galactopyranosyl series (56:32:12 (29), 61:21:18 (30),
71.6:13.9:14.5 (31), 65:21:14 (33), and 67:30:3(32)). A de-
tailed review of the conformational properties of the ω angle
has been provided by Bock and Duus (29).
In contrast to the condensed-phase experimental data, gas-
phase quantum mechanical calculations fail to predict the
correct rotamer populations unless the formation of internal
hydrogen bonds among hydroxyl groups at C4 and C6 or be-
tween the hydroxyl group at C6 and the ring oxygen atom is
prevented (34, 23, 31). A Boltzman analysis of the gas-phase
quantum mechanical energies, computed at the B3LYP/6-
31G(2d,2p) level has been performed for methyl α-D-
glucopyranosides and galactopyranosides. A theoretical pop-
ulation ratio of 43:29:28 was obtained for the glucopy-
ranoside when internal hydrogen bonding was allowed,
whereas qualitative agreement with experiment (62:0:37)
was only obtained when intramolecular hydrogen bonds
were prevented from forming. Similarly for the galacto-
pyranoside, ratios of 38:5:57 (with internal hydrogen bonds)
X-ray crystallography, NMR spectroscopy, and theoretical
calculations are the three techniques most commonly used in
the conformational analysis of carbohydrates. X-ray crystal-
lography has been very successful in characterizing protein–
carbohydrate complexes, provided that the carbohydrate is
limited in size (13, 14). However, single crystal X-ray dif-
fraction structures of oligo- and poly-saccharides are ex-
tremely rare; a direct result of difficulties in crystallizing the
generally flexible oligosaccharides (15). For solution
conformational analysis, NMR spectroscopy is the preferred
technique. However, the deconvolution of the observed
NMR properties, such as chemical shifts, scalar J couplings,
nuclear Overhauser effects (NOEs), and residual dipolar
couplings, into the 3-D structures that are present in solution
presents particular difficulties for these flexible molecules
(16, 17). The challenge in deriving a mixture of physically
real conformations from a spectroscopic signal that arises
from the average of the conformational ensemble is gener-
ally met by introducing some aspect of theoretical modeling
into the NMR data analysis. Unlike the situation with pro-
tein structure determinations by NMR, the use of NMR
observables as restraints during the modeling of flexible car-
bohydrates can lead to the generation of nonphysical virtual
conformations (18). To avoid this problem, theoretical calcu-
lations are generally employed in one of two ways, either to
provide discrete models for the presumed conformations
present in solution or to independently predict the confor-
mational states. In the former application, a limited number
of theoretical structures are generated, usually by some form
of gas-phase energy minimization (quantum or classical me-
chanical energies) of initial conformations. Based on the
computed NMR properties for each individual structure it
may then be possible to establish the populations of the indi-
vidual structures that give the best agreement with experi-
ment. Perhaps the most significant limitation to the accuracy
of this approach is the selection of the individual structures,
which is often based on scientific intuition. In contrast,
molecular dynamics (MD) simulations or Monte Carlo sam-
pling may be employed to predict the ensemble of confor-
mations that the carbohydrate populates at room temperature
in the appropriate solvent, independent of investigator bias
(19). The accuracy of this latter approach is limited by the
© 2006 NRC Canada

Gonzalez-Outeiriño et al.
571
and 41:59:0 (without) were obtained (23). In neither deriva-
tive were the gas-phase calculations able to give quantitative
agreement with experiment, although they suggested a role
for internal hydrogen bonding. It should be noted that in all
structures examined quantum mechanically, prevention of
internal hydrogen bonding led to significantly higher ener-
gies relative to those of conformations in which intramole-
cular hydrogen bonds were present (23). Nonetheless, the
observation that qualitative agreement with experiment was
obtained only in the absence of intramolecular hydrogen
bonds is consistent with an interpretation of rotamer popula-
tions as arising from the minimization of unfavorable 1,3-
diaxial electrostatic interactions among O4, O5, and O6 (29,
23).
(J) were measured in Hertz (Hz). Matrix assisted laser
desorption ionization – time-of-flight (MALDI-TOF) mass
spectra were recorded using a Hewlett Packard G2025A sys-
tem with 2,5-dihydroxybenzoic acid (gentisic acid)
(10 mg/mL in 50% acetonitrile) as the matrix. Melting points
were recorded on a Kofler hot stage apparatus and are un-
corrected.
Chemicals and reagents were purchased from Sigma-
Aldrich, Fluka, or VWR and used without further purifica-
tion unless otherwise stated. The reaction solvent
(acetonitrile) was distilled from calcium hydride and stored
over 4 Å molecular sieves prior to use. Ethanol was used
without any further purification.
To obtain quantitative agreement between theoretical cal-
culations and experiment, it was found necessary to include
water explicitly in the MD simulations of the glucopyrano-
sides and galactopyranosides (23). The mechanism whereby
solvent influences ω population in methyl α-D-glucopyrano-
side has previously been examined experimentally in an
NMR analysis of the 2,3,4,6-tetra- and 2,3,4-tri-O-methyl-
ated derivatives in several solvents (27). By examining the
effect of solvent polarity on rotamer population of these
methylated derivatives, Rockwell and Grindley (27) made
two important observations, namely that permethylation re-
moved any dependence of rotamer population on solvent po-
larity (maintaining approximately a ratio of 61:38:0), and, in
contrast, that when O6 was unsubstituted there was a strong
dependence on polarity. They inferred that polar solvents led
to a decrease of intramolecular hydrogen bonding, but also
that hydrogen bond donation from the hydroxymethyl group
to water was not important in determining rotational prefer-
ences.
Here we extend the work of Rockwell and Grindley to de-
rivatives of both glucopyranosides and galactopyranosides in
which we elected to examine derivatives that, like the natural
carbohydrates, have free hydroxyl groups at either O4 or O6.
Thus, we have synthesized methyl 2,3-di-O-methyl-α-D-glu-
copyranoside (3) and methyl 2,3-di-O-methyl-α-D-galacto-
pyranoside (6), and determined the rotational properties of
each of their ω angles by NMR in a variety of solvents. By
employing these simple models, an assessment of the impor-
tance of hydrogen bond donation from HO4 to O6 may be
made, as well as of the effect of configuration at C4 on
rotamer population as a function of solvent polarity. In addi-
tion, a detailed examination of the hydrogen bond properties
for these glycosides and related disaccharides was under-
taken by performing the longest explicitly solvated MD sim-
ulations reported to date in both water and dichloromethane.
