Aromatic-carbohydrate interactions: An NMR and computational study of model systems

Article abstract of DOI:10.1002/chem.200800247

The interactions of simple carbohydrates with aromatic moieties have been investigated experimentally by NMR spectroscopy. The analysis of the changes in the chemical shifts of the sugar proton signals induced upon addition of aromatic entities has been interpreted in terms of interaction geometries. Phenol and aromatic amino acids (phenylalanine, tyrosine, trypto-phan) have been used. The observed sugar-aromatic interactions depend on the chemical nature of the sugar, and thus on the stereochemistries of the different carbon atoms, and also on the solvent. A preliminary study of the sol_vation state of a model monosaccharide (methyl β-galactopyranoside) in aqueous solution, both alone and in the presence of benzene and phenol, has also been carried out by monitoring of intermolecular homonuclear solvent-sugar and aromatic-sugar NOEs. These experimental results have been compared with those obtained by density functional theory methods and molecular mechanics calculations.

Full text of DOI:10.1002/chem.200800247

DOI: 10.1002/chem.200800247
Aromatic–Carbohydrate Interactions:
An NMRand Computational Study of Model Systems
Sophie Vandenbussche,[a] Dolores Díaz,[b] María Carmen Fernµndez-Alonso,[b]
Weidong Pan,[c] StØphane P. Vincent,[c] Gabriel Cuevas,[d] Francisco Javier CaÇada,[b]
Jesffls JimØnez-Barbero,*[b] and Kristin Bartik*[a]
Abstract: The interactions of simple
carbohydrates with aromatic moieties
have been investigated experimentally
by NMR spectroscopy. The analysis of
the changes in the chemical shifts of
the sugar proton signals induced upon
addition of aromatic entities has been
interpreted in terms of interaction geo-
metries. Phenol and aromatic amino
acids (phenylalanine, tyrosine, trypto-
phan) have been used. The observed
sugar–aromatic interactions depend on
the chemical nature of the sugar, and
thus on the stereochemistries of the
different carbon atoms, and also on the
solvent. A preliminary study of the sol-
vation state of a model monosacchar-
ide (methyl b-galactopyranoside) in
aqueous solution, both alone and in the
presence of benzene and phenol, has
also been carried out by monitoring of
intermolecular homonuclear solvent–
sugar and aromatic–sugar NOEs. These
experimental results have been com-
pared with those obtained by density
functional theory methods and molecu-
lar mechanics calculations.
Keywords:
carbohydrates
·
molecular modeling
·
molecular
recognition · NMR spectroscopy ·
protein–carbohydrate interactions
Introduction
tion, at the molecular level, of the events involved at the
heart of biological phenomena. Many of these are mediated
by interactions between proteins and the carbohydrates
present on the surfaces of cells.[1] Proteins that recognize
carbohydrates are found in a wide variety of organisms,
from viruses and bacteria to plants and animals, and include
enzymes, glycoproteins (antibodies, for example), and lec-
tins. These protein–carbohydrate interactions are highly se-
lective—lectins can distinguish between carbohydrates that
differ only in the stereochemistry at one carbon atom[2]—
and are characterized by affinity constants typically of the
order of 103 mÀ1 for monosaccharides and up to 107 mÀ1, or
even higher, for complex carbohydrates. Hydrogen-bonding
and nonpolar interactions[3,4] play important roles in the
binding process. In the last few years, the presence of aro-
matic rings in the binding sites of lectins has been highlight-
ed as essential for recognition of neutral sugars, especially
of the Gal/GalNAc and Glc/GlcNAc families.[5–14] The im-
portance of these aromatic amino acids has been confirmed
by site-directed mutagenesis.[5,6] For the molecular recogni-
tion process, both the structure and the conformation of the
carbohydrate,[9] as well as the nature and orientation of the
aromatic rings, are of importance.[10,11] This so-called “stack-
ing interaction” between the carbohydrate and aromatic
amino acid side chains is also referred to as a “CH–p inter-
Understanding of molecular recognition processes is of par-
amount importance for the life sciences and for the elucida-
[a] Ir. S. Vandenbussche,+ Prof. Dr. K. Bartik
Molecular & Biomolecular Engineering
UniversitØ Libre de Bruxelles, Brussels (Belgium)
Fax : (+32)26503606
E-mail: kbartik@ulb.ac.be
[b] Dr. D. Díaz,+ Dr. M. C. Fernµndez-Alonso,+ Dr. F. J. CaÇada,
Prof. Dr. J. JimØnez-Barbero
Centro de Investigaciones Biológicas, CSIC
Ramiro de Maeztu 9, 28040 Madrid (Spain)
Fax : (+34)915-360-432
E-mail: jjbarbero@cib.csic.es
[c] Dr. W. Pan, Prof. Dr. S. P. Vincent
Laboratoire de chimie bio-organique
FacultØs Universitaires ND de la Paix
Namur (Belgium)
[d] Dr. G. Cuevas
Instituto de Química
Universidad Nacional Autónoma de MØxico
Ap Postal 70213
04510, Coyoacµn, Circuito Exterior
MØxico DF (MØxico)
[+] Contributed equally to this work.
Supporting information for this article is available on the WWW
under http://www.chemeurj.org/ or from the author.
7570
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2008, 14, 7570 – 7578

FULL PAPER
action”,[5,12–14] and has also been shown to occur in the gas
phase.[15]
tions with the target saccharides to be established. Millimo-
lar affinities were obtained with synthetic receptors for the
first time.
On this basis, in order to allow better understanding of
CH–p interactions, it seems interesting to study—not only
theoretically but also experimentally—simple model sys-
tems. Here we present a methodology for exploration of car-
bohydrate–aromatic interactions from the structural view-
point.
Sometimes regarded as a hydrogen bond, the CH–p inter-
action is much weaker than the “classical” hydrogen bond
(ca. 30% according to QM calculations).[16–18] In the comput-
ed benzene–methane and benzene–chloromethane systems,
the most stable configuration is always the one in which the
CH bond is perpendicular to the benzene ring plane, al-
though the interaction energy does not seem to be much
modified by the position of the hydrogen over the aromatic
ring (i.e., above the center of the aromatic cycle or at the
edge of the cycle).[17–19]
Ab initio HF, MP2, and CCSD(T) calculations undertak-
en on several simple systems (benzene–methane, -ethane,
and others[16,17]) have suggested that the dispersion interac-
tion between the aromatic entity and the carbon atom of
the CH bond is largely responsible for the CH–p interac-
tion, with the electron correlation greatly enhancing the cal-
culated binding energy.[17] The electrostatic contribution in-
creases when the acidity of the hydrogen increases.[17] The
presence of an electrostatic contribution to the attraction
might be responsible for the directionality of the CH–p
bond.
