Combining weak affinity chromatography, NMR spectroscopy and molecular simulations in carbohydrate-lysozyme interaction studies

Article abstract of DOI:10.1039/c2ob07066a

By examining the interactions between the protein hen egg-white lysozyme (HEWL) and commercially available and chemically synthesized carbohydrate ligands using a combination of weak affinity chromatography (WAC), NMR spectroscopy and molecular simulations, we report on new affinity data as well as a detailed binding model for the HEWL protein. The equilibrium dissociation constants of the ligands were obtained by WAC but also by NMR spectroscopy, which agreed well. The structures of two HEWL-disaccharide complexes in solution were deduced by NMR spectroscopy using ¹H saturation transfer difference (STD) effects and transferred ¹H,¹H-NOESY experiments, relaxation-matrix calculations, molecular docking and molecular dynamics simulations. In solution the two disaccharides β-d-Galp-(1→4) -β-d-GlcpNAc-OMe and β-d-GlcpNAc-(1→4)-β-d-GlcpNAc-OMe bind to the B and C sites of HEWL in a syn-conformation at the glycosidic linkage between the two sugar residues. Intermolecular hydrogen bonding and CH/π-interactions form the basis of the protein-ligand complexes in a way characteristic of carbohydrate-protein interactions. Molecular dynamics simulations with explicit water molecules of both the apo-form of the protein and a ligand-protein complex showed structural change compared to a crystal structure of the protein. The flexibility of HEWL as indicated by a residue-based root-mean-square deviation analysis indicated similarities overall, with some residue specific differences, inter alia, for Arg61 that is situated prior to a flexible loop. The Arg61 flexibility was notably larger in the ligand-complexed form of HEWL. N,N′-Diacetylchitobiose has previously been observed to bind to HEWL at the B and C sites in water solution based on ¹H NMR chemical shift changes in the protein whereas the disaccharide binds at either the B and C sites or the C and D sites in different crystal complexes. The present study thus highlights that protein-ligand complexes may vary notably between the solution and solid states, underscoring the importance of targeting the pertinent binding site(s) for inhibition of protein activity and the advantages of combining different techniques in a screening process. The Royal Society of Chemistry 2012.

Full text of DOI:10.1039/c2ob07066a

View Article Online / Journal Homepage / Table of Contents for this issue
Dynamic Article Links
Organic &
Biomolecular
Chemistry
Cite this: Org. Biomol. Chem., 2012, 10, 3019
www.rsc.org/obc
PAPER
Combining weak affinity chromatography, NMR spectroscopy and molecular
simulations in carbohydrate–lysozyme interaction studies†
Jens Landström,a Maria Bergström,b Christoffer Hamark,a Sten Ohlsonb and Göran Widmalm*a
Received 9th December 2011, Accepted 1st February 2012
DOI: 10.1039/c2ob07066a
By examining the interactions between the protein hen egg-white lysozyme (HEWL) and commercially
available and chemically synthesized carbohydrate ligands using a combination of weak affinity
chromatography (WAC), NMR spectroscopy and molecular simulations, we report on new affinity data as
well as a detailed binding model for the HEWL protein. The equilibrium dissociation constants of the
ligands were obtained by WAC but also by NMR spectroscopy, which agreed well. The structures of two
HEWL–disaccharide complexes in solution were deduced by NMR spectroscopy using 1H saturation
transfer difference (STD) effects and transferred 1H,1H-NOESY experiments, relaxation-matrix
calculations, molecular docking and molecular dynamics simulations. In solution the two disaccharides
β-D-Galp-(1→4)-β-D-GlcpNAc-OMe and β-D-GlcpNAc-(1→4)-β-D-GlcpNAc-OMe bind to the B and C
sites of HEWL in a syn-conformation at the glycosidic linkage between the two sugar residues.
Intermolecular hydrogen bonding and CH/π-interactions form the basis of the protein–ligand complexes
in a way characteristic of carbohydrate–protein interactions. Molecular dynamics simulations with explicit
water molecules of both the apo-form of the protein and a ligand–protein complex showed structural
change compared to a crystal structure of the protein. The flexibility of HEWL as indicated by a residue-
based root-mean-square deviation analysis indicated similarities overall, with some residue specific
differences, inter alia, for Arg61 that is situated prior to a flexible loop. The Arg61 flexibility was notably
larger in the ligand-complexed form of HEWL. N,N′-Diacetylchitobiose has previously been observed to
bind to HEWL at the B and C sites in water solution based on 1H NMR chemical shift changes in the
protein whereas the disaccharide binds at either the B and C sites or the C and D sites in different crystal
complexes. The present study thus highlights that protein–ligand complexes may vary notably between
the solution and solid states, underscoring the importance of targeting the pertinent binding site(s) for
inhibition of protein activity and the advantages of combining different techniques in a screening process.
chromatography with affinity chromatography.1 It was shown
Introduction
that an efficient separation system could be obtained by immobi-
lizing low affinity ligands (KD > 10−5 M) onto microparticulate
Studies of biomolecular interactions are of great importance in
order to understand biological phenomena and for guidance in
the quest for new drugs. A variety of different biophysical
methods are available for research involving molecular inter-
actions. The different methods have their own advantages and
disadvantages, making combinations of techniques an attractive
approach to extract reliable data in an effective manner.
porous silica. The WAC technique has been shown to be a valu-
able tool especially in characterizing protein–carbohydrate inter-
actions since these often are of low affinity.2–4 Normally the
protein (for example an antibody, lectin or toxin) is immobilized
and small amounts of carbohydrate solutions are injected into
the column. From chromatography theory the retention volume
of an injected substance should be directly related to the affinity
and the number of binding sites under standardized conditions.5
The KD value of the interaction can therefore be easily extracted
from the retention in combination with the number of binding
sites, or by comparing the retention in relation to a compound of
known affinity. The advantages with the WAC technique include
low consumption of analyte (injected substance) and short analy-
sis time, since KD might be obtained from a single injection at
low concentration. Another benefit is that racemates, sample
Weak affinity chromatography (WAC) was introduced about
twenty years ago, combining high performance liquid
aDepartment of Organic Chemistry, Arrhenius Laboratory, Stockholm
University, S-106 91 Stockholm, Sweden. E-mail: gw@organ.su.se
bSchool of Natural Sciences, Linnaeus University, S-391 82 Kalmar,
Sweden
†Electronic supplementary information (ESI) available. Experimental
details, characterizations and additional results. See DOI: 10.1039/
c2ob07066a
This journal is © The Royal Society of Chemistry 2012
Org. Biomol. Chem., 2012, 10, 3019–3032 | 3019

View Article Online
Fig. 1 The binding cleft of HEWL is divided into six subsites depicted
as A–F. Each subsite can accommodate one monosaccharide unit. The
hydrolysis site is situated between subsites D and E. The HEWL struc-
ture is from ref. 23.
mixtures and impure substances can be analyzed, and that the
KD of the individual components in the mixture can be deter-
mined concomitantly.
During the past decade, NMR spectroscopy has become an
important tool in molecular interaction studies.6 A major advan-
tage is that the molecules can be studied label-free in solution.
NMR spectroscopy gives valuable dynamic as well as structural
information of both the target protein and the ligand on an
atomic level. In order to predict, understand and interpret exper-
imental results it is often necessary to implement computational
methods. Molecular docking simulations are being used to
predict the structure of intermolecular complexes7,8 and are often
used to screen libraries of compounds in silico. Docking gives a
static picture of the binding event and if information on dynamic
properties is required, more computationally demanding molecu-
lar dynamics (MD) simulations can be carried out.9,10 MD simu-
lations give information about the intra- and intermolecular
motions of molecules over a time period. Results from docking
and MD simulations can be translated to NMR spectroscopy
data by relaxation-matrix calculations11,12 and compared to
experimental results. By combining these two experimental
methods, WAC and ligand detected NMR spectroscopy, with
molecular docking, relaxation-matrix calculations and molecular
dynamics simulations, the high throughput characteristics of
WAC can be complemented with qualitative and quantitative
information on an atomic level.
Hen egg-white lysozyme (HEWL) is being used as a model
protein in a great number of structural13–18 and mechanistic19,20
studies. HEWL is known to catalyze the hydrolysis of the glyco-
sidic bond of the β-(1→4)-linkage between N-acetyl-muramic
acid (MurNAc) and N-acetyl-D-glucosamine (GlcNAc) in pepti-
doglycans that constitute the cell wall of Gram-positive bacteria.
It is also known to hydrolyze the β-(1→4)-linkages in chitin.21
HEWL has a large binding cleft that can be divided into six sub-
sites, A–F, each one can contain one monosaccharide unit and
the hydrolysis is known to take place between subsites D and E
(see Fig. 1).19 Furthermore, HEWL binds chitin polymers of
different lengths and it has been reported that GlcNAc binds to
subsites C and E,22 N,N′-diacetylchitobiose in the CD site23 and
chitotriose in the BCD as well as in the ABC sites.23
Fig. 2 Flowchart for the generation of HEWL–ligand structure com-
plexes in solution.
We have chosen HEWL as a model system to investigate
protein–ligand interactions with a combination of weak affinity
chromatography (WAC), NMR spectroscopy and computational
methods. This study provides the scientific community with
efficient quantitative and qualitative methodology for ligand–
protein interaction studies.
Results and discussion
Protocol
A process of ligand screening, in combination with binding and
structure analysis, is outlined in Fig. 2. In the present study the
interactions between HEWL and small carbohydrates were ana-
lyzed. The ligands were obtained either by chemical synthesis or
from commercial sources and together they make up a library of
small carbohydrate compounds (Table 1). The screening of the
ligand-binding to HEWL first employed the high-throughput
technique weak affinity chromatography which resulted in KD
values for the ligands. For the subsequent NMR experiments
1
they were characterized with respect to, in particular, H NMR
chemical shifts. By subsequent NMR titration experiments using
a T2 relaxation experiment it was possible to obtain KD values
from an orthogonal technique. The interactions at an atomic
level were then investigated by STD (saturation transfer differ-
ence) and trNOE (transfer nuclear Overhauser effect) NMR
experiments. In silico molecular docking was able to propose
binding poses which were possible to evaluate by comparison to
theoretically calculated STD build-up curves. Finally, the
protein–ligand solution structures may then be compared to
crystal structures and refined using MD simulations.
3020 | Org. Biomol. Chem., 2012, 10, 3019–3032
This journal is © The Royal Society of Chemistry 2012