Data from the theoretical simulations is used to provide
atomic level insight into the validity of the inferences made
on the basis of the experimental data.
Synthesis
Methyl 4,6-O-benzylidene-␣-D-glucopyranoside (1)
To a solution of methyl α-D-galactopyranoside (2.11 g,
10.87 mmol) in dry acetonitrile (60 mL) was added
benzaldehyde dimethyl acetal (2.80 g, 18.40 mmol) and the
solution acidified with camphorsulfonic acid (approx.
0.05 g, 0.22 mmol) to pH 3. After stirring at room tempera-
ture under an atmosphere of argon for 15 h, the mixture was
neutralized with triethylamine and concentrated in vacuo to
dryness using toluene as a cosolvent. The resulting residue
was recrystallized from ethyl acetate – hexane to afford
compound 1 as a white crystalline solid (2.22 g, 76%); mp
1
165–167 °C (lit. value (35) mp 166 to 167 °C). H NMR
(300 MHz, CDCl3) δ: 7.53–7.45 (m, 2H, Ph), 7.41–7.33 (m,
3H, Ph), 5.53 (s, 1H, PhCH), 4.80 (d, 1H, H1), 4.30 (dd, 1H,
H6a), 3.93 (t, 1H, H3), 3.85–3.77 (m, 1H, H5), 3.73 (t, 1H,
H6b), 3.63 (dd, 1H, H2), 3.49 (t, 1H, H4), 3.46 (s, 3H,
OCH3). 13C NMR (75 MHz, CDCl3) δ: not observed (Cq,
Ph), 129.2, 126.7 (5 × Ph), 101.7 (PhCH), 99.2 (C1), 81.7
(C4), 72.9 (C2), 71.7 (C3), 69.2 (C6), 62.9 (C5), 55.4
(OCH3). MALDI-TOF MS m/z: 305.2 [M + Na]+.
Methyl 4,6-O-benzylidene-2,3-di-O-methyl-␣-D-
glucopyranoside (2)
Methyl 4,6-O-benzylidene-α-D-glucopyranoside (500 mg,
1.77 mmol) was dissolved in dry DMF (20 mL) under a ni-
trogen atmosphere. To this was added 60% NaH suspended
in oil (260 mg, 6.50 mmol) followed by the dropwise addi-
tion of methyl iodide (340 µL, 5.46 mmol) and the reaction
was left to stir for 16 h under a nitrogen atmosphere at room
temperature. After this time, the reaction was quenched with
water (40 mL) and the aqueous layer was washed with ethyl
acetate (3 × 50 mL). The combined organic layers were then
washed with satd. NaHCO3 (3 × 50 mL) and water (3 ×
50 mL). The combined organic layers were dried (MgSO4),
filtered, and the solvent removed in vacuo and the residue
recrystallized from ethanol to furnish 2 as a white solid
(270 mg, 49%); mp 121 to 122 °C (lit. value (36) mp 122 to
Experimental
1
123 °C). H NMR (300 MHz, CDCl3) δ: 7.53–7.45 (m, 2H,
Ph), 7.40–7.31 (m, 3H, Ph), 5.54 (s, 1H, PhCH), 4.85 (d,
1H, H1), 4.28 (dd, 1H, H6a), 3.88–3.78 (m, 1H, H5), 3.74
(dd, 1H, H6b), 3.69 (t, 1H, H3), 3.64 (s, 3H, OCH3), 3.56 (s,
3H, OCH3), 3.53 (t, 1H, H4), 3.45 (s, 3H, OCH3), 3.30 (t,
1H, H2). 13C NMR (75 MHz, CDCl3) δ: not observed (Cq,
Ph), 129.4, 126.9 (5 × Ph), 101.9 (PhCH), 98.1 (C1), 81.9
General
1H and 13C NMR and 2-D correlation spectra were re-
corded on Varian 300 and 500 MHz spectrometers, each
equipped with Sun off-line editing workstations. Chemical
shifts were measured in parts per million (δ) relative to the
appropriate solvent peaks and the scalar coupling constants
© 2006 NRC Canada

572
Can. J. Chem. Vol. 84, 2006
(C2), 81.9 (C4), 80.6 (C3), 69.4 (C6), 61.9 (C5), 60.6, 59.4,
Methyl 2,3-di-O-methyl-␣-D-galactopyranoside (6)
55.6 (3 × OCH3). MALDI-TOF MS m/z: 332.8 [M + Na]+.
The synthesis of methyl 2,3-di-O-methyl-α-D-galacto-
pyranoside was carried out following the procedure de-
scribed for the preparation of the methyl 2,3-di-O-methyl-α-
D-glucopyranoside. Thus, methyl 4,6-O-benzylidene-2,3-di-
O-methyl-α-D-galacto-pyranoside (230 mg, 0.74 mmol) was
treated with 80% acetic acid (15 mL) to give the desired
product 6 as a white solid after recrystallization (49 mg,
30%). 1H NMR (500 MHz, CDCl3) δ: 4.90 (d, 1H, H1), 4.13
(dd, 1H, H4), 3.94 (dd, 1H, H6a), 3.81 (dd, 1H, H6b), 3.76
(t, 1H, H5), 3.56 (dd, 1H, H2), 3.52 (dd, 1H, H3), 3.49 (s,
3H, OCH3), 3.48 (s, 3H, OCH3), 3.41 (s, 3H, OCH3), 2.01
(s, 3H, OCH3). 13C NMR (75 MHz, CDCl3) δ: 97.7 (C1),
78.9 (C3), 77.2 (C2), 69.0 (C5), 67.4 (C4), 62.5 (C6), 58.7,
57.7, 55.1 (3 × OCH3). MALDI-TOF MS m/z: 245.2 [M +
Na]+.
