Insights into the dynamics and molecular recognition features of glycopeptides by protein receptors: The 3D solution structure of hevein bound to the trisaccharide core of N-glycoproteins

Article abstract of DOI:10.1002/chem.201000939

Protein-carbohydrate interactions are at the heart of a variety of essential molecular recognition events. Hevein, a model lectin related to the superantigen family, recognizes the trisaccharide core of N-glycoproteins (1). A combined approach of NMR spectroscopy and molecular modeling has permitted us to demonstrate that an Asn-linked Man(GlcNAc)₂ (2) is bound with even higher affinity than (GlcNAc)₃. The molecular recognition process entails conformational selection of only one of the possibilities existing for chitooligosaccharides. The deduced 3D structure of the hevein/2 complex permits the extension of polypeptide chains from the Asn moiety of 2, as well as glycosylation at Man O-3 and Man O-6 of the terminal sugar. Given the ubiquity of the Man(GlcNAc)₂ core in all mammalian N-glycoproteins, the basic recognition mode presented herein might be extended to a variety of systems with biomedical importance.

Full text of DOI:10.1002/chem.201000939

FULL PAPER
DOI: 10.1002/chem.201000939
Insights into the Dynamics and Molecular Recognition Features of
Glycopeptides by Protein Receptors: The 3D Solution Structure of Hevein
Bound to the Trisaccharide Core of N-Glycoproteins
Josꢀ Juan Hernꢁndez-Gay,[a] Ana Ardꢁ,[a, c] Steffen Eller,[b] Stefano Mezzato,[b]
Bas R. Leeflang,[c] Carlo Unverzagt,[b] F. Javier CaÇada,*[a] and
Jesffls Jimꢀnez-Barbero*[a]
Dedicated to Professor Horst Kessler on the occasion of his 70th birthday
Abstract: Protein-carbohydrate interac-
tions are at the heart of a variety of es-
sential molecular recognition events.
Hevein, a model lectin related to the
superantigen family, recognizes the tri-
saccharide core of N-glycoproteins (1).
A combined approach of NMR spec-
troscopy and molecular modeling has
permitted us to demonstrate that an
(GlcNAc)3. The molecular recognition
process entails conformational selec-
tion of only one of the possibilities ex-
isting for chitooligosaccharides. The de-
duced 3D structure of the hevein/2
complex permits the extension of poly-
peptide chains from the Asn moiety of
2, as well as glycosylation at Man O-3
and Man O-6 of the terminal sugar.
Given the ubiquity of the Man-
AHCTUNGRTEN(GNUN GlcNAc)2 core in all mammalian N-
glycoproteins, the basic recognition
mode presented herein might be ex-
tended to a variety of systems with bio-
medical importance.
Keywords: allergies · carbohydrate
binding · glycoproteins · hevein ·
NMR spectroscopy
Asn-linked ManACHTUNGTRENNUNG(GlcNAc)2 (2) is
bound with even higher affinity than
Introduction
scopic, biophysical, and theoretical protocols is a powerful
tool to obtain detailed structural information of molecules
with atomic resolution. The features that mediate the inter-
actions between biomolecules related to a variety of biologi-
cal and biomedical problems can be understood through this
tool box. In this context, and within glycosciences,[1–6] lectins
are carbohydrate-specific binding proteins that mediate the
transfer of biological information from the sugar code and
have been widely used as tools in different areas of (bio)-
chemical investigations.[1–6] Some of these lectins from plant
origin, dubbed hevein domains, bind reversibly to chitin, a
key structural component of the fungi cell wall and inverte-
brate exoskeletons.[7–12] Many of the proteins containing
hevein domains have been associated with antimicrobial and
plant defense functions.[1–6,9] Additionally, hevein itself (43
amino acids; Scheme 1), found in Hevea brasiliensis
The study of molecular recognition processes from the
chemical perspective is of paramount importance to under-
stand and modulate key processes in nature. Chemical syn-
thesis methods employed in combination with other spectro-
[a] Dr. J. J. Hernꢀndez-Gay,+ Dr. A. Ardꢀ,+ Prof. Dr. F. J. CaÇada,
Prof. Dr. J. Jimꢁnez-Barbero
Chemical and Physical Biology
Centro de Investigaciones Biolꢂgicas, CSIC
Ramiro de Maeztu 9, 28040 Madrid (Spain)
Fax : (+34)91-536-0432
E-mail: jjbarbero@cib.csic.es
jcanada@cib.csic.es
[b] S. Eller, S. Mezzato, Prof. Dr. C. Unverzagt
Bioorganische Chemie, Gebꢃude NWI
Universitꢃt Bayreuth, 95440 Bayreuth (Germany)
[c] Dr. A. Ardꢀ,+ Dr. B. R. Leeflang
Bijvoet Center, Faculty of Sciences
Utrecht University, Padualaan 8
NL-3584 CH Utrecht (The Netherlands)
[+] These two authors have equally contributed to this work
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/chem.201000939.
Scheme 1. Heveinꢄs sequence. Disulfide bridges: 3–18, 12–24, 17–31, and
37–41 are indicated.
Chem. Eur. J. 2010, 16, 10715 – 10726
ꢅ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
10715

latex,[9,13–15] and some other very close homologues present
in several fruits have been identified as major allergens re-
sponsible for allergic syndromes to latex and fruits.[7,10–12]
X-ray crystallography,[11,12] NMR spectroscopy,[13–16] and
microcalorimetry[15,17,18] studies have provided basic structur-
al information about hevein and its binding features to dif-
ferent glycoligands. Indeed, conformational and dynamic
features, as well as thermodynamic data, have been de-
duced.[11–18] From the chemical perspective, there are several
key amino acid residues involved in the recognition of chi-
tooligosaccharides. The aromatic residues placed in the rela-
tive positions Trp 21, Trp 23, and Tyr 30 stabilize the com-
plexes by CH–p interactions[11–23] and van der Waals con-
tacts. Additionally, the hydroxyl groups of the conserved
residues Ser and Tyr (19 and 30 in hevein) are involved in
hydrogen bonding with the acetamide carbonyl group and
the hydroxyl group at position 3 of a key GlcNAc residue
(Figure 1).
Figure 2. Representation of the two possible binding modes of (GlcNAc)3
to hevein. A) Orientation +3, +2, +1. B) Orientation +2, +1, ꢀ1 (see
main text).
bond), whereas the terminal nonreducing end displays some
contacts with the lectin at the so-called subsite ꢀ1.
For longer chitin chains, the process is multivalent and,
depending on the length of the chitooligosaccharide chain,
different hevein domains may simultaneously bind to the
same chitin chain.[15]
Interestingly, and in relation with its allergenic character,
a conformational immunoglobulin binding epitope that in-
cludes the aromatic residues has been described in
hevein[11,12,24,25] and hevein recognition by neutrophils is in-
hibited by chitooligosaccharides and glycoproteins contain-
ing N-glycosidically linked glycans.[26] Furthermore it has
been described that the Urtica dioica lectin (UDA), a cova-
lent hevein-homologue dimer that equally binds chitooligo-
saccharides,[26,27] behaves as a superantigen to T cells, induc-
ing exclusive proliferation of Vbeta 8.3 lymphocytes.[28] In
this case, it seems that a key interaction for the biological
response takes place through the N-glycan chains of mam-
malian glycoproteins.[28] It is well known that the N-glycan
chains in glycoproteins have a common core pentasacchar-
ide. This core contains the N,N’-diacetyl chitobiose unit
(minimal element recognized by hevein) that connects the
glycan chain to the protein through an N-glycosidic bond to
an asparagine side chain (Scheme 2).
Figure 1. Structure of the complex between hevein and N,N’-diacetyl chi-
tobiose (amino acids in the binding site are highlighted).
One of the key features of the molecular recognition pro-
cess of hevein by chitooligosaccharides is that it displays in-
teresting dynamic features. Indeed, when passing from chito-
biose to chitotriose, it has been demonstrated that hevein
domains recognize the chitin trimer in two different man-
ners (Figure 2).[15] There is a first binding mode in which the
terminal nonreducing GlcNAc residue is placed at the so-
called subsite +1, interacting with Trp 23 (CH–p stacking),
with Ser 19 (hydrogen bond), and with Tyr 30 (CH3–p stack-
ing and hydrogen bond). The intermediate residue is located
at subsite +2 interacting with Trp 21 (CH–p stacking),
whereas the reducing GlcNAc residue provides very few
contacts with the lectin, at subsite +3. In the second orien-
tation, the reducing residue is placed at subsite +2, interact-
ing with Trp 21, and the middle residue is placed at subsite
+1, interacting with Trp 23 (CH–p stacking), Ser 19 (hydro-
gen bond), and with Tyr 30 (CH3–p stacking and hydrogen
Scheme 2. General structure of the trisaccharide core region of an N-
glycan chain. In mammalian glycoproteins, the terminal mannose displays
additional substitutions.
To further understand a possible recognition of the sugar
moiety of mammalian N-glycoproteins by hevein domains,
and thus its relationship to the described super-antigen be-
havior of UDA, we have initiated the study of the molecular
recognition features of the interaction process between
hevein and the trisaccharide core of the N-glycoproteins at-
10716
www.chemeurj.org
ꢅ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2010, 16, 10715 – 10726