Ab initio studies on CH–p interactions in sugar–protein
complexes have also been reported.[14,19–22] An extensive
study of protein–carbohydrate complexes obtained from six
high-resolution (1.3 Š or better) X-ray structures has been
undertaken by Spiwok et al.[14] The estimated interaction en-
ergies fall between À11.7 and À26.8 kJmolÀ1, except for one
complex in which the higher interaction energy
(À51.4 kJmolÀ1) was explained by the additional presence
of a classical H-bond between a hydroxyl group of the car-
bohydrate and the hydroxyl group of a tyrosine. Within
these complexes, a wide range of angles was observed be-
NMR is a particularly interesting tool for studying geo-
metrical features and molecular interactions. Provided that a
CH–p interaction takes place, the chemical shifts of the pro-
tons involved in the interaction should be modified accord-
ing to their spatial positions in the field created by the aro-
matic ring current. Here we report an NMR study of the in-
teractions between different d-monosaccharides (the a- and
b-methyl anomers of glucopyranose, galactopyranose, and
ribofuranose, as well as a-methyl mannopyranose) and sev-
eral aromatic entities (phenol and the l-amino acids phenyl-
alanine, tyrosine, and tryptophan), in an attempt to general-
ize and to extend the previously obtained conclusions. Meas-
urements of the chemical shift variations of the sugar pro-
tons in the absence of each aromatic moiety and in the pres-
ence of excesses (10–20 molar equivalents) have been
undertaken. Titration experiments have been performed for
some of these systems.
We have also undertaken the analysis of the possible
changes in the solvation sphere of one simple carbohydrate
(methyl b-d-galactopyranoside) upon addition of phenol,
through intermolecular homonuclear NOEs. More precisely,
we have investigated whether intermolecular sugar–water
and sugar–aromatic NOEs can provide experimental infor-
mation relating to the existence of the complex and to the
solvation of this carbohydrate in the absence and in the
presence of aromatic compounds.
À
tween the donor C H bond and the aromatic plane (318 to
nearly perpendicular). The perpendicular distances calculat-
ed between the hydrogen atom and the plane of the aromat-
ic ring ranged between 2.6 and 3.3 Š, fairly similar to those
found in the experimentally determined X-ray structures.[23]
In comparison, for the benzene–methane and benzene–
methanol complexes, the distances between the hydrogen
atom and the aromatic cycle were only 2.7 and 2.5 Š, re-
spectively.[17]
All the observed data indicate that, depending on the
chemical nature of the sugar, aromatic–carbohydrate inter-
actions indeed take place, and can be easily monitored by
NMR. These NMR data have also been interpreted in terms
of geometrical models with the aid of molecular mechanics
calculations.
The group in Madrid has previously performed calcula-
tions (at the MP2/6–31GACHTRE(UGN d,p) level with counterpoise cor-
Results and Discussion
rection) relating to the complex formed by b-d-fucose (6-
deoxy-b-d-galactose) and benzene, and modeled the geome-
try of the interaction.[24] The interaction energy was
À10.9 kJmolÀ1, falling in the lower limit of the interaction
energies obtained by Spiwok et al.,[14] but of the same order
of magnitude as that computed for the methanol–benzene
complex[17] (taking the known 20% overestimation of MP2
calculations relative to CCSD(T) ones into account[16]).
In an application of all this knowledge, a synthetic lectin
has very recently been chemically prepared[25] through the
combination of different simple molecular fragments that si-
multaneously allow hydrogen bonds and stacking interac-
The most commonly used reference compound for NMR
studies of biological and bioorganic molecules in water is
2,2-dimethyl-2-silapentane-5-sulfonate (DSS); the chemical
shift of the methyl protons is set to zero. An alternative ref-
erence is sodium 3-trimethylsilylpropionate (TSP), with one
CH2 group fewer than DSS, and thus less prone to provide
interactions with the solutes.[26] Initial experiments demon-
strated that minor, but appreciable, changes in the chemical
shift of the CH2 adjacent to the silicon atom took place
when a 0.1 mm concentration of DSS in D2O was mixed
with a 220 mm concentration of phenol. In view of these ob-
servations, and although the employed DSS concentration
Chem. Eur. J. 2008, 14, 7570 – 7578
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
7571

J. JimØnez-Barbero, K. Bartik et al.
for reference purposes is one order of magnitude lower (ca.
10 mm), it was obvious that DSS was not appropriate for use
as an internal reference for the aromatic interaction experi-
ments. Significantly smaller, but still observable, variations
took place for TSP. Different referencing methods were em-
ployed in the two labs. In Brussels, experiments were under-
taken with an external reference in a spherical insert. The
bulk magnetic susceptibility varies with the solvent, so a
spherical insert, which has a shape factor of zero, was there-
fore essential.[27] In Madrid, TSP was employed as an inter-
nal standard at a 10 mm concentration (see Experimental
Section).
protons of methyl a-glucopyranoside, and even larger ones
between the protons of its b-epimer. The differences in the
proton chemical shift variations are similar for the methyl
a- and b-ribofuranosides.
The differences in the chemical shift variations of the pro-
tons of the methyl a- and b-galactopyranosides were signifi-
cantly larger than for the other sugars, well beyond the max-
imum estimated experimental error. In both sugars H2 is
the proton that experiences the least shielding, followed by
one of the H6 atoms and the O-methyl protons. H3, H4, and
H5 for the a-methyl derivative, and H1, H3, H4, and H5 for
The interaction of methyl gly-
Table 1. Shielding (in Hz) of the protons of methyl pyranosides and furanosides upon addition of phenol or an
aromatic amino acid.
cosides with phenol—chemical
shift perturbation data: Table 1
shows the variations in the
1H NMR resonance frequencies
(Hz) of the signals of a variety
of methyl glycosides in the
presence of variable amounts of
phenol. Shielding of all the res-
onance signals is observed. In
order to allow easy comparison
of the results obtained in the
two laboratories, the sugar
proton that experienced the
smallest shielding was used as
an internal reference (its chemi-
cal shift variation was set to
zero). The experimental errors
in Dn reported in Table 1 are
smaller than 3 Hz (calculated
on the basis of the spectral res-
olution and with comparison of
results obtained from several
experiments on the same
system). No variations in the
chemical shifts of methyl b-gal-
actopyranoside were observed
when the experiments with an
excess of phenol were conduct-
ed in DMSO or acetonitrile,
highlighting the importance of
water for the interaction to
take place.