View Article Online
Table 1 Schematic structures of compounds studied by weak affinity chromatography and NMR spectroscopy (11) with corresponding KD values
Structure
No.
1
KD/mM
Structure
No.
9
KD/mM
α = 27
β = 47
0.27
2
46
10
0.64
3
4
43
42
11
12
0.62
0.4a
0.29
5
α = 31
β = 90
13
>100
6
7
61
38
14
15
38
>100
8
24
16
17
>100
>100
a Measured by NMR spectroscopy.
Synthesis
α-D-GlcpNAc is possible as mutarotation of monosaccharides is
a rather slow process at ambient temperature.28,29 Break-through
curves in the frontal analysis were determined in less than ten
minutes after the sample was dissolved and the derived values
were therefore judged to be a good approximation of α-D-
GlcpNAc binding. It was found that the concentration of binding
sites in the column was 4.8 mM when determined with the
monosaccharide (1) and 3.0 mM with the disaccharide (10).
These values were compared with the immobilized amount of
HEWL that were estimated to 133 mg g–1, corresponding to a
HEWL concentration of 4.7 mM (silica density = 0.5 g mL–1).
Previous immobilizations of other proteins have shown that
about 50% of the binding sites are inactivated after immobiliz-
ation, probably due to the random orientations in coupling.2,3
This implies that the expected number of sites on the column is
about 2.3 mM (assuming one site per HEWL), a value that is
close to what was obtained with 10, and it seems reasonable to
assume a 1 : 1 interaction between 10 and HEWL. This result
contrasts with the situation with monosaccharide 1 where an
interaction model with two independent sites on HEWL is most
likely, as the concentration of binding sites in this analysis was
The thio-analogue of the oxazoline derivative of N-acetylgluco-
samine,24 also known as thiazoline (7), is an inhibitor of
NAGases25 and hence a potential inhibitor of HEWL. To test
this hypothesis, thiazoline and its seleno-analogue 8, were syn-
thesized (Scheme S1†). Synthesis of selenazoline 8 has not been
previously published, so a new synthetic procedure based on the
synthesis of 7 was developed. The only difference between the
syntheses of 7 and 8 was that Woollins’ reagent26 was used in
the synthesis of 8 instead of Lawesson’s reagent.27 The disac-
charide methyl glycosides, 11 and 12, were synthesized accord-
ing to procedures that are described in Schemes S2 and S3.†
Frontal chromatography
The number of active binding sites on the HEWL silica was
assessed by frontal chromatography at pH 5.5. The analysis was
performed with newly dissolved monosaccharide (1, α-anomer)
and disaccharide (10, anomeric mixture) in order to determine
if these saccharides interacted differently. Analysis of
This journal is © The Royal Society of Chemistry 2012
Org. Biomol. Chem., 2012, 10, 3019–3032 | 3021

View Article Online
compared with the α- and β-methyl glycosides of GlcNAc
(2 and 3), which exhibited KD values in the same range as β-D-
GlcpNAc. A reason for this could be that the anomeric hydroxyl
proton in the α-anomeric form of 1 is involved in hydrogen
bonding with the protein.31 The affinities of the other monosac-
charides and the thio- and selenoanalogues of the oxazoline
derivatives of N-acetylglucosamine (7 and 8) were in the same
range as 1 with KD values ranging from 24 to 90 mM. The
group of N-acetylglucosamine-containing disaccharides, on the
other hand, had a high variation in HEWL affinity. The
suggested reason for the difference in behavior is that the small
monosaccharides, which all contain the key N-acetyl group,
interact in a similar way with HEWL, while the interaction with
larger N-acetyl-containing saccharides are more selective. Com-
pounds 9–12 being β-(1→4)-linked disaccharides, having at least
one N-acetyl-D-glucosamine residue, binds with <1 mM affinity
whereas disaccharides 13–17, which also have a D-GlcpNAc
residue but where anomeric or ring-carbon configurations as well
as linkage positions are different, do not represent compounds
with any significant binding affinity, i.e., KD ≥ 38 mM. The low
affinity of the latter compounds are presumably due to steric
clashes in the binding site, thereby preventing any tighter
binding even though the monosaccharide D-GlcpNAc residue of
these compounds binds to site C in HEWL.
Fig. 3 Elution profiles of disaccharides 11 (dashed line) and 12 (solid
line) on a WAC HEWL column (3 × 0.21 cm). See Materials and
methods section for chromatography conditions.
2.1 times higher than the calculated concentration of HEWL.
The frontal analysis therefore indicates that disaccharides and
monosaccharides interact differently with HEWL. A number of
studies have shown that N-acetylglucosamine-containing oligo-
saccharides may bind in several positions close to subsite C in
the binding cleft.13,23,30 X-ray crystallography studies have
found 1 mainly in site C31,32 but other studies, performed in sol-
ution with NMR, has indicated several binding possibilities.22
The WAC results imply that the monosaccharide is small enough
to interact with at least two of these sites at the same time.
The concentrations of the injected saccharide in weak affinity
chromatography should be less than the KD of the interaction in
order to not overload the column. Overriding this prerequisite
will induce the compound to elute earlier resulting in an erro-
neous (higher) KD value. Due to detection limitations the con-
Weak affinity chromatography (WAC)
Lysozyme has previously been reported to have an optimal
binding between pH 4 and 6.33 This was confirmed by determin-
ing the retention of 1 and 5 with mobile phases buffered to pH
3.0, 5.5 and 7.0 (0.1 M sodium phosphate, data not shown),
using the 50 × 0.46 cm HEWL column. Both compounds had a
maximal retention at pH 5.5 and all library compounds were
therefore analyzed using this mobile phase. The 50 × 0.46 cm
HEWL column was used for all analytes except for 9–12 that
were run on the 3 × 0.21 cm HEWL column because of their
higher affinity. The WAC analysis of 11 and 12 is shown in
centration of saccharides
9 and 12 was analyzed at a
concentration close to the obtained KD value (0.3 mM), which
means that the affinity could be slightly underestimated. The
analysis error due to a nonlinear behavior is however likely to be
small because of the sample dilution in the column (band-broad-
ening effects). The obtained KD values of 1 as well as 9 do also
agree well with previous studies performed at similar conditions
using NMR and microcalorimetry.22,34–36 The reported KD
values are ≈30 mM for N-acetyl-D-glucosamine (1) and
≈0.3 mM for N,N′-diacetylchitobiose (9) (cf. Table 1).
Fig. 3. The affinity was calculated using the relation KD = Bmax
/
Δv where KD is the equilibrium dissociation constant, Δv is the
retention volume and Bmax is equal to the number of binding
sites present in the column.3 All monosaccharides as well as 7
and 8 were assumed to recognize the same number of sites as
monosaccharide 1, which was calculated from the frontal analy-
sis to be 40.2 µmol regarding the 50 × 0.46 cm column. All dis-
accharides were assumed to be similar to 10, resulting in 24.6
µmol and 0.31 µmol binding sites for the 50 × 0.46 cm and 3 ×
0.21 cm HEWL columns, respectively. KD values of all com-
pounds are presented in Table 1.
The thio- and selenoanalogues 7 and 8 of the oxazoline
derivatives of N-acetylglucosamine were bound only weakly to
HEWL (Table 1) indicating that they do not represent the
hydrolysis reaction intermediate25,37 thereby supporting the
mechanism with a covalently linked intermediate, proposed by
Vocadlo et al.20 and confirmed by QM/MM simulations per-
formed by Bowman et al.38 instead of participation by the aceta-
mido group.
The KD values of the α- and β-anomeric forms of 1 were
determined concomitantly by analyzing the anomeric mixture at
equilibrium. This is possible since the chromatographic separ-
ation is faster than the anomeric re-equilibration.28,29 Repeated
analysis at 15 minute intervals of newly dissolved α-D-GlcpNAc
NMR spectroscopy
NMR assignments of H and 13C chemical shifts at 343 K and
1
1H chemical shifts at 278 K were performed for 11 and 12. H
1
chemical shifts and coupling constants were refined using the
NMR spin simulation software PERCH and are summarized in
Table S1† for 11 and Table S2† for 12. Subsequently, the
protein–ligand dissociation constant, KD, was measured for 11
using ligand observed transverse relaxation rates, R2.39 The
HEWL concentration was kept constant and T2 CPMG
1
(the α-anomeric form was determined by H NMR) on the 50 ×
0.46 cm HEWL column confirmed that the anomeric equilibrium
was reached after approximately one hour (data not shown). It
was found that α-D-GlcpNAc interacts with HEWL with higher
affinity compared to the β-anomeric form. The analysis of 1 was
3022 | Org. Biomol. Chem., 2012, 10, 3019–3032
This journal is © The Royal Society of Chemistry 2012