The 3D Solution Structure of Hevein
FULL PAPER
tached to Asn (Scheme 3) by using a combination of organic
synthesis, NMR spectroscopy, and molecular modeling pro-
cedures. Thus, herein we describe the 3D solution structure
Scheme 3. The structure of the target synthetic models of the trisacchar-
ide N-glycan core: Compound 1 is a trisaccharide N-acetylated at the re-
ducing end and compound 2 a trisaccharide N-linked asparagine at the
reducing end.
at atomic resolution of the corresponding complex, as well
as key atomic and thermodynamic parameters for the inter-
action. We aimed at understanding how the presence of the
mannose residue at the nonreducing end of chitobiose af-
fects the binding to hevein, as well as the effect of the
amino acid moiety at the reducing end. It should be interest-
ing to know how the presence of the additional mannose
residue affects (by reducing or enhancing) the dynamic fea-
tures of the binding process, and the possibility of the
formed complex to further interact with other receptors.
Thus, extension of the N-glycan chain at the mannose resi-
due with natural branches and of the Asn moiety with a
polypeptide chain could allow for an interaction of the N-
glycan-containing glycoprotein with additional receptors.
Scheme 4. Synthesis of compound 1. Bzl=benzyl; DIPEA=N,N-diiso-
propylethylamine.
natural linkage of the core trisaccharide to asparagine com-
bined with two minimal amide bonds of the peptide back-
bone mimicking the linkages of native N-glycopeptides. Ad-
ditionally, by using the N-acetylated glycosylasparagine N-
methylamide instead of the corresponding free asparagine,
electrostatic interactions with the lectin were precluded.
In the synthesis of 2, the introduction of the desired N-
acetylated asparagine N-methyl amide proved unexpectedly
demanding. It was initially planned to couple commercially
available amino acid building blocks to the glycosylamine
generated from the N-acetylated trisaccharide 7. Compound
7 was obtained from 3 in a three-step one-pot dephthaloyla-
tion sequence. After reduction of the azido group of 7 with
propanedithiol (Z)-Asp-OBzl was coupled to the glycosyla-
mine by using benzotriazol-1-yl-oxytripyrrolidinophosphoni-
um hexafluorophosphate (PyBOP) activation (data not
shown). Reaction with aqueous methylamine cleaved O-acy-
lated side products and converted the benzyl ester to the
methyl amide. Removal of the residual protecting groups by
catalytic hydrogenation showed that the main product was
N-methylated by in situ generated formaldehyde as shown
by NMR spectroscopy and mass spectrometry. To avoid N-
methylation, the Boc-protected aspartic acid benzyl ester 5
was employed instead for the reaction sequence (data not
shown). However, after coupling to the glycosylamine fol-
lowed by N-methyl amide formation, catalytic hydrogena-
tion, cleavage of the Boc group, and subsequent N-acetyla-
tion two products of the desired mass were obtained, which
could not be separated. Thus, the N-methyl amide was intro-
duced first by reacting Boc aspartic acid benzyl ester 5 with
methylamine, giving compound 6. Amino acid 6 was activat-
Results and Discussion
Synthesis of trisaccharide 1 and the glycosylamino acid 2:
By starting from synthetic core trisaccharide 3,[29] compound
4 was obtained after a sequence of five reactions[30] per-
formed as a one-pot conversion (Scheme 4). After selective
reduction of the anomeric azido group of trisaccharide 3 by
using propanedithiol, the intermediate glycosylamine was
acetylated. Removal of the phthalimido and O-acetyl groups
followed by complete acetylation and selective O-deacetyla-
tion gave the benzylated trisaccharide 4 in 84% isolated
yield over five steps. The benzyl groups were cleaved by cat-
alytic hydrogenation furnishing the anomerically N-acetylat-
ed core trisaccharide 1 in 60% yield after purification by gel
filtration.
As a step towards the investigation of potential interac-
tions of hevein with glycopeptides containing the core trisac-
charide, a second model compound 2 was designed and syn-
thesized (Scheme 5). The rationale for compound 2 was the
Chem. Eur. J. 2010, 16, 10715 – 10726
ꢅ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
10717

J. Jimꢁnez-Barbero, F. J. CaÇada et al.
was 6 ns and the GB/SA (generalized Born solvent-accessi-
ble surface area) solvent model was used.[33]
The glycosidic torsions were defined as F(H1-C1-O-C4)
and
Y(C1-O-C4-H4)
for
Manb1!4GlcNAc
and
GlcNAcb1!4GlcNAc bonds, and F(H1-C1-N-Cg) and
Y(C1-N-Cg-Cb) for the GlcNAcb1!Asn glycosylamide
bond.
These simulations (Figure 3A) indicated that both Man-
GlcNAc and GlcNAc-GlcNAc linkages present a typical
syn-F/syn-Y conformation, with F and Y values of around
60 and 08, respectively, thus in agreement with the exo-
anomeric effect.[34] In contrast, for the GlcNAc-Asn bond,
the F torsion did not show well-defined minima, but dis-
played fluctuations between 60 and ꢀ608, whereas Y torsion
remained in the 1808 anti-region as expected for an amide-
type bond. The behavior of this glycosyl amide linkage is in
accordance with the torsion angles most frequently found
(Figure 3B) for the GlcNAc-Asn linkage present in N-glyco-
proteins deposited in the PDB.[35]
Conformational study by NMR spectroscopy: The obtained
simulation results were compared to the experimental data
obtained for both compounds 1 and 2 by NMR spectroscopy
to assess the modeling conclusions. Thus, in a first step, the
1H NMR spectroscopic resonances were assigned by using
standard COSY, TOCSY, and ROESY/NOESY experiments
(see Tables S1 and S2 in the Supporting Information).
The proton scalar coupling constant analysis indicated
4
that all the pyranose rings adopt a typical C1 chair geome-
try. The observed J5,6 intermediate values indicate that the
hydroxymethyl moieties are in concordance with a confor-
mational equilibrium between gg:gt conformers in Glc and
Man rings, as described for these residues.[36]
Scheme 5. Synthesis of compound 2. Boc=tert-butoxycarbonyl;
DEPBT=3-(diethoxyphosphoryloxy)-1,2,3-benzotriazin-4(3H)-one;
TFA=trifluoroacetic acid.
In a second step, ROESY and NOESY experiments were
performed to obtain the relevant 3D information.[37,38] In
these spectra, the unique observed inter-residual contacts
were those corresponding to cross-peaks between H1 from
one ring and H4 from the contiguous one. These peaks are
exclusive for a syn-exo-anomeric conformation, which per-
mits us to confirm that this is the adopted geometry by both
glycosidic bonds. The presence of an anti-type disposition
between the anomeric proton of the reducing end GlcNAc1
and the Asn-NHd proton was confirmed by the existence of
an NOE between the Asn-NHd and H2 from GlcNAc1,
thus in accordance to the calculated most-stable conforma-
tion of compound 2 (F angle of around 08 in the right panel
of Figure 3A).
ed with DEPBT to avoid O-acylation of the trisaccharide 8.
Because of residual propanedithiol in the reaction mixture
an excess of activated 6 was added. Evaporation of the sol-
vent gave some O-acylation, which was removed by reaction
with methylamine. After purification by flash chromatogra-
phy, product analysis by NMR spectroscopy and HPLC-MS
showed an anomeric mixture of glycosylamides (a/b 1:2),
which could only be separated by RP-HPLC and led to a re-
duced overall yield (28%). The aromatic protecting groups
of the b-linked trisaccharide-asparagine 8 were removed by
catalytic hydrogenation. The Boc group was cleaved by TFA
followed by selective N-acetylation, leading to the desired
core trisaccharide–asparagine conjugate 2.
The interaction with hevein: Thermodynamic analysis of the
binding: The interaction of 1 and 2 with hevein was studied
by NMR spectroscopy by using a well-established methodol-
Conformations of compounds 1 and 2 in the free state
1
ogy.[13–16] The association was studied by 1D H NMR titra-
Molecular modeling: MD calculations: A conformational
study of both compounds 1 and 2 was carried out by using
molecular dynamics (MD) calculations, which were per-
formed by using the MM3* force field,[31] as implemented in
MAESTRO software.[32] The total time of the simulations
tions by following the procedure described in the Materials
and Methods Section, which is based on the analysis of 1D
spectra recorded for series of samples containing a constant
concentration of polypeptide with increasing ligand concen-
trations.
10718
www.chemeurj.org
ꢅ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2010, 16, 10715 – 10726

