Multivalent gold glycoclusters: High affinity molecular recognition by bacterial lectin PA-IL

Article abstract of DOI:10.1002/chem.201102034

Multivalent protein-carbohydrate interactions are involved in the initial stages of many fundamental biological and pathological processes through lectin-carbohydrate binding. The design of high affinity ligands is therefore necessary to study, inhibit an

Full text of DOI:10.1002/chem.201102034

DOI: 10.1002/chem.201102034
Multivalent Gold Glycoclusters: High Affinity Molecular Recognition by
Bacterial Lectin PA-IL
Michael Reynolds,[a, b] Marco Marradi,[c, d] Anne Imberty,[a] Soledad Penadꢀs,[c, d] and
Serge Pꢀrez*[a, b]
Abstract: Multivalent protein-carbohy-
drate interactions are involved in the
initial stages of many fundamental bio-
logical and pathological processes
through lectin–carbohydrate binding.
The design of high affinity ligands is
therefore necessary to study, inhibit
and control the processes governed
through carbohydrate recognition by
their lectin receptors. Carbohydrate-
functionalised gold nanoclusters (glyco-
nanoparticles, GNPs) show promising
potential as multivalent tools for stud-
ies in fundamental glycobiology re-
search as well as biomedical applica-
tions. Here we present the synthesis
and characterisation of galactose func-
tionalised GNPs and their effectiveness
as binding partners for PA-IL lectin
from Pseudomonas aeruginosa. Interac-
tions were evaluated by hemagglutina-
tion inhibition (HIA), surface plasmon
resonance (SPR) and isothermal titra-
tion calorimetry (ITC) assays. Results
show that the gold nanoparticle plat-
form displays a significant cluster gly-
coside effect for presenting carbohy-
drate ligands with almost a 3000-fold
increase in binding compared with a
monovalent reference probe in free so-
lution. The most effective GNP exhibit-
ed
a dissociation constant (Kd) of
50 nm per monosaccharide, the most ef-
fective ligand of PA-IL measured to
date; another demonstration of the po-
tential of glyco-nanotechnology to-
wards multivalent tools and potent
anti-adhesives for the prevention of
pathogen invasion. The influence of
ligand presentation density on their
recognition by protein receptors is also
demonstrated.
Keywords: gold · lectin · multiva-
lence · nanoparticle · protein–car-
bohydrate interactions
Introduction
bohydrate ligand. In nature, this low affinity is compensated
by the architecture of the lectin itself, containing typically
two or more carbohydrate binding sites, and by the host pre-
senting the carbohydrate ligands multivalently or as clusters
on the cell surface.[3] The resulting interaction thus being a
combination of several binding events, yet the overall bind-
ing is significantly greater than the combination of the indi-
vidual binding events, that is, the “whole” of the interaction
is greater than the sum of its parts. This effect is known as
the multivalence or cluster–glycoside effect,[4–6] and is well
documented for lectin–carbohydrate interactions for increas-
ing ligand affinity and selectivity. Multivalence has been at-
tributed to increasing structural complementarities between
ligand and receptors,[3] as well as statistical effects such as
effective concentrations and statistical rebinding.[7–12]
Carbohydrates at the cell surface (glycolipids and glycopro-
teins) play key roles in the recognition mechanisms of cells
and their external environment.[1] Specific, reversible inter-
actions between these carbohydrates and protein receptors
(lectins) are of great importance in many biological and
pathological processes ranging from fertilisation and inter-
cellular communication to bacterial invasion and tumour
metastasis.[2] These lectin–carbohydrate interactions typically
exhibit high specificity and weak affinities toward their car-
[a] Dr. M. Reynolds, Dr. A. Imberty, Dr. S. Pꢀrez
CERMAV-CNRS, affiliated with the Universitꢀ Joseph Fourier
BP 53, F-38041, Grenoble Cedex 9 (France)
Fax : (+33)476-547-203
With this in mind, many multivalent “scaffolds” with vari-
ous valencies and modes of ligand presentation have been
developed. Synthetic oligosaccharides, themselves function-
alised with suitable “spacer” molecules have been conjugat-
ed to a number of multivalent scaffolds ranging from cyclo-
peptides,[13,14] glycoproteins,[15–19] dendrimers and den-
drons,[20–24] and polymers[25–27] to polymeric nanoparti-
cles,[28,29] fullerenes,[30,31] calixarenes[32–35] and quantum
dots.[36–40] Over the last decade, an integrated approach for
the construction of a well-defined platform for presenting
carbohydrates was developed. This strategy was based on
the self-assembly of carbohydrate monolayers on the surface
of gold particles, termed glyconanoparticles (GNPs).[41] Sev-
E-mail: serge.perez@cermav.cnrs.fr
[b] Dr. M. Reynolds, Dr. S. Pꢀrez
European Synchrotron Radiation Facility
6 Rue Jules Horowitz, BP 220, 38043
Grenoble Cedex 9 (France)
[c] Dr. M. Marradi, Dr. S. Penadꢀs
CIC biomaGUNE
PoMiramꢁn 182, 20009 San Sebastian (Spain)
[d] Dr. M. Marradi, Dr. S. Penadꢀs
Biomedical Research Networking Centre in Bioengineering
Biomaterials and Nanomedicine (CIBER-BBN)
PoMiramꢁn 182, 20009 San Sebastian (Spain)
Supporting Information for this article is available on the WWW
under http://dx.doi.org/10.1002/chem.201102034.
4264
ꢂ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2012, 18, 4264 – 4273