Methyl 2,3-di-O-methyl-␣-D-glucopyranoside (3)
To methyl 4,6-O-benzylidene-2,3-di-O-methyl-α-D-gluco-
pyranoside (250 mg, 0.81 mmol) was added a solution of
80% acetic acid in water (15 mL) and the mixture was
heated to 50 °C. After a period of 4 h, the reaction mixture
was cooled, concentrated in vacuo, and then coevaporated
with toluene (3 × 20 mL). The residue was recrystallized
from ethyl acetate – hexane to give the desired compound 3
as a white solid (73 mg, 41%); mp 81–84 °C (lit. value (37)
mp 84 °C). 1H NMR (500 MHz, CDCl3) δ: 4.84 (d, 1H, H1),
3.85 (dd, 1H, H6a), 3.80 (dd, 1H, H6b), 3.65–3.61 (m, 1H,
H5), 3.63 (s, 3H, OCH3), 3.49 (s, 3H, OCH3), 3.50 (dd, 1H,
H4), 3.43 (s, 3H, OCH3), 3.46 (dd, 1H, H3), 3.21 (dd, 1H,
H2). 13C NMR (75 MHz, D2O) δ: 97.3 (C1), 83.0 (C3), 80.6
(C2), 72.2 (C5) 69.8 (C4), 61.2 (C6), 60.6, 58.5, 55.4 (3 ×
OCH3). MALDI TOF m/z: 245.2 [M + Na]+.
Molecular dynamics simulation
Molecular dynamics (MD) simulations were performed
using the SANDER module in the AMBER5 program suite
(39) with the GLYCAM (40) parameters for carbohydrates.
The 1–4 electrostatic and van der Waals interactions were
unscaled (SCEE = 1.0 and SCNB = 1.0), in contrast to the
traditional AMBER formalism, but as demonstrated to be es-
sential for carbohydrate simulations (23). The MD simula-
tions were performed using an nPT ensemble with TIP3P
water (41) and a dielectric constant set to unity. A long
range cutoff of 8 Å was employed. For solvation in water,
the carbohydrates were immersed in a theoretical box of 484
(3) and 473 (6) water molecules with approximate dimen-
sions of 28 Å × 27 Å × 23 Å. For simulations employing di-
chloromethane, a cutoff of 12 Å was employed with a
slightly larger theoretical box (34 Å × 32 Å × 29 Å) of 294
dichloromethane molecules (42) for both carbohydrates. The
SHAKE algorithm was applied to all hydrogen-containing
bonds, using the default settings. Initial conjugate-gradient
energy minimization (20 000 cycles), with restrained solute
coordinates, was performed on all systems studied, employ-
ing a 0.01 kcal mol–1 Å (1 cal = 4.184 J) convergence
criterion in the energy gradient. Subsequently, energy
minimization of the entire system was performed followed
by a simulated annealing period for the solvent, during
which the systems were heated from 5 to 300 K for a period
of 50 ps, held at 300 K for 250 ps in the case of water, and
1350 ps in dichloromethane, and then the systems were
cooled to 5 K over 50 ps. Lastly, the systems were heated
from 5 to 300 K over 50 ps prior to the production run MD
simulations. Between 100–120 ns of data were collected for
each system at 300 K. In all simulations, a 2 fs time step
was employed in the integration of the equations of motion.
The extremely long MD simulation times were required to
achieve adequate sampling of the rotational states. The post-
simulation analyses were performed using the CARNAL
module of the AMBER suite of programs.
Methyl 4,6-O-benzylidene-␣-D-galactopyranoside (4)
The synthesis of methyl 4,6-O-benzylidene-α-D-
galactopyranoside was carried out following the procedure
described for the preparation of the methyl 4,6-O-benzyl-
idene-α-D-glucopyranoside. Thus, methyl α-D-galactopy-
ranoside (1.00 g, 5.15 mmol) in dry acetonitrile (160 mL)
was reacted with benzaldehyde dimethyl acetal (2.20 g,
14.45 mmol) and camphorsulfonic acid (approx. 0.03 g,
0.13 mmol) to afford 4 as a white crystalline solid after
recrystallization (0.73 g, 50%); mp 169–171 °C (lit. value
1
(35) mp 168 to 169 °C). H NMR (300 MHz, CDCl3) δ:
7.52–7.35 (m, 5H, Ph), 5.55 (s, 1H, PhCH), 4.92 (d, 1H,
H1), 4.28 (dd, 1H, H6a), 4.25 (dd, 1H, H4), 4.08 (dd, 1H,
H6b), 3.93 (dd, 1H, H2), 3.89 (dd, 1H, H3), 3.69 (ddd, 1H,
H5), 3.46 (s, 3H, OCH3). 13C NMR (75 MHz, CDCl3) δ: not
observed (Cq, Ph), 129.4, 128.2, 126.3 (5 × Ph), 101.3
(PhCH), 100.0 (C1), 75.7 (C4), 70.0 (C2), 70.0 (C3), 69.4
(C6), 62.5 (C5), 55.7 (OCH3). MALDI-TOF MS m/z: 305.1
[M + Na]+.
Methyl 4,6-O-benzylidene-2,3-di-O-methyl-␣-D-
galactopyranoside (5)
The synthesis of methyl 4,6-O-benzylidene-2,3-di-O-
methyl-α-D-galactopyranoside was carried out following the
procedure described for the preparation of the methyl 4,6-O-
benzylidene-2,3-di-O-methyl-α-D-glucopyranoside.
Thus,
methyl 4,6-O-benzylidene-α-D-galactopyranoside (700 mg,
2.48 mmol) was reacted with NaH (300 mg, 7.50 mmol) and
methyl iodide (900 µL, 14.46 mmol) in dry DMF (20 mL) to
give 5 as a white solid after recrystallization (230 mg, 40%);
1
mp 126 to 127 °C (lit. value (38) mp 127 to 128 °C. H
NMR (300 MHz, CDCl3) δ: 7.56–7.48 (m, 2H, Ph), 7.38–
7.29 (m, 3H, Ph), 5.53 (s, 1H, PhCH), 4.97 (d, 1H, H1), 4.30
(d, 1H, H4), 4.22 (d, 1H, H6a), 4.03 (d, 1H, H6b), 3.77 (dd,
1H, H2), 3.66 (dd, 1H, H3), 3.57 (s, 1H, H5), 3.51 (s, 3H,
OCH3), 3.50 (s, 3H, OCH3), 3.43 (s, 3H, OCH3). 13C NMR
(75 MHz, CDCl3) δ: 137.9 (Cq, Ph), 128.9, 128.1, 126.3
(5 × Ph), 101.1 (PhCH), 98.5 (C1), 77.6 (C2), 77.6 (C3),
73.7 (C4), 69.4 (C6), 62.5 (C5), 59.0, 57.8, 55.4 (3 ×
OCH3). MALDI-TOF MS m/z: 332.8 [M + Na]+.
