Synthesis of multivalent carbohydrate-centered glycoclusters as nanomolar ligands of the bacterial lectin LecA from pseudomonas aeruginosa

Article abstract of DOI:10.1002/chem.201300135

A family of fifteen glycoclusters based on a cyclic oligo-(1→6)- β-D-glucosamine core has been designed as potential inhibitors of the bacterial lectin LecA with various valencies (from 2 to 4) and linkers. Evaluation of their binding properties towards LecA has been performed by a combination of hemagglutination inhibition assays (HIA), enzyme-linked lectin assays (ELLA), and isothermal titration microcalorimetry (ITC). Divalent ligands displayed dissociation constants in the sub-micromolar range and tetravalent ligands displayed low nanomolar affinities for this lectin. The influence of the linker could also be demonstrated; aromatic moieties are the best scaffolds for binding to the lectin. Copyright

Full text of DOI:10.1002/chem.201300135

DOI: 10.1002/chem.201300135
Synthesis of Multivalent Carbohydrate-Centered Glycoclusters as Nanomolar
Ligands of the Bacterial Lectin LecA from Pseudomonas aeruginosa
Marina L. Gening,[a] Denis V. Titov,[a] Samy Cecioni,[b, c] Aymeric Audfray,[b]
Alexey G. Gerbst,[a] Yury E. Tsvetkov,[a] Vadim B. Krylov,[a] Anne Imberty,*[b]
Nikolay E. Nifantiev,*[a] and Sꢀbastien Vidal*[c]
Abstract: A family of fifteen glycoclus-
ters based on a cyclic oligo-(1!6)-b-d-
glucosamine core has been designed as
potential inhibitors of the bacterial
lectin LecA with various valencies
(from 2 to 4) and linkers. Evaluation of
their binding properties towards LecA
has been performed by a combination
of hemagglutination inhibition assays
(HIA), enzyme-linked lectin assays
(ELLA), and isothermal titration mi-
crocalorimetry (ITC). Divalent ligands
displayed dissociation constants in the
sub-micromolar range and tetravalent
ligands displayed low nanomolar affini-
ties for this lectin. The influence of the
linker could also be demonstrated; aro-
matic moieties are the best scaffolds
for binding to the lectin. The affinities
observed in vitro were then correlated
with molecular models to rationalize
the possible binding modes of these
glycoclusters with the bacterial lectin.
Keywords: click chemistry · glucos-
amines
· glycocluster · lectins ·
Pseudomonas aeruginosa · oligosac-
charides
Introduction
Even though the monovalent interaction of a carbohy-
drate to its specific lectin is usually quite weak, with dissoci-
ation constants (Kd) in the millimolar range, multivalent in-
teractions are commonly used in biological systems to reach
higher affinities and to mediate several important natural or
pathological recognition processes. The synthesis of multiva-
lent glycoconjugates based on a multivalent central core
scaffold and a carbohydrate epitope has been thoroughly in-
vestigated over the past decades and has clearly demonstrat-
ed the influence of valency on the affinity towards a lectin.[4]
More recently, studies have been focused on demonstrating
the influence of the macromolecule topology: mismatched
geometries tend to provide weak binding motifs whereas
well-adapted architectures can yield potent binding proper-
ties towards the lectin.
Pseudomonas aeruginosa is an opportunistic bacterium re-
sponsible for hospital-acquired diseases and pulmonary in-
fections that are particularly threatening for immune-com-
promised individuals and cystic fibrosis patients. Two soluble
lectins, LecA (also called PA-IL) and LecB (also called PA-
IIL) have been identified in P. aeruginosa and have specifici-
ty towards galactose and fucose, respectively.[5,6] LecA ap-
pears to play an important role in the infection process and
methyl a-d-galactoside has a strong protective effect in the
murine model of infection.[7] LecA is a tetrameric calcium-
dependent lectin[8] and structure/specificity studies have
helped to elucidate the molecular basis for preference to-
wards the a-galactosylated Gb3 sphingolipid.[9,10] a-Galacto-
sylated oligosaccharides have weak affinities for LecA (Kd
of 50 mm), whereas b-galactosides with aromatic aglycones
perform better.[11,12]
The inhibition of carbohydrate–lectin interactions is of
prime importance for the design of potential new ap-
proaches against bacterial or viral infections.[1] The first step
of these infection occurs through the highly selective recog-
nition of carbohydrate receptors (i.e., lectins) and oligosac-
charidic epitopes.[2] The design of multivalent ligands target-
ing bacterial or viral lectins is therefore appearing as a pow-
erful approach to develop anti-adhesive drugs as active
treatments against pathogenic infections.[3]
[a] Dr. M. L. Gening, Dr. D. V. Titov, Dr. A. G. Gerbst,
Dr. Y. E. Tsvetkov, Dr. V. B. Krylov, Prof. N. E. Nifantiev
Laboratory of Glycoconjugate Chemistry
N. D. Zelinsky Institute of Organic Chemistry
Russian Academy of Sciences
Leninsky prospect 47, 119991 Moscow (Russia)
Fax : (+7)499-1358784
E-mail: nen@ioc.ac.ru
[b] Dr. S. Cecioni, Dr. A. Audfray, Dr. A. Imberty
Centre de Recherche sur les Macromolꢀcules Vꢀgꢀtales
(CERMAV–CNRS UPR 5301), affiliated with
Grenoble Universitꢀ and ICMG
BP 53, 38041 Grenoble (France)
Fax : (+33)476-037-636
E-mail: imberty@cermav.cnrs.fr
[c] Dr. S. Cecioni, Dr. S. Vidal
Institut de Chimie et Biochimie Molꢀculaires et Supramolꢀculaires
Laboratoire de Chimie Organique 2–Glycochimie
UMR 5246, CNRS, Universitꢀ Claude Bernard Lyon 1
43 Boulevard du 11 Novembre 1918, 69622 Villeurbanne (France)
Fax : (+33)472-448-109
E-mail: sebastien.vidal@univ-lyon1.fr
Several groups have proposed the use of multivalency for
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/chem.201300135.
the synthesis of high-affinity ligands for LecA. Fullerenes,[13]
9272
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2013, 19, 9272 – 9285