FULL PAPER
eral reviews have also been published on GNP synthesis,
characterisation and application.[42–44] Functionalising nano-
particles with oligosaccharides has several advantages over
other multivalent scaffolds: Their synthesis and functionali-
sation is a simple, one step process that allows the tuning of
various physical properties (stability, water solubility, cyto-
toxic activity, particle core composition, etc.). Nanoparticle
dimensions are comparable to biomacromolecules and can
be adjusted by modifying particular experimental conditions.
They are also globular in shape, making them ideal for pro-
viding a high valence glycocalyx-like surface for presenting
oligosaccharide molecules. Furthermore, several molecules
of interest can be attached to the same nanoparticle, giving
rise to multifunctionality, and by altering the ratios of the
molecules to be conjugated to the GNP surface, their pre-
sentation density can be controlled. Ligand presentation
density has previously been shown to influence lectin-medi-
ated recognition processes.[45–50] Finally, as well as their
chemical and biochemical properties, nanoparticles of this
size exhibit interesting physical properties due to quantum
size effects, which could also be exploited in numerous ap-
plications.[51,52]
Due to their function, lectins are important targets for
many analytical, diagnostic and therapeutic applications. Re-
cently, our laboratory has characterised several bacterial lec-
tins, which exhibit high affinities for monosaccharide li-
gands. In this report, the PA-IL (LecA) lectin from Pseudo-
monas aeruginosa is considered.[53] P. aeruginosa is an oppor-
tunistic bacteria responsible for numerous nosocomial infec-
tions in patients suffering from cystic fibrosis as well as the
immunocompromised.[54,55] The virulence and host invasion
of this bacterium is mediated by PA-IL and PA-IIL (LecB)
lectins, which exhibit specificities for galactose and fucose,
respectively. Therefore the design of high affinity ligands for
these lectins is required as part of a combined approach for
blocking lectin binding sites and inhibiting infection. This
study presents galactose-functionalised GNPs for binding to
PA-IL. Several GNPs were synthesised exhibiting varying
galactose valencies and presentation densities and their
binding properties studied by hemagglutination inhibition
(HIA), surface plasmon resonance (SPR) and isothermal ti-
tration calorimetry (ITC) assays.
Results and Discussion
Synthesis: We have prepared and characterised a series of
water-soluble gold GNPs functionalised with galactose, glu-
cose and mannose neoglycoconjugates. The GNPs prepared
exhibit varying presentation densities of the galactose and
mannose ligands. The neoglycoconjugates synthesised and
the GNPs prepared are summarised in Figure 1 and Table 1.
Synthesis of neoglycoconjugates 1–3: The thiol-functional-
ised neoglycoconjugate of glucose 1 was synthesised as pre-
viously reported.[56] The galactose neoglycoconjugate 2 was
prepared as for its a-d-mannoside equivalent 3,[56] and is
shown in Scheme 1S (see the Supporting Information). In
this study, the glucose neoglycoconjugate 1 has a short five-
carbon linear aliphatic chain and acts as an inert ligand with
respect to lectin interactions, allowing the modification of
“active” galactoside/mannoside ligand presentation at the
GNP surface. The galactose neoglycoconjugate 2 was conju-
gated to the GNP surface through the thiol-ending group of
the amphiphilic linker as in previous studies.[56] This linker
imparts good chemisorption to the GNP surface as well as
Figure 1. Ligands 1–3 synthesised for protecting Au clusters GNP-1–GNP-10. For clarity, the neoglycoconjugates 1, 2 and 3 are named and depicted as
disulfides.
Chem. Eur. J. 2012, 18, 4264 – 4273
ꢂ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
4265