The 3D Solution Structure of Hevein
FULL PAPER
Figure 3. A) F versus Y representations for the different glycosidic torsions of 2. From left to right, Man-GlcNAc, GlcNAc-GlcNAc, and GlcNAc-Asn
linkages. B). Frequency distribution of GlcNAc-Asn bond torsion angles F and Y found in the PDB.[35]
The observed perturbations in the chemical shifts of the
peptide upon sugar addition (see Figure S1 in the Support-
ing Information) clearly proved the formation of specific
complexes between hevein and 1 or 2. The continuous varia-
tion of the chemical shifts when increasing quantities of
ligand were added indicated that these changes can be used
to determine the association constant of the equilibrium be-
tween the free and bound species. Since the observed
proton chemical shift changes of the protein are proportion-
al to the molar fraction of the present complex in solution,
the association constants (Ka) were then determined by non-
linear least-square fitting of the observed chemical shifts
perturbations (Figure 4) versus different ligand/receptor
molar ratios. Several amidic protein protons changed their
chemical shifts after ligand addition in a noteworthy way.
For this case, the signal belonging to the Trp 21 side-chain
NH was monitored as a function of the added ligand con-
centration to determine the association constant values.
The titrations were performed at four different tempera-
tures, 298, 303, 308, and 313 K, providing, in the case of the
trisaccharide amino acid 1, Ka values of around 12000,
10000, 8500, and 7000mꢀ1, respectively (Table 1). These
values were used to obtain the thermodynamic parameters
through a van’t Hoff approximation (Figure 4) by the plot of
ln(Ka) versus 1/T.
approximations regarding the lack of heat capacity depend-
ence with temperature, which have not been demonstrated
for these systems, although NMR spectroscopy and microca-
lorimetry data have been shown to be fairly similar for chi-
tobiose and chitotriose binding to hevein and related do-
mains.[13–18] In any case, the observed negative values of DH8
and DS8 are in agreement with an enthalpy-driven process.
The Ka data analysis and the obtained thermodynamic pa-
rameters for the 1/hevein and 2/hevein complexes (Table 1)
permitted us to conclude that the hevein affinity for these
new ligands is much higher than for chitobiose and even
higher than for chitotriose. This fact indicates that the pres-
ence of the mannose residue is indeed stabilizing. In con-
trast, the presence of the Asn moiety is not beneficial for
the interaction, since the affinity for 2 is lower than for 1.
Nevertheless, the origin of this extra stabilization of 2 and
especially 1 versus the (GlcNAc)3 seems to be entropic,
since the interaction enthalpy is better for (GlcNAc)3. In
fact, the change of the nonreducing terminal GlcNAc resi-
due in chitotriose to Man precludes for 1 and 2 the existence
of one of the two existing binding modes of chitotriose to
hevein,[13–16] (Figure 2), with the corresponding loss of en-
thalpic stabilization. Additionally the improvement of the
affinity for compounds 1 and 2 relative to the (GlcNAc)3 tri-
saccharide, existing as a mixture of anomers a and b, could
partially arise from the fixing of the b configuration at the
reducing end in both compounds 1 and 2.
It should be recognized that the use of van’t Hoff plots
should be considered with caution, since there are several
Chem. Eur. J. 2010, 16, 10715 – 10726
ꢅ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
10719

J. Jimꢁnez-Barbero, F. J. CaÇada et al.
(500 and 900 MHz, 1:8 molar
ratio). From the technical view-
point, and within the frame-
work of the FP6 project EU-
ROCarbDB, new tools were in-
troduced to the CCPN data-
model[40] and its derived soft-
ware tools for the knowledge of
carbohydrates down to the
atomic level. In this study, the
feasibility of the CCPN Analy-
sis package[40] was demonstrat-
ed for the computer-assisted
book-keeping of assignments of
these complex (glycopeptide
and protein) molecular systems.
Regarding the cross-peaks cor-
responding to intraprotein pro-
tons, 446 cross-peaks (Figure 5)
could be unequivocally as-
signed, which were basically
identical to those previously
found for hevein when com-
plexed to a variety of chitooli-
gosaccharides.[13–16] Thus, it
could be safely assessed that
the topology, folding and 3D
shape of the lectin is preserved
upon binding to glycosylamino
acid 2. Also, the perturbation of
chemical shifts in the protein
protons was particularly fo-
cused on the region between
1
Figure 4. Titration experiments: A) Amide region of the H NMR spectra at 298 K of hevein at increasing con-
centrations of 2. The chemical shift perturbation of the Trp 21 side-chain NH signal is highlighted. B) Titration
curves at different temperatures for the Trp 21 NH signal: chemical shift variation versus concentrations of 2.
C) van’t Hoff representation for the hevein/2 interaction.
Table 1. Binding affinities and thermodynamic parameters for the interaction of hevein with 1 and 2, com-
pared with those previously determined for chitooligosaccharides (GlcNAc)2, methyl-b (GlcNAc)2, and
(GlcNAc)3.[14]
Ka [mꢀ1
]
Thermodynamic parameters
298 K
303 K
308 K
313 K
DH8 [kJmolꢀ1
]
DS8 [Jmolꢀ1
]
amino acids 19 and 31, thus ex-
clusively corresponding to the
N,N’-diacetyl chitobiose
N,N’-diacetyl methyl b-chitobiose
N,N’,N’’-triacetyl chitotriose
glycosylamino acid 2
620
1225
11500
11863
14898
460
1069
8700
9842
–
381
882
6900
8437
10111
337
647
5700
6792
8245
ꢀ31.3
ꢀ32.1
ꢀ36.4
ꢀ28.3
ꢀ30.4
ꢀ52.5
ꢀ53.5
ꢀ45.1
ꢀ17.0
ꢀ22.2
chitooligosaccharide
binding
region (see Figure S1 in the
Supporting Information).
The use of a very high field
(900 MHz) permitted the detec-
tion of the presence of many in-
trisaccharide 1
When the estimated DH8 and DS8 are compared with
those previously described for the other ligands, it is evident
that the data show an enthalpy/entropy compensation phe-
nomenon, typical in association processes between proteins
and carbohydrates.[39] Better binding enthalpy is often ac-
companied by a larger entropic loss. The enthalpy driven
processes are probably due to van der Waals interactions
and hydrogen bonds in the ligand-receptor complex, as has
been demonstrated in the case of hevein domains. The cor-
responding analysis was corroborated by determining the
3D structure of the complex in solution by NMR spectro-
scopic methods and modeling protocols.
termolecular NOE peaks between the lectin and the ligand
(Table 2). In fact, the intermolecular cross-peaks were ob-
served between the sugar ring protons and the correspond-
ing acetamide methyl groups of the GlcNAc residues with
different protons of the hevein amino acids Trp 21, Trp 23,
Tyr 30, and Ser 19. No cross-peaks between the Asn side
chain of the glycosylamino acid and protons of the protein
could be detected. These experimental data indeed assessed
that the recognition of the glycosylamino acid 2 involves the
same binding site as for chitotriose. More importantly, some
of the intermolecular NOEs permitted the unequivocal loca-
tion of the glycosylamino acid in the binding site of the
lectin, in a +2, +1, ꢀ1 disposition, as shown in Figure 2B;
cross-peaks involving H3 and H4 protons of the GlcNAc
residues, and z3, h2, and d1 protons of Trp 23 and h2 of Trp
21 required some interpretation due to certain overlapping.
NMR spectroscopic structure: The analysis of the NOESY
experiments (100 and 500 ms mixing times) of the complex
between hevein and glycosylamino acid 2 was performed
10720
www.chemeurj.org
ꢅ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2010, 16, 10715 – 10726