FULL PAPER
gold nanoparticles,[14] glycodendrimers,[15] and polymers[16]
decorated with a-Gal or b-Gal demonstrated effective clus-
ter effects with high aggregation power. However, smaller
glycoclusters based on calix[4]arenes,[11,17] resorcin[4]ar-
enes,[18] calix[6]arenes,[19] b-peptoids,[19] porphyrins,[19] carbo-
hydrates,[20] and peptidodendrimers[12] were also very effi-
cient, some of them with dissociation constants below
200 nm. A molecular modeling approach combined with
atomic force microscopy established that clustering can be
obtained when two adjacent binding sites of the lectin, 30 ꢂ
apart, are targeted by two galactose residues from the same
glycocluster.[21] The peptidodendrimers were tested for their
anti-biofilm properties against P. aeruginosa infection,[12] and
the galactosylated porphyrins have applications in the nano-
detection of lectin–carbohydrate interactions.[22,23] Neverthe-
less, the search for additional multivalent ligands of LecA
with improved solubility or better binding properties is still
needed for the design of a potential anti-adhesive agent
against bacterial infection by P. aeruginosa.
conformational flexibility[27] but lack a distinct hydrophobic
cavity. The presence of amino groups suitable for conjuga-
tion allows these cyclic oligosaccharides to be regarded as a
new type of functionalized scaffolds for the development of
multivalent glycoconjugates. The conjugation of the cyclo-
oligosaccharides with carbohydrate epitopes leads to carbo-
hydrate-centered glycoclusters, of which the binding proper-
ties towards LecA have been studied by hemagglutination
inhibition assays (HIA), enzyme-linked lectin assays
(ELLA), and isothermal titration microcalorimetry (ITC).
The affinities observed in vitro were then correlated with
molecular models in order to rationalize the possible bind-
ing modes of the glycoclusters with bacterial lectin.
Results and Discussion
Synthesis of glycoclusters: Multivalency[28–31] is known to in-
crease the affinity of a ligand for lectin through the glyco-
side cluster effect.[32,33] To evaluate the effect of the valency
of glycoclusters on binding properties, the cyclic di-, tri-, and
tetra-(1!6)-b-d-glucosamines 1a, 4a, and 5a[24,25] were
chosen as core structures (Figure 1). Cyclotetrasaccharides
2a and 3a, which comprise of alternate or adjacent glucose
Recently, we designed a family of cyclic oligo-(1!6)-b-d-
glucosamines consisting of two to seven monosaccharide
units[24,25] and two cyclotetrasaccharides comprising both
glucosamine and glucose residues.[26] These cyclic oligomers
differ from the well-known cyclodextrins, which have more
Figure 1. Schematic representation of the studied glycoclusters based on cyclic di-, tri-, and tetra-(16)–d-glucosamines (1, 4, 5), and the mixed cyclo-
tetrasaccharides (2, 3).
ACHTUNGTRENNUNG
Chem. Eur. J. 2013, 19, 9272 – 9285
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
9273

A. Imberty, N. E. Nifantiev, S. Vidal et al.
and glucosamine units, were used to prepare glycoclusters
with different relative orientations of epitopes (i.e., longitu-
dinal or angular) and to study the impact of topology on
their affinity towards lectin.
tion to conventional Zemplꢀn deacetylation to afford the
desired tri- and tetravalent glycoclusters 4b,c and 5b,c after
purification by gel-permeation chromatography (Scheme 2).
To assemble glycoclusters in which the core cyclic oligo-
saccharides and sugar ligands were connected through link-
ers d and e, a combination of N-acylation and CuAAC was
used. The order of these reactions may be different. Thus,
compounds 1e, 4d,f, and 5d,f were synthesized by initial N-
acylation followed by multiple-CuAAC. In the first step, the
cyclic cores 1a, 4a, and 5a were acylated with 4-pentynoic
acid activated ester 17[38] to provide the di-, tri-, and tetra-
Glycoconjugates 1–5 contain different linkers connecting
carbohydrate epitopes (Gal or Man) with oligovalent scaf-
folds (Figure 1). Three types of linkers (L) with different hy-
drophobicities and conformational mobilities were used in
this study. The most hydrophilic and flexible linkers b and c
comprise of only oligoethyleneglycol chains. The less hydro-
philic and flexible linkers d and f include triethyleneglycol
moieties and a 1,2,3-triazole ring,[34] which arises upon con-
necting sugar ligands with the cyclic core by copper-cata-
lyzed azide–alkyne cycloaddition (CuAAC).[35–37] The most
hydrophobic and rigid linker e, which consists of phenyl and
1,2,3-triazole rings,[11] was used to prepare selected types of
conjugates. The effect of the linker structure and length is
taken into account when discussing the binding properties
of the multivalent glycoclusters.
N-Acylation or a combination of N-acylation and CuAAC
were employed for the conjugation of the peripheral sugar
ligands to the core cyclic oligosaccharides. Tri- and tetrava-
lent glycoclusters 4b,c and 5b,c were obtained by N-acyla-
tion of cyclic oligomers 4a and 5a with activated esters de-
rived from glycosylated oligoethyleneglycols (Scheme 1).
Thus, pentafluorophenyl (Pfp)-activated esters 15 and 16
with tri- and hexaethyleneglycol chains were synthesized.
Two derivatives of this type were prepared to evaluate the
effect of the linker length on the binding properties towards
the lectin. Tri- and hexaethyleneglycols 6 and 7 were mono-
alkylated with methyl diazoacetate to afford the w-hydrox-
yesters 8 and 9, respectively. Glycosylation with galactosyl
bromide 10 under Helferich conditions provided the b-galac-
tosides 11 and 12. Selective cleavage of the methyl ester
with sodium iodide in boiling pyridine gave carboxylic acids
13 and 14. The acids were activated by trans-esterification
with Pfp-trifluoroacetate to afford the corresponding esters
15 and 16.
AHCTUNGTREGaNNNU lkynylated derivatives 18–20 (Scheme 3). These compounds
were purified by gel-permeation chromatography and the
more hydrophobic amide 18 was also subjected to purifica-
tion on a C-18 Sep-Pak cartridge.
In the second step, the cycloaddition of azido-functional-
ized carbohydrate epitopes to synthesize alkynylated scaf-
folds was carried out. The addition of water-insoluble pro-
tected azide 21[11] to cyclo-disaccharide 18 was performed in
acetonitrile in the presence of CuI/DIPEA as the catalytic
system (Scheme 4). Application of an azido-functionalized,
acetylated sugar residue usually allows for purification of
the resulting glycoconjugates by conventional silica-gel
column chromatography, a strategy that proved successful in
our recent reports.[11,17–19,39] Final O-deacetylation of the
pure precursors with MeONa/MeOH provided the desired
conjugates. Thus, the divalent glycocluster 1e was obtained
in high purity after the removal of inorganic salts by filtra-
tion through a C-18 Sep-Pak cartridge and gel-permeation
chromatography; however, it was formed in moderate yield
(52%) after two steps. The conjugation of the alkynylated
core scaffolds 19 and 20 by CuAAC was initially attempted
with an acetylated azido-functionalized galactoside (i.e., ace-
tylated galactoside 26). The disappearance of the core scaf-
folds was clearly indicated by TLC and new products were
observed, although they were not identified. The attempted
purification by column chromatography on silica gel or basic
alumina, or by gel-permeation chromatography on Sephadex
G-10, failed. Although the purification of highly polar mole-
cules can be quite difficult, the unprotected galactoside 22
was conjugated to the alkynylated core scaffolds 19 and 20
The coupling of esters 15 and 16 with the cyclic-core oli-
gosaccharides 4a and 5a readily produced the corresponding
amide intermediates, which were subjected without purifica-
Scheme 1. Synthesis of the Pfp-activated esters 15 and 16 (Pfp=pentafluorophenyl).
9274
www.chemeurj.org
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2013, 19, 9272 – 9285