S. Pꢀrez et al.
Table 1. Summary of GNPs synthesised.
GNP
Average core Shell
Average no. of Average
Average galactose Average galactose Average
diameter [nm] thickness [nm] Au atoms[a]
molecular formula
valency
valency [%]
Mw ACHTUNGTRENNUNG
[a][kDa]
gluco
GNP-1
galacto/gluco
GNP-2
GNP-3
GNP-4
GNP-5
GNP-6
manno/
gluco
U
0.8
100
(C11H21O6S)41Au100
0
0
31.2
1.9
2.2
120
79
140
100
70
(C28H55N2O10S2)12(C11H21O6S)59Au120 12
(C28H55N2O10S2)15(C11H21O6S)30Au79 15
(C28H55N2O10S2)65(C11H21O6S)16Au140 65
(C28H55N2O10S2)57(C11H21O6S)7Au100 57
17
33
80
90
100
47.9
33.6
73.9
58.3
56.9
2.6[b]
2.6[b]
2.6
(C28H55N2O10S2)67Au70
67
manno
valency
GNP-7
GNP-8
GNP-9
GNP-10
N
1.7
2.2
2.7
2.6
125
140
140
70
(C28H55N2O10S2)11(C11H21O6S)72Au125 11 (Man)
(C28H55N2O10S2)24(C11H21O6S)72Au140 24 (Man)
(C28H55N2O10S2)39(C11H21O6S)46Au140 39 (Man)
13 (Man)
25 (Man)
46 (Man)
100 (Man)
51.9
63.3
65.6
56.9
(C28H55N2O10S2)67Au70
67 (Man)
[a] The average number of gold atoms in the cluster, the molecular formulae, and the molecular weights were calculated according to the average gold
core diameter by TEM and elemental analysis.[57] [b] Shell thickness extrapolated from GNP SAXS data.
ligand flexibility and aqueous solubility. The mannose neo-
glycoconjugate 3 and mannose-GNPs were used as negative
controls to account for non-specific PA-IL-–GNP interac-
tions.
tionally small core size of the GNPs, surface plasmon reso-
nance bands were not seen in the UV/Vis spectra. This is
typical of AuNPs of core diameter <2 nm and has been ob-
served for similar GNPs.[51,56] SAXS was used to confirm the
size and size distribution of the core and, in combination
with other techniques, give an estimate of the ligand shell
radii in solution. The corresponding curves can be seen in
the Supporting Information. Curves show scattered X-ray
intensities due to particle shape, size and size distribution.
Data were fitted with a spherical core-shell model with core
radii and dispersion constrained to that measured by
TEM.[60] Results from SAXS data will be discussed in detail
in a separate paper. At low presentation densities of active
ligands 2 or 3, the organic shell thickness around the gold
core is significantly thinner than the length of the active li-
gands in a linear conformation (by up to 50%). This sug-
gests the active ligands fold and collapse on to the glucose
shell perhaps to shield the hydrophobic gold core or form
non-covalent contacts to other surface bound ligands. As
active ligand density increases at the GNP surface, the li-
gands experience increasing steric effects from each other
and are therefore encouraged to adopt a more linear display,
perpendicular to the GNP surface (Figure 2, a more refined
diagram can be found in the Supporting Information).
Preparation of gold glyconanoparticles: The galactose- and
mannose-protected glyconanoparticles (GNP-6 and GNP-
10) exhibit high ligand presentation densities for investigat-
ing lectin–carbohydrate interactions. GNP-2–GNP-5 were
prepared as hybrids of neoglycoconjugates 1 and 2 to study
the influence of galactose presentation density on molecular
recognition by PA-IL. Glucose GNP-1 and mannose/glucose
hybrids GNP-7–GNP-10 were prepared as negative controls
for the assays. All GNPs proved to be stable and soluble in
aqueous environments.
All GNPs were prepared and characterised using proce-
dures previously developed.[58,59] Briefly, a methanolic solu-
tion of the corresponding neoglycoconjugates was added to
an aqueous solution of tetrachloroauric acid (HAuCl4). In
situ reduction of the resulting mixture with NaBH4 gave a
dark brown mixture, which was shaken for 2 h. The solvent
was removed and the aggregates re-suspended in water. Ex-
haustive dialysis against water, followed by centrifugation
and lyophilisation gave the GNPs as amorphous brown
solids. They were characterised by 1H NMR, FT-infrared
spectroscopy, UV/Vis spectroscopy, elemental analysis,
transmission electron microscopy (TEM) and small angle X-
ray scattering (SAXS). The GNPs prepared are water-solu-
ble, stable and can be treated as macromolecules. A summa-
ry of the GNPs produced can be found in Figure 1, Table 1
and in the Supporting Information.
Usually GNPs are soluble in water and stable for several
years in solution. However, some galactose containing
The 1H NMR spectra of the resulting GNPs featured
broader peaks with regard to the neoglycoconjugates in free
solution. From TEM, GNPs show an exceptionally small
average core size (1.2–1.7 nm), with a uniform monomodal
dispersion. Elemental analysis confirms for all GNPs the
coverage density estimated by NMR. From GNP core size
and elemental analysis, an average molecular formula was
estimated according to previous work.[57] Due to the excep-
Figure 2. Schematic representation showing the variation of shell thick-
ness with active ligand presentation density (1.8–2.2–2.6 nm for 10–25–
100% respective presentation density). Solid hexagons represent active
ligands, hollow hexagons represent inactive glucose ligands. Diagram not
to scale.
4266
www.chemeurj.org
ꢂ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2012, 18, 4264 – 4273