NMR
Each carbohydrate (5–10 mg per sample) was dissolved in
each of the different deuterated solvents (D2O, DMSO-d6,
CD3CN, CD2Cl2, CDCl3, toluene-d8, benzene-d6) and intro-
duced into a 5 mm NMR tube. The NMR experiments were
performed on Varian 600 and 500 MHz spectrometers. One-
© 2006 NRC Canada

Gonzalez-Outeiriño et al.
573
1
Fig. 2. Expansion of the experimental and simulated H NMR spectra for 6 in toluene-d8: (A) simulated spectrum; (B) homodecoupled
experimental spectrum with irradiation of H4; (C) experimental spectrum.
1
dimensional H spectra and homodecoupled experiments with
irradiation of the H4 and hydroxyl proton, OH6, were per-
Scheme 1. Synthesis of methyl 2,3-di-O-methyl-pyranosides. Re-
action conditions: (i) PhCH(OMe)2, CSA, MeCN; (ii) MeI, NaH,
DMF, RT; (iii) 80% AcOH, H2O, 50 °C.
3
formed to obtain the initial JH5-H6(R or S) couplings (Fig. 2).
Final values for the scalar couplings were obtained by itera-
tive spectral simulation using the SPINWORKS (43) and
PERCH NMR software (44).
R2
R2
OH
OH
i, ii, iii
R1
MeO
R1
HO
O
O
MeO
OMe
HO
OMe
Results and discussion
R1 = OH, R2 = H
R1 = H, R2 = OH
3: R1 = OH, R2 = H
6: R1 = H, R2 = OH
Synthesis of methyl 2,3-di-O-methyl-␣-D-gluco-
pyranoside and galactopyranoside
The desired 2,3-di-O-methylated α-D-glucopyranosides
and galactopyranosides were prepared in three steps follow-
ing established procedures (Scheme 1). Thus, regioselective
benzylidenation of the commercially available methyl α-
pyranosides at the O4 and O6 positions, followed by methyl-
ation of the hydroxyls at the C2 and C3 positions, resulted in
the fully protected intermediates (2 and 5). The final step in-
volved the acid-mediated cleavage of the benzylidene acetal
to give the desired compounds (3 and 6) as white solids after
recrystallization.
average of contributions from the individual states involved
in the process. In the case of ω-angle populations, at least
three conformations should be considered as being in fast
equilibrium on the NMR timescale (29, 27, 47). The NMR
data for the JHH couplings related to the C5-C6 rotational
equilibrium are generally deconvoluted into rotamer popula-
tions using eqs. [1]–[3]:
3
3
3
[1]
3JH5-H6R = ngt J(H5-H6R, gt) + ntg J(H5-H6R, tg)
3
3
Calculation of JHH couplings
+ ngg J(H5-H6R, gg)
Due to the rapid rotational timescale for the ω angle (45,
46), the rotamer population from NMR spectroscopy may
only be indirectly inferred. Any experimental measurement
corresponding to a fast equilibrium with respect to the
timescale of the technique can be interpreted as a weighted
3
3
[2]
[3]
3JH5-H6S = ngt J(H5-H6S, gt) + ntg J(H5-H6S, tg)
3
+ ngg J(H5-H6S, gg)
ngt + ntg + ngg = 1
© 2006 NRC Canada

574
Can. J. Chem. Vol. 84, 2006
Fig. 3. Expansion of the experimental and simulated spectra for 3 in CDCl3: (A) simulated spectrum; (B) experimental spectrum.
3
3
3
in which JH5-H6R and JH5-H6S are the experimentally mea-
Table 1. Limiting J values employed in the present analysis.
sured couplings between protons H5 and each of the
prochiral H6 protons, and J(H5-H6, ii) are the limiting values
of the specific coupling constants in each conformational
state ii with population nii.
3
Empirically derived
(48)
Quantum derived
(32)
Coupling
constant
3JH5-H6R
3JH5-H6S
gt
tg
gg
gt
tg
gg
There are numerous sources of uncertainty in this ap-
proach, both in terms of the experimental measurements and
in the data interpretation. Second-order effects are com-
monly observed in this type of spectra (Fig. 3), making the
10.7
3.1
5
0.9
2.8
9.9
1.5
4.5
10.8
0.8
1.3
10.7
Note: J values are in Hertz.
3
use of directly measured J values inappropriate. To remove
the second-order effects, the spectra were simulated via an
iterative procedure in which the first-order 3J couplings were
varied, until optimal agreement between the experimental
and the predicted spectra was achieved (43).
thousands of conformations typically generated during MD
simulations.
In the parameterization of the any given Karplus-type re-
3
There are two further significant limitations of this
method, related to the calculation of the 3J couplings.
Firstly, 3J couplings are generally computed using a
Karplus-type equation that correlates coupling constants
with dihedral angles and many approaches to the
parameterization of these relationships have been proposed,
including empirical methods based on experimental data
(48) and theoretical methods based on quantum mechanical
data (32). In the case of the empirical equation reported by
Haasnoot et al. (48), the equation included the effects of
lationship, a set of limiting J values must be derived for
each rotamer. The accuracy of the limiting J values is critical
when applying eqs. [1]–[3]. Previously, it was assumed that
the ω angle would adopt idealized values of 60°, 180°, and
–60° in each of the rotameric states; however, this approach
often led to small unrealistic negative rotamer populations.
More recently, gas-phase energy minimized geometries have
been employed for each rotamer. Notably, these optimized
geometries differed markedly from idealized values (29, 27).
As part of this study, the extent to which the omission of sol-
vent in the geometry optimizations biases the conformations
away from those actually present in solution will be exam-
ined. Gas-phase energy minimization always leads to an ex-
cessive degree of internal hydrogen bonding (23), which is
expected to introduce some level of distortion into the ω an-
gle relative to its orientation in water. To assess the effect of
the limiting J value on the derivation of the rotamer popula-
tions we have employed both the empirical and quantum-
derived values (see Table 1).