Multivalent Carbohydrate-Centered Glycoclusters as Nanomolar Ligands
FULL PAPER
Scheme 2. Synthesis of the tri- and tetravalent glycoclusters 4b,c and 5b,c.
by CuAAC (Scheme 4). The reactions were performed in
aqueous methanol with a CuSO4/sodium ascorbate catalytic
system under microwave irradiation. The crude mixtures
were then purified by reverse-phase C-18 silica-gel column
chromatography, and the excess azide derivative 22 could be
easily separated from the glycoclusters. The target glycocon-
jugates 4d and 5d were obtained in good yields of 67 and
86%, respectively. The presence of under-substituted conju-
gates in 4d and 5d could be ruled out based on mass spec-
trometry analyses. The same reaction was conducted for the
introduction of the mannose ligands by using unprotected
azido-functionalized mannoside 23. The synthesized manno-
sylated glycoclusters 4 f and 5 f were used as negative con-
trols for HIA and ELLA binding studies.
active ester 17 provided amide 25 (Scheme 5). Subsequent
conjugation with the acetylated azide 26[34] through CuAAC
followed by O-deacetylation resulted in the formation of the
desired monovalent ligand 27 (Scheme 5).
CuAAC and N-acylation reactions were applied in the re-
verse order for the preparation of glycoclusters 1d, 2d,e,
and 3d,e. At first, glycosylated linkers were assembled using
a single cycloaddition reaction, then the obtained structures
were connected to the cyclic oligosaccharides by using N-
acylation.
Succinimidyl 4-pentynoate (17) was used to obtain the ac-
tivated esters 28 and 29, which have linkers containing tri-
AHCTUNGTREGeNNNU thyleneglycol or phenyl moieties (Scheme 6). [3+2]Cyclo-
addition of the known azides 21 and 26[34] to alkyne 17
under CuI/DIPEA (DIPEA=N,N-diisopropylethylamine)
catalysis afforded succinimidyl esters in moderate yields.
Some of the low yields observed may be attributed to partial
A reference monovalent ligand was also prepared to
reveal the effect of multivalency upon binding to the lectin.
Thus, the acylation of N-methyl-d-glucamine (24) with the
Chem. Eur. J. 2013, 19, 9272 – 9285
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
9275

A. Imberty, N. E. Nifantiev, S. Vidal et al.
Both approaches were accepta-
ble, but the second method
seems to be preferable because
the most valuable compounds
among the starting materials,
the cyclic oligosaccharide scaf-
folds, are involved in only one
synthetic step, thus providing a
more effective transformation
into the final glycoclusters.
HIA and ELLA binding stud-
ies: The library of prepared gly-
coclusters and their correspond-
ing monovalent reference com-
pounds were studied as ligands
for LecA with a combination of
hemagglutination
inhibition
assay (HIA), enzyme-linked
lectin assay (ELLA), and the
most informative, the isother-
mal titration microcalorimetry
(ITC). All glycoclusters were
evaluated with a HIA-test to
obtain primary data regarding
binding to the lectin. Multiva-
lent lectins cross-link to red
blood cells through interactions
of the carbohydrate groups at
the surface of the cell mem-
brane and their multiple bind-
Scheme 3. Synthesis of the alkynylated cyclic oligosaccharides 18–20.
ing sites. The resulting aggluti-
nation of the erythrocytes can
hydrolysis of the esters during their purification on silica
gel.
therefore be inhibited by multivalent glycoclusters, which
would prevent this lectin–carbohydrate interaction. The po-
tency of these ligands can be expressed as the minimum in-
hibitory concentration (MIC) required for inhibiting the ag-
glutination of erythrocytes. This method is based on visual
observation of agglutination and thus gives semi-quantita-
tive data concerning binding activity; it is commonly used as
a first trial for adjustment of the efficient concentration for
further experiments. An important advantage of HIA is the
biological origin of ligands from the surface of the cell in-
volved in interaction with LecA; in the other types of
assays, non-natural, synthetic, galactosylated molecular
probes are used. The synthesized multivalent glycoclusters
and monovalent reference compounds were characterized
with their corresponding MIC values (Table 1).
Monovalent methyl b-d-galactopyranoside, para-nitro-
phenyl b-d-galactopyranoside, and compound 27 displayed
MIC values towards the LecA-induced agglutination of er-
ythrocytes in the millimolar range. Monovalent compound
30[11] (Figure 2), with the hydrophobic aglycone of the type
e, exhibited an MIC value in the high micromolar range.
The multivalent glycoclusters inhibited the agglutination of
erythrocytes with MIC values in the micromolar to low mi-
cromolar range.
Subsequent N-acylation of the cyclic oligosaccharide cores
1a, 2a, 3a, and 5a with succinimidyl esters 28 and 29 gave
the corresponding amide intermediates, which were then O-
deacetylated, without purification of the intermediate, to
give the desired di- and tetravalent glycoclusters 1d, 2d,e,
3d,e, and 5e (Scheme 7). The synthesized glycoconjugates
were purified by gel-permeation chromatography. The
poorly water soluble glycoconjugate 5e was also subjected
to reverse-phase C-18 flash chromatography to remove inor-
ganic salts.
The synthesis of one monovalent, six divalent, four triva-
lent, and five tetravalent glycoconjugates has been per-
formed by either direct amide-bond formation (in the case
of the oligoethyleneglycol-only linkers), or by using the
CuAAC conjugation strategy (in the case of triazole-con-
taining linkers). Two methodologies were used for the as-
sembly of glycoclusters with the triazole ring: 1) N-acylation
of cyclic oligosaccharides with 4-pentynoic acid, then attach-
ment of azido-functionalized sugar ligands with the CuAAC
reaction, or 2) CuAAC reaction of azido-functionalized
sugar ligands with succinimidyl 4-pentynoate, then N-acyla-
tion of cyclic-core oligosaccharides with activated esters.
9276
www.chemeurj.org
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2013, 19, 9272 – 9285

Multivalent Carbohydrate-Centered Glycoclusters as Nanomolar Ligands
FULL PAPER
Scheme 4. Synthesis of di-, tri-, and tetravalent glycoclusters 1e, 4d,f, and 5d,f with CuAAC.
microtiter plate. The quantity of LecA binding in the well
was monitored by incubating with horseradish-peroxidase-
labeled streptavidin, followed by treatment with ortho-phe-
nylenediamine (OPD) then UV/Vis analysis.
The monovalent probes displayed IC50 values in the high
micromolar range, and the inhibition of LecA adhesion was
improved with the multivalent glycoclusters (Table 1,
Figure 3). As previously shown in the HIA study, the rela-
tive potency (b) of the multivalent inhibitors compared with
the monovalent compound clearly demonstrated a cluster
effect. This tendency is easily demonstrated by comparison
of the binding activity of mono-, di-, tri-, and tetravalent an-
alogues (27, 1d, 4d, and 5d) with the same ethyleneglycol–
Figure 2. Monovalent reference 30 with an aromatic aglycone.
In perfect agreement with the HIA experiments, the
ELLA results revealed that the galactosylated glycoclusters
display a potent inhibition of LecA adhesion to the galacto-
sylated surface. The binding properties of the glycoclusters
were evaluated by measuring their IC50 values for the inhibi-
tion of binding of biotin-labeled LecA to a galactose-coated
Chem. Eur. J. 2013, 19, 9272 – 9285
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
9277