The 3D Solution Structure of Hevein
FULL PAPER
Figure 5. Amide region of the NOESY spectrum of hevein/2 (1:8) acquired at 900 MHz with 500 ms mixing time. Inside circles: Some selected intermo-
lecular NOE cross-peaks are listed in Table 2. Inside squares: Intramolecular NOEs of 2, two of which are positive and correspond to the Asn residue.
M=Man, GNAc2=GlcNAc2, and GNAc1=GlcNAc1 (Scheme 3).
Table 2. Intermolecular NOEs between compound
2
and hevein
HdY30 or Me GlcNAc2Hz3W21) would indicate minor
contributions of a different disposition of the ligand in the
binding site in which GlcNAc1 would be located at site +1
or GlcNAc2 on site ꢀ1.
(900 MHz NOESY, 100 and 500 ms mixing), supporting the +2, +1, ꢀ1
sites occupancy, and the corresponding distance in that binding mode.
2 proton
Hevein proton
Distance in final model [ꢆ]
mannose
H3
H5
z3 W23
z3 W23
e3 W23
e3 W23
NHC24
e3 W23
d1 W23
e1 W23
d W23
z2 W23
z2 W23
e1 W23
da Y30
NHS19
z2 W21
e W21
3.8
3.1
2.4
3.6
3.0
2.9
1.9
3.5
3.8
3.8
3.1
3.6
3.1
3.9
3.4
3.3
2.6
From the molecular recognition point of view and the
ligand mobility in the complex, it has to be emphasized that
the intraresidual NOESY cross-peaks for the GlcNAc and
Man residues displayed the same phase as the diagonal
peaks, which indicated that this part of the molecule moves
with a global motion rotational correlation time similar to
that of the protein with relatively slow mobility. On the con-
trary, the NOESY cross-peaks for the amide protons of the
Asn moiety showed the opposite phase to the diagonal,
which indicated a faster effective correlation time, as a pos-
sible consequence of weaker interactions with the protein.
This fact shows that the major interaction of the ligand and
the lectin occurs in the oligosaccharide part with the Asn
moiety being more exposed to the solvent.
H6a
H6b
GlcNAc2
GlcNAc1
H2
H4
H6
Me
H1
H5
e1 W21
The nature of the cross-peaks permitted the confirmation
that the binding of the glycosylamino acid 2 to hevein
occurs mainly in the binding mode with a +2, +1, ꢀ1 dispo-
sition (see Figure 6 and below in the molecular modeling
section).
Thus, all assigned NOE interactions shown in Table 2
indeed indicate the recognition site +2, +1, ꢀ1 as the main
one. Besides, other intermolecular NOEs (Me GlcNAc1/
Chem. Eur. J. 2010, 16, 10715 – 10726
ꢅ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
10721

J. Jimꢁnez-Barbero, F. J. CaÇada et al.
Molecular modeling: Finally,
molecular-modeling calculations
were also performed to inde-
pendently confirm the NMR
spectroscopic results, which
point to the existence of one
unique binding mode, and to
also substantiate the possibility
of adopting a modeling proto-
col to study this type of protein/
carbohydrate complexes. First,
docking calculations were per-
formed with AutoDock.[41] Ad-
ditionally, the modeling ap-
proach should also permit the
the ranges expected for stabilizing interactions. In contrast,
fluctuations with time were observed for the Asn moiety, as
also deduced from the NMR spectroscopic data, by monitor-
ing the distance between its terminal groups and the NH of
Trp 23, also ruling out any kind of important interaction be-
tween the Asn fragment and the lectin binding site.
In addition, the glycosidic torsions for the ligand re-
mained fairly stable during the simulation, close to those de-
duced in the free-state. All these data support the idea of
preorganization of both hevein and its ligands for the molec-
ular recognition process. There are no important changes in
the global shape of the lectin or of ligands 1 and 2 between
the free and bound states, only some restriction to motion
due to the intermolecular protein–ligand contacts.
Figure 6. Binding
hevein (amino acids Ser 19,
Gln 20, Trp 21, Trp 23, Cys 24,
Tyr 30) with compound
docked on site +2, +1, ꢀ1.
site
of
2
Finally, the obtained 3D structure of the complex was em-
ployed to predict the possibility of extensions at the Asn
and Man moieties of the glycosylamino acid 2. Inspection of
Figure 7 permits the assessment that, in the complex with
assessment of a further extension of the polypeptide at the
Asn moiety (within a glycoprotein) and of additional glyco-
sylations at the terminal Man residue.
The two major clusters of structures obtained from the
calculations for both ligands 1 and 2 satisfactorily fitted the
binding orientation described in Figure 6 with different ori-
entations for the Asn moiety.
The major complexes of 1 and 2 with hevein completely
backed up the existence of only one binding orientation
(+2, +1, ꢀ1). The glycosylamino acid 2 is located so that
the reducing GlcNAc residue is placed at subsite +2, the
middle GlcNAc at subsite +1, and the nonreducing Man
ring at ꢀ1, leaving the Asn residue very exposed to the sol-
vent, at subsite +3. The change of the chemical nature of
the sugar and the variation of the stereochemistry at posi-
tion C2 at the nonreducing end moiety (from GlcNAc to
Man) precludes the possibility of establishing proper inter-
actions between the terminal Man and subsite +1. Thus, of
the two possibilities existing for the recognition of chitooli-
gosaccharides, only one possibility of sugar–lectin interac-
tion remains possible in the case of the core trisaccharide of
N-glycoproteins.
Molecular dynamics: Finally, to assess the conformational
stability of the proposed complex, the 3D models obtained
from the docking calculations, in agreement with the NMR
spectroscopic data, were used as input structures for molec-
ular dynamics simulations,[42] by using explicit solvent with
the AMBER 9 program,[43] with no experimental NMR
spectroscopic restraints. The simulation time was beyond
four ns. It was observed that the complex remained com-
pletely stable during the complete simulation, showing
minor motion of the lateral side chains of the essential
amino acids for the molecular recognition; Trp 21, Trp 23,
Ser 19, and Tyr 30. Regarding the key intermolecular con-
tacts between the ligand and the protein (i.e., the hydrogen
bond between Ser 19 OH and the carbonyl group of the in-
termediate GlcNAc, the hydrogen bond between Tyr 30 OH
and OH3 of the intermediate GlcNAc, and the stacking in-
teraction between the acetamide methyl group and the
center of Tyr 30), the corresponding distances were also
monitored, showing a very good stability with time, within
Figure 7. The 3D structure of the complex of hevein/2, showing the posi-
tions for both peptide elongation from the asparagine residue and the
mannose branching at positions 3 and 6 of the terminal Man residue that
would exist in N-glycoproteins.
hevein, both the Asn and Man residues adopt orientations
that allow the extension of the polypeptide chain from the
Asn moiety with no steric hindrance, as well as branching
glycosylation at O-3 and O-6 of the terminal mannose. Thus,
in principle, hevein could also recognize N-linked glycopro-
teins by using the same binding mode outlined above.
Conclusion
NMR spectroscopy and modeling calculations have permit-
ted us to demonstrate that hevein, a model lectin related to
the superantigen family, recognizes the trisaccharide core of
10722
www.chemeurj.org
ꢅ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2010, 16, 10715 – 10726