Sugar
Aromatic
moiety
Chemical shift variation [Dd (Hz)]
H1
H2
H3
H4
H5
H6A
H6B
OMe
methyl a-galacto-
pyranoside
phenol[a]
phenol[b]
phenol[c]
l-Phe[d]
l-Phe[e]
À12
À2
À5
–
À10
À17
À7
À24
À14
À14
–
À26
À16
À15
À3
À12
À2
À6
À1
À2
À12
–
À15
À8
À9
À2
À2
À2
–
À17
À12
À6
À2
À1
À12
À5
À5
À1
À1
–
À16
À6
À7
À1
À2
0
À12
À2
À3
0
À1
À4
–
À10
À3
À4
0
0
0
0
À15
À1
À5
0
À5
À4
À7
l-Trp[f]
–
À5
À0.2
À7
0
0
0
À8
À10
À11
calcd Dd[g]
phenol[a]
phenol[b]
phenol[c]
l-Phe[d]
l-Trp[f]
À0.2
À0.5
À0.6
À2.7
–
methyl b-galacto-
pyranoside
À21
À14
À14
À2
À20
À13
À12
À1
À20
À13
À12
À2
À28
À21
À19
À3
À12
À5
À6
À1
À1
À2
–
À11
À6
À5
À2
À1
À11
À5
À3
À1
À1
–
À6
0
0
À7
À6
À8
0
l-Tyr[h]
À4
À5
À4
À7
À1
calcd Dd[g]
phenol[a]
phenol[b]
phenol[c]
l-Phe[d]
l-Trp[f]
À1.9
À0.3
À0.5
À0.4
À1.7
–
methyl a-gluco-
pyranoside
À13
À8
À10
–
À11
À5
À9
À4
À3
À1
À1
À8
À1
À2
À1
À1
À0.2
À3
À11
À6
À5
À1
À1
À20
À13
À11
À2
–
À11
À6
À4
À1
À1
À11
À4
À4
À1
0
À6
À5
À2
À2
À7
0
0
–
0
0
0
0
0
À9
À2
À3
0
–
methyl b-gluco-
pyranoside
phenol[a]
phenol[b]
phenol[c]
l-Phe[d]
l-Trp[f]
À20
À13
À12
À2
À4
–
calcd Dd[g]
phenol[b]
À2.9
À0.3
À0.4
À0.6
–
À7
methyl a-manno-
pyranoside
methyl a-ribo-
furanoside
methyl b-ribo-
furanoside
À5
À4
0
À1
À3
À1
phenol[a]
phenol[b]
phenol[a]
phenol[b]
l-Phe[d]
À24
À10
À21
À6
À22
À20
À16
À19,À20
À6,À7
À20,À21
À6
–
–
–
–
–
–
–
–
–
–
À16
0
À21
À6
0
À8
À20
À7
À21
0
À3
À15
À5
–
À5,À6
–
–
À1,
À1,–
[a] Absolute shielding (Hz) measured for the resonance signals of the sugars (10 mm) upon addition of a 20-
fold excess of phenol, as determined at 600 MHz and 298 K, pH 5–6. DSS is used as external standard.All
other data shown were obtained by using the least shielded proton as reference (its chemical shift variation is
Analysis of the data in
Table 1 clearly shows that the set to zero); [b] relative shielding in the same conditions as above; [c] sugars 15 mm, addition of a 15-fold
excess of phenol, 500 MHz, 298 K, pH 5–6; [d] sugars 5 mm, addition of a 20-fold excess of l-Phe, 600 MHz,
298 K, pH 5–6; [e] sugars 15 mm, addition of an 11-fold excess of l-Phe, 500 MHz, 298 K, pH 6–7; [f] sugars
4 mm, addition of a sevenfold excess of l-Trp, 500 MHz, 298 K, pH 6–7; [g] estimated shielding (ppm) due to
differences in the chemical shift
variations of the protons of a
sugar depend on the chemical
nature of the sugar. Negligible
differences in the variations
were observed between the var-
ious protons of methyl a-man-
nopyranoside. Larger differen-
the presence of an aromatic moiety, calculated by the modified Bovey–Johnson equation[28] for the protons of
both anomers of methyl galactopyranoside and for methyl b-glucopyranoside, according to the geometry calcu-
lated by molecular mechanics calculations. In the case of methyl b-galactopyranoside the shielding calculated
for the so-called 1, 3, 5 and 3, 4, 5 arrangements (see text) was 50:50 averaged. The estimated proportion of
complex was deduced from the averaged ratio of the experimentally observed absolute shifts at 600 MHz and
298 K to those estimated theoretically; [h] methyl b-galactopyranoside 4 mm, addition of an eightfold excess of
l-Tyr, 500 MHz, 298 K, pH 11–12. Although this pH is not physiologically relevant, and is above the pK for
ces were noted between the the different tyrosine acid/basic groups, the data are given for purposes of comparison.
7572
www.chemeurj.org
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2008, 14, 7570 – 7578

Aromatic–Carbohydrate Interactions
FULL PAPER
the b-epimer showed the largest chemical shift variations.
As an example, Figure 1 shows the 1H NMR spectra of
methyl a- and b-galactopyranosides in water and in the
presence of phenol. As can be seen, a remarkable effect on
the chemical shifts of the sugar protons is revealed. The H4
and H5 resonance signals undergo the largest chemical shift
variation. An upfield shift of the H3 signals is also mea-
sured. For the a-epimer, the H1 chemical shift does not
show a significant shielding, in contrast with the observa-
tions for the b-epimer, in which H1 showed remarkable
changes.
methyl b-galactopyranoside on phenol addition up to a 55:1
molar ratio. A 1:1 model under fast-exchange conditions
was fitted to the experimental data by use of a nonlinear
equation for the analysis, yielding a value of about 1mÀ1 for
the binding constant. Similar behavior was observed for all
the non-exchangeable protons of both sugars. However,
since no plateau value was reached during the titrations, the
affinity constant values should be regarded as qualitative, as
a higher limit. Analogous behavior was observed when both
methyl ribofuranosides were titrated with phenol.
Titration experiments were undertaken in order to try to
quantify the interactions of phenol with methyl b-galacto-
pyranoside (10 mm) and methyl a-glucopyranoside (10 mm).
As an example, Figure S1 in the Supporting Information
shows the behavior of the chemical shift of the H3 proton of
The interaction of methyl glycosides with aromatic amino
acids—chemical shift perturbation data: In a further step,
the interactions between sugars and aromatic amino acids
were studied. Obviously, the CH–p interactions that occur
in the active sites of lectins are mediated by the lateral
chains of the aromatic amino acids, and so we decided to in-
vestigate the interactions of methyl a- and b-galactopyrano-
side and methyl a- and b-glucopyranoside with the naturally
occurring aromatic amino acids l-phenylalanine (Phe), l-ty-
rosine (Tyr), and l-tryptophan (Trp). The NMR experiments
were carried out at different pH values, and the results were
found to be independent of pH. Table 1 shows the variation
in the resonance frequencies of the different sugars in the
presence of Phe, Tyr, and Trp. The most significant differen-
ces between the chemical shift variations are observed for
the methyl galactopyranoside epimers. For the methyl b-
epimer, for instance, although the observed trend is similar
for the three aromatic moieties, the major effect on the
sugar resonances occurs when Trp is added to the monosac-
charide solution. Because only lower aromatic/sugar ratios
were accessible, because of the solubility limit of the aro-
matic amino acids, the observed chemical shift variations
are smaller than those observed in the presence of phenol.