A. Imberty, N. E. Nifantiev, S. Vidal et al.
Scheme 5. Synthesis of the glucamine-based reference monovalent ligand
27.
triazole linker. Thus, b values of the inhibitors 27, 1d, 4d,
and 5d in ELLA experiments are 0.63, 23.6, 189, and 279 re-
spectively (Table 1).
Although the tetravalent glycoconjugates were better in-
hibitors in the above-described series, valency did not great-
ly influence the affinity of the glycoclusters towards LecA,
contrary to the influence of linker properties. Comparison
of the binding properties of tetravalent glycoclusters with
the different linkers (5b–e) demonstrated the strong advant-
age of a rigid aromatic linker over a flexible and hydrophilic
ethyleneglycol linker. Thus, b values of a tetravalent inhibi-
tor (5b,c) with a different-length ethyleneglycol linker in
ELLA are 104 and 118, respectively, whereas the compound
with triazole-containing linker (5d) displayed a b value of
279. The highest activity was demonstrated for compounds
(5e) with a phenyltriazole linker (b=1209). Moreover, all
studied compounds with this phenyltriazole linker were
more active than compounds with the other types of linkers
in the synthesized series.
Figure 3. Inhibition measured by ELLA for glycoclusters. Lines are sig-
moidal fits.
pounds (2d,e) with diagonal orientation of linkers when
compared to compounds (3d,e) with neighboring orienta-
tion.
Based on a combination of HIA and ELLA results, we
can divide the synthesized, galactosylated clusters into three
groups according to their activity. The group with higher ac-
tivity consists of compounds with the rigid phenyltriazole
linker (1e, 2e, 3e, 5e). Moderate activity was observed for
tri- and tetravalent glycoclusters (4b–d, 5b–d) with the
other types of linkers. The tri-
Divalent glycoclusters (1–3) with a different spatial orien-
tation of galactose epitopes demonstrated approximately
equal inhibition activity. Slightly higher values for com-
valent glycocluster 4b with the
shortest triethyleneglycol linker
showed the lowest activity and
stood out in this group. The dis-
tance between galactose resi-
dues was probably inadequate
(see below) to allow simultane-
ous binding to two sites of
LecA without stretching the
molecular architecture and
therefore reducing its affinity.
The group with the lowest ac-
Scheme 6. Synthesis of the succinimidyl-activated esters 28 and 29.
tivity includes divalent glyco-
9278
www.chemeurj.org
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2013, 19, 9272 – 9285

Multivalent Carbohydrate-Centered Glycoclusters as Nanomolar Ligands
FULL PAPER
tions as high as 781 mm, which
can be attributed to a non-spe-
cific activity of the glycoclus-
ters. However, the mannosylat-
ed glycoclusters, when used as
negative controls, did not dis-
play any inhibition at the high-
est concentration used in
ELLA experiments. The rele-
vance of the weak inhibition
observed by HIA can therefore
be discarded when comparing
with these ELLA data.
Thus, we can conclude that
the use of different linkers can
change the binding properties
quite dramatically. It was also
observed that the core organi-
zation and the valency of the
glycocluster influence the orien-
tation of the galactose residues
and therefore their binding
properties. All of these parame-
ters influence the binding ob-
served towards LecA and
a
more detailed study was then
performed with microcalorime-
try.
ITC binding studies: Isothermal
titration
microcalorimetry
(ITC) measures the heat re-
leased or absorbed during the
association process between
two molecular entities by
means of titration of ligand so-
A
ACHTUNGTRENNUNGlu-
AHCTUNGTRENNUNG
ner (lectin). ITC provides an
overview of the thermodynam-
ics involved in this type of mul-
tivalent interaction, such as
1) the stoichiometry of binding
(n) for the formed complex by
determining the concentration
in each partner at the inflexion
point of the titration curve,
2) the variation of enthalpy
(DH) from the amplitude of the
titration curve, and 3) the disso-
ciation constant (Kd) related to
the slope of the curve. This bio-
analytical technique does not
Scheme 7. Synthesis of di- and tetravalent glycoclusters 1d, 2d,e, 3d,e, and 5e.
clusters with ethyleneglycol linkers (1d, 2d, 3d). Mannosy-
lated glycoclusters (4 f, 5 f) did not exhibit any activity
toward LecA, as measured in ELLA experiments, but some
inhibition of hemagglutination was observed at concentra-
require the proteins or ligands to be labelled or immobilized
as for ELLA experiments (or surface plasmon resonance;
SPR) and therefore relates to natural binding properties of
the interacting partners in solution.
Chem. Eur. J. 2013, 19, 9272 – 9285
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
9279

A. Imberty, N. E. Nifantiev, S. Vidal et al.
Table 1. HIA, ELLA, and isothermal titration microcalorimetry (ITC) measurements for the binding of multivalent glycoclusters to LecA.
Compound[a]
Linker[b]
HIA
MIC[c]
[mm]
ELLA
ITC
b[d]
IC50
[mm]
b[d]
n[e]
DH
[kJmolꢀ1
TDS
Kd
[mm]
b[d]
G
]
[kJmolꢀ1
]
monovalent compounds
b-d-Gal-OMe
none
6250
1000
6250
250[h]
1
6.3
1
69
28
1
0.8[h]
1.0[i]
1[g]
39[h]
34[i]
15[h]
6[i]
70[h]
1
b-d-GalOpPhNO2
none
2.5
0.6
1.5
13[i]
5.3
0.5
12
27
EG3-triazole
Ph-triazole
83
33.1ꢁ0.6
11.1ꢁ0.5
140ꢁ5
30[h]
25
46[h]
0.81[h]
532[h]
23[h]
5.80.9[h]
divalent compounds
1d
1e
2d
2e
3d
3e
EG3-triazole
Ph-triazole
EG3-triazole
Ph-triazole
EG3-triazole
Ph-triazole
625
19.5
625
1.22
1250
9.8
10
320
10
5122
5
638
2.9
0.12
2.1
0.057
2.9
0.15
23.6
574
32.3
1200
23.6
460
0.34ꢁ0.1
92ꢁ11
60ꢁ11
2.7ꢁ0.6
25
209
33
318
23
132
0.50ꢁ0.03
0.32ꢁ0.02
0.55ꢁ0.03
0.32ꢁ0.03
0.50ꢁ0.09
70.3ꢁ0.6
84.8ꢁ0.9
75.2ꢁ1.5
90ꢁ4
33.4ꢁ0.6
52.4ꢁ0.9
37.0ꢁ0.3
58ꢁ3
0.34ꢁ0.02
2.15ꢁ0.07
0.22ꢁ0.04
3.0ꢁ0.3
76ꢁ9
40ꢁ9
0.53ꢁ0.06
trivalent compounds
4b
4c
4d
4 f
EG3-amide
EG6-amide
EG3-triazole
EG3-triazole
391
195
98
16
32
64
8
1.29
0.81
0.27
45
71
189
–/–
0.36ꢁ0.06
0.40ꢁ0.02
0.45ꢁ0.03
n/d
98ꢁ8
90ꢁ2
90ꢁ6
65ꢁ9
56ꢁ2
54ꢁ6
1.59ꢁ0.65
0.98ꢁ0.27
0.46ꢁ0.07
47
71
151
781
>2500[f]
tetravalent compounds
5b
5c
5d
5e
5 f
EG3-amide
EG6-amide
EG3-triazole
Ph-triazole
EG3-triazole
195
98
49
1.2
781
32
64
128
5122
8
0.59
0.50
0.15
104
118
279
1210
–/–
0.39ꢁ0.02
0.31ꢁ0.01
0.29ꢁ0.01
0.46ꢁ0.02
n/d
117ꢁ5
122ꢁ6
145ꢁ6
87ꢁ2
83ꢁ6
86ꢁ6
108ꢁ2
47ꢁ3
1.45ꢁ0.03
0.45ꢁ0.03
0.31ꢁ0.02
0.079ꢁ0.002
48
157
226
886
0.057
>2500[f]
[a] The cores of compounds 1–5 are depicted in Figure 1. [b] EG=ethyleneglycol, Ph=phenyl. [c] Minimum inhibitory concentration (MIC) required to
inhibit the agglutination of erythrocytes. [d] Calculated using monovalent methyl a-d-galactopyranoside as a reference. [e] Stoichiometry of binding is de-
fined as the number of glycoclusters per monomer of LecA. [f] No inhibition was observed at the highest concentration indicated. [g] The value was pre-
determined during the fitting procedure. [h] Results from ref. [11]. [i] Results from ref. [16]. n/d=not determined.
ITC measurements performed with monovalent probes
(b-d-Gal-OMe, b-d-GalOpPhNO2, and compounds 27 and
30) showed Kd values in the micromolar range (Table 1,
Figure 4). The introduction of an EG3-triazole linker with a
terminal glucamine residue (compound 27) did not improve
the binding to LecA, but was instead slightly unfavorable.
On the contrary, aromatic aglycones (b-d-GalOpPhNO2 and
compound 30) provided better binding, apparently due to
their hydrophobicity. The Kd values of synthesized glycoclus-
ters measured with ITC confirmed the in binding activity
Figure 4. Raw ITC data (top) obtained by the injection of glycocompounds in a solution of LecA and the corresponding integrated titration curve
(bottom). Left: 175mm of 2d in 35mm of LecA; middle: 125mm of 5d in 50mm of LecA; right: 100mm of 5e in 20mm of LecA.
9280
www.chemeurj.org
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2013, 19, 9272 – 9285