The 3D Solution Structure of Hevein
FULL PAPER
nol 20:1). Yield: 221 mg (84.2%); Rf (glycosylamine)=0.26 (hexane/ace-
tone 1.5:1), Rf (acetamide)=0.56 (CH2Cl2/methanol 15:1), Rf (diamine)=
0.25 (CH2Cl2/methanol 15:1), Rf (peracetate)=0.31 (CH2Cl2/methanol
15:1), Rf (4)=0.27 (CH2Cl2/methanol 15:1); [a]2D5 =ꢀ29.4 (c=0.5 in
N-glycoproteins (1) and also an Asn-linked glycosylamino
acid 2 with even higher affinity than (GlcNAc)3. The binding
mode is similar to those described for regular chitooligosac-
charides. However, for the core trisaccharide, the mode of
binding selects only one of the possibilities existing for chi-
tooligosaccharides. The deduced 3D structure of the com-
plex permits the extension of polypeptide chains from the
Asn moiety, as well as branching glycosylation at O-3 and
O-6 of the terminal mannose. Given the ubiquity of the
CH2Cl2); ESI-MS (100% acetonitrile): m/z: calcd for C59H69N3O16
1075.5; found: 1098.3 [M+Na]+.
:
O-b-d-Mannopyranosyl
G
ACHTUNGTRENNUNGpyran-
A
ACHTUNGTRENNUNG
Trisaccharide 4 (125 mg, 116.2 mmol) was dissolved in a mixture of freshly
distilled methanol (16 mL) and acetic acid (4 mL). Palladium(II) oxide
hydrate (120 mg) was added and the suspension was stirred under a hy-
drogen atmosphere for 3 d. After disappearance of 4 (TLC: isopropanol/
1m ammonium acetate 4:1), the solution was diluted with methanol and
the catalyst was removed by filtration. After evaporation of the solvents,
the crude product was purified by gel-filtration chromatography (Phar-
macia Hi Load Superdex 30 prep grade (600ꢇ16 mm), 0.1m ammonium
hydrogen carbonate in water, flow rate=0.75 mLminꢀ1). Yield: 43.8 mg
(60.0%), Rf =0.33 (isopropanol/1m ammonium acetate 4:1); [a]2D5 =ꢀ17.9
(c=1.43 in water); ESI-MS (water): m/z: calcd for C24H41N5O16: 627.3;
found: 650.8 [M+Na]+; 1H NMR (360 MHz, D2O+[D6]DMSO as an in-
ternal standard): d=4.84 (d, J1,2 =9.5 Hz, 1H; H-11), 4.57 (d, J1,2 <1 Hz,
1H; H-13), 4.41 (d, J1,2 =7.2 Hz, 1H; H-12), 3.86 (dd, J1,2 <1, J2,3 =3.0 Hz,
1H; H-23), 3.76–3.33 (m, 16H; H-6a3, H-6a2, H-21, H-6a1, H-22, H-6b2, H-
32, H-31, H-42, H-6b3, H-6b1, H-33, H-41, H-52, H-43, H-51), 3.22 (m, 1H;
H-53), 1.86 (s, 3H; NAc), 1.80 ppm (s, 6H; NAc); 13C NMR (90 MHz,
D2O+[D6]DMSO as an internal standard): d=176.4, 176.2, 176.0 (C=O
NAc), 102.9 (C-12), 101.6 (C-13), 80.3 (C-41), 80.2 (C-42), 79.9 (C-11), 77.9
(C-53), 77.8 (C-51), 76.2 (C-52), 74.3 (C-33), 74.3 (C-31), 73.5 (C-32), 72.1
(C-23), 68.2 (C-43), 62.5 (C-63), 61.6 (C-62), 61.4 (C-61), 56.6 (C-22), 55.3
(C-21), 23.7, 23.6, 23.5 ppm (NAc).
ManACHTUNGTRENNUNG(GlcNAc)2 core in all mammalian N-glycoproteins, the
basic recognition mode presented herein might be extended
to a variety of systems with increased complexity and with
biomedical importance. In any case, and due to the very im-
portant issue that latex allergies bring in to public health,
the fact that the carbohydrate binding site of hevein could
accept N-glycoproteins and, at the same time, is part of the
immunoglobuling binding epitope,[24–26] makes it interesting
to explore the potential beneficial use of chitin or its oligo-
saccharide derivatives,[26] mimics, and analogues in protec-
tive formulations against latex allergies (or detrimental be-
havior acting as local concentrators of the hevein antigen).
It should be pointed that chitin itself is already present in di-
verse commercial skincare formulations.
Experimental Section
O-(4,6-O-Benzylidene-b-d-mannopyranosyl)
di-O-benzyl-2-deoxy-b-d-glucopyranosyl)
(1!4)-2-acetamido-3,6-di-O-
benzyl-2-deoxy-b-d-glucopyranosylazide (7): Trisaccharide (200 mg,
G
AHCTUNGTRENNUNG
General methods: Solvents were dried according to standard methods.
Optical rotations were measured on a Perkin–Elmer 241 polarimeter at
589 nm. NMR spectra were recorded on a Bruker Avance 360 instru-
ment. Coupling constants are reported in Hz. ESI-TOF mass spectra
were recorded on a Micromass LCT instrument coupled to an Agilent
1100 HPLC. Flash chromatography was performed on silica gel 60 (230–
400 mesh, Merck Darmstadt). The reactions were monitored by TLC on
coated aluminum plates (silica gel 60 GF254, Merck Darmstadt). Spots
were detected by UV light or by charring with a 1:1 mixture of 2n
H2SO4 and 0.2% resorcine monomethylether in ethanol.
0.16 mmol) was dissolved in n-butanol (26 mL). Ethylenediamine
(6.5 mL, 97.2 mmol) was added and the solution was stirred at 808C.
After 20 h (TLC: CH2Cl2/methanol 15:1), the volatiles were removed
under reduced pressure. The residue was co-distilled with toluene (3ꢇ)
and dried under high vacuum. Subsequently, the solid was treated with
pyridine/acetic anhydride (20 mL, 2:1) for 3 h at room temperature.
After complete acetylation (TLC: CH2Cl2/methanol 15:1), the reagents
were removed in vacuo followed by co-distillation with toluene (3ꢇ).
The dried residue was dissolved in methanol (10 mL) and aqueous meth-
ylamine (13 mL, 40%) was added. After disappearance of the starting
material (13 h, TLC: CH2Cl2/methanol 15:1), the mixture was concentrat-
ed in vacuo and the residue co-distilled with toluene (3ꢇ) and dried
under high vacuum. The crude product was purified by flash chromatog-
raphy (CH2Cl2/methanol 20:1). Yield: 127 mg (74.1%); Rf (amine)=0.40
(CH2Cl2/methanol 15:1), Rf (peracetate)=0.46 (CH2Cl2/methanol 15:1),
Rf (7)=0.18 (CH2Cl2/methanol 15:1); [a]2D5 =ꢀ45.4 (c=0.5 in CH2Cl2);
ESI-MS (50% acetonitrile): m/z: calcd for C57H65N5O15: 1059.5; found:
1060.5 [M+H]+, 1082.6 [M+Na]+, 1098.5 [M+K]+; 1H NMR (360 MHz,
Synthesis
O-(4,6-O-Benzylidene-b-d-mannopyranosyl)
di-O-benzyl-2-deoxy-b-d-glucopyranosyl)
(1!4)-2-acetamido-N1-acetyl-
3,6-di-O-benzyl-2-deoxy-b-d-glucopyranosylamine (4): Trisaccharide
(1!4)-O-(2-acetamido-3,6-
AHCTUNGTRENNUNG
3
(301 mg, 0.16 mmol) was dissolved in freshly distilled methanol
(12.5 mL). DIPEA (420 mL) and 1,3-propanedithiol (1.25 mL, 12.4 mmol)
were added and the solution was stirred at room temperature. After 2 h
(TLC: hexane/acetone 1.5:1), the solvent was removed under reduced
pressure and the residue was dried under high vacuum. The remaining
mixture was treated with pyridine/acetic anhydride (30 mL, 2:1) at room
temperature. After complete acetylation (TLC: CH2Cl2/methanol 15:1),
the solvent was removed in vacuo and the residue was co-distilled with
toluene (3ꢇ) and subsequently dried under high vacuum. The residue
was dissolved in mixture of n-butanol (28 mL) and ethylenediamine
(7 mL, 0.10 mol) and the solution was stirred at 908C. After 17 h (TLC:
CH2Cl2/methanol 15:1), the volatiles were removed under reduced pres-
sure and the residue was co-distilled with toluene (3ꢇ) and dried under
high vacuum. The remaining solid was treated with pyridine/acetic anhy-
dride (30 mL, 2:1) for 1 h (TLC: CH2Cl2/methanol 15:1). After complete
acetylation, the solvent was removed in vacuo and the residue was co-dis-
tilled with toluene (3ꢇ) and dried under high vacuum. The residue was
dissolved in aqueous methylamine (20 mL, 40%) and stirred for 1 h at
room temperature. After disappearance of the starting material (TLC:
CH2Cl2/methanol 15:1), the volatiles were removed in vacuo and the re-
mainder was co-distilled with toluene (3ꢇ) and dried under high vacuum.
The crude product was purified by flash chromatography (CH2Cl2/metha-
[D6]DMSO): d=8.03 (d, JNH,2 =8.9 Hz, 2H; NH), 7.49–7.13 (m, 25H;
3
ꢀ
Ar), 5.51 (s, 1H; Ph CH), 5.00–4.89 (m, 3H; CH2O, CH2O, OH-3 ), 4.86
(d, JOH,2 =4.3 Hz, 1H; OH-23), 4.66–4.41 (m, 8H; CH2O, H-12, CH2O, H-
11, CH2O, H-13, CH2O, CH2O), 4.38 (d, Jgem =12.5 Hz, 1H; CH2O), 3.96–
3.42 (m, 16H; H-6a3, H-41, H-42, H-6a1, H-21, H-23, H-6b1, H-43, H-6a2,
H-32, H-22, H-51, H-31, H-6b2, H-6b3, H-33), 3.27–3.20 (m, 1H; H-52),
3.07–2.99 (m, 1H; H-53), 1.81 (s, 3H; OAc), 1.79 ppm (s, 3H; OAc);
13C NMR (90 MHz, [D6]DMSO): d=169.3, 169.2 (C=O NAc), 139.3,
139.2, 138.5, 138.3, 137.9, 128.7, 128.2, 127.7, 127.2, 127.0, 126.3 (C Ar),
3
2
1
1
ꢀ
101.0 (Ph CH), 100.6 (C-1 ), 100.0 (C-1 ), 88.0 (C-1 ), 80.5 (C-3 ), 79.9
(C-32), 78.3 (C-43), 77.0 (C-42), 76.3 (C-51), 74.9 (C-41), 74.4 (C-52), 73.6
(CH2O), 73.4 (CH2O), 72.2 (CH2O), 71.9 (CH2O), 70.9 (C-23), 70.0 (C-
33), 68.5 (C-62), 68.1 (C-61), 67.9 (C-63), 66.8 (C-53), 55.3 (C-22), 53.6 (C-
21), 22.9, 22.8 ppm (NAc).
N2-tert-Butyloxycarbonyl-l-aspartic acid-1-methylamide (6): Boc aspartic
acid benzl ester 5 (200 mg, 0.69 mmol) was dissolved in freshly distilled
methanol (4 mL). Aqueous methylamine (4 mL, 40%) was added and
Chem. Eur. J. 2010, 16, 10715 – 10726
ꢅ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
10723