Because the CH–p interactions are probably a conse-
quence of the dispersion of the electronic density of p-mo-
lecular orbitals of the aromatic rings interacting with the hy-
drophobic faces of carbohydrates, we also examined a mix-
ture of methyl a-galactopyranoside and hexafluorobenzene,
a strongly deactivated aromatic ring because of the electron-
withdrawing capacity of the six fluorine atoms. The NMR
spectrum of the methyl a-galactopyranoside/hexafluoroben-
zene mixture did not show any significant differences from
the spectrum of the sugar in the absence of the aromatic
moiety. The data clearly indicate that p-electron-rich aro-
matic rings seem to be required in order to establish stabiliz-
ing CH–p interactions with the carbohydrates.[4]
All the experimental data presented above suggest that
À
three C H vectors pointing into the same spatial region are
required in order to produce significant deviation in the ob-
served NMR chemical shifts. This is clearly highlighted by
the data pertaining to the methyl a- and b-galactopyrano-
sides. Phenol—and also the aromatic amino acids—interacts
with the pyranose chairs in a specific (although weak)
manner, by establishing CH–p interactions. Indeed, for the
1
Figure 1. A) 500 MHz H NMR spectra (D2O, 298 K) of a methyl a-galac-
topyranoside solution (a, above) in the presence of 0.75 equivalent of
phenol (b, middle panel), and in the presence of 6.5 equivalent of phenol
(c, below). B) 600 MHz 1H NMR spectra (D2O, 298 K) of a 10 mm solu-
tion of methyl b-galactopyranoside (above) and upon addition of 21.2
molar equivalent in phenol (below).
À
interaction to take place, three C H vectors of the sugar
with suitable orientations need to exist.
Chem. Eur. J. 2008, 14, 7570 – 7578
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
7573

J. JimØnez-Barbero, K. Bartik et al.
The chemical shifts of the sugar protons were shifted up-
field in the presence of the studied aromatic moieties. These
observed perturbations are the result of the effect of aro-
matic ring currents[28] and consequently yield information on
the geometry of interaction (see below).
Further demonstration of the existence of sugar–aromatic
complexes—intermolecular solvent–sugar and aromatic–
sugar NOEs: Several studies have previously underlined the
importance of the solvent for the formation of carbohy-
drate–protein complexes.[29,30] Consequently, a preliminary
examination of the solvation state of the sugars in the pres-
ence of phenol or benzene was undertaken. We expected
that the solvation spheres of carbohydrates would be modi-
fied somehow when the aromatic rings interacted with the
hydrophobic faces of the sugars. Examples of NMR-based
studies of the solvation of small organic molecules, based on
the detection of intermolecular NOEs between solvent and
solute, have been reported.[31,32] In particular, the solvent
compositions of the first solvation shells of carbohydrates in
binary mixtures have been studied by detection of intermo-
lecular NOEs from the solvent to the solute[31b] or vice
versa.[33]
Figure 2. Build-up curves of the experimentally determined intermolecu-
lar homonuclear NOE enhancements measured between water and pro-
tons of methyl b-galactopyranoside (10 mm), as a function of mixing time
in a water/phenol mixture at 20:1 phenol/sugar molar ratio. Different
slopes are evident for the different protons, in contrast to the observa-
tions in the absence of phenol.
To identify interactions between water and the CH groups
of methyl b-galactopyranoside, 1D-DPFGSE-NOE experi-
ments[34] were performed on samples of the sugar with and
without phenol (phenol/methyl b-galactopyranoside molar
ratio 20:1). A semiquantitative analysis of the NOE build-
up curves for the different protons of the free sugar in
water, by the initial build-up rate method,[35] showed that
the slopes of the curves at short mixing times were basically
identical, suggesting that methyl b-galactopyranoside is es-
sentially solvated in an isotropical manner when dissolved in
water. In contrast, strikingly different results were deduced
for the same sugar protons (Figure 2 and Figure S2 in the
Supporting Information) when the sugar was dissolved in a
water/phenol mixture. Indeed, H1, H2, and H6 seem to be
more water-exposed than H3, H4, and H5, indicating a cer-
tain degree of protection from the solvent for the latter set
of protons. Although these preliminary data should be con-
sidered with caution, they suggest the formation of a sugar/
aromatic complex, indirectly probing the existence of the
CH–p interaction, which partially protects H3, H4, and H5
from water. This result is also consistent with the reported
preference for phenol to be surrounded by the more hydro-
phobic cosolvent when dissolved in aqueous binary mixtur-
es.[31b]
If the measurement of intermolecular NOE is a valid ap-
proach for extracting information about weak interactions
(i.e., solvent–solute interactions), one might also consider
the detection of sugar/aromatic interactions (i.e., solute–
solute interactions) when the complex is formed. Thus, the
search for intermolecular NOEs between methyl b-galacto-
pyranoside (as model compound, given the lack of overlap-
ping of the key NMR resonance signals) and benzene was
attempted. Intermolecular NOEs between two solutes of
low molecular weight are usually too small to be detected,
unless a large binding constant exists, which is not the case
for the system under study. As a result, no sugar–aromatic
NOEs were detected under standard conditions, with sugar
concentrations of 10–30 mm and a 10–20-fold excess of
phenol. However, as shown above, solvent–solute interac-
tions, which are intrinsically weak, became measurable
thanks to the convergent addition of the single contributions
of the huge number of solvent molecules, which greatly
exceed the number of sugar molecules.[31d] A sample con-
taining a 400 mm concentration of sugar in water saturated
with benzene (solubility ca. 22 mm) was prepared. Different
sugar protons were inverted and the benzene signal was ob-
served. Intermolecular NOEs were indeed observed (Fig-
ure S3 in the Supporting Information), especially when H1
or H4 of the sugar ring were inverted, while no NOE was
observed when H2 was the target-inverted proton. These
observations also point towards the existence of sugar–aro-
matic complexes. The experimentally measured NOE build-
up curve for the methyl b-galactopyranoside/benzene pair is
shown in Figure S4 in the Supporting Information.
A 3D model—theoretical calculations: As additional sup-
port to verify the existence of stabilizing CH–p interactions,
density functional and molecular mechanics calculations
were carried out on simple sugar–aromatic systems, as de-
scribed in our previous study of the fucose/benzene com-
plex.[24,36] Studies on the complex formed by a b-fucopyrano-
side (Fuc) residue interacting with the aromatic ring of tryp-
tophan (Trp) were performed. For purposes of comparison,
and also to judge the influence of deactivation of the aro-
matic ring, calculations for the interaction of a simple sugar
(methyl 2,3,4-tri-O-methyl a-fucopyranoside, FucMe4) with
difluorobenzene (C6H4F2), and hexafluorobenzene (C6F6)
were also performed.
7574
www.chemeurj.org
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2008, 14, 7570 – 7578

Aromatic–Carbohydrate Interactions
FULL PAPER
DFT calculations: Firstly, the previously studied Fuc/ben-
zene system[24] was modified as necessary, for preparing Fuc/
Trp, FucMe4/C6H4F2, and FucMe4/C6F6. In the cases of the
last two complexes it was necessary to substitute the differ-
ent hydroxyl groups of the sugar to avoid hydrogen bonding
of the hydroxyl groups with the fluorine atoms.
the six fluorine atoms attached to the aromatic moiety pro-
duced a strong modification of the electronic density of the
ring, thus precluding the existence of the CH–p interaction.