Multivalent Carbohydrate-Centered Glycoclusters as Nanomolar Ligands
FULL PAPER
tendencies previously observed with HIA and ELLA. Thus,
a higher valency improved the affinity towards LecA when
comparing divalent 1d, trivalent 4d, and tetravalent 5d gly-
coclusters. The introduction of a phenyltriazole linker great-
ly improved the affinity. The reason for these increases in
binding affinity is mostly related to entropy contributions.
Although the enthalpy of binding of glycoclusters with the
phenyltriazole linker was less than for compounds contain-
ing EG-triazole linkers (see data for 1–3 in Table 1), the en-
tropic factor strongly favored the rigid phenyltriazole linker.
It also can be concluded that the diagonal orientation of the
spacer arms in core 2 is better than the neighboring orienta-
tion in core 3.
All high-affinity inhibitors demonstrate a stoichiometry
(n) close to one lectin monomer per galactose moiety, with
the exception of 5e, in which only two out of four galactose
epitopes were involved in the protein interaction. For low-
affinity inhibitors of LecA, some deviation from theoretical
stoichiometry was observed.
A complete analysis of lectin interactions with multivalent
ligands requires a combination of different analytical techni-
ques to obtain a general view of the binding properties
through complementary experimental setups. In the present
study, the different bioanalytical techniques that were used
agree on a strong affinity of the galactosylated glycocluster
5e for LecA. It was confirmed that the phenyltriazole
moiety is highly efficient as a linker for designing LecA
blockers.[11,12] Indeed, the tetravalent compound 5e has the
highest reported affinity for LecA (Kd 80 nm), and the corre-
sponding divalent compounds 1e and 2e are also highly effi-
cient ligands (Kd of 340 and 220 nm).
Molecular modeling studies: The crystal structure of LecA
in a complex with aGalACHTUNGTRENNUNG
from the Protein Data Bank (PDB code 2WYF).[40] The tet-
ramer forms a rectangle with distances between binding
sites of ꢂ90 ꢂ for the long sides and 32 ꢂ for the shorter
ones. From this architecture, it does not appear possible for
a tetravalent glycoconjugate to interact with all four binding
sites of the same tetramer. Bivalent binding of two neigh-
boring sites, however, seems probable; in this case, two
modes are possible for a tetravalent glycoconjugate: binding
with two adjacent carbohydrate residues (couples N1–N4,
Figure 5A); and binding of two epitopes attached to diamet-
rically opposed positions on the scaffold (couples D1–D2,
Figure 5A).
Molecular dynamics (MD) calculations were performed
for a selection of cyclic di-, tri-, and tetrasaccharide-based
glycoclusters to determine their conformational behavior.
Here, we discuss the most representative results obtained
for a set of ligands of LecA, such as: tetravalent glycoclus-
ters with hexaethyleneglycol and the phenyltriazole linkers
(5c and 5e, respectively); the more rigid bivalent conjugate
1e; and the intermediate case, a trisaccharide-based conju-
gate with the phenyltriazole linker, hypothetical structure
4e. The length of MD trajectory was 10 ns for each struc-
ture, structural snapshots taken every 2 ps, yielding 5000
snapshot conformations in each case.
Figure 5. A) Definition of galactose residue pairs analyzed in theoretical calculation. B) Percentage of conformations as function of C1–C1 distances for
5c, couples D1 and N1; C) same for 5e, couples D1 and N1; D) same for 1e for one possible couple; E) same for 4e, couples N1, N2, N3. The vertical
arrows mark the distance of 29 ꢂ corresponding to the C1–C1 distance in the crystal structure of the LecA–galactose complex (PDB code 2WYF). For
other distance distributions, refer to the Supporting Information, Figures S4–S6.
Chem. Eur. J. 2013, 19, 9272 – 9285
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
9281

A. Imberty, N. E. Nifantiev, S. Vidal et al.
Trajectories were first analyzed by calculating the distan-
ces between C1 atoms of a pair of galactose residues to
compare with the 29 ꢂ distance measured in the LecA–gal-
actose complex (PDB code 2WYF). Charts of the popula-
tion percentage with respect to C1–C1 distance are shown in
Figures 5B–E. Due to tetravalent molecule symmetry, only
N1 and D1 are represented for compounds 5c and 5e. Popu-
lation charts for the other distances are very similar and are
given in the Supporting Information. For conjugate 5c, the
most populated distances are around 13 ꢂ for proximal gal-
actose (N) and 28 ꢂ for distal residues (D), and the 29 ꢂ
value can be reached by a significant population of galactose
pairs (Figure 5B). For conjugate 5e, the distribution is dif-
ferent, with a higher population at around 23 ꢂ (N) and
32 ꢂ (D), and a stronger population around 29 ꢂ (Fig-
ure 5C), which was further confirmed experimentally by
using the specially synthesized conjugates 2e and 3e (see
Table 1). For both tetravalent compounds, distal pairs of gal-
actose are more likely to interact with two LecA binding
sites than proximal pairs. For compound 1e (Figure 5D), the
population able to interact with two galactose residues at a
distance of 29 ꢂ is significantly smaller but is still present,
which explains its moderate inhibition activity against PA-
1L. For the hypothetical structure 4e, the distribution of the
C1–C1 distance is shown in Figure 5E. Its shape is similar to
the distribution curve for the N1 distance of the correspond-
ing tetravalent conjugate 5e. This suggests that the inhibito-
ry activity of the trisaccharide-based conjugate 4e should
not exceed that of its counterpart 5e, and therefore conju-
gate 4e was not synthesized.
In addition to the distance, the mutual orientation of gal-
actose residues in the conjugates is also crucial for binding.
To compare the orientations of galactose residues in solu-
tion with that observed in complex with LecA, the root
mean square deviations (RMSD) were calculated, taking
into account oxygen atoms of the galactose residues of the
glycocluster and of the galactose residues in the binding
sites (Figures S1–S3 in the Supporting Information).
The graphs obtained for glycocluster 5c with a very flexi-
ble ethyleneglycol spacer arm did not display any significant
maximum and had a very broad shape. This confirms that
these compounds are rather flexible, with no conformational
preference. On the contrary, the more rigid phenyltriazole
linker demonstrated the presence of a maximum at about
4 ꢂ in histograms of the RMSD distributions of oppositely
attached epitopes. This confirms that molecule 5e mostly
adopted conformations close to that which is necessary for
binding two neighboring active sites in LecA with the oppo-
sitely attached galactose residues. This correlates with exper-
imentally observed fact that the conjugate 2e with distally
attached galactose ligands exhibits similar activity as tetra-
dendate conjugate 5e, whereas the conjugate 3e with proxi-
mally attached Gal ligands is remarkably less active
(Table 1). In the case of bivalent conjugate 1e, the shape of
the RMSD distribution was very similar to that of the rigid
tetravalent conjugate 5e. Even though rigidity can be claim-
ed as a beneficial parameter for improving the affinity to
the lectin, the relatively increased hydrophobicity of the
phenyltriazole linker in comparison to the ethyleneglycol
spacer can also contribute to binding.
Further illustration of the semi-rigid behavior of 5e is
given in Figure 6. From the superimposition of several con-
formations of glycocluster 5e generated during MD simula-
Figure 6. Top: Two views of the superimposition of several representative
conformations for glycocluster 5e generated during MD simulations (two
orientations). Bottom: Two views of the complex between the MD-gener-
ated structure of glycocluster 5e and LecA with the position of the galac-
tose residue in neighboring binding sites best-fitted to those of galactoses
in the crystal structure.
tion, it is clear that some flexibility is observed but that it is
limited to the extremity of the galactosylated branches. An
attempt to model the complex between the conformation
which exhibited the lowest RMSD and LecA structure con-
firmed that two opposite galactose residues (couple D1) can
reach the neighboring binding sites and that the cyclic tetra-
saccharide scaffold does not present steric conflicts with the
protein surface (Figure 6).
Conclusion
The efficient synthesis of fifteen synthetic multivalent glyco-
clusters with a set of cyclic oligosaccharidic core scaffolds
and carbohydrate epitopes has been developed. The detailed
analysis of their binding properties with the lectin LecA was
performed by HIA, ELLA, and ITC. These molecules are
candidates for the design of anti-infective agents and, more
particularly, for the development of anti-adhesive treatments
against bacterial infection by P. aeruginosa. The conjugation
was performed by using a combination of amide-bond for-
mation and click chemistry to provide a series of glycoconju-
gates in good yields and purities. The synthetic steps in-
9282
www.chemeurj.org
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2013, 19, 9272 – 9285