J. Jimꢁnez-Barbero, F. J. CaÇada et al.
the solution was stirred at room temperature. After 2 h, the solvents
were removed by lyophilization and the residue was purified by ion-ex-
change chromatography (DOWEX 50WX8–100, column: 20ꢇ500 mm,
eluent: water). Yield: 141.7 mg (84.1%); Rf =0.35 (CH2Cl2/methanol 10:1
with 0.1% acetic acid); [a]2D4 =+6.3 (c=0.9 in water); IR (KBr): n˜ =
1732, 1695, 1643 cmꢀ1 (C=O); ESI-MS (100% water): m/z: calcd for
C10H18N2O5: 246.1; found: 493.2 [2M+H]+, 515.2 [2M+Na]+, 531.2
[2M+K]+; 1H NMR (360 MHz, [D6]DMSO): d=12.22 (s, 1H; CO2H),
The filtrate was lyophilized and used in the next step without further pu-
rification. Rf =0.60 (isopropanol/1m ammonium acetate 4:1); ESI-MS
(100% water): m/z: calcd: 813.4; found: 814.3 [M+H]+, 836.3 [M+Na]+,
1649.8 [2M+Na]+. The lyophilized solid was dissolved in trifluoroacetic
acid (0.3 mL) and kept for 5 min at room temperature. The trifluoroace-
tic acid was removed under reduced pressure followed by high vacuum.
The remainder was dissolved in aqueous acetic acid (10%) and lyophi-
lized. Rf (amine)=0.40 (isopropanol/1m ammonium acetate 2:1); ESI-
MS (100% water): m/z: calcd: 713.3; found: 736.4 [M+Na]+.
7.72 (d,
JNH,Me =4.6 Hz, 1H; NHMe), 6.99 (d, JNH,a =8.2 Hz, 1H;
NHBoc), 4.24–4.16 (m, 1H; aCH-Asn), 2.62–2.56 (m, 1H; bCHa-Asn),
The amine was dissolved in a mixture of freshly distilled methanol
(2 mL) and water (0.3 mL). Acetic anhydride (0.2 mL) was added and
the reaction mixture was stirred for 90 min. After complete conversion of
the starting material (TLC: isopropanol/1m ammonium acetate 2:1), the
mixture was concentrated and the residue was dried under high vacuum.
The crude product was purified by gel-filtration chromatography (Bio-
Gel P-4 fine (750ꢇ15 mm), eluent: water, flow rate=1.5 mLminꢀ1).
Yield: 5.1 mg (80.3%); Rf (2)=0.56 (isopropanol/1m ammonium acetate
2:1); [a]2D3 =ꢀ11.2 (c=0.5 in water); ESI-MS (100% water): m/z: calcd
for C29H49N5O18: 755.3; found: 778.6 [M+Na]+; 1H NMR (360 MHz,
D2O): d=4.89 (d, J1,2 =9.7 Hz, 1H; H-11), 4.61 (d, J1,2 <1 Hz, 1H; H-13),
4.50 (dd, Ja,b =6.6 Hz, 1H; aCH-Asn), 4.45 (d, J1,2 =7.7 Hz, 1H; H-12),
3.90 (dd, J1,2 <1, J2,3 =2.7 Hz, 1H; H-23), 3.80–3.71 (m, 2H; H-6a3, H-
6a2), 3.70–3.54 (m, 8H; H-21, H-6a1, H-22, H-32, H-6b2, H-31, H-42, H-
6b3), 3.54–3.37 (m, 6H; H-33, H-41, H-6b1, H-52, H-43, H-51), 3.30–3.22
(m, 1H; H-53), 2.67–2.51 (m, 5H; bCHa,b-Asn, Me), 1.90 (s, 3H;
NHAc), 1.87 (s, 3H; NHAc), 1.85 ppm (s, 3H; NAc); 13C NMR
(90 MHz, D2O): m/z: d=174.3, 174.1, 173.7, 172.3, 172.2 (C=O NHAc,
Asn), 100.8 (C-12), 99.7 (C-13), 78.2 (C-42), 78.2 (C-41), 77.8 (C-11), 76.0
(C-53), 75.7 (C-51), 74.2 (C-52), 72.4 (C-31), 72.3 (C-33), 71.5 (C-32), 70.1
(C-23), 66.2 (C-43), 60.5 (C-63), 59.7 (C-62), 59.5 (C-61), 54.6 (C-22), 53.2
(C-21), 49.9 (C-a Asn), 36.5 (C-b Asn), 25.5 (Me), 21.7, 21.4 ppm
(NHAc).
2.55 (d,
JNH,Me =4.6 Hz, 1H; Me), (dd, Ja,b =8.6, Jgem =16.1 Hz, 1H;
bCHb-Asn), 1.36 ppm (s, 9H; tBu); 13C NMR (90 MHz, [D6]DMSO): d=
171.9, 171.3, 155.1 (C=O), 78.2 (qC tBu), 51.0 (C-a Asn), 36.5 (C-b Asn),
28.2 (tBu), 25.8 ppm (Me).
N4-[O-(4,6-O-Benzylidene-b-d-mannopyranosyl)
(1!4)-O-(2-acetamido-
3,6-di-O-benzyl-2-deoxy-b-d-glucopyranosyl)
AHCTUNGTRENNUNG
O-benzyl-2-deoxy-b-d-glucopyranosyl)]-N2-tert-butyloxycarbonyl-l-aspar-
agine methylamide (8): Trisaccharide 7 (25 mg, 23.6 mmol) was dissolved
in freshly distilled methanol (2 mL) under an argon atmosphere. Triethyl-
amine (131 mL, 0.94 mmol) and 1,3-propanedithiol (473 mL, 4.7 mmol)
were added and the solution was stirred at room temperature. After 4 h
(TLC: CH2Cl2/methanol 15:1), the mixture containing the glycosylamine
was concentrated under reduced pressure and dried under high vacuum.
Boc aspartic acid methylamide (6) (116 mg, 0.47 mmol) and DEPBT
(141 mg, 0.47 mmol) were dissolved in freshly distilled CH2Cl2 (1 mL)
and DIPEA (202 mL, 1.2 mmol) was added. After shaking for 10 min, this
solution was added to the freshly prepared glycosylamine under an argon
atmosphere and the reaction mixture was stirred for 40 min. After com-
plete reaction (TLC: CH2Cl2/methanol 10.1) the solvent was removed
under reduced pressure and the residue was dried under high vacuum.
The remaining mixture was dissolved in methanol (2.5 mL) and treated
with aqueous methylamine (2.5 mL, 40%). After disappearance of the
faster migrating spots (TLC: CH2Cl2/methanol 10:1), the mixture was
lyophilized. The crude product was purified by flash chromatography
(CH2Cl2/methanol 20:1), followed by HPLC (Agilent C8 XBD 15ꢇ
4.65 mm, gradient: 50!65% acetonitrile/water (0.1% formic acid), flow
rate=1 mLminꢀ1). Yield: 8.2 mg (27.5%); Rf (glycosylamine)=0.30
Conformational analysis
NMR spectroscopy: The corresponding spectra for structure determina-
tion of the complex were recorded at 800 MHz in a Bruker Avance spec-
trometer. The samples for free and bound Hev32S19D (0.5 mm) were
prepared in
a buffer (90% H2O/10% D2O, 100 mm NaCl, 20 mm
(CH2Cl2/methanol 10:1), Rf (8)=0.36 (CH2Cl2/methanol 10:1); [a]D23
=
NaH2PO4, pH 5.6). TOCSY[44] (50 and 70 ms of mixing time) experiments
were performed by using standard sequences at 298 K, by using the Wa-
tergate module for water suppression. NOESY[45] experiments were ac-
quired with 200 and 300 ms of mixing times at 298 K, by using the Water-
gate module for water suppression.
ꢀ17.7 (c=0.3 in methanol); ESI-MS (50% acetonitrile): m/z: calcd for
C67H83N5O19: 1261.6; found: 1262.5 [M+H]+, 1284.6 [M+Na]+; 1H NMR
(360 MHz, [D6]DMSO): d=8.29 (d, JNH,1 =8.7 Hz, 1H; NH), 8.01 (d,
J
J
NH,2 =8.4 Hz, 1H; NHAc2), 7.85 (d, JNH,2 =8.9 Hz, 1H; NHAc1), 7.63 (d,
NH,Me =4.21 Hz, 1H; NHMe), 7.44–7.13 (m, 15H; Ar), 6.60 (d, JNH,a
=
Titration experiments: Titration experiments were performed by record-
ing a series of 1D 1H NMR spectra, in a Bruker Avance 500 MHz spec-
trometer, for different mixtures of hevein with the glycosylamino acid 2,
by following the procedure previously described.[13–15] Firstly, the
1H NMR spectra of two samples were recorded: for one 0.5 mL aliquot
of a 5 mL solution of hevein, as a zero-point of the titration, and for a
0.5 mL aliquot of a 5 mL solution of a mixture of hevein (0.28 mm) and
the corresponding ligand (8 mm), as final points of the titration, corre-
sponding to highest ligand/peptide ratio (ca. 29:1). To build up the titra-
tion curve, small aliquots of the highest ligand/peptide ratio sample were
added to the ligand-free peptide sample in a systematic way, as previous-
ly described.[13–15] For each titration point, with different concentrations
of carbohydrate, but maintaining a constant concentration of hevein, the
1H NMR spectra were acquired at four different temperatures (298, 303,
308, and 313 K). These data allowed the qualitative estimation of the
thermodynamic parameters (DS0 and DH0) of the interaction of hevein
with both oligosaccharides, by using van’t Hoff plots. It should be recog-
nized that the use of van’t Hoff plots should be considered with caution,
since there are several approximations regarding the lack of heat capacity
dependence with temperature that have not been demonstrated for these
systems.
ꢀ
8.4 Hz, 1H; NHBoc), 5.50 (s, 1H; Ph CH), 4.99–4.88 (m, 4H; CH2O, H-
11, CH2O, OH-33), 4.81 (d, JOH,2 =4.5 Hz, 1H; OH-23), 4.68–4.36 (m, 8H;
H-12, H-13, CH2O, CH2O, CH2O, CH2O, CH2O, CH2O), 4.23–4.14 (m,
1H; aCH-Asn), 3.95–3.85 (m, 2H; H-6a3, H-41), 3.83–3.42 (m, 13H; H-
42, H-21, H-6a1, H-43, H-23, H-32, H-6b1, H-31, H-22, H-6a2, H-6b3, H-6b2,
H-33), 3.42–3.17 (m, 2H; H-51, H-52), 3.08–2.99 (m, 1H; H-53), 2.54 (d,
J
NH,Me =4.2 Hz, 3H; Me), 2.46–2.28 (m, 2H; bCHa,b-Asn), 1.82 (s, 3H;
NHAc), 1.76 (s, 3H; NHAc), 1.35 ppm (s, 9H; tBu), 13C NMR (90 MHz,
[D6]DMSO): d=171.6, 169.9, 169.4, 165.8, 165.1 (C=O NHAc, Asn, Boc),
139.3, 138.5, 138.4, 137.9, 128.7, 128.2, 127.9, 127.8, 127.6, 127.3, 127.2,
127.1, 127.0, 126.9, 126.3 (C Ar), 100.8 (Ph-CH), 100.4 (C-13), 99.6 (C-12),
81.4 (C-31), 79.6 (C-32), 78.2 (qC tBu), 78.1 (C-11b, 1JC-1,H-1 =156.6 Hz),
78.1 (C-43), 76.8 (C-42), 75.9 (C-51), 74.4 (C-41), 74.2 (C-52), 73.5 (CH2O),
73.3 (CH2O), 72.2 (CH2O), 71.8 (CH2O), 71.0 (C-23), 70.0 (C-33), 68.6 (C-
62), 67.9 (C-61), 66.8 (C-63), 66.8 (C-53), 55.4 (C-22), 53.3 (C-21), 50.9 (C-a
Asn), 37.4 (C-b Asn), 28.1 (CH3 tBu), 25.7 (Me), 22.9, 22.8 ppm (NHAc).
N4-[O-b-d-Mannopyranosyl
(1!4)-O-(2-acetamido-2-deoxy-b-d-gluco-
A
ACHTUNGTRENNUNG
asparagine methylamide (2): Palladium(II) oxide hydrate (23 mg) was
suspended in freshly distilled methanol (0.5 mL) containing acetic acid
(50 mL) and stirred under a hydrogen atmosphere. After 3.5 h, a solution
of glycosylated amino acid 8 (10.6 mg, 8.4 mmol) in freshly distilled meth-
anol (1.2 mL) containing acetic acid (120 mL) was added and the suspen-
sion was stirred for 2 d under a hydrogen atmosphere. After disappear-
ance of 8 (TLC: isopropanol/1m ammonium acetate 4:1), the catalyst was
removed by centrifugation and washed with methanol and water (3ꢇ).
Structure determination: The 3D structure of free and ligand-bound
hevein has already been determined.[13–15] In any case,
a complete
NOESY cross-peak assignment was again performed to deduce the inter-
molecular ligand–hevein cross-peaks. In a first step, the spin systems of
all amino acids and sugar residues that constitute the protein and the gly-
10724
www.chemeurj.org
ꢅ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2010, 16, 10715 – 10726