3
electronegative substituents on the JHH values, but was
parameterized using rigid model compounds. A single equa-
3
tion was developed for computing each of the JH5-H6R and
3JH5-H6S values. This approach can lead to the prediction of
small negative rotamer populations, particularly for dynamic
systems. The equation introduced by Stenutz et al. (32) was
parameterized using 3J couplings computed from density
functional theory (DFT) for a set of model compounds. By
considering variations in the ω conformations, as well as in
the orientation of the hydroxyl groups, Stenutz et al. arrived
at a separate equation for the couplings between H5 and
each of the prochiral H6 protons, reflecting both geometric
and electronic variations in the pro-R and pro-S environ-
MD simulations
The rotamer populations of the ω torsion angle were pre-
dicted (Fig. 4) from explicitly solvated simulations of 3 and
6 in both water and dichloromethane (see Fig. 4). It may be
seen that conformational convergence for these systems was
not reached until simulation times of between 40 and 80 ns.
A reoccurring question when running MD simulations is,
“When has a simulation been run long enough that conclu-
3
ments. A promising alternative is to compute the J cou-
plings directly from quantum mechanical calculations, without
employing an empirical Karplus-type equation (49). This
method has many benefits; however, its computationally in-
tensity makes it challenging to employ when analyzing the
© 2006 NRC Canada

Gonzalez-Outeiriño et al.
575
Fig. 4. MD trajectories (120 and 100 ns) of the ω angle in (A) methyl 2,3-di-O-methyl-α-D-glucopyranoside (3) and (B) methyl 2,3-di-
O-methyl-α-D-galactopyranoside (6) in water, with resulting histograms and population % on the right-hand side of each figure.
Conformational convergence is plotted below the respective trajectories.
sions about the dynamic properties of a system can be
drawn?” The amount of simulation time required to reach
conformational equilibration will vary depending on the life-
time of the possible conformations, which itself is directly
related to the barrier heights among the states. A molecule
that has long conformational lifetimes, exhibiting infrequent
conformation transitions, will require a long simulation time
period to reach dynamic convergence. Conversely, a mole-
cule that possesses short conformational lifetimes and fre-
quent conformation transitions, will require a shorter
simulation to reach convergence. In terms of the rotational
states of the ω angle in 3, there are frequent transitions be-
tween the gt and tg rotamers, and infrequent transitions to
the gg rotamer, as seen in Fig. 4. The infrequent transitions
of the gg rotamer, coupled with the large populations of the
gt and gg rotamers, resulted in the slow convergence of the
gt and gg rotamers. A relatively low energy barrier between
gt and tg would explain the rapid transitions, while a higher
energy barrier between tg and gg is consistent with the long
lifetime for each of these rotamers. These results are in qual-
itative agreement with the quantum mechanical rotational
energy curves computed for the rotation of the ω angle in
methyl α-D-glucopyranoside (23).
gg rotamer. In contrast to the case of 3, the rotamer distribu-
tion in 6 reached convergence in less than 50 ns, with a
rotamer ratio of 62:27:11. Again, in terms of rotational bar-
riers, the frequency of transitions between the gt and tg
states indicates a low energy barrier between these rotamers
and high energy barriers around the gg. The rotamer conver-
gence was achieved more rapidly than in 3 because the two
rotamers with higher populations are in frequent transition
with each other.
In water, the rotamer distributions for methyl 2,3-di-O-
methyl-α-D-glucopyranosides (3) and methyl 2,3-di-O-
methyl-α–D–galactopyranoside (6) were predicted to be
34:6:61 and 62:32:6, respectively, which are in reasonable
agreement with the previously reported experimental data
for the underivatized methyl glycosides (41:1:58 and
61:18:21, respectively) (30). The results from the present
MD simulations are also similar to those from the 50 ns
fully solvated MD simulations previously reported (23) for
methyl α-D-glucopyranoside (40:6:54) and methyl α-D-
galactopyranoside (64:28:8). These observations confirm
that methylation of oxygens O2 and O3 does not make a re-
markable impact on the rotamer population of the ω torsion
angle in either 3 or 6 in water.
A very different behavior for rotamer interconversions
was observed for 6, displaying frequent transitions between
the gt and tg rotamers, and only infrequent transitions to the
In an attempt to examine the ability of MD simulations to
predict the effect of solvent polarity on rotamer population,
MD simulations of 3 and 6 in CH2Cl2 were performed. It
© 2006 NRC Canada

576
Can. J. Chem. Vol. 84, 2006
Fig. 5. MD trajectories (100 ns) of the ω angle in (A) methyl 2,3-di-O-methyl-α-D-glucopyranoside (3) and (B) methyl 2,3-di-O-
methyl-α-D-galactopyranoside (6) in dichloromethane, with resulting histograms and population % on the right-hand side of each fig-
ure. Conformational convergence is plotted below the respective trajectories.
should be noted that there are at present no force field pa-
Table 2. Average ω torsion angles from the MD simulations of 3
rameters available for deuterated solvents, and simulations
in nonaqueous solvents are uncommon. Thus, the solvent
model for CH2Cl2 is relatively untested; however, the initial
results are encouraging, particularly in the case of 3 (Fig. 5).
The simulation produced a population conformation ratio
of 49:21:30 for 3, which indicates a significant increase in
the population of the tg and gt rotamers at the expense of the
gg rotamer, relative to the system in water (34:6:61,
gt:tg:gg). A negligible effect was predicted for 6 (66:28:6 in
CH2Cl2 vs. 64:28:8 in water); however, an examination of
the convergence plots indicates that even after 100 ns the
populations have not fully converged.
and 6 in water, CH2Cl2, and vacuum.
3
6
Environment
gt
tg
gg
gt
tg
gg
H2O
70.1
66.4
72.6
158.3
161.2
171.3
291.5
293.7
289.4
70.5
69.6
—
168.6
170.1
—
298.2
296.3
—
CH2Cl2
Vacuuma
aCalculated using MM3(94) with modified O-C-C-O torsion parameters
(27).
with quantum-derived limiting J values (32) did not result in
any nonphysical (negative) rotamer populations. In the case
of 3, regardless of the choice of the methodology, the popu-
lation of the tg rotamer was shown to increase approxi-
mately linearly with the decreasing polarity of the small
solvent molecules. This observation is consistent with
expectations based on gas-phase quantum mechanical calcu-
lations of the rotational properties of the ω angle in methyl
α-D-glucopyranoside (23), which implies that there would be
a preferential enhancement of internal hydrogen bonding in
low polarity environments, leading, in the limiting case (ε =
1), to a population ratio of approximately 43:28:28. In the
larger aromatic solvents, the tg population is again high, but
it does not follow a direct relationship with solvent polarity.