Multivalent Carbohydrate-Centered Glycoclusters as Nanomolar Ligands
FULL PAPER
400 MHz, 500 MHz, or 600 MHz spectrometer. Chemical shifts are refer-
enced relative to deuterated solvent residual peaks. NMR spectra of free
oligosaccharides were measured for solutions in D2O by using acetone
(dH =2.225, dC =31.45 ppm) as an internal standard. The following abbre-
viations are used to explain the observed multiplicities: s, singlet; d, dou-
blet; t, triplet; m, multiplet and br t, broad triplet. Assignments were de-
duced from 2D experiments (COSY, HSQC, HMBC, and TOCSY). High
resolution mass spectra (HR-ESI-QTOF) were recorded with a Bruker
MicrOTOF-Q II XL spectrometer. Thin-layer chromatography (TLC)
was carried out on aluminum sheets coated with silica gel 60 F254
(Merck). TLC plates were inspected under UV light (l=254 nm) and de-
veloped by treatment with a mixture of 10% H2SO4 in EtOH/H2O (95:5
v/v) followed by heating. Silica-gel column chromatography was per-
formed with Silica Gel 60 (40–63 mm, E. Merck). C-18 Reverse-phase
column chromatography was performed on a pre-packed 4 g-cartridge
with a flow rate of 10 mLminꢀ1. Gel-permeation chromatography of free
oligosaccharides was performed on a column of TSK HW-40 (S) gel (25ꢃ
800 mm) in 0.1m AcOH. Optical rotation was measured by using a
Perkin–Elmer polarimeter or a JASCO DIP-360 polarimeter at 20–248C
and values are given in 10ꢀ1 degcm2 gꢀ1. Recombinant LecA was pro-
duced in E. coli as described previously.[9]
volved in the construction of the glycoclusters (amide bond
and click chemistry) can be easily inverted to address tech-
nical problems (reactivity and/or purification) encountered
during their preparation. The synthetic strategy designed
here and the use of a water-soluble cyclic oligo-glucosamine
core allowed for the synthesis of the first water-soluble phe-
nyltriazole-containing glycoclusters. The introduction of aro-
matic moieties at the anomeric position of galactose proved
beneficial for binding to LecA,[12] but solubility issues were
then detrimental for the calix[4]arene-based glycoclusters
designed earlier in our group.[11]
The data obtained from HIA and ELLA experiments are
in good agreement with each other and allowed for the
standard study of these multivalent lectin–carbohydrate in-
teractions. The best ligands were the tetravalent glycoclus-
ters, although the divalent derivatives also had interesting
binding properties towards LecA. Indeed, the designed diva-
lent glycoclusters demonstrated a high affinity for LecA,
which brings them into the group of the most practical and
promising potential anti-adhesive agents against bacterial in-
fection by P. aeruginosa. Further analysis of the binding
properties with ITC experiments allowed for detailed analy-
sis of the thermodynamic parameters of these interactions
and for an in-depth discussion of the advantages of each
molecular architecture designed. Again, the data obtained
by ITC was in agreement with the HIA and ELLA assays
performed. In addition, molecular modeling studies have
been performed to rationalize the binding properties of the
best ligands. The obtained data confirmed the preferred che-
late binding mode of the best ligands, with two galactose
epitopes bound at two binding sites of the same LecA tet-
ramer.
Hemagglutination inhibition assays (HIA): HIA was performed in U-
shaped 96-well microtiter plates. Rabbit erythrocytes were purchased
from BioMꢀrieux and used without further washing. The erythrocytes
were diluted to a 4% solution in NaCl (150 mm). Solutions of lectin
(2 mgmLꢀ1) were prepared in Tris/HCl (20 mm), NaCl (100 mm), and
CaCl2 (100 mm). The hemagglutination unit (HU) was first obtained by
the addition of 25 mL of the 4% erythrocyte solution to 25 mL aliquots of
sequential (two-fold) lectin dilutions. The mixture was incubated at 258C
for 60 min. The HU was measured as the minimum lectin concentration
required for observing hemagglutination. The LecA concentration used
during inhibition assays was equal to four times the HU. In this protocol,
the concentration of LecA was 8 mgmLꢀ1. Inhibition assays were then
carried out by the addition of 12.5 mL lectin solution (at the required con-
centration) to 25 mL of sequential dilutions of glycoclusters, monovalent
molecules, and controls. These solutions were then incubated at 258C for
2 h before the addition of 12.5 mL of 4% erythrocyte solution, followed
by an additional incubation at 258C for 30 min. The minimum inhibitory
concentration for each molecule was determined for each duplicate.
The best ligand designed in this study is composed of a
cyclic tetraglucosamine core scaffold with four galactose
epitopes displaying a phenyltriazole aglycone. The Kd value
(79 nm) measured by ITC is the lowest reported to date for
LecA, although it is in the same range as similar scaffolds
reported in the recent literature by us and other groups. A
general tendency for an observed maximum inhibition in
Enzyme-linked lectin assays (ELLA): ELLA were conducted by using
96-well microtiter plates (Nunc Maxisorb) coated with PAA-a-d-Gal for
LecA (100 mL of a 5 mgmLꢀ1 solution in carbonate buffer, pH 9.6 for 1 h
at 378C, then blocking at 378C for 1 h with 100 mL per well of 3% (w/v)
BSA in PBS). Inhibitor solutions (50 mL) were subjected to serial dilu-
tions (three-fold) with PBS-BSA (0.3% w/v). Then, 50 mL of biotinylated
LecA solution (0.5 mgmLꢀ1) was added to each well and the plates were
incubated for 1 h at 378C. After 3 washes with 50 mL of T-PBS (PBS con-
taining 0.05% Tween 20) for 5 min, 100 mL of horseradish peroxidase
(HRP)/streptavidin conjugate (dilution 2:5000; Boehringer-Mannheim)
was added and left for 1 h at 378C. After three more washes, coloration
was developed by using 100 mL per well of phosphate/citrate buffer
the 80 to 200 nm range is seen in these reports.[11,12,15,17,20]
A
minimum value determined by ITC seems to have been
reached and questions such as “can better LecA ligands be
designed?” or “are better LecA ligands needed for potential
biomedical applications?” arise. The next question can
therefore be enunciated as: “Is there a limit for the design
of ligands of LecA and glycoclusters with sub-nanomolar Kd
values might not be possible to design?”
(0.05m)
containing
)
ortho-phenylenediamine
dihydrochloride
OPD kit,
(0.4 mgmLꢀ1
and urea hydrogen peroxide (0.4 mgmLꢀ1
;
Sigma–Aldrich) for 15 min, and the reaction was stopped with 50 mL of
30% sulfuric acid. Absorbance at 490 nm was then read at 490 nm using
a microtiter-plate reader (BioRad 680) and transformed into inhibition
percentages with the help of positive and negative controls. Plots (inhibi-
tion percentage versus inhibitor concentration) and sigmoidal fitting al-
lowed determination of IC50.
Experimental Section
Isothermal titration microcalorimetry (ITC): Purified and lyophilized
LecA was dissolved in buffer (0.1m Tris-HCl buffer containing CaCl2
(6 mm), pH 7.5) at a concentration of 0.05 mm and degassed. Protein con-
centration was checked by measurement of optical density using a theo-
retical molarity extinction coefficient of 28000 (1 cm). Carbohydrate li-
gands were dissolved directly in the same buffer at a concentration of
0.12 mm, degassed, and placed in the injection syringe. Isothermal titra-
tion calorimetry was performed with a VP-ITC MicroCalorimeter from
MicroCal Incorporated. LecA was placed into the 1.4478 mL sample cell,
General procedures: All reagents for synthesis were commercially availa-
ble and used without further purification. Solvents were distilled over
CaH2 (CH2Cl2), Mg/I2 (MeOH), Na/benzophenone (THF), or purchased
dry. All reactions involving air- or moisture-sensitive reagents were car-
ried out by using dry solvents under dry Argon. Reactions under micro-
wave activation were performed on a Biotage Initiator system. NMR
spectra were recorded at 293 K, unless otherwise stated, with a 300 MHz,
Chem. Eur. J. 2013, 19, 9272 – 9285
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
9283