À
The sugar moiety was displaced from the ring, and C H···F
interactions were observed (see Figure S5 in the Supporting
Information).
Calculations were performed by density functional theory.
Although it is well known that the hybrid B3LYP method[37]
is not fully adequate for study of long-distance interactions,
it was used to obtain approximations of the geometries of
the different complexes. It has been reported that its consid-
eration of the dispersion term is somehow deficient,[38] al-
though recent developments may overcome these draw-
backs.[39]
The obtained geometry for the basic Fuc/Trp complex
(Figure 3A) shows that the hydrogen atoms at the 3-, 4-, and
5-positions are those closest to the aromatic ring, as experi-
mentally deduced for a variety of sugar–aromatic com-
plexes,[23] including those described above, while H1 is the
furthest away. H3 is at 2.95 Š from C’3, the equatorially ori-
ented H4 is at 3.07 Š from the nitrogen atom, whereas H5
is at 3.15 Š from the bridging C’7 atom.
Interaction energies were calculated by consideration of
the BSSE correction. For the Fuc/Trp complex, the mini-
mum-energy conformer was obtained with a stabilizing in-
teraction energy of À1.90 kJmolÀ1. For the complex
FucMe4/C6H4F2, a transition state was obtained, character-
ized by one imaginary frequency (À17.90) and associated
with the motion of approximation of the F1 atom to the C3
methoxyl group, with no change in the 1,3,5-type sugar–aro-
matic interaction. It could be expected that the geometry of
the minimum associated with this transition state should not
change from the reported one (and the same with regard to
the energy). Interestingly, in this case, the interaction energy
was increased to +2.72 kJmolÀ1, corresponding to an endo-
thermic process and thus supporting a destabilizing role of
the fluorine atoms. A similar trend has been experimentally
observed for mutant hevein domains, with non-natural fluo-
rinated aromatic amino acid residues, interacting with oligo-
saccharides.[4] The increment is even bigger when the
FucMe4/C6F6 complex is considered. In this case, the inter-
action energy was strongly unfavorable, at +43.85 kJmolÀ1.
AMBER* molecular mechanics calculations: In a further
step, we decided to resort to molecular mechanics calcula-
tions, which are much less computationally time-consuming
(than the much more time-consuming MP2 used previous-
ly[24]) and because dispersion of electronic density and van
der Waals forces seem to be the key factors responsible for
the interactions. The corresponding terms are properly para-
meterized in the force fields currently used to deal with bio-
molecules, such as AMBER*,[40] as integrated in the MAES-
TRO package.[41] Furthermore, long-distance interactions
are well described by Newtonian Mechanics, employed in
Molecular Mechanics. The results indicate that the distances
are significantly smaller than those observed in the geome-
tries obtained at the B3LYP level, which is in agreement
with the observations relating to the Fuc/Ben complex at
Figure 3. Stationary states of the supramolecular complexes formed from
Fuc/Trp (A), and FucMe4/C6H4F2 (B). The key distances (in Š) from CH
groups of the sugar to the aromatic ring atoms are shown. Calculations
were performed at the B3LYP/6-31GACHTER(UNG d,p) level. According to the nota-
tion, the ring carbon atoms from the aromatic are denoted by primes
(C1’–C6’), while the sugar atoms keep their regular numbering.
(d,p) level of calculation[24] (see Table 2). For
the MP2/6–31GACHTREUNG
the interaction between methyl b-galactopyranoside and
benzene, clear minimum-energy geometries (with very simi-
lar energy values) were obtained for the two possible geo-
metrical arrangements, for which either H1, H3, and H5
(Figure 4B) or H3, H4, H5 (Figure 4C) interact with the
benzene ring. In fact, this last geometry was the only one
provided when the calculations were performed for the ben-
zene/methyl a-galactopyranoside complex (Figure 4A). Ac-
cording to the calculations, two geometries are also possible
for the methyl b-glucopyranoside complex, but in this case,
the interaction may take place through the lower or the
upper face of the pyranose ring, interacting either with H1,
H3, and H5 or with H2, H4, and one of the H6 protons. The
two possible arrangements are shown in Figure 4D.
A different situation is observed for the FucMe4/C6H4F2
complex. In this case, proximity between the axially oriented
protons at the 1-, 3-, and 5-positions of the sugar ring and
the aromatic ring is observed (Figure 3B) but the obtained
distances are clearly longer. The sugar H1 is located at
3.36 Š from carbon C’1, whereas H3 and H5 are at 3.28 Š
and 3.52 Š from C’3 and C’4, respectively. The obtained dis-
tances are also longer than those previously calculated for
the Fuc/benzene complex—3.21, 3.09, and 3.28 Š, respec-
tively—at the same theory level. In principle, this could be
explained by the deactivating role of the fluorine atoms,
which was confirmed when the geometry of the FucMe4/
C6F6 complex was calculated. In this case, the presence of
Chem. Eur. J. 2008, 14, 7570 – 7578
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
7575

J. JimØnez-Barbero, K. Bartik et al.
Table 2. Calculated distances (Š) between selected protons of the sugar
moiety (1,3,5 and 3,4,5 CH-arrangements) and the aromatic ring carbons.
[a]
Distance
FucMe4/C6H4F2
B3LYP AMBER*
Fuc/C6H6
B3LYP
AMBER*
MP2
H1ax–C’1
H1ax–C’5
H1ax–C’6
H3ax–C’2
H3ax–C’3
H5ax–C’4
H5ax–C’5
H4eq–C’4
3.363
3.997
3.572
3.429
3.279
3.516
3.626
4.831
3.041
2.906
2.938
2.982
2.961
3.310
4.060
5.308
3.212
3.639
4.176
3.294
3.087
3.396
3.280
4.207
3.641
4.882
4.269
2.916
2.956
3.042
2.905
3.021
3.150
4.556
3.896
2.894
2.896
2.793
2.735
3.229
[a] For purposes of comparison, distances for the Fuc/benzene complex
from ref [24] are included. 6-Deoxygalactose (fucose) was used to cir-
À
cumvent the possibility of rotamers around the C5 C6 bond.
Figure 4. Minimum-energy geometries obtained by molecular mechanics
calculations for (from left to right) methyl a-galactopyranoside/benzene
(A), methyl b-galactopyranoside/benzene with the aromatic ring interact-
ing with H1, H3, and H5 (B), methyl b-galactopyranoside/benzene with
the aromatic ring interacting with H3, H4, and H5 (C), and methyl b-glu-
copyranoside with two benzene moieties (D). Calculations were per-
formed with AMBER*, as implemented in the MAESTRO package.