In contrast to 3, little sensitivity to solvent polarity is dis-
played by 6. This is an unexpected result in that quantum
It is interesting to note that the ω torsion angle displayed
sensitivity to the solvent, as determined from the values of
the average O5-C5-C6-O6 torsion angles for each rotamer
extracted from each simulation (Table 2). Clearly, idealized
values of 60°, 180°, and 300° are in poor agreement with
our averaged solution values. Solvation appears to signifi-
cantly perturb the value of the ω angle in the tg rotamer
away from the idealized value of 180°.
NMR analysis
Presented in Tables 3 and 4 are the experimental J values
3
and the rotamer populations for 3 and 6, respectively, pre-
dicted from these values using eqs. [1]–[3] and two sets of
3
limiting J values (32). Unlike the results from the empiri-
cally determined limiting J values (48), the results obtained
© 2006 NRC Canada

Gonzalez-Outeiriño et al.
Table 3. Experimental J values and rotamer populations for 3, listed in order of decreasing solvent polarity.
577
3
3
a
Experimental JHH
Empirical fit (48)
Quantum mechanical fit (32)
Solvent
3JH5-H6R
3JH5-H6S
gt
tg
gg
gt
tg
gg
7
D2O
5.44
5.87
5.40
4.87
4.34
4.32
3.74
2.08
2.11
2.92
3.27
3.47
3.10
2.91
51 (49)
55 (53)
46
–11(0)b
60 (51)
56 (47)
54
48
53
44
37
30
31
26
45
40
40
43
48
50
58
DMSO-d6
CD3CN
CD2Cl2
–11(0)b
7
16
20
22
18
16
0
4
7
3
0
39
57
CDCl3
32
61
Toluene-d8
Benzene-d6
34
64
29
71
Note: Dielectric constants (at 20°): D2O (79.8), DMSO (48.9), CH3CN (37.5), CH2Cl2 (9.1), CHCl3 (4.8), toluene (2.4),
benzene (2.3) (ref. 51).
a 3
J
from simulated spectra. The maximum error allowed between simulated and experimental spectra was 0.5 Hz.
HH
bValues in parentheses were calculated by assuming a zero population for the tg rotamer.
3
Table 4. Experimental J values and rotamer populations for 6.
3
a
Experimental JHH
Empirical equation (48)
DFT equation (32)
Solvent
3JH5-H6R
3JH5-H6S
gt
tg
gg
gt
tg
gg
D2O
7.13
5.99
6.20
6.37
6.31
4.30
5.19
4.24
4.88
5.22
57
40
47
46
43
17
29
16
25
29
27
31
36
30
28
57
41
47
46
44
30
40
30
37
40
12
19
23
17
16
CD2Cl2
CDCl3
Toluene-d8
Benzene-d6
Note: Due to signal overlap, coupling constants could not be measured in DMSO-d6 and CD3CN.
a 3
J
from simulated spectra. The maximum error allowed between simulated and experimental spectra was 0.5 Hz.
HH
calculations predict a decrease in polarity should lead to a
significant decrease in the population of the tg rotamer with
an increase in the gg (23). This change in population was
predicted on the basis of the formation of strong internal hy-
drogen bonds in the limiting case of a purely apolar environ-
ment. However, based on the data in Table 4, it appears that
an interpretation of rotamer behavior, based on the view that
polar solvents disrupt the internal hydrogen bonds and there-
fore lead to the rotamer distribution being determined by re-
pulsive 1,3-diaxial electrostatic interactions among O4, O5,
and O6, may be correct (23). It should be noted that hydro-
gen bonds among conformationally restricted 1,3-diaxial
hydroxyl groups appear to persist even in water (50).
Overall, the agreement with the predictions from the MD
simulations both in terms of the response of the populations
to polarity and in quantitative terms is notable. Thus, it is
reasonable to probe the simulational results further to extract
quantitative information pertaining to the level of intra-
molecular hydrogen bonding in each rotational state in both
water and CH2Cl2. For each simulation the periods during
which hydrogen bonds were present between O4, O5, and
O6 were determined for 3 and 6 and are presented in Ta-
ble 5.
internuclear distances is very weak. Similarly, the reverse
interaction, that is, one in which HO6 donates a proton to
HO4 (O6(H)···O4) is also weak; however, this hydrogen
bond shows a strong sensitivity in both 3 and 6 to solvent
polarity. In CH2Cl2, the O4(H)···O6 interaction was pre-
dicted to increase both in occupancy and strength. This may
be understood given that as a primary alcohol, the proton at
O6 is preferentially exposed to solvent and would generally
be able to act as either a proton donor or acceptor with water
molecules. In the low polarity environment, HO6 has little
choice but to interact with either HO4 or HO5. In contrast,
HO4 may continue to interact with HO3, regardless of the
solvent. Similar arguments may be employed to explain the
enhanced strength and occupancy in CH2Cl2 of the
O6(H)···O5 hydrogen bond in 3 and 6. It is notable that
CH2Cl2 does not enhance all internal hydrogen bonding to
an equal extent. When HO6 acts as a proton donor with O5,
the occupancy time for this interaction is approximately
doubled in both 3 and 6 in CH2Cl2. Notably, in the galacto-
derivative 6, the low polarity environment of the CH2Cl2
doubles the occupancy of the O6(H)···O4 hydrogen bond,
relative to an aqueous environment, whereas for the gluco-
derivative 3, the occupancy of this hydrogen bond in CH2Cl2
increased nearly fourfold. Therefore, in 6, although a low
polarity environment increases the probability of interactions
between O6(H) and O4 and O5, the relative strengths and
occupancies of these interactions are unaffected by the sol-
vent. On this basis, it would be predicted that the rotamer
populations for 6 would not strongly influenced by solvent
polarity, as confirmed by analysis of the NMR data. Since in
3 the O6(H)···O4 hydrogen bond is preferentially favored in
Considerable insight into the sensitivity of internal hydro-
gen bonding to solvent polarity is provided by the MD data.
Some overall trends are apparent; for example, the hydrogen
bond between HO6 and O5 is consistently the most occupied
(12.9%–18.0%) and the strongest, as indicated by the inter-
oxygen distances. In contrast, hydroxyl group HO4 is con-
sistently a very poor proton donor to HO6 (O4(H)···O6); dis-
playing the lowest occupancies (2.9%–5.8%) and long
© 2006 NRC Canada

578
Can. J. Chem. Vol. 84, 2006
Table 5. Average hydrogen bond distances and occupancies for 3 and 6 from the MD data.