The 3D Solution Structure of Hevein
FULL PAPER
cosylamino acid 2 were assigned through the CcpNmr Analysis pro-
gram.[40] Subsequently, the cross-peak volumes were determined by
manual peak integration. The CYANA program (version 2.1)[46] was used
to calculate the structure, by following the standard protocol through
seven iterative cycles, by starting with 100 randomized conformers. The
20 best conformers, with the lowest final CYANA target function values,
were retained for analysis and used as starting geometries for the next
cycle. The obtained folding was basically identical to that previously re-
ported.
Integrated Infrastructure Initiative—Contract # RII3-026145. C.U. thanks
the Deutsche Forschungsgemeinschaft and Fonds der Deutschen Chemi-
schen Industrie for financial support.
[1] C. Bewley, Protein–Carbohydrate Interactions and Infectious Diseas-
es, RSC, London, 2005.
In any case, the best CYANA structure was minimized in a box of explic-
it water molecules, by using the conjugated gradient method, with the
AMBER 9 program.[43] The free peptide was immersed in a TIP3P water
box (3000 molecules), with a thickness of 12 ꢆ. The restrained energy
minimization process was carried out as follows: Initially, to eliminate
the bad contacts between the water molecules and the polypeptide, a
500 steps minimization was performed only to the water molecules, keep-
ing the position of the peptide atoms fixed, and by using a force constant
of 100 kcalmolꢀ1 and constant volume. A subsequent minimization was
then carried out for which the peptide was relaxed with the NOE-based
experimental restrictions and the water molecules were kept fixed. Final-
ly, the restrained energy minimization was performed by taking into ac-
count both the solvent and the peptide, by using 3000 steps with the force
field of Cornell et al.[47]
[2] H.-J. Gabius, The Sugar Code: Fundamentals of Glycosciences,
Wiley-VCH, Weinheim, 2009.
[3] H.-J. Gabius, H.-C. Siebert, S. Andrꢁ, J. Jimꢁnez-Barbero, H. Rꢈdig-
er, ChemBioChem 2004, 5, 740.
[4] H. L. N. Sharon, in Glycosciences: Status and Perspectives (Eds.: H.-
J. Gabius), Chapman & Hall, New York, 1997, p. 133.
[5] O. Seitz, ChemBioChem 2000, 1, 214.
[6] M. R. Wormald, A. J. Petrescu, Y. L. Pao, A. Gilthero, T. Elliott,
R. A. Dwek, Chem. Rev. 2002, 102, 371.
[7] H. Alenius, N. Kalkkinen, T. Reunala, K. Turjanmaa, T. Palosuo, J.
Immunol. 1996, 156, 1618.
[8] B. L. Archer, Biochem. J. 1960, 75, 236.
[9] J. J. Beintema, FEBS Lett. 1994, 350, 159.
[10] A. Diaz-Perales, C. Collada, C. Blanco, R. Sanchez-Monge, T. Car-
rillo, C. Aragoncillo, G. Salcedo, J. Allergy Clin. Immunol. 1998,
102, 127.
[11] C. A. Reyes-Lꢂpez, A. Hernꢀndez-Santoyo, M. Pedraza-Escalona,
G. Mendoza, A. Hernꢀndez-Arana, A. Rodrꢉguez-Romero, Bio-
chem. Biophys. Res. Commun. 2004, 314, 123.
[12] A. Rodrꢉguez-Romero, K. G. Ravichandran, M. Soriano-Garcꢉa,
FEBS Lett. 1991, 291, 307.
[13] J. L. Asensio, F. J. CaÇada, M. Bruix, C. Gonzalez, N. Khiar, A. Ro-
drꢉguez-Romero, J. Jimenez-Barbero, Glycobiology 1998, 8, 569.
[14] J. L. Asensio, F. J. CaÇada, M. Bruix, A. Rodrꢉguez-Romero, J. Jime-
nez-Barbero, Eur. J. Biochem. 1995, 230, 621.
[15] J. L. Asensio, F. J. CaÇada, H.-C. Siebert, J. Laynez, A. Poveda,
P. M. Nieto, U. M. Soedjanaamadja, H.-J. Gabius, J. Jimꢁnez-Bar-
bero, Chem. Biol. 2000, 7, 529.
[16] N. H. Andersen, B. Cao, A. Rodrꢉguez-Romero, B. Arreguin, Bio-
chemistry 1993, 32, 1407.
Molecular modeling
The glycosylamino acid 2: The low-energy conformers of the glycosylami-
no acid 2 were calculated by using the MM3* force field, in MAESTRO.
The F/Y torsion angles were defined as Man H-1’’-Man C-1’’-O-GlcNAc
C-4ꢄ and GlcNAc H-4’-GlcNAc C-4’-O-Man C-1’’and GlcNAc H-1’-
GlcNAc C-1’-O-GlcNAc C-4 and GlcNAc H-4-GlcNAc C-4-O-GlcNAc
C-1’, respectively, for the terminal nonreducing and the intermediate gly-
cosidic linkages, and GlcNAc H-1-GlcNAc C-1-N-CH2 and GlcNAc C-1-
N-CH2-CH for the glycosylamino acid linkage. Potential-energy maps
were generated in a systematic manner as previously described.[34] In all
the molecular mechanics and dynamics calculations, the GB/SA solvation
model for water was used.
Also, molecular dynamics simulations were performed by using the
MM3* force field. A temperature of simulation of 300 K was employed
with a time step of 1.5 fs and an equilibration time of 100 ps. The total
simulation time was 5 ns.
[17] E. Garcꢉa-Hernꢀndez, R. A. Zubillaga, A. Rojo-Domꢉnguez, A. Ro-
drꢉguez-Romero, A. Hernꢀndez-Arana, Proteins 1997, 29, 467.
[18] A. Hernꢀndez-Arana, A. Rojo-Domꢉnguez, M. Soriano-Garcꢉa, A.
Rodrꢉguez-Romero, Eur. J. Biochem. 1995, 228, 649.
[19] M. C. Fernandez-Alonso, F. J. CaÇada, J. Jimenez-Barbero, G.
Cuevas, J. Am. Chem. Soc. 2005, 127, 7379.
For the C5-C6 torsion of the GlcNAc and Man moieties, only the gt ge-
ometry (w, defined as C4-C5-C6-O6, ca. 1808) was considered.
Autodock: The most stable conformers of the glycosylamino acid, in
agreement with the exo-anomeric effect for both glycosidic linkages (as
also observed by NMR spectroscopy) was docked into the deduced NMR
spectroscopic structure of hevein for the complex. AutoDock 4.0 simula-
tions with the multiple Lamarckian Genetic Algorithm were performed.
Only local searches were performed centered around the known GlcNAc
binding site.[11,12] Grids of probe atom interaction energies and electro-
static potential were generated by the AutoGrid program present in Au-
toDock 4.0.[41] A grid spacing of 0.275 ꢆ was used. 100 docking runs were
performed by using a population of 200 individuals and an energy evalua-
tion number of 3ꢇ106.
[20] M. Muraki, Protein Pept. Lett. 2002, 9.
[21] M. Muraki, H. Morii, K. Harata, Protein Eng. 2000, 13, 385.
[22] G. Terraneo, D. Potenza, A. Canales, J. Jimenez-Barbero, K. K. Bal-
dridge, A. Bernardi, J. Am. Chem. Soc. 2007, 129, 2890.
[23] S. Vandenbussche, D. Dꢉaz, M. C. Fernꢀndez-Alonso, W. Pan, S. P.
Vincent, G. Cuevas, F. J. CaÇada, J. Jimꢁnez-Barbero, K. Bartik,
Chem. Eur. J. 2008, 14, 7570.
[24] M. Pedraza-Escalona, B. Becerril-Lujꢀn, C. Agundis, L. Domꢉnguez-
Ramꢉrez, A. Pereyra, L. RiaÇo-Umbarila, A. Rodrꢉguez-Romero,
Mol. Immunol. 2009, 46, 668.
[25] E. Rojas, P. Llinas, A. Rodrꢉguez-Romero, C. Hernꢀndez, M. Li-
nares, E. Zenteno, R. Lascurain, Glycoconjugate J. 2001, 18, 339.
[26] R. T. Lee, H.-J. Gabius, Y. C. Lee, Glycoconjugate J. 1998, 15, 649.
[27] K. Harata, M. Muraki, J. Mol. Biol. 2000, 297, 673.
[28] F. A. Saul, P. Rovira, G. Boulot, E. J. M. Van Damme, W. J. Peu-
mans, P. Truffa-Bachi, G. A. Bentley, Structure 2000, 8, 593.
[29] C. Unverzagt, Chem. Eur. J. 2003, 9, 1369.
Molecular dynamics: Finally, 4 ns MD were run of the complex deduced
by Audock employing the protocol described above for the NMR spec-
troscopic structure determination but without the experimental restraints.
The complex structure was found to be completely stable during the MD
run, as described in the text.
Acknowledgements
[30] C. Unverzagt, Carbohydr. Res. 1997, 305, 423.
We thank Prof. Dr. Hans Kamerling for his interest and for helpful dis-
cussions. We thank the EU for financial support (MRTN CT2006-
035546) and the MICINN (Spain) for grants CTQ2006-10874-C02-01 and
CTQ2009-8536. We also thank the EU for RIDS Contract number
011952 (for EUROCarbDB: http://www.eurocarbdb.org) and EU-NMR
(http://www.eu-nmr.eu) for the EC FP6 Research Infrastructure action—
[31] N. L. Allinger, Y. H. Yuh, J. H. Lii, J. Am. Chem. Soc. 1989, 111,
8551.
[32] Maestro, A powerful, all-purpose molecular-modeling environment,
Schrodinger LLC, 2005.
[33] W. C. Still, A. Tempczyk, R. C. Hawley, T. Hendrickson, J. Am.
Chem. Soc. 1990, 112, 6127.
Chem. Eur. J. 2010, 16, 10715 – 10726
ꢅ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
10725