View Article Online
experiments were performed on 11 at eight different concen-
trations. By plotting the inverse of the difference between R2 of
the ligand in exchange with lysozyme and R2 of the free ligand
versus ligand concentrations a KD of 0.4 mM was extracted
(ESI†). This value is in very good agreement with that measured
by WAC (0.6 mM). The small difference between the techniques
may be due to immobilization effects and minor pH differences
because the NMR experiments were run in D2O solution instead
of water.
Saturation transfer difference (STD) experiments40 in which
binding epitopes on the ligand may be identified were performed
on HEWL in complex with 11 and 12, using saturation times of
0.5, 1, 2 and 4 s and two different saturation frequencies, 0.5 and
7.0 ppm. The STD NMR effects for 11 and 12 in complex with
HEWL at different saturation times and saturation frequencies
are compiled in Tables S4 and S5.† We also carried out control
STD experiments without protein present and did not observe
any STD effects. In Fig. 4, conspicuous STD effects are
observed for the methyl resonance of the N-acetyl group at the
reducing end residue of both disaccharides when a 0.5 s
irradiation is set at 0.5 ppm (Fig. 5a and 5b), indicating that the
protons are in close proximity to methyl groups in the protein.
The methyl resonance of the N-acetyl group at the reducing end
residue and the H6pro-S′ in the terminal residue of both disacchar-
ides give notable STD effects when irradiation is set at 7 ppm
(Fig. 5c and 5d), indicating that these protons are in close
contact with aromatic residues, e.g. Trp, in the protein complex
(Fig. 6a and 6b; poses obtained from docking, vide infra). These
results may be compared to a crystal structure of the HEWL–9
complex,23 where the methyl group of the N-acetyl group in the
terminal residue is in close contact with methyl and aromatic
protons in the protein (Fig. 6c). A published crystal structure of
HEWL in which 10 as a glycoside is covalently attached as an
ester to Asp52 via a three-carbon linker, pdbid: 1UC0,41 displays
a disaccharide binding mode (Fig. 6d) similar to that obtained
herein based on experimental NMR data in solution. Disacchar-
ides 11 and 12 show very similar STD patterns, indicating that
they have similar binding modes.
In solution a 1D 1H,1H-NOESY experiment at 298 K and
700 MHz of β-D-GlcpNAc-(1→4)-β-D-GlcpNAc-OMe (12) with
selective excitation of the resonance from H1′ and a mixing time
of 300 ms revealed a prominent NOE to H4 across the glycosidic
linkage, consistent with a syn conformation.42 This result is in
complete agreement with the transient NOE experiment per-
formed on 12 at 293 K and 400 MHz which showed a strong
NOE between H1′ and H4, besides an intra-residue NOE
between H1′ and H5′.43 An MD simulation of 12 with explicit
water molecules as solvent at 290 K43 had average torsion angles
ϕH ≈ 59° and ψH ≈ 15° over the 50 ps duration where the mol-
ecule was confined to a single conformational state at the glyco-
sidic linkage, i.e., a syn conformation.
Fig. 4 STD NMR and 1H NMR reference spectra of HEWL and 12
(a–c) and 11 (d–f) showing two different irradiation (0.5 s) positions.
1
(a) H reference spectrum of 12 and HEWL in a ratio of 50 : 1; (b) STD
spectrum of 12 and HEWL with irradiation at 7 ppm; (c) STD spectrum
of 12 and HEWL with irradiation at 0.5 ppm; (d) 1H reference spectrum
of 11 and HEWL in a ratio of 30 : 1; (e) STD spectrum of 11 and HEWL
with irradiation at 7 ppm; (f) STD spectrum of 11 and HEWL with
irradiation at 0.5 ppm.
Further information on ligand-bound conformation comes
from trNOE experiments44,45 of both complexes that show
strong negative cross-peaks between H1′ and H4, indicating a
syn conformation also in the bound state. This is exemplified for
12 at 289 K and 700 MHz showing a stronger inter-residue NOE
from H1′ to H4 than the intra-residue NOEs to H3′ and H5′
(Fig. 7). At this temperature and magnetic field strength the
NOE was close to zero for a ligand preparation of 12 without
protein. Thus, in the tr-NOESY experiments the contribution
from the ligand in solution should be negligible compared to that
originating from the protein–ligand complex. This trNOE result
is fully consistent with a syn conformation of 12 when bound to
HEWL. A weak cross-peak between the methyl resonance of the
N-acetyl group of the reducing end sugar residue and the methyl
This journal is © The Royal Society of Chemistry 2012
Org. Biomol. Chem., 2012, 10, 3019–3032 | 3023