Hydrogen bond
MD solvent
Total simulation time (ns)
Molecule
O4(H)···O6
O6(H)···O4
O6(H)···O5
Water
CH2Cl2
Water
122
101
100
114
1
1
2
2
3.41, 3.9
3.19, 2.9
3.36, 5.8
3.36, 2.7
3.39, 4.8
3.14, 19.0
3.34, 6.3
3.35, 12.3
2.98, 18.0
2.96, 31.4
3.01, 12.9
3.00, 26.8
CH2Cl2
Note: Bond distances are between non-hydrogen atoms (in angstroms) and occupancies are presented as a % of total simu-
lation time.
CH2Cl2, and as it can only form in the tg rotamer, a low po-
larity solvent should induce a shift in rotamer populations in
favor of the tg state, as confirmed experimentally.
this work, we find that in a low polarity environment HO6
prefers to interact with O4, but that in water these interac-
tions are markedly weakened, indicating that HO6 acts as a
hydrogen bond donor to water. That this conclusion was not
supported by the studies on methyl 2,3,4-tri-O-methyl-α-D-
glucopyranoside is perhaps related to the absence of a
methyl group at O4 in the present work. It may be that the
steric bulk of the methyl groups at the 2-, 3-, and 4-positions
detracts from the ability of O4 to act as a hydrogen bond ac-
ceptor for HO6. The simulations performed in water have
been shown to reproduce the rotamer populations to within
experimental error; however, despite the long simulation
times, the results from those in CH2Cl2 were less accurate.
The MD simulations illustrate that not only are the rotamer
populations sensitive to the configuration at C4 in hexopy-
ranosides, but so are the lifetimes and pathways for transi-
tions among the rotameric states. This latter observation
indicates that MD simulations of flexible carbohydrates may
require very long simulation times (>100 ns) to reach statis-
tical convergence.
In water, the internal hydrogen bonding networks are dis-
rupted by competition with hydrogen bonds to the solvent.
When the internal hydrogen bonds are differentially dis-
rupted, that is, when some become relatively weaker than
others, the rotamer populations associated with the ω angle
may be altered. In the case of 3, in water the preferential dis-
ruption of the O6(H)···O4 interaction destabilized the tg
rotamer, leading to the observed preference for gauche
rotamers. In 6, the O6(H)···O4 hydrogen bond forms most
strongly in the gg orientation, an orientation in which the
O6(H)···O5 hydrogen bond may also form. Thus, in the hy-
drogen bond strengthening environment of low polarity sol-
vents, a higher ratio of gg:gt rotamers would be expected
than in water. This trend was observed in the NMR data
(see, for example, the ratios 27:57 in D2O vs. 36:47 in
CDCl3). Unlike the case of 3, the shift in rotamer popula-
tions in 6 is less profound, presumably reflecting the fact
that the hydrogen bonds in 6 are approximately equally
prone to disruption by water.
Acknowledgments
Conclusion
This paper is dedicated to Professor W.A. Szarek in hon-
our of his innumerable contributions to carbohydrate chem-
istry and in recognition of his inspiration as a mentor. We
thank the National Institutes of Health (RR05357 and
GM55230) for financial support, and JG-O wishes to thank
Dr. J. Glushka for helpful discussions regarding the optimi-
zation of the NMR experiments.
The joint NMR and MD analysis of 3 and 6 leads to the
conclusion that the balance between inter- and intra-
molecular interactions involving HO6 plays a direct role in
determining the rotational profiles for the ω angles. In a low
polarity environment, internal hydrogen bonds are shown to
be strengthened in both 3 and 6, however, in 3, such an envi-
ronment disproportionately enhances the O6(H)···O4 interac-
tion, leading to an increase in the population of the tg state.
Without the hydrogen bond enhancement offered by a low
polarity environment, both 3 and 6 display rotamer popula-
tions that are consistent with expectations based on the
minimization of repulsive intramolecular oxygen–oxygen in-
teractions. That is, when in an aqueous environment, the wa-
ter disrupts stabilizing intramolecular hydrogen bonding and
leads to a system whose bond rotational properties are domi-
nated by internal electrostatic repulsions.
The results of this work are generally consistent with the
conclusions reached by Rockwell and Grindley (27), to the
extent that both studies illustrate that low polarity environ-
ments strengthen internal hydrogen bonds in carbohydrates.
The data from the MD simulations confirm and quantify the
strengthening of the intramolecular hydrogen bonds in
CH2Cl2 inferred by the NMR analysis. A significant differ-
ence between the two studies pertains to the role of interac-
tions between the solvent and HO6 in glucopyranosides. In
References
1. A. Varki. Glycobiology, 3, 97 (1993).
2. R.A. Dwek. Chem. Rev. 96, 683 (1996).
3. H. Lis and N. Sharon. Chem. Rev. 98, 637 (1998).
4. H.C. Joao, I.G. Scragg, and R.A. Dwek. FEBS Lett. 307, 343
(1992).
5. D.F. Wyss, J.S. Choi, J. Li, M.H. Knoppers, K.J. Willis,
A.R.N. Arulanandam, A. Smoylar, E.L. Reinherz, and G. Wag-
ner. Science (Washington, D.C.), 269, 1273 (1995).
6. D.M. Coltart, R.A.K. Royyuru, L.J. Williams, P.W. Glunz, D.
Sames, S.D. Kuduk, J.B. Schwarz, X.-T. Chen, S.J.
Danishefsky, and D.H. Live. J. Am. Chem. Soc. 124, 9833
(2002).
7. C.J. Bosques, S. Tschampel, R.J. Woods, and B. Imperiali. J.
Am. Chem. Soc. 126, 8421 (2004).
8. P.M. Rudd, H.C. Joao, E. Coghill, P. Fiten, M.R. Saunders, G.
Opdenakker, and R.A. Dwek. Biochemistry, 33, 17 (1994).
© 2006 NRC Canada

Gonzalez-Outeiriño et al.
579
9. P.M. Rudd, M.R. Wormald, R.L. Stanfield, M. Huang, N.
Mattsson, J.A. Speir, J.A. DiGennaro, J.S. Fetrow, R.A. Dwek,
and I.A. Wilson. J. Mol. Biol. 293, 351 (1999).