A. Imberty, N. E. Nifantiev, S. Vidal et al.
at 258C, with 10 mL injections of carbohydrate every 300 s. Carbohydrate
ligands were also titrated alone against the buffer. Data were fitted with
MicroCal Origin 7 software according to standard procedures. Fitted data
yielded the stoichiometry (n), the association constant (Ka) and the en-
thalpy of binding (DH). Other thermodynamic parameters (i.e., changes
in free energy, DG, and entropy, DS) were calculated from the equation
DG=DHꢀTDS=ꢀRTlnKa, in which T is the absolute temperature and
[5] N. Gilboa-Garber, Methods Enzymol. 1982, 83, 378–385.
[6] A. Imberty, M. Wimmerova, E. P. Mitchell, N. Gilboa-Garber, Mi-
crobes Infect. 2004, 6, 221.
[7] C. Chemani, A. Imberty, S. de Bentzmann, M. Pierre, M. Wimmero-
va, B. P. Guery, K. Faure, Infect. Immun. 2009, 77, 2065–2075.
[8] G. Cioci, E. P. Mitchell, C. Gautier, M. Wimmerova, D. Sudakevitz,
S. Pꢀrez, N. Gilboa-Garber, A. Imberty, FEBS Lett. 2003, 555, 297–
301.
[9] B. Blanchard, A. Nurisso, E. Hollville, C. Tꢀtaud, J. Wiels, M.
Pokornꢅ, M. Wimmerovꢅ, A. Varrot, A. Imberty, J. Mol. Biol. 2008,
383, 837–853.
[10] A. Nurisso, B. Blanchard, A. Audfray, L. Rydner, S. Oscarson, A.
Varrot, A. Imberty, J. Biol. Chem. 2010, 285, 20316–20327.
[11] S. Cecioni, J. P. Praly, S. E. Matthews, M. Wimmerovꢅ, A. Imberty,
S. Vidal, Chem. Eur. J. 2012, 18, 6250–6263.
R=8.314 Jmolꢀ1 Kꢀ1
. Two or three independent titrations were per-
formed for each ligand tested.
Molecular dynamics (MD) calculations: All calculations were performed
by using the TINKER Version 5.0 software suit with an implemented mo-
lecular mechanics MM3 force field. The Solvent Accessible Solvent Area
(SASA) approach was used for the modeling of solution media. Global
minima conformations of the di- and tetrasaccharide matrices as de-
ACHTUNGTRENNUNG
scribed[25] were used for the construction of initial structures of the conju-
gates. After adding galactose-linker moieties, the structures were partially
optimized, leaving the cyclic core frozen until the RMS gradient reached
0.001 kcalmolꢀ1e·ꢂ. The resulting conformations were used as starting
points for MD calculations. The length of MD trajectory was 10 ns for
each structure; structural snapshots were written every 2 ps, yielding 5000
snapshot conformations in each case.
[12] R. U. Kadam, M. Bergmann, M. Hurley, D. Garg, M. Cacciarini,
M. A. Swiderska, C. Nativi, M. Sattler, A. R. Smyth, P. Williams, M.
Cꢅmara, A. Stocker, T. Darbre, J.-L. Reymond, Angew. Chem. 2011,
123, 10819–10823; Angew. Chem. Int. Ed. 2011, 50, 10631–10635.
[13] S. Cecioni, V. Oerthel, J. Iehl, M. Holler, D. Goyard, J.-P. Praly, A.
Imberty, J.-F. Nierengarten, S. Vidal, Chem. Eur. J. 2011, 17, 3252–
3261.
Calculation of RMSD values: RMS deviations (RMSD) between orienta-
tions of galactose residues in each snapshot and in the crystal structure of
2WYF were calculated by using in-house software. The pair of galactose
molecules was taken from the crystal structure of the complex. For each
of the two molecules, the Cartesian coordinates of all oxygen atoms O1
through O6 were extracted and the center of mass of the resulting system
was translated to the origin. Then, the same operation sequence was ap-
plied for each couple of galactose residues from a sample conjugate
structure from MD trajectories. The order of oxygen atoms was kept ex-
actly the same as in the case of galactoses from the crystal structure. This
gave two sets of twelve atoms each, with their centers of mass both
placed at the origin. After that, three random rotations around coordi-
nate axis were applied to the coordinates of the atom set corresponding
to the galactose residues from the MD structure. The RMS difference be-
tween the resulting coordinates and the intact coordinates of the oxygen
atoms from the crystal structure was calculated and averaged across all
12 atoms, giving the RMSD per atom (RMSDpa) value. The procedure
was automatically repeated by using a simple script-like, home-made pro-
gram. On the first step, the calculated value was simply stored, on each
of the next steps the calculated RMSDpa value was compared with the
previous one: if it was greater, it was discarded, if smaller, it was stored
as a replacement value. The program stopped when the difference be-
tween the newly stored and the previous RMSDpa value was less than
0.2 ꢂ and not less than 500000 iterations had previously passed. This pro-
cedure was performed for each structure from the MD trajectories.
[14] M. Reynolds, M. Marradi, A. Imberty, S. Penadꢀs, S. Pꢀrez, Chem.
Eur. J. 2012, 18, 4264–4273.
[15] Y. M. Chabre, D. Giguꢆre, B. Blanchard, J. Rodrigue, S. Rocheleau,
M. Neault, S. Rauthu, A. Papadopoulos, A. A. Arnold, A. Imberty,
R. Roy, Chem. Eur. J. 2011, 17, 6545–6562.
[16] I. Otsuka, B. Blanchard, R. Borsali, A. Imberty, T. Kakuchi, Chem-
BioChem 2010, 11, 2399–2408.
[17] S. Cecioni, R. Lalor, B. Blanchard, J.-P. Praly, A. Imberty, S. E. Mat-
thews, S. Vidal, Chem. Eur. J. 2009, 15, 13232–13240.
[18] Z. H. Soomro, S. Cecioni, H. Blanchard, J.-P. Praly, A. Imberty, S.
Vidal, S. E. Matthews, Org. Biomol. Chem. 2011, 9, 6587–6597.
[19] S. Cecioni, S. Faure, U. Darbost, I. Bonnamour, H. Parrot-Lopez, O.
Roy, C. Taillefumier, M. Wimmerovꢅ, J.-P. Praly, A. Imberty, S.
Vidal, Chem. Eur. J. 2011, 17, 2146–2159.
[20] F. Pertici, R. J. Pieters, Chem. Commun. 2012, 48, 4008–4010.
[21] D. Sicard, S. Cecioni, M. Iazykov, Y. Chevolot, S. E. Matthews, J.-P.
Praly, E. Souteyrand, A. Imberty, S. Vidal, M. Phaner-Goutorbe,
Chem. Commun. 2011, 47, 9483–9485.
[22] Y. Chen, H. Vedala, G. P. Kotchey, A. Audfray, S. Cecioni, A. Imber-
ty, S. Vidal, ACS Nano 2012, 6, 760–770.
[23] H. Vedala, Y. Chen, S. Cecioni, A. Imberty, S. Vidal, A. Star, Nano
Lett. 2011, 11, 170–175.
[24] M. L. Gening, Y. E. Tsvetkov, G. B. Pier, N. E. Nifantiev, Russ. J.
Bioorg. Chem. 2006, 32, 389–399.
[25] M. L. Gening, D. V. Titov, A. A. Grachev, A. G. Gerbst, O. N.
Yudina, A. O. Chizhov, Y. E. Tsvetkov, N. E. Nifantiev, Eur. J. Org.
Chem. 2010, 2465–2475.
[26] D. V. Titov, M. L. Gening, A. G. Gerbst, A. O. Chizhov,Y. E. Tsvet-
kov, N. E. Nifantiev, Carbohydr. Res., DOI: http://dx.doi.org/
10.1016/j.carres.2012.12.005.
[27] A. A. Grachev, A. G. Gerbst, M. L. Gening, D. V. Titov, O. N.
Yudina, Y. E. Tsvetkov, A. S. Shashkov, G. B. Pier, N. E. Nifantiev,
Carbohydr. Res. 2011, 346, 2499–2510.
[28] M. Mammen, S.-K. Choi, G. M. Whitesides, Angew. Chem. 1998,
110, 2908–2953; Angew. Chem. Int. Ed. 1998, 37, 2754–2794.
[29] N. Jayaraman, Chem. Soc. Rev. 2009, 38, 3463–3483.
[30] Architectures of Multivalent Glycomimetics for Probing Carbohy-
drate–Lectin Interactions: M. Lahmann in Glycoscience and Micro-
bial Adhesion (Eds.: T. K. Lindhorst, S. Oscarsson), Springer, Berlin,
2009, pp. 17–65.
Acknowledgements
The authors thank the Universitꢀ Claude Bernard Lyon 1 and the
CNRSfor financial support. Travels and visits were funded by the collab-
orative project CNRS-Russian Academy of Science (no. 23964). A.I. ac-
knowledges support from GDR Pseudomonas and Labex ARCANE. Dr
F. Albrieux, C. Duchamp, and N. Henriques are gratefully acknowledged
for mass spectrometry analyses. S.V. and A.I. are grateful to financial
support from the COST Action CM-1102 MultiGlycoNano. S.C. thanks
the Rꢀgion Rhꢄne-Alpes (Cluster de Recherche Chimie) for additional
funding.
[31] R. J. Pieters, Org. Biomol. Chem. 2009, 7, 2013–2025.
[32] Y. C. Lee, R. T. Lee, Acc. Chem. Res. 1995, 28, 321–327.
[33] J. J. Lundquist, E. J. Toone, Chem. Rev. 2002, 102, 555–578.
[34] S. Cecioni, M. Almant, J.-P. Praly, S. Vidal in Carbohydrate Chemis-
[1] A. Imberty, Y. M. Chabre, R. Roy, Chem. Eur. J. 2008, 14, 7490–
7499.
[2] A. Imberty, A. Varrot, Curr. Opin. Struct. Biol. 2008, 18, 567–576.
[3] N. Sharon, Biochim. Biophys. Acta Gen. Subj. 2006, 1760, 527–537.
[4] Y. M. Chabre, R. Roy, Adv. Carbohydr. Chem. Biochem. 2010, 63,
165–393.
ˇ
try: Proven Synthetic Methods, (Ed.: P. Kovꢅc), CRC Press, Boca
Raton, 2012, pp. 175–180.
9284
www.chemeurj.org
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2013, 19, 9272 – 9285