View Article Online
(ranked 7th place by Autodock), shows a similar binding mode
as 12. Both disaccharides populate the B and C subsites. These
results contrast the crystal structure of HEWL with bound N,N′-
diacetylchitobiose23 in which the disaccharide is present in sub-
sites C and D (shown in red in Fig. 8). Theoretical STD build-up
curves, calculated using CORCEMA-ST,12 of both the proposed
docked structure of 12 and the crystal structure with bound N,N′-
diacetylchitobiose are presented in Fig. 9. The docked confor-
mation shows much better agreement with experimental STD
NMR data (cf. Fig. 4) compared to the crystal structure, giving
credence to the interpretations presented herein.
The docked structures of disaccharides 11 and 12 in the BC
binding site of HEWL are shown in Fig. 10. The characteristic
interactions between the ligands with hydrogen bonding and
CH/π-stacking49–52 are present to different extents. The π-stack-
ing interaction for benzene dimers53 has been described as paral-
lel-stacked or T-shaped and such arrangements were observed in
the crystal structures of molecules that were synthesis intermedi-
ates in the present study, viz., ethyl 3,6-di-O-benzyl-2-deoxy-2-
phthalimido-1-thio-β-D-glucopyranoside54 having an intramole-
cular parallel-stacked arrangement and ethyl 4,6-O-benzylidene-
2-deoxy-2-phthalimido-1-thio-β-D-glucopyranoside55 showing
an intermolecular T-shape-like arrangement. The disaccharide
interactions with HEWL differ slightly where the tighter binder
compound 12 (cf. Table 1) shows a closer stacking (Fig. 10a)
than compound 11 (Fig. 10b) which has a lower binding affinity.
However, in the latter complex a hydrogen bond is present
between HO4′ and the carboxylate group of Asp101 (Fig. 6b
and 10b). A corresponding hydrogen bond from hydroxyl
groups in 12 that are proximate to Asp101 was not present,
neither from HO6′ nor from HO4′. The interaction strength of a
classical hydrogen bond and a CH/π-stacking interaction is of
similar magnitude.56,57 The CH/π-stacking interaction between
the terminal sugars in the two complexes were further analyzed
with respect to distances and angles. In order to compare the two
structure complexes, centroids for the H1′/H3′/H5′ protons and
centroids for each ring of the indole part of Trp62 were defined
as well as a plane based on the three sugar protons and another
plane for the indole part. The distance separation in the HEWL-
12 complex (Fig. 11a) is significantly shorter than for the
HEWL-11 complex (Fig. 11b) and whereas the planes are almost
parallel in the first case (16°) the angle between the planes is
large (43°) in the second case. The geometrical arrangements for
both hydrogen bonds and CH/π-stacking interactions are impor-
tant descriptors to characterize interaction energies for com-
plexes. The former are highly directional whereas the latter show
only weak directionality.58,59 The distance between protons and
carbon atoms in the π-plane are on the order of 2.5–3.0 Å for
1
Fig. 5 Normalized levels of saturation (%) from H STD NMR spec-
troscopy of HEWL in complex with (a) 11 and (b) 12 (on-resonance
irradiation δ = 0.5 ppm, off-resonance irradiation δ = 60 ppm), (c) 11
and (d) 12 (on resonance irradiation δ = 7.0 ppm, off-resonance
irradiation δ = 60 ppm). Protons included in the figures indicate detected
STD effects.
region of the protein could also be observed in the 2D-NOESY
spectra.46 Similar results were obtained by 2D-NOESY exper-
iments for the complex between compound 11 and HEWL.
1
Hydrolysis of the methyl glycosides was not detected by H
NMR spectroscopy, which could have been the case if binding
had occurred at site D. Furthermore, hydrolysis of the glycosidic
linkage between the sugar residues in either of the disaccharides
was not detected, indicating that binding at sites D and E is of
limited importance. Moreover, a 1H NMR spectrum of an
anomeric mixture of 1 and HEWL showed a difference in the
chemical shifts of the methyl resonances of the N-acetyl groups
of the α- and β-anomeric forms compared to the chemical shifts
in absence of HEWL (data not shown). This indicates different
binding modes for α- and β-anomeric forms of 1, which is in
agreement with WAC data (see Table 1) and previously pub-
lished results.22,31
Molecular docking and simulation
TrNOE results indicate that the major conformation in the bound
state is syn for both of the disaccharides, 11 and 12. Molecular
docking of disaccharides 11 and 12 into the crystal structure
1SF4 were performed with the ϕ-torsion angles restrained at
ϕH = 50° using Autodock 4.47 The top ranked structure clusters
from docking were compared with NMR data and theoretical
STD build-up curves, calculated for the complexes using COR-
CEMA-ST.12 Fig. 8 shows the docking result for 12 (in blue)
that was in best agreement with experimental data. This structure
was ranked 6th place by the Autodock scoring function. The
ranking of docked ligands still poses significant problems in the
evaluation of the resulting poses due to insufficiently developed
scoring functions,7 but in some cases illuminating results on the
carbohydrate–protein interactions have been obtained for the
most favorable poses,48 stressing that Autodock is still an attrac-
tive tool for carbohydrates.7 We emphasize that in the present
study the selection of reasonable poses is based on docking in
conjunction with experimental NMR data. The docked structure
for 11, which is in best agreement with experimental data
geometrically optimized structures with
a favorable CH/
π-stacking.49–52 The somewhat tighter binding of 12 compared
to 11 may thus be reasonable to attribute to a stronger CH/
π-stacking interaction with Trp62 in HEWL.
To study structure further, as well as dynamics, MD simu-
lations with explicit solvent were performed on HEWL, HEWL-
11 complex from docking, HEWL-12 complex from docking
and HEWL–N,N′-diacetylchitobiose complex from a crystal
structure (1SF4). A CHARMM22 force field with CMAP60 was
used to describe the protein; the carbohydrates were described
by PARM22/SU01.61 The durations of the simulations were
3024 | Org. Biomol. Chem., 2012, 10, 3019–3032
This journal is © The Royal Society of Chemistry 2012

View Article Online
Fig. 6 Possible hydrogen bonding and hydrophobic interactions between HEWL and disaccharide ligands 12 (a), 11 (b), N,N′-diacetylchitobiose
from crystal structure 1SF4 (c) and β-D-Galp-(1→4)-β-D-GlcpNAc-(1→1)-Gro-3-Asp52 from crystal structure 1UC0 (d). The numerical values are the
distances in Å between the donor and acceptor atoms responsible for the interaction.
175 ns for HEWL and the HEWL-11 complex and 50 ns for the
HEWL-12 complex from docking and the HEWL–N,N′-diacetyl-
chitobiose complex from a crystal structure. The conformational
preference of the glycosidic linkage of 11 in complex with
HEWL from the 175 ns simulation is shown in Fig. 12, where
the ligand exists in two conformations during the simulation.
The second conformation appears after approximately 50 ns and
is then stable for the remaining part of the simulation
(Table S3†). The MD simulation then suggests that the confor-
mation after 50 ns of simulation is close to the conformation of
the disaccharide in complex with HEWL in solution. The glyco-
sidic torsion angles of the docked HEWL-12 complex adopt one
conformation, close to the starting structure, during the 50 ns
simulation. Table S3† shows the torsion angle averages of 12
where the ψH torsion angle shows a higher flexibility compared
to the ϕH torsion angle.
The complex of HEWL and N,N′-diacetylchitobiose (9) from
crystal structure 1SF4 dissociates during the MD simulation by
losing contact between the reducing end residue and the protein
binding site D, making the reducing end to protrude into
This journal is © The Royal Society of Chemistry 2012
Org. Biomol. Chem., 2012, 10, 3019–3032 | 3025

View Article Online
Fig. 7 Part of the 1H NMR spectrum of 12 (1 mM) in the presence of HEWL (25 μM) at 289 K (top); the corresponding 1D DPFGSE
1H,1H-NOESY spectrum with selective excitation of the resonance from H1′ and a mixing time of 300 ms (bottom). Pertinent resonances are
annotated.
Fig. 8 (left) Complex between HEWL and β-D-GlcpNAc-(1→4)-
β-DGlcpNAc-OMe from docking (blue) and HEWL and N,N′-diacetyl-
chitobiose from crystal structure 1SF4 (red); (right) time dependence of
the RMSD of the atoms in the disaccharides in complex with HEWL,
referenced from docking (blue) and crystal structure (red), during 10 ns
production runs of the MD simulations.
Fig. 9 Theoretical CORCEMA-ST STD build-up curves of 12 in
complex with HEWL from proposed structure from docking (a, in blue)
and crystal structure 1SF4 (b, in red); irradiation was set at the aromatic
protons in the protein. The build-up curves describe the methyl protons
solution. In the right panel of Fig. 8, the time dependency of the
RMSD of the ligand compared to the starting structure illustrates
that the crystal structure complex (red) breaks early in the simu-
lation, but the docked complex (blue) remains intact. The fact
of the N-acetyl group at the reducing end residue (solid), H6pro-S in the
that the crystal structure complex, which has an anti-ϕ confor-
mation at the glycosidic linkage of the ligand,23 does not survive
the MD simulation is a strong indication that the crystal structure
complex with the disaccharide is not likely to be present in
solution.
reducing end residue (dashed), methyl protons of the N-acetyl group at
the terminal residue (dots) and H6pro-S in the terminal residue (dashed-
dots).
and somewhat more for the HEWL-11 complex where the fluctu-
ations after 50 ns (vide infra) take place around new structural
equilibria. As an experimental evaluation of the simulated
From the MD simulation it is also evident that the overall
backbone structure of HEWL in solution changes compared to
the crystal structure (Fig. 13). This occurs both for the apo-form
3026 | Org. Biomol. Chem., 2012, 10, 3019–3032
This journal is © The Royal Society of Chemistry 2012

View Article Online
Fig. 12 Scatter plot (top) and time dependence (bottom) of the glyco-
sidic torsion angles of disaccharide 11 in complex with HEWL from the
175 ns MD simulation.
Fig. 10 Intermolecular interactions at sites B and C for 12 (a) and 11
(b) docked into complex with HEWL.
consists of 50 structures that fulfil the NMR data and the RMSD
of this data can be used as a model of the dynamics of the
protein. The backbone RMSDs, referenced for the MD to the
average trajectory structure during 50–175 ns of the simulations,
for every residue of the HEWL-11 complex (black) and the apo-
form of HEWL (brick red) from simulations as well as the NMR
structure (green) have been plotted in the lower panel of Fig. 13.
Experimental data from NMR is in good agreement with MD
simulations, which supports the reliability of the simulations.
Conspicuously higher RMSDs are observed for Arg61 and
Arg68 in the HEWL-11 complex. The three structural models
are presented in Fig. 14.
It may be noted that the higher flexibility in the loop region,
e.g., amino acids 68–70 (Fig. 14), observed in the MD simu-
lations as well as from the RMSD-analysis of the 50 NMR struc-
tures (Fig. 13) determined by Schwalbe et al.18 agrees well with
the 15N NMR relaxation data previously reported by Buck
et al.,17 which show main-chain values of S2 ≈ 0.7 whereas
most residues have S2 ≥ 0.8 and some helices have S2 ≈ 0.9. A
compilation of B-factors in crystal structures of the apo-form of
the protein showed that they were anti-correlated to the general-
ized order parameters, e.g., larger for residues in the region
around Pro70. The accessible surface area was likewise larger in
this region. However, for many of the amino acid residues
involved in direct or close contact with the ligands in the present
study (Fig. 6 and 10) the S2 values from the NMR relaxation
study are high in the apo-form of HEWL with one exception,
viz., Asn103 which has S2 ≈ 0.5, attributed to its high surface
accessibility.17
Fig. 11 The intermolecular stacking interactions of the indole part of
Trp62 in HEWL with the terminal sugars in 12 (a) and 11 (b). The dis-
tances between the centroids are given in Å and the angle between the
planes, given in degrees, is denoted by θ.
HEWL structures, a previously published NMR structure ensem-
ble of HEWL (pdbid: 1E8L)18 was used. The NMR result
This journal is © The Royal Society of Chemistry 2012
Org. Biomol. Chem., 2012, 10, 3019–3032 | 3027