J. Jimꢁnez-Barbero, F. J. CaÇada et al.
[34] J. L. Asensio, F. J. CaÇada, A. Garcia-Herrero, M. T. Murillo, A. Fer-
nandez-Mayoralas, B. A. Johns, J. Kozak, Z. Zhu, C. R. Johnson, J.
Jimenez-Barbero, J. Am. Chem. Soc. 1999, 121, 11318.
[35] Calculated with Glytorsion (http://www.glycosciences.de/tools); T.
Lꢈtteke, M. Frank, C.-W. von der Lieth, Nucleic Acids Res. 2005, 33,
D242.
[36] K. Bock, J. O. Duus, J. Carbohydr. Chem. 1994, 13, 513.
[37] A. Imberty, S. Perez, Chem. Rev. 2000, 100, 4567.
[38] J. Jimꢁnez-Barbero, J. L. Asensio, F. J. CaÇada, A. Poveda, Curr.
Opin. Struct. Biol. 1999, 9, 549.
[43] AMBER 9, D. A. Case, T. A. Darden, T. E. Cheatham III, C. L.
Simmerling, J. Wang, R. E. Duke, R. Luo, K. M. Merz, D. A. Pearl-
man, M. Crowley, R. C. Walker, W. Zhang, B. Wang, S. Hayik, A.
Roitberg, G. Seabra, K. F. Wong, F. Paesani, X. Wu, S. Brozell, V.
Tsui, H. Gohlke, l. Yang, C. Tan, J. Mongan, V. Hornak, G. Cui, P.
Beroza, D. H. Mathews, C. Schafmeister, W. S. Ross, P. A. Kollman,
Univesrity of California, San Francisco, 2006.
[44] A. Bax, D. G. Davis, J. Magn. Reson. 1985, 65, 355.
[45] A. Kumar, R. R. Ernst, K. Wꢈthrich, Biochem. Biophys. Res.
Commun. 1980, 95, 1.
[39] E. J. Toone, Curr. Opin. Struct. Biol. 1994, 4, 719.
[40] W. F. Vranken, W. Boucher, T. J. Stevens, R. H. Fogh, A. Pajon, M.
Llinas, E. L. Ulrich, J. L. Markley, J. Ionides, E. D. Laue, Proteins
Struct. Funct. Bioinf. 2005, 59, 687.
[46] P. Gꢈntert, Methods Mol. Biol. 2004, 278, 353.
[47] W. D. Cornell, P. Cieplak, C. I. Bayly, I. R. Gould, K. M. Merz,
D. M. Ferguson, D. C. Spellmeyer, T. Fox, J. W. Caldwell, P. A. Koll-
man, J. Am. Chem. Soc. 1995, 117, 5179.
[41] G. M. Morris, D. S. Goodsell, R. S. Halliday, R. Huey, W. E. Hart,
R. K. Belew, A. J. Olson, J. Comput. Chem. 1998, 19, 1639.
[42] K. N. Kirschner, A. B. Yongye, S. M. Tschampel, J. Gonzꢀlez-Outeir-
iÇo, C. R. Daniels, B. L. Foley, R. J. Woods, J. Comput. Chem. 2008,
29, 622.
Received: April 13, 2010
Published online: July 22, 2010
10726
www.chemeurj.org
ꢅ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2010, 16, 10715 – 10726