Figure 5. Minimum-energy geometry structures of the supramolecular
complexes made up by Fuc/Trp (A) and FucMe4/C6H4F2 (B). The key
distances (in Š) from CH groups of the sugar to the aromatic ring are
shown. Molecular Mechanics calculations were performed with the
AMBER* force field. The ring carbon atoms from the aromatic are de-
noted by primes (C1’–C6’), while the sugar atoms keep their regular
numbering.
For less simple aromatic systems, such as the Fuc/Trp
complex (Figure 5A), the sugar–aromatic distances are
clearly shorter than those obtained by the B3LYP calcula-
tions. The H3–C’3 atom pair is separated by 2.83 Š, H4eq–
N1 and H5–C’7 by only 2.96 and 2.77 Š, respectively. The
AMBER* energy estimation for this stable complex is
À23.28 kJmolÀ1. In the case of the FucMe4/C6H4F2 complex
(Figure 5B), H1 is located 2.91 Š from the closest aromatic
carbon atom, whereas H3 is 2.96 and 2.98 Š from C’3 and
C’2, respectively. H5 axial is the furthest away, at 3.31 Š
from C’4. The AMBER-based interaction energy for this
complex was heavily positive, at +128.87 kJmolÀ1.
the semiquantitative character of the model, are taken into
consideration, the agreement can be considered as satisfac-
tory. If the calculated shieldings are compared with those
observed experimentally, the estimated proportion of com-
plex in solution for the two methyl galactopyranoside sam-
ples containing a 1:20 sugar/phenol ratio should be around
10%.
The geometries obtained from the molecular mechanics
calculations for the methyl a- and b-galactopyranosides
were used to calculate, with the aid of the MOLMOL pro-
gram, the expected ring current shifts from the Johnson and
Bovey equation.[28a] Johnson and Bovey calculated the
shielding increment in the space region around benzene
using a current loop model, and in the program the equation
takes account of the fact that the ring current is different for
each amino acid.[42] Calculations were performed with the
parameters for tyrosine. A comparison between the estimat-
ed and calculated data is given in Table 1. The computed
data are in agreement with the experimental observations:
the experimentally most shielded protons in the NMR spec-
tra are those that are also predicted to experience the maxi-
mum shielding. If the weakness of the complexes, as well as
Conclusion
The study of the interactions between carbohydrates and
proteins is a field of current interest. Aromatic amino acids
are nearly always involved in these interactions. In order to
gain insight relating to this experimental observation, we
have studied model systems using different monosaccharides
and different aromatic moieties. We showed that, depending
on the nature of the sugar and the stereochemistry of the
asymmetric carbon atoms, simple carbohydrates may inter-
act with aromatic moieties, including aromatic amino acids,
although with rather low affinity constants (below 1mÀ1). A
hydrophobic component to this interaction is detected, since
7576
www.chemeurj.org
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2008, 14, 7570 – 7578

Aromatic–Carbohydrate Interactions
FULL PAPER
For the experiments with the aromatic amino acids, the same method
was used, except that the 1D 1H NMR spectra were recorded on a
400 MHz Varian spectrometer with a digital resolution of 1.25 Hz per
point before zero-filling, in 10 mm tubes and with an external reference
(DSS) in a Wilmad spherical bulb (5 mm), which was maintained in the
center of the detection area.
the interaction cannot be detected in other polar solvents
(DMSO, acetonitrile). Provided that three sugar protons
point towards the same spatial region, the CH–p interaction
indeed exists and can be detected by simple NMR experi-
ments in water. Both theoretical calculations and experi-
mental NMR data support this interaction model. In fact,
the corresponding sugar ring protons experience shielding
due to a relative orientation that places these protons above
the aromatic system. As it is well known that carbohydrate-
binding proteins have high selectivities and high affinities
for their carbohydrates (103 mÀ1 to 107 mÀ1, depending on the
complexity of the carbohydrate), these results show that
more than one type of interaction is necessary in order to
achieve a high affinity constant and to observe appropriate
selectivity towards the distinct carbohydrates. Spatial organi-
zation of multiple aromatic entities may be responsible for
the selectivity of proteins towards carbohydrates, together
with additional interactions (i.e., hydrogen bonding) be-
tween the amino acids and the carbohydrates.
In Madrid: The NMR spectra were recorded at 500 MHz on a Bruker
Avance spectrometer at 298 K. The spectra were processed by use of
Topspin software (Bruker, Inc.). For all the experiments, the high-field
resonance of [D4](trimethylsilyl)propionic acid sodium salt (TSP, 10 mm)
was used as an internal chemical shift reference. All the samples were
prepared as mixtures of deuterated and normal water (10:90) as the over-
all solvent composition. Sample pH values were kept around 8 and tested
with a thin electrode (Wilmad) fitted directly in a 5 mm NMR tube. The
sample concentrations were 4–20 mm in carbohydrate, and the aromatic/
sugar molar ratio varied from 1:1 to 20:1. The solutions were not de-
gassed. One-dimensional high-resolution experiments were recorded with
32k complex data points, and 16–32 scans were collected at a spectral
width of 4500 Hz. Water suppression was accomplished by use of the
WATERGATE pulse sequence.[43] The original FID was zero-filled to
64 k, and Fourier transformation with use of an exponential window func-
tion was applied (exponential multiplication, lb=1 Hz).
The DPFGSE-NOE method was used for the 1D intermolecular NOE
measurements with subsequent solvent suppression as described else-
where.[34] The measurements were recorded at different mixing times
from 50 ms to 2 s with a relaxation delay of 30 s and typically 1 K transi-
ents.
Experimental Section
Theoretical calculations: The calculations were performed on a “HP
Cluster Superdome” computer, at the Supercomputing Center of Galicia,
Spain (CESGA).
Full geometry optimizations were done with Gaussian98[44] for Fuc/Trp,
FucMe4/C6H4F2, and FucMe4/C6F6 complexes with use of Density Func-
tional Theory (DFT), at the B3LYP/6-31GCAHTRE(UNG d,p) level. Vibrational fre-
quency calculations were done in order to characterize the nature of the
stationary points. To determine the interaction energy precisely, basis set
superposition error correction (BSSE)[45] was calculated. The counter-
poise method proposed by Boys and Bernardi was used,[46] so the proper
correction for changes in geometry of the components of the complex
was considered.
All the sugars are from the d series, except the l-fucose. All O-methylat-
ed sugars, except for the ribose derivatives, as well as the l-amino acids
were purchased from Aldrich, with purities higher than 98%. Deuterated
water 99.9% (D2O) was purchased from Cambridge Isotope Laboratories
(CIL). The ribose derivatives were synthesized by the protocol described
in the Supporting Information.