View Article Online
Fig. 13 (top) Time dependence of the RMSD of the protein backbone
of HEWL in complex with 11 (black) and in the apo-form (brick red)
during the 175 ns MD simulation, referenced to crystal structure 1SF4;
(bottom) RMSD per amino acid residue for the protein backbone of
HEWL in complex with 11 (black) and in the apo-form (brick red)
during the production run between 50 and 175 ns of the MD simulation,
referenced to the trajectory average structure as well as 50 NMR struc-
tures18 (green), referenced to its average structure.
Fig. 14 HEWL colored by RMSD relative to its pertinent average
structure, where blue represents low RMSD and red represents high
RMSD; (a) HEWL in complex with 11 between 50 and 175 ns of the
MD simulation, (b) HEWL between 50 and 175 ns of the MD simu-
lation and (c) HEWL from the 50 experimental NMR structures (pdbid:
1E8L).
Conclusions
We have described a combination of techniques to study the
interaction between HEWL and carbohydrate ligands that give
both qualitative and quantitative information and the same pro-
cedures can be applied to many types of protein–ligand systems.
The high through-put qualities of WAC have been demonstrated
and it complemented structural information obtained from NMR
spectroscopy, docking and MD simulations. The methodology
does not require isotopically enriched samples nor is a large mol-
ecular mass of the protein a limitation. The dissociation con-
stants differ for ligands that bind by two orders of magnitude
with KD ≈ 30 mM for monosaccharides related to GlcNAc
whereas KD ≈ 0.3 mM for GlcNAc-containing disaccharides.
Two new solution structures of protein–ligand complexes have
been proposed and this shows that results obtained from a crystal
structure, disaccharide binding at C and D sites discussed herein,
are not always present for solution complexes where the corre-
sponding disaccharide binding occurs at sites B and C. The
occurrence of binding in this region is fully consistent with the
results of Dobson and co-workers30 who analyzed 1H NMR
chemical shifts changes in HEWL upon addition N,N′-diacetyl-
chitobiose (Fig. 15). The proposed structure poses did not show
the highest ranking from the docking simulation. This may be
due to conformational changes of the protein upon complex for-
mation, which were not taken into consideration in the docking
procedure and show that a combination of docking and MD
simulations may be needed in order to obtain good agreement
with experimental data. Furthermore, the results of this study
underscore the importance of targeting the appropriate binding
site(s), e.g., when small molecule fragment-based libraries are
employed62 in the process of generating ligands with high
affinity for protein receptors.
3028 | Org. Biomol. Chem., 2012, 10, 3019–3032
This journal is © The Royal Society of Chemistry 2012

View Article Online
5012 pump, a Rheodyne manual injector with a 1 mL injection
loop and a 9050 single UV wavelength detector, were used in
the experiment. The mobile phase was 0.1 M sodium phosphate
buffer pH 5.5 and the temperature was 23 1 °C. Two different
compounds were used in separate sets of experiments: newly dis-
solved GlcNAc (1, α-anomeric form) and LacNAc (10, anomeric
mixture). Each compound was dissolved in 0.1 M sodium phos-
phate buffer pH 5.5 at several concentrations and injected until
the column was saturated; the break-through curve (front) was
detected at 220 nm. The concentrations of 1 were varied from 4
to 320 mM and concentrations of 10 were varied from 1.5 to
5.0 mM. A negative spike from the injection of pure water was
used to determine the void volume of the system. Chromato-
graphy data were evaluated with EZChrom software version 6.8
(Scientific Software, San Ramon, CA, USA) and the midpoint of
each front was determined from the first derivative of the curve
(peak apex). The number of binding sites on the column and the
dissociation constant (KD) of the analyzed compound were
obtained from the experimental data by non-linear regression
analysis using GraphPad Prism 5 (SanDiego, CA, USA) apply-
ing a one-site binding model as been described previously;3 see
the ESI† for additional information.
Fig. 15 HEWL with amino acid residues colored in red based on
chemical shift changes upon formation of an N,N′-diacetylchitobiose
complex in a study by Dobson and co-workers.30 ΔδH limit for coloring:
0.1 ppm. Due to signal overlap the authors were not able to detect
chemical shift changes of a number of proton resonances in the B–C
sites. The docked structure of compound 12 is shown in dark blue
colour. The HEWL structure is from ref. 23.
Weak affinity chromatography (zonal chromatography)
Zonal chromatography experiments were performed on an
Agilent 1100 series HPLC system controlled by ChemStation
chromatography data system (Agilent Technologies, Santa Clara,
CA, USA). The mobile phase was 0.1 M sodium phosphate
buffer pH 5.5, and the temperature was 23 1 °C. The flow rate
was 0.1 mL min−1 when using the 3 × 0.21 cm column and
0.5 mL min−1 when using the 50 × 0.46 cm column. Com-
pounds (see Table 1) were injected individually and the retention
was recorded. An average from three separate injections was
used in the calculations. The 3 × 0.21 cm HEWL column was
used for compounds with moderate affinity (KD values in the
range of 0.1–5 mM) and the 50 × 0.46 cm column was used for
compounds with very low affinities (KD values above 5 mM).
Sample concentration was typically between 25–100 µg mL−1
(corresponding to a concentration of about 0.1 mM, depending
on the molecular weight of each compound).
Material and methods
Synthesis
Organic synthesis procedures and characterization of compounds
are presented in the ESI.†
Preparation of HEWL columns
HEWL silica was prepared by coupling the protein to aldehyde
functionalized silica, essentially as been described previously.3
In short, spherical and porous silica Kromasil® silica (EKA
chemicals AB, Bohus, Sweden) with a diameter of 10 µm and
pore size of 100 Å was derivatized with (3-glycidyloxypropyl)-
trimethoxysilane to obtain diol silica which was converted to
aldehyde silica by periodate oxidation. Aldehyde silica (6.81 g)
was mixed with HEWL (1.35 g in 70 mL of 0.1 M sodium phos-
phate, pH 7.0) and sodium cyanoborohydride (0.7 g). The coup-
ling was allowed to proceed at 4 °C for 48 h. The coupling
efficiency was determined by measuring the absorbance at
280 nm in the supernatant and 133 mg HEWL was found to be
immobilized per gram of silica. The HEWL silica was packed
into a 50 × 0.46 cm stainless steel column (Hichrom, Berkshire,
UK) and a 3 × 0.21 cm PEEK column (Poros, Applied Biosys-
tems, Carlsbad, CA, USA), using an air-driven liquid pump
(Haskel, Burbank, CA, USA) and an ordinary HPLC pump
(Varian, Walnut Creek, CA, USA), respectively.
NMR spectroscopy
For the proton R2 relaxation rate measurements to obtain KD
a
Bruker Advance 500 MHz spectrometer equipped with a 5 mm
PFG triple-resonance CryoProbe was used. The experiments
were carried out at 294 K in D2O solutions with 20 μM TSP
(sodium 3-trimethylsilyl-(2,2,3,3-2H4)-propanoate), 25 mM
sodium phosphate buffer, pH 5.5, and 50 mM NaCl. Disacchar-
ide 11 was titrated into a solution of 30 μM HEWL (kept con-
stant) and R2 measurements were performed at eight
concentrations (0.2–13.4 mM) with the CPMG spin-echo pulse
sequence. The methyl resonance at 2.04 ppm was observed
using 12 delay times (10 ms–10 s). A decaying curve, based on
the peak heights, was generated for each concentration. These
curves were fitted to an exponentially decaying function to give
relaxation rate R2 values that for each concentration were plotted
Frontal chromatography
Frontal chromatography3,63 was performed in order to determine
the number of binding sites on the HEWL-immobilized silica. A
Varian HPLC system (Walnut Creek, CA, USA) equipped with a
This journal is © The Royal Society of Chemistry 2012
Org. Biomol. Chem., 2012, 10, 3019–3032 | 3029