Multivalent Carbohydrate-Centered Glycoclusters as Nanomolar Ligands
FULL PAPER
[35] V. V. Rostovtsev, L. G. Green, V. V. Fokin, K. B. Sharpless, Angew.
Chem. 2002, 114, 2708–2711; Angew. Chem. Int. Ed. 2002, 41, 2596–
2599.
[36] C. W. Tornoe, C. Christensen, M. Meldal, J. Org. Chem. 2002, 67,
3057–3064.
[39] S. Cecioni, S. E. Matthews, H. Blanchard, J.-P. Praly, A. Imberty, S.
Vidal, Carbohydr. Res. 2012, 356, 132–141.
[40] H. M. Berman, J. Westbrook, Z. Feng, G. Gilliland, T. N. Bhat, H.
Weissig, I. N. Shindyalov, P. E. Bourne, Nucl. Acids Res. 2000, 28,
235–242.
[37] M. Meldal, C. W. Tornoe, Chem. Rev. 2008, 108, 2952–3015.
[38] M. Galibert, P. Dumy, D. Boturyn, Angew. Chem. 2009, 121, 2614–
2617; Angew. Chem. Int. Ed. 2009, 48, 2576–2679.
Received: January 14, 2013
Revised: April 17, 2013
Published online: June 11, 2013
Chem. Eur. J. 2013, 19, 9272 – 9285
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
9285