NMR—general aspects
In Brussels: Stock solutions of the different carbohydrates, phenol, and
the amino acids were prepared in D2O. The pH of each solution was ad-
justed by addition of small amounts of concentrated DCl or NaOD and
measured with a thin electrode (Wilmad) fitted directly in the NMR tube
or in the preparation test-tube. No corrections were made for isotopic ef-
fects. Typical experimental conditions used were sugar concentrations be-
tween 2.5 and 10 mm, 20 molar excesses of aromatic moiety, pH between
5 and 6 (far enough from the pKa of phenol (9.95), the pKa values of the
carbohydrate hydroxyl groups (16), and the pKa values of the different
functions of the amino acids). Experiments with the amino acids at pH
between 11 and 12 were also performed for purposes of comparison. For
titration experiments, a separate NMR tube was prepared for each titra-
tion point with use of the stock solutions. The carbohydrate concentra-
tion was kept constant throughout each titration (1 or 10 mm, depending
on the sugar). Phenol concentration was varied between 0 and 650 mm
(and verified by comparison of the integrals of the phenol protons with
those of the carbohydrate protons).
Moreover, molecular mechanics minimizations were performed with the
AMBER* force field (as implemented in the Maestro Program[41]).
Acknowledgements
The group in Madrid thanks the Ministry of Education and Science of
Madrid for funding (Grant CTQ2006-10874-C02-01) and the CESGA for
computing time. The group in Brussels thanks the “CommunautØ fran-
Åaise de Belgique” for funding (ARC2002-7). D.D. thanks the Ramón y
Cajal programme for her contract. S.V.D.B. thanks the Belgian FNRS for
a PhD fellowship. J.J.B. and S.P.V. thank the European Commission
through the MRTN-CT2005-019561 DYNAMIC project.
For the experiments with phenol, the 1H NMR experiments were ac-
quired at 298 K on a 600 MHz Varian spectrometer with a digital resolu-
tion of 0.35 Hz per point before zero-filling, with 5 mm high-resolution
tubes. In each set of experiments, the 908 pulse was adjusted, and the re-
laxation delay was set so as to ensure a minimum of 95% relaxation be-
tween two acquisitions. 16 to 512 scans were recorded, depending on the
sample concentration. Two levels of zero-filling were used for the data
processing. DSS was used as an internal reference, but its chemical shift
was adjusted according to an external reference through a calibration ex-
periment: 1D 1H NMR spectra were recorded at 298 K on a 400 MHz
Varian spectrometer with a digital resolution of 1.25 Hz per point before
zero-filling, with 10 mm tubes and an external reference (methanol) in a
Wilmad spherical bulb (5 mm), which was maintained in the center of
the detection area. The chemical shift of DSS present in the outer area
was recorded as a function of phenol concentration.
[1] H.-J. Gabius, H.-C. Siebert, S. Andre, J. JimØnez-Barbero, H. Rudig-
er, ChemBioChem 2004, 5, 740–764.
[2] H. Lis, N. Sharon, Chem. Rev. 1998, 98, 637–674.
[3] a) J. V. Pratap, A. A. Jeyaprakash, P. G. Rani, K. Sekar, A. Surolia,
M. Vijayan, J. Mol. Biol. 2002, 317, 237–247; b) D. Solís, P. Fernµn-
dez, T. Díaz-MauriÇo, J. JimØnez-Barbero, M. Martín-Lomas, Eur. J.
Biochem. 1993, 214, 677–683.
[4] a) M. S. Sujatha, P. V. Balaji, Proteins 2004, 55, 44–65; b) J. L. Asen-
sio, F. J. CaÇada, H. C. Siebert, J. Laynez, A. Poveda, P. M. Nieto,
U. M. Soedjanaamadja, H. J. Gabius, J. JimØnez-Barbero, Chem.
Biol. 2000, 7, 529–543; c) J. L. Asensio, H. C. Siebert, C. W. von der
Lieth„ J. Laynez, M. Bruix, U. M. Soedjanaamadja, J. J. Beintema,
F. J. CaÇada, H. J. Gabius, J. JimØnez-Barbero, Proteins 2000, 40,
Chem. Eur. J. 2008, 14, 7570 – 7578
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
7577

J. JimØnez-Barbero, K. Bartik et al.
218–236; d) N. Aboitiz, M. Vila-Perello, P. Groves, J. L. Asensio, D.
Andreu, F. J. CaÇada, J. JimØnez-Barbero, ChemBioChem 2004, 5,
1245–1255; e) M. I. Chavez, C. Andreu, P. Vidal, N. Aboitiz, F.
Freire, P. Groves, J. L. Asensio, G. Asensio, M. Muraki, F. J. CaÇada,
J. JimØnez-Barbero, Chem. Eur. J. 2005, 11, 7060–7074.
[5] M. Nishio, M. Hirota, Y. Umezawa, The CH–Pi Interaction: Evi-
dence, Nature, and Consequences, Wiley-VCH, New York, 1998.
[6] M. Muraki, K. Harata, N. Sugita, K. Sato, Biochemistry 2000, 39,
292–299.
[7] T. Ponyi, L. Szabo, T. Nagy, L. Orosz, P. J. Simpson, M. P. William-
son, H. J. Gilbert, Biochemistry 2000, 39, 985–991.
[8] F. Fusetti, T. Pijning, K. H. Kalk, E. Bos, B. W. Dijkstra, J. Biol.
Chem. 2003, 278, 37753–37760.
[9] D. Solís, H. Kogelberg, J. JimØnez-Barbero, Curr. Opin. Struct. Biol.
2003, 13, 646–653.
[10] K. K. S. Ng, A. R. Kolatkar, S. Park-Snyder, H. Feinberg, D. A.
Clark, K. Drickamer, W. I. Weis, J. Biol. Chem. 2002, 277, 16088–
16095.
[11] G. Colombo, M. Meli, J. CaÇada, J. L. Asensio, J. JimØnez-Barbero,
Carbohydr. Res. 2004, 339, 985–994.
[12] K. Maenaka, G. Kawai, K. Watanabe, F. Sunada, I. Kumagai, J. Biol.
Chem. 1994, 269, 7070–7075.
[28] a) F. Bovey, Nuclear Magnetic Resonance Spectroscopy, Academic
Press, 1969; b) R. J. Abraham, M. Reid, J. Chem. Soc. Perkin Trans.
2, 2002, 1081–1091; c) N. H. Martin, N. W. Allen, J. C. Moore, J.
Mol. Graphics Mod. 2000, 18, 242–246.
[29] K. N. Kirschner, R. J. Woods, Proc. Natl. Acad. Sci. USA 2001, 98,
10541–10545.
[30] C. Clarke, R. J. Woods, J. Gluska, A. Cooper, M.A. Nutley, G.-J.
Boons, J. Am. Chem. Soc. 2001, 123, 12238–12247.
[31] a) T. Brand, E. J. Cabrita, S. Berger, Prog. Nucl. Magn. Reson. Spec-
trosc. 2005, 46, 159–196; b) A. Bagno, J. Phys. Org. Chem. 2002, 15,
790–795; c) A. Bagno, M. Campulla, M. Pirana, G. Scorrano, S. Stiz,
Chem. Eur. J. 1999, 5, 1291–1300; d) A. Bagno, F. Rastrelli, G.