View Article Online
and the KD value was extracted according to eqn (1)39 (plots are
displayed in the ESI†):
scans were acquired per increment and 32 dummy scans were
used.
Disaccharides 11 and 12 were dissolved in D2O to a concen-
tration of 15 and 26 mM respectively, and assigned using stan-
dard 1H,1H-homonuclear and 1H,13C-heteronuclear 2D
experiments at 343 K and 278 K. Proton chemical shifts and
coupling constants were refined using the NMR spin simu-
lation66 software PERCH (PERCH Solutions Ltd, Kuopio,
Finland). Assignments of the NMR resonances of the protected
compounds utilized a limited number of 1D and 2D NMR exper-
iments but were complemented with chemical shift predictions
½Pꢀ0ΔRm2 ax
½Lꢀ0 ¼
ꢁ KD;
ð1Þ
ΔRo2bs
where [L]0 and [P]0 are the total concentrations of ligand and
obs
protein, respectively, ΔR2 is the observed change in spin–spin
relaxation rate constant for a given concentration relative to that
max
of the free ligand, and ΔR2
is the limiting case for bound and
free states. For processing of the KD data Excel and Matlab
programs were used. The FID from each experiment was
acquired with 32–96 scans and the spectral width used was 8
kHz which together with the 16 k data points corresponds to an
acquisition time of 1 s. The waiting period between scans was
set to 16 s.
1
made by PERCH. The H and 13C chemical shift assignments
and JHH are presented in the ESI.† The data from the NMR
experiments were processed and analyzed with Bruker TOPSPIN
2.1 software.
NMR assignments of disaccharides 11 and 12, STD and
trNOE NMR experiments were carried out on a Bruker Avance
III 700 MHz NMR spectrometer with a 5 mm TCI Z-gradient
high resolution CryoProbe. The STD and trNOE NMR exper-
iments were carried out using Norell S-3-HT-7 3 mm tubes at
278 K with an oligosaccharide concentration of 1 mM and a
protein concentration of 20 µM. Reference samples with free
ligand (26 mM, 5 mm tube) were analyzed with STD, 1D and
2D NOESY NMR experiments. The buffer was 25 mM sodium
phosphate, 50 mM sodium chloride pH 5.5. The buffer was lyo-
philized from H2O and subsequently exchanged with D2O.
The STD experiments were recorded using the standard pulse
sequence40 with a 60 ms 5 kHz spin-lock pulse to reduce the
background protein resonances, excitation sculpting64 with a 224
Hz square π-pulse to suppress residual HDO and purge pulses to
remove unwanted magnetization. Saturation of the protein NMR
signals of the enzyme was performed using a train of selective
65 Hz Gaussian pulses with duration of 50 ms, adding up to
total saturation times of 0.5, 1.0, 2.0 and 4.0 s. The on-resonance
frequency was set to 0.5 and 7.0 ppm and off-resonance
irradiation was applied at 60 ppm. STD NMR spectra were
acquired with a total of 2048, 1536, 1024 and 1024 transients
for the four different saturation times respectively, in addition to
32 dummy scans. Spectra were recorded with a spectral width of
11 kHz and 18 k data points corresponding to an acquisition
time of 0.8 s followed by a relaxation delay of 4.2 s.
Computer simulations
Molecular dynamics (MD) simulations used NAMD67,68 (paral-
lel version, 2.6) employing a CHARMM22 force field with
CMAP60 for the protein. The carbohydrates were described by
the PARM22/SU01 force field.61 Initial coordinates for the
protein were taken from the crystal structure 1SF423 and top-
ology files, TIP3P solvent boxes and ions were generated in
VMD.69 Pdb and psf files of the oligosaccharides were created
using VEGA ZZ70 and assigned CHARMM partial charges. The
MD simulations were carried out with multiple-time-stepping of
2 fs, 2 fs, and 6 fs were used as the inner, middle and outer time
steps in the NPT ensemble (P = 1 atm, 294 K) with a cutoff dis-
tance for non-bonded interactions set at 12 Å and periodic
boundary conditions giving box sizes of approximately 50 Å ×
44 Å × 51 Å. The smooth particle-mesh Ewald (SPME) method
was used to calculate the full electrostatic interactions. The temp-
erature and pressure were kept constant using a Langevin ther-
mostat and a Langevin barostat, respectively. All bonds to
hydrogen atoms were kept rigid. After energy minimization,
heating and 1 ns of equilibration, production runs were carried
out during 175 ns for HEWL and for the complex between di-
saccharide 11 and HEWL. For the crystal structure and docked
complex of disaccharide 12 and HEWL, simulations were
carried out for 50 ns. Data were saved every 1000 time steps for
analysis. Trajectory analyses were carried out with VEGA ZZ
and VMD.
For molecular docking simulations the crystal structure of the
complex between HEWL and N,N′-diacetylchitobiose (pdbid:
1SF4),23 with the disaccharide removed, was used as the receptor
with AutoDock 4.0.47 Coordinates for oligosaccharides 11 and
12 were obtained by the Glycam Biomolecule Builder (www.
glycam.com). The ϕ-torsions were restrained in an exo-anomeric
conformation at ϕH = 50° and all other torsions were flexible.
The grid dimensions were 50 × 50 × 70 points, with points sep-
arated by 0.375 Å. The grids were chosen to be centered on
binding site C and sufficiently large to cover the whole pocket.
In total, 150 runs with the Lamarckian genetic algorithm were
performed and the maximum number of free energy evaluations
was set to 5 × 106. Other parameters were set to default values.
In the complexes obtained by docking, the centroids used for
CH/π-stacking analyses were based on the H1′/H3′/H5′ protons
For the trNOE experiments standard 1D and phase-sensitive
2D 1H,1H-NOESY pulse sequences were used together with
excitation sculpting, with a 224 Hz square π-pulse, to suppress
residual HDO and purge pulses to remove unwanted magnetiza-
tion. In order to determine suitable experimental conditions 1D
NOESY experiments were performed on the free ligand 12 at
different temperatures and it was found that the NOE was close
to zero at 289 K (at 16.4 T). The 1D experiments were per-
formed with 32 k points with a sweep width of 11 kHz giving an
acquisition time of 2.1 s. Selective excitation of the H1′ reson-
ance was enabled using a 30 ms long r-SNOB shaped pulse.65
A
mixing time of 300 ms was used with a relaxation delay of 2.9 s,
giving a recycle time between scans of 5.3 s. Ten thousand scans
were acquired and 64 dummy scans were used to reach a steady
state. The 2D experiments were performed with 4096 points in
the direct dimension, 320 increments, with a sweep width of 11
kHz in both dimensions, and a relaxation delay of 1.1 s. Eighty
3030 | Org. Biomol. Chem., 2012, 10, 3019–3032
This journal is © The Royal Society of Chemistry 2012