10. M. Tomana, J. Novak, B.A. Julian, K. Matousovic, K.
Konecny, and J. Mestecky. J. Clinical Investigation, 104, 73
(1999).
32. R. Stenutz, I. Carmichael, G. Widmalm, and A.S. Serianni. J.
Org. Chem. 67, 949 (2002).
33. N.K. de Vries and H.M. Buck. Carbohydr. Res. 165, 1 (1987).
34. D.A. Cumming and J.P. Carver. Biochemistry, 26, 6676 (1987).
35. K. Yoza, N. Amanokura, Y. Ono, T. Akao, H. Shinmori, H.
Takeuchi, S. Shinkai, and D.N. Reinhoudt. Chem. Eur. J. 5,
2722 (1999).
11. G.J. Davies, V.M.-A. Ducros, A. Varrot, and D.L. Zechel.
Biochem. Soc. Trans. 31, 523 (2003).
12. S. Sabesan, S. Neira, and Z. Wasserman. Carbohydr. Res. 267,
239 (1995).
13. Y. Bourne, B. Bolgiano, D.-I. Liao, G. Strecker, P. Cantau, O.
Herzberg, T. Feizi, and C. Cambillau. Nature Struct. Biol. 1,
863 (1994).
14. A. Zdanov, Y. Li, D.R. Bundle, S.-J. Deng, C.R. MacKenzie,
S.A. Narang, N.M. Young, and M. Cygler. Proc. Natl. Acad.
Sci. U.S.A. 91, 6423 (1994).
15. M.R. Wormald, A.J. Petrescu, Y.-L. Pao, A. Glithero, T. Elliot,
and R.A. Dwek. Chem. Rev. 102, 371 (2001).
16. J. Jiménez-Barbero and T. Peters. NMR of glycoconjugates.
Wiley-VCH, Weinheim. 2002.
36. R.A. Edington, E.L. Hirst, and E.E. Percival. J. Chem. Soc.
2281 (1955).
37. D. Trimnell, W.N. Doane, C.R. Russell, and C.E. Rist.
Carbohydr. Res. 11, 497 (1969).
38. D.J. Bell and G.D. Greville. J. Chem. Soc. 1136 (1955).
39. D.A. Case, D.A. Pearlman, J.W. Caldwell, T.E. Cheatham III,
W.S. Ross, C.S. Simmerling, T.A. Darden, K.M. Merz, R.V.
Stanton, A.L. Cheng, J.J. Vincent, M. Crowley, D.M. Fergu-
son, R.J. Radmer, G.L. Seibel, U.C. Singh, P.K. Weiner, and
P.A. Kollman. AMBER 5.0 [computer program]. University of
California San Francisco, San Francisco, California. 1997.
40. R.J. Woods, R.A. Dwek, C.J. Edge, and B. Fraser-Reid. J.
Phys. Chem. 99, 3832 (1995).
17. W.A. Bubb. Concept. Magn. Res. 19A, 1 (2003).
18. D.A. Cumming and J.P. Carver. Biochemistry, 26, 6664 (1987).
19. J. Gonzalez-Outeriño, R. Kadirvelraj, and R.J. Woods.
Carbohydr. Res. 340, 1007 (2005).
20. M. Martín-Pastor, J.L. Asensio, R. López, and J. Jiménez-
Barbero. J. Chem. Soc. Perkin Trans. 2, 713 (1995).
21. M. Martin-Pastor and C.A. Bush. Biochemistry, 39, 4674
(2000).
41. W.L. Jorgensen. J. Am. Chem. Soc. 103, 341 (1981).
42. T. Fox and P.A. Kollman. J. Phys. Chem. B, 102, 8070 (1998).
43. K. Marat. SPINWORKS 2.4 [computer program]. Department
of Chemistry, University of Manitoba, Winnipeg, Manitoba.
2004
44. PERCH Solutions Ltd. PERCH NMR software [computer pro-
gram]. PERCH Solutions Ltd., Kuopio, Finland. 2005
45. J.W. Brady. J. Am. Chem. Soc. 111, 5155 (1989).
46. P.J. Hajduk, D.A. Horita, and L.E. Lerner. J. Am. Chem. Soc.
115, 9196 (1993).
22. E.W. Sayers and J.H. Prestegard. Biophys. J. 79, 3313 (2000).
23. K.N. Kirschner and R.J. Woods. Proc. Natl. Acad. Sci. U.S.A.
98, 10541 (2001).
47. M. Kraszni, Z. Szakacs, and B. Noszal. Analyt. Bioanalyt.
24. H.F. Azurmendi and C.A. Bush. Carbohydr. Res. 337, 905
(2002).
25. R.U. Lemieux and S. Koto. Tetrahedron, 30, 1933 (1974).
26. S. Wolfe, M.-H. Whangbo, and D.J. Mitchell. Carbohydr. Res.
69, 1 (1979).
Chem. 378, 1449 (2004).
48. C.A.G. Haasnoot, M. De Leeuw, and C. Altona. Tetrahedron,
36, 2783 (1980).
49. J. Gonzalez-Outeriño, J. Glushka, A. Siriwardena, and R.J.
Woods. J. Am. Chem. Soc. 126, 6866 (2003).
27. G.D. Rockwell and B.T. Grindley. J. Am. Chem. Soc. 120,
10953 (1998).
28. R.H. Marchessault and S. Perez. Biopolymers, 18, 2369 (1979).
29. K. Bock, and J.Ø. Duus. J. Carbohydr. Chem. 13, 513 (1994).
30. Y. Nishida, H. Hori, H. Ohrui, and H. Meguro. J. Carbohydr.
Chem. 7, 239 (1988).
50. J. Hawley, N. Bampos, N. Aboitiz, J. Jiménez-Barbero, M.
López de la Paz, J.K.M. Sanders, P. Carmona, and C. Vicent.
Eur. J. Org. Chem. 1925 (2002).
51. R.C. Weast (Editor). CRC handbook of chemistry and physics.
59th ed. CRC Press, Inc., Boca Raton, Florida. 1978.
31. I. Tvaroska, F.R. Taravel, J.P. Utille, and J. Carver. Carbohydr.
Res. 337, 353 (2002).
© 2006 NRC Canada