Saielli, Prog. Nucl. Magn. Reson. Spectrosc. 2005, 47, 41–93.
[32] a) A. Guerrero-Martinez, S. Berger, G. Tardajos, ChemPhysChem
2006, 7, 2074–2076; b) M. Angulo, C. Hawat, H.-J. Hofmann, S.
Berger, Org. Biomol. Chem. 2003, 1, 1049–1052; c) S. Berger, M. D.
Díaz, Ch. Hawat, Pol. J. Chem. 1999, 73, 193–197.
[33] A. Bagno, F. Rastrelli, G. Scorrano, J. Magn. Reson. 2004, 167, 31–
35.
[34] M. D. Díaz, S. Berger, Magn. Reson. Chem. 2001, 39, 369–373.
[35] D. Neuhaus, M. P. Williamson, The Nuclear Overhauser Effect in
structural and conformational analysis, VCH, Weinheim, 1989.
[36] J. JimØnez-Barbero, F. J. CaÇada, G. Cuevas, J. L. Asensio, N. Aboi-
tiz, A. Canales, M. I. Chavez, M. C. Fernµndez-Alonso, A. Garcia-
Herrero, S. Mari, P. Vidal, in NMR Spectroscopy and Computer
Modeling of Carbohydrates: Recent Advances. ACS Symposium
Series 2006, 930, 60–80.
[13] M. S. Sujatha, Y. U. Sasidhar, P. V. Balaji, Biochemistry 2005, 44,
8554–8562.
[14] V. Spiwok, P. Lipovova, T. Skalova, E. Vondrackova, J. Dohnalek, J.
Hasek, B. Kralova, J. Comput.-Aided Mol. Des. 2006, 19, 887–901.
[15] E. C. Stanca-Kaposta, D. P. Gamblin, J. Screen, B. Liu, L. C. Snoek,
B. G. Davis, J. P. Simons, Phys. Chem. Chem. Phys. 2007, 9, 4444–
4451.
[16] S. Tsuzuki, K. Honda, T. Uchimaru, M. Mikami, K. Tanabe, J. Am.
Chem. Soc. 2000, 122, 3746–3753.
[17] S. Tsuzuki, K. Honda, T. Uchimaru, M. Mikami, A. Fujii, J. Phys.
Chem. A 2006, 110, 10163–10168.
[18] T. Steiner, Angew. Chem. 2002, 114, 50–80; Angew. Chem. Int. Ed.
2002, 41, 49–76.
[19] V. S. R. Rao, K. Lam, P. K. Qasba, Int. J. Biol. Macromol. 1998, 23,
295–307.
[20] V. Spiwok, P. Lipovova, T. Skalova, E. Buchtelova, J. Hasek, B. Kra-
lova, Carbohydr. Res. 2004, 339, 2275–2280.
[21] D. Neumann, C.-M. Lehr, H.-P. Lenhof, O. Kohlbacher, Adv. Drug
Delivery Rev. 2004, 56, 437–457.
[37] A. D. Becke, J. Chem. Phys. 1993, 98, 5648–5652.
[38] a) J. Sponer, K. Leszczynski, P. Hobza, J. Comput. Chem. 1996, 17,
841–850; b) P. Hobza, J. Sponer, Chem. Rev. 1999, 99, 3247–3276.
[39] Y. Zhao, D. G. Truhlar, J. Chem. Theory Comput. 2007, 3, 289–300.
[40] D. A. Pearlman, D. A. Case, J. W. Caldwell, W. S. Ross, T. E. Cheat-
ham, S. DeBolt, D. Ferguson, G. Siebal, P. A. Kollmann, Comput.
Phys. Commun. 1995, 91, 1–41.
[41] MacroModel, version 9.1, Schrçdinger, LLC, New York, NY, 2005.
[42] D. A. Case, J. Biomol. NMR 1995, 6, 341–346.
[43] M. Piotto, V. Saudek, V. Sklenµr, J. Biomol. NMR 1992, 2,661–665.
[44] Gaussian 98 (Revision A.11.3), M. J. Frisch, G. W. Trucks, H. B.
Schlegel, G. E. Scuseria, M. A. Robb, J. R. Cheeseman, V. G. Zakr-
zewski, J. A. Montgomery, R. E. Stratmann, J. C. Burant, S. Dap-
prich, J. M. Millam, A. D. Daniels, K. N. Kudin, M. C. Strain, O.
Farkas, J. Tomasi, V. Barone, M. Cossi, R. Cammi, B. Mennucci, C.
Pomelli, C. Adamo, S. Clifford, J. Ochterski, G. A. Petterson, P. Y.
Ayala, Q. Cui, K. Morokuma, D. K. Malick, A. D. Rabuck, K. Ra-
ghavachari, J. B. Foresman, J. Cioslowski, J. V. Ortiz, B. B. Stefanov,
G. Liu, A. Liashenko, P. Piskorz, I. Komaromi, R. Gomperts, R. L.
Martin, D. J. Fox, T. Keith, M. A. Al-Laham, C. Y. Peng, A. Na-
nayakkara, C. Gonzalez, M. Challacombe, P. M. W. Gill, B. G. John-
son, W. Chen, M. W. Wong, J. L. Andres, M. Head-Gordon, E. S.
Replogle, J. A. Pople, Gaussian, Inc., Pittsburgh, PA, 1998
[45] S. Simon, M. Duran, J. J. Dannenberg, J. Chem. Phys. 1996, 105,
11024–11031.
[22] M. S. Sujatha, Y. U. Sasidhar, P. V. Balaji, Protein Sci. 2004, 13,
2502–2514.
[23] a) For the 3D-lectin database, see Lectines (http://webenligne.cer-
mav.cnrs.fr/lectines); b) For the Protein Data Bank, see H. M.
Berman, J. Westbrook, Z. Feng, G. Gilliland, T. N. Bhat, H. Weissig,
I. N. Shindyalov, P. E. Bourne, Nucleic Acids Res. 2000, 28, 235–242;
c) R. Loris, R. H. De Greve, M.-H. Dao-Thi, J. Messens, A. Imberty,
L. Wyns, J. Mol. Biol. 2000, 301, 987–1002.
[24] M. C. Fernµndez-Alonso, F. J. CaÇada, J. JimØnez-Barbero, G.
Cuevas, J. Am. Chem. Soc. 2005, 127, 7379–7386.
[25] Y. Ferrand, M. P. Crump, A. P. Davis, Science 2007, 318, 619–622.
[26] A. Shimizu, M. Ikeguchi, S. Sugai, J. Biomol. NMR, 1994, 4, 859–
862.
[46] S. F. Boys, F. Bernardi, Mol. Phys. 1970, 19, 553–566.
[27] H. Günther, NMR spectroscopy—Basic principles, concepts, and ap-
plications in chemistry, Wiley, New York, 1995, pp. 56–59.
Received: February 8, 2008
Published online: May 15, 2008
7578
www.chemeurj.org
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2008, 14, 7570 – 7578