View Article Online
10 H. Alonso, A. A. Bliznyuk and J. E. Gready, Combining docking and
molecular dynamics simulations in drug design, Med. Res. Rev., 2006,
26, 531–568.
11 B. A. Borgias, M. Gochin, D. J. Kerwood and T. L. James, Relaxation
matrix analysis of 2D NMR data, Prog. Nucl. Magn. Reson. Spectrosc.,
1990, 22, 83–100.
in the terminal sugar residues and 4/6/7a and 2/3/7a atoms in the
indole part of Trp62. The pertinent planes were based the H1′/
H3′/H5′ protons in the terminal sugar residues and the 2/5/6
atoms in the indole part of Trp62.
Theoretical STD build-up curves were calculated from the
docking results and the crystal structure 1SF4 using a modified
version of CORCEMA-ST.12,71 KD values were taken from the
WAC results. We have assumed the diffusion limited on-rate as
108 s–1 M–1, a generalized order parameter S2 of 0.85 and a
uniform leakage relaxation of 0.25 s–1. The ligand correlation
time (τL) and the methyl group internal correlation time (τm)
were chosen as 0.12 ns and 5 ps, respectively, since this resulted
in the best agreement with experimental T1 values. The protein
correlation time (τP) was calculated to 12 ns using Stokes’ law.72
The conformation of the ligand was assumed to be the same in
both the free and the bound state. The SHIFTX software73 was
12 V. Jayalakshmi and N. Rama Krishna, Complete relaxation and confor-
mational exchange matrix (CORCEMA) analysis of intermolecular satur-
ation transfer effects in reversibly forming ligand-receptor complexes,
J. Magn. Reson., 2002, 155, 106–118.
13 C. C. Blake, D. F. Koenig, G. A. Mair, A. C. North, D. C. Phillips and V.
R. Sarma, Structure of hen egg-white lysozyme: a three-timensional
Fourier synthesis at 2 Å resolution, Nature, 1965, 206, 757–761.
14 L. N. Johnson and D. C. Phillips, Structure of some crystalline lysozyme-
inhibitor complexes determined by X-ray analysis at 6 Å resolution,
Nature, 1965, 206, 761–763.
15 K. Kurachi, L. C. Sieker and L. H. Jensen, Structures of triclinic mono-
and di-N-acetylglucosamine: lysozyme complexes – A crystallographic
study, J. Mol. Biol., 1976, 101, 11–24.
16 C. B. Post, B. R. Brooks, M. Karplus, C. M. Dobson, P. J. Artymiuk,
J. C. Cheetham and D. C. Phillips, Molecular dynamics simulations of
native and substrate-bound lysozyme. A study of the average structures
and atomic fluctuations, J. Mol. Biol., 1986, 190, 455–479.
17 M. Buck, J. Boyd, C. Redfield, D. A. MacKenzie, D. J. Jeenes, D.
B. Archer and C. M. Dobson, Structural determinants of protein
dynamics: analysis of 15N NMR relaxation measurements for main-chain
and side-chain nuclei of hen eggwhite lysozyme, Biochemistry, 1995, 34,
4041–4055.
1
used to calculate the H chemical shifts of the protein and the
protons resonating between 0.3 and 0.7 ppm were assumed to be
saturated when irradiation was set at 0.5 ppm. All aromatic
protons were assumed to be saturated when irradiation was set at
7 ppm.
18 H. Schwalbe, S. B. Grimshaw, A. Spencer, M. Buck, J. Boyd, C.
M. Dobson, C. Redfield and L. J. Smith, A refined solution structure of
hen lysozyme determined using residual dipolar coupling data, Protein
Sci., 2001, 10, 677–688.
19 D. M. Chipman and N. Sharon, Mechanism of lysozyme action, Science,
1969, 165, 454–465.
20 D. J. Vocadlo, G. J. Davies, R. Laine and S. G. Withers, Catalysis by hen
egg-white lysozyme proceeds via a covalent intermediate, Nature, 2001,
412, 835–838.
21 L. R. Berger and R. S. Weiser, The β-glucosaminidase activity of egg-
white lysozyme, Biochim. Biophys. Acta, 1957, 26, 517–521.
22 S. J. Perkins, L. N. Johnson, D. C. Phillips and R. A. Dwek, The binding
of monosaccharide inhibitors to hen egg-white lysozyme by proton mag-
netic resonance at 270 MHz and analysis by ring-current calculations,
Biochem. J., 1981, 193, 553–572.
Acknowledgements
This work was supported by grants from the Swedish Research
Council, The Knut and Alice Wallenberg Foundation, and Magn.
Bergvalls Stiftelse. Computing resources were kindly provided
by the Center for Parallel Computers (PDC), Stockholm,
Sweden. We are also grateful for the contribution from Emma
Fång regarding the preparation of the HEWL WAC columns.
References
23 R. B. Von Dreele, Binding of N-acetylglucosamine oligosaccharides to
hen egg-white lysozyme: a powder diffraction study, Acta Crystallogr.,
Sect. D: Biol. Crystallogr., 2005, 61, 22–32.
24 M. Noguchi, T. Tanaka, H. Gyakushi, A. Kobayashi and S.-I. Shoda,
Efficient synthesis of sugar oxazolines from unprotected N-acetyl-2-
amino sugars by using chloroformamidinium reagent in water, J. Org.
Chem., 2009, 74, 2210–2212.
1 D. Zopf and S. Ohlson, Weak-affinity chromatography, Nature, 1990,
346, 87–88.
2 S. Ohlson, M. Bergström, P. Påhlsson and A. Lundblad, Use of mono-
clonal antibodies for weak affinity chromatography, J. Chromatogr., A,
1997, 758, 199–208.
3 M. Bergström, S. Liu, K. L. Kiick and S. Ohlson, Cholera toxin inhibitors
studied with high-performance liquid affinity chromatography: a robust
method to evaluate receptor–ligand interactions, Chem. Biol. Drug Des.,
2009, 73, 132–141.
4 P. Cheshev, L. Morelli, M. Marchesi, Č. Podlipnik, M. Bergström and
A. Bernardi, Synthesis and affinity evaluation of a small library of biden-
tate cholera toxin lignads: towards nonhydrolyzable ganglioside mimics,
Chem.–Eur. J., 2010, 16, 1951–1967.
5 S. Ohlson and D. Zopf, Weak affinity chromatography, in Handbook of
Affinity Chromatography, ed. T. Kline, Marcel Dekker, New York, NY,
1993, pp. 299–314.
6 M. Pellecchia, I. Bertini, D. Cowburn, C. Dalvit, E. Giralt, W. Jahnke, T.
L. James, S. W. Homans, H. Kessler, C. Luchinat, B. Meyer,
H. Oschkinat, J. Peng, H. Schwalbe and G. Siegal, Perspectives on NMR
in drug discovery: a technique comes of age, Nat. Rev. Drug Discovery,
2008, 7, 738–745.
7 D. Neumann and O. Kohlbacher, Structural glycomics – Molcular details
of protein–carbohydrate interactions and their prediction. In Proceedings
of the International Beilstein Symposium in Glyco-Bioinformatics, ed. M.
G. Hicks and C. Kettner, Beilstein-Institut, Frankfurt/Main, 2009, pp.
101–122.
25 S. Knapp, D. Vocadlo, Z. N. Gao, B. Kirk, J. P. Lou and S. G. Withers,
NAG-thiazoline,
an
N-acetyl-β-hexosaminidase
inhibitor
that
implicates acetamido participation, J. Am. Chem. Soc., 1996, 118, 6804–
6805.
26 P. Bhattacharyya and J. D. Woollins, Selenocarbonyl synthesis using
Woollins reagent, Tetrahedron Lett., 2001, 42, 5949–5951.
27 T. Ozturk, E. Ertas and O. Mert, Use of Lawesson’s reagent in organic
syntheses, Chem. Rev., 2007, 107, 5210–5278.
28 G. Livingstone, F. Franks and L. J. Aspinall, The effects of aqueous
solvent structure on the mutarotation kinetics of glucose, J. Solution
Chem., 1977, 6, 203–216.
29 P. W. Wertz, J. C. Garver and L. Anderson, Anatomy of complex mutaro-
tation. Kinetics of tautomerization of α-D-galactopyranose and β-D-galac-
topyranose in water, J. Am. Chem. Soc., 1981, 103, 3916–3922.
30 K. J. Lumb, J. C. Cheetham and C. M. Dobson, 1H nuclear magnetic res-
onance studies of hen lysozyme-N-acetylglucosamine oligosaccharide
complexes in solution. Application of chemical shifts for the comparison
of conformational changes in solution and in the crystal, J. Mol. Biol.,
1994, 235, 1072–1087.
31 C. C. F. Blake, L. N. Johnson, G. A. Mair, A. C. T. North, D. C. Phillips
and V. R. Sarma, Crystallographic studies of the activity of hen egg-white
lysozyme, Proc. R. Soc. London, Ser. B, 1967, 167, 378–388.
32 R. B. Von Dreele, Binding of N-acetylglucosamine to chicken egg lyso-
zyme: a powder diffraction study, Acta Crystallogr., Sect. D: Biol.
Crystallogr., 2001, 57, 1836–1842.
8 I. Halperin, B. Ma, H. Wolfson and R. Nussinov, Principles of docking:
an overview of search algorithms and a guide to scoring functions, Pro-
teins, 2002, 47, 409–443.
9 S. A. Adcock and J. A. McCammon, Molecular dynamics: survey of
methods for simulating the activity of proteins, Chem. Rev., 2006, 106,
1589–1615.
This journal is © The Royal Society of Chemistry 2012
Org. Biomol. Chem., 2012, 10, 3019–3032 | 3031

View Article Online
33 S. K. Banerjee and J. A. Rupley, Temperature and pH dependence of the
binding of oligosaccharides to lysozyme, J. Biol. Chem., 1973, 248,
2117–2124.
54 C. Hamark, J. Landström, L. Eriksson and G. Widmalm, Ethyl 3,6-di-O-
benzyl-2-deoxy-N-phthalimido-1-thio-β-D-glucopyranoside, Acta Crystal-
logr., Sect. E: Struct. Rep. Online, 2010, 66, o3250–o3251.
34 F. W. Dahlquist and M. A. Raftery, A nuclear magnetic resonance
study of association equilibria and enzyme-bound environments of N-
acetyl-D-glucosamine anomers and lysozyme, Biochemistry, 1968, 7,
3269–3277.
35 B. D. Sykes and C. Parravano, A nuclear magnetic resonance study of the
inhibition of lysozyme by N-acetyl-D-glucosamine and di-N-acetyl-D-glu-
cosamine, J. Biol. Chem., 1969, 244, 3900–3904.
55 C. Hamark, J. Landström, L. Eriksson and G. Widmalm, Ethyl 4,6-O-
benzylidene-2-deoxy-N-phthalimido-1-thio-β-D-glucopyranoside,
Crystallogr., Sect. E: Struct. Rep. Online, 2010, 66, o3249.
Acta
56 M. S. Sujatha, Y. U. Sasidhar and P. V. Balaji, Insights into the role of the
aromatic residue in galactose-binding sites: MP2/6–311G++** study on
galactose- and glucose-aromatic residue analogue complexes, Biochemis-
try, 2005, 44, 8554–8562.
36 C. Bjurulf and I. Wadsö, Thermochemistry of lysozyme-inhibitor
binding, Eur. J. Biochem., 1972, 31, 95–102.
37 Y. He, M. S. Macauley, K. A. Stubbs, D. J. Vocadlo and G. J. Davies,
Visualizing the reaction coordinate of an O-GlcNAc hydrolase, J. Am.
Chem. Soc., 2010, 132, 1807–1809.
57 R. Sharma, J. P. McNamara, R. K. Raju, M. A. Vincent, I. H. Hillier and
C. A. Morgado, The interaction of carbohydrates and amino acids with
aromatic systems studied by density functional and semi-empirical mol-
ecular orbital calculations with dispersion corrections, Phys. Chem.
Chem. Phys., 2008, 10, 2767–2774.
38 A. L. Bowman, I. M. Grant and A. J. Mulholland, QM/MM simulations
predict a covalent intermediate in the hen egg white lysozyme reaction
with its natural substrate, Chem. Commun., 2008, 4425–4427.
39 L. Fielding, NMR methods for the determination of protein–ligand dis-
sociation constants, Prog. Nucl. Magn. Reson. Spectrosc., 2007, 51, 219–
242.
58 S. Tsuzuki and A. Fujii, Nature and physical origin of CH/π-interaction:
significant difference from convetional hydrogen bonds, Phys. Chem.
Chem. Phys., 2008, 10, 2584–2594.
59 S. Tsuzuki, T. Uchimaru and M. Mikami, Magnitude and nature of carbo-
hydrate–aromatic interactions: ab initio calculations of fucose–benzene
complex, J. Phys. Chem. B, 2009, 113, 5617–5621.
40 M. Mayer and B. Meyer, Characterization of ligand binding by saturation
transfer difference NMR spectroscopy, Angew. Chem., Int. Ed., 1999, 38,
1784–1788.
41 M. Muraki and K. Harata, X-Ray structural analysis of the ligand-recog-
nition mechanism in the dual-affinity labeling of c-type lysozyme with
2′,3′-epoxypropyl β-glycoside of N-acetyllactosamine, J. Mol. Recognit.,
2003, 16, 72–82.
42 G. Widmalm, General NMR spectroscopy of carbohydrates and confor-
mational analysis in solution, in Comprehensive Glycoscience, ed. J.
P. Kamerling, Elsevier, Oxford, 1st edn, 2007, pp. 101–132.
43 M. Aida, Y. Sugawara, S. Oikawa and K. Umemoto, Structural fluctuation
of methyl N,N′-diacetyl-β-D-chitobioside in vacuo and in aqueous sol-
ution: molecular dynamics simulations and proton NMR spectroscopy,
Int. J. Biol. Macromol., 1995, 17, 227–235.
44 G. M. Clore and A. M. Gronenborn, Theory and applications of the trans-
ferred nuclear Overhauser effect to the study of the conformations of
small ligands bound to proteins, J. Magn. Reson., 1982, 48, 402–417.
45 C. B. Post, Exchange-transferred NOE spectroscopy and bound ligand
structure determination, Curr. Opin. Struct. Biol., 2003, 13, 581–588.
46 T. Scherf and J. Anglister, A T1ρ-filtered two-dimensional transferred
NOE spectrum for studying antibody interactions with peptide antigens,
Biophys. J., 1993, 64, 754–761.
47 G. M. Morris, D. S. Goodsell, R. S. Halliday, R. Huey, W. E. Hart, R.
K. Belew and A. J. Olson, Automated docking using a Lamarckian
genetic algorithm and an empirical binding free energy function, J.
Comput. Chem., 1998, 19, 1639–1662.
48 J. Landström, E.-L. Nordmark, R. Eklund, A. Weintraub, R. Seckler and
G. Widmalm, Interaction of a Salmonella enteritidis O-antigen octasac-
charide with the phage P22 tailspike protein by NMR spectroscopy and
docking studies, Glycoconjugate J., 2008, 25, 137–143.
49 V. Spiwok, P. Lipovová, T. Skálová, E. Buchtelová, J. Hašek and
B. Králova, Role of CH/π interactions in substrate binding by Escherichia
coli β-galactosidase, Carbohydr. Res., 2004, 339, 2275–2280.
50 M. del Carmen Fernández-Alonso, F. J. Cañada, J. Jiménez-Barbero and
G. Cuevas, Molecular recognition of saccharides by proteins. Insights on
the origin of carbohydrate–aromatic interactions, J. Am. Chem. Soc.,
2005, 127, 7379–7386.
51 V. Spiwok, P. Lipovová, T. Skálová, E. Vondráčková, J. Dohnálek,
J. Hašek and B. Králova, Modelling of carbohydrate–aromatic inter-
actions: ab initio energetics and force field performance, J. Comput.-
Aided Mol. Des., 2006, 19, 887–901.
60 A. D. Mackerell Jr., M. Feig and C. L. Brooks III, Extending the treat-
ment of backbone energetics in protein force fields: limitations of gas-
phase quantum mechanics in reproducing protein conformational distri-
butions in molecular dynamics simulations, J. Comput. Chem., 2004, 25,
1400–1415.
61 R. Eklund and G. Widmalm, Molecular dynamics simulations of an
oligosaccharide using a force field modified for carbohydrates, Carbo-
hydr. Res., 2003, 338, 393–398.
62 C. Rademacher, J. Landström, N. Sindhuwinata, M. M. Palcic,
G. Widmalm and T. Peters, NMR-based exploration of the acceptor
binding site of human blood Group B galactosyltransferase with molecu-
lar fragments, Glycoconjugate J., 2010, 27, 349–358.
63 K.-I. Kasai, Y. Oda, M. Nishikata and S.-I. Ishii, Frontal affinity-chrom-
atography: theory for its application to studies on specific interactions of
biomolecules, J. Chromatogr., Biomed. Appl., 1986, 376, 33–47.
64 T. L. Hwang and A. J. Shaka, Water suppression that works. Excitation
sculpting using arbitrary waveforms and pulsed field gradients, J. Magn.
Reson., Ser. A, 1995, 112, 275–279.
65 Ē. Kupče, J. Boyd and I. D. Campbell, Short selective pulses for bio-
chemical applications, J. Magn. Reson., Ser. B, 1995, 106, 300–303.
66 R. Laatikainen, M. Niemitz, U. Weber, J. Sundelin, T. Hassinen and
J. Vepsäläinen, General strategies for total-lineshape-type spectral analy-
sis of NMR spectra using integral-transform iterator, J. Magn. Reson.,
Ser. A, 1996, 120, 1–10.
67 L. Kalé, R. Skeel, M. Bhandarkar, R. Brunner, A. Gursoy, N. Krawetz,
J. Phillips, A. Shinozaki, K. Varadarajan and K. Schulten, NAMD2:
greater scalability for parallel molecular dynamics, J. Comput. Phys.,
1999, 151, 283–312.
68 J. C. Phillips, R. Braun, W. Wang, J. Gumbart, E. Tajkhorshid, E. Villa,
C. Chipot, R. D. Skeel, L. Kalé and K. Schulten, Scalable molecular
dynamics with NAMD, J. Comput. Chem., 2005, 26, 1781–1802.
69 W. Humphrey, A. Dalke and K. Schulten, VMD: visual molecular
dynamics, J. Mol. Graphics, 1996, 14, 33–38, 27–28.
70 A. Pedretti, L. Villa and G. Vistoli, VEGA: a versatile program to
convert, handle and visualize molecular structure on Windows-based
PCs, J. Mol. Graphics Modell., 2002, 21, 47–49.
71 S. Kemper, M. K. Patel, J. C. Errey, B. G. Davis, J. A. Jones and
T. D. W. Claridge, Group epitope mapping considering relaxation of the
ligand (GEM-CRL): including longitudinal relaxation rates in the analysis
of saturation transfer difference (STD) experiments, J. Magn. Reson.,
2010, 203, 1–10.
52 S. Kozmon, R. Matuška, V. Spiwok and J. Koča, Three-dimensional
potential energy surface of selected carbohydrates’ CH/π dispersion inter-
actions calculated by high-level quantum mechanical methods, Chem.–
Eur. J., 2011, 17, 5680–5690.
53 G. B. McGaughey, M. Gagné and A. K. Rappé, π-Stacking interactions,
J. Biol. Chem., 1998, 273, 15458–15463.
72 J. Cavanagh, W. J. Fairbrother, A. G. Palmer III and N. J. Skelton, Classi-
cal NMR spectroscopy, in Protein NMR Spectroscopy: Principles and
Practice, Academic Press, Boston, 1st edn, 1996, p. 17.
73 S. Neal, A. M. Nip, H. Zhang and D. S. Wishart, Rapid and accurate cal-
culation of protein 1H, 13C and 15N chemical shifts, J. Biomol. NMR,
2003, 26, 215–240.
3032 | Org. Biomol. Chem., 2012, 10, 3019–3032
This journal is © The Royal Society of Chemistry 2012