Multivalent Gold Glycoclusters
FULL PAPER
GNPs studied here, although water-dispersible, were not
soluble in sufficient quantity in buffer solutions such as
GNP-4 and GNP-5. ITC experiments involving these GNPs
proved difficult.
portant than presentation density. This is again confirmed
upon a reduction in inhibition power between GNP-5 and
GNP-6. This could be due to subtle alterations in ligand pre-
sentation and organisation at the GNP surface from 80 to 90
to 100% galactose presentations. For GNP-6, despite an in-
crease in galactose valency, no further galactose ligands may
be available for binding PA-IL units–the “effective valency”
of a GNP with a high ligand presentation density maybe
similar to the actual valency of a GNP with a lower presen-
tation density. Therefore, GNP-6 may in fact have the same
inhibition power, per “effective ligand”, which is then aver-
aged over all galactosides on the GNP surface, available or
not.
Interaction studies: Binding inhibition by hemagglutination
inhibition assay (HIA): Hybrid galactose-GNP-2–6 were
evaluated as ligands for PA-IL in a hemagglutination assay
with rabbit erythrocytes. Glucose GNP-1 and hybrid man-
nose-GNP-8 were used as negative controls. Lectins are well
known for their ability to agglutinate red blood cells. Meas-
uring the minimum concentration of GNPs required to pre-
vent this agglutination gives an indication of their average
inhibition power. Hemagglutination assays were performed
using PA-IL (5 mgmLÀ1) with GNPs in serial dilutions vary-
ing from 2 mgmLÀ1 to 1 mgmLÀ1. Galactose (and mannose)
concentrations on the GNPs were confirmed by a modified
phenol–sulphuric acid method.[61] Looking at the photo-
graphs of the assay (see the Supporting Information, Fig-
ure 5S), a yellow discolouring of the solutions can be seen
at 2 mgmLÀ1 GNP concentrations due to haemolysis. We
suspect there maybe cytotoxicity at such high concentrations
possibly caused by the ethylene glycol units of the spacer
molecule resulting in cell membrane damage;[62] however, at
lower concentrations this is not seen. The inhibition power
of GNPs 2–6 can be seen in Table 2. The corresponding
monomer 2SAc (protected thiol form of the galactose con-
jugate 2 in free solution) was assigned a relative activity
value of 1.
Binding kinetics by surface plasmon resonance: Surface
plasmon resonance (SPR) was used to study the kinetics
and quantify the interaction between the galactose-GNPs
and PA-IL. 300 RUs of PA-IL were immobilised to a CM5
sensor chip (Biacore/GE, Uppsala, Sweden). Ethanolamine
was immobilised to a second channel of the CM5 sensor
chip to serve as a blank surface. A third channel was func-
tionalised with 500 RUs of the lectin BC2L-A,[63] a mannose
specific lectin, as a negative control for non-specific galac-
tose–GNP interactions and a positive control for mannose–
GNP activity. Solutions of GNP-1–10 in HEPES buffer were
flowed over the sensor chip with concentrations of
20 mgmLÀ1 and serial dilutions thereof to 300 ngmLÀ1. The
galactose containing GNP-2–6 showed significant binding to
the PA-IL surface (see the Supporting Information, Fig-
ure 6S). A typical sensorgram for GNP-6 can be seen in
Figure 3. A one site binding model was applied in the BIA
evaluation software (Uppsala, Sweden) to fit the association
and dissociation of the GNPs to the PA-IL surface. A one
site model was used as this gave the best fit and the lowest
c2 value. The association (ka) and dissociation (kd) rate con-
stants as well as the calculated association (Ka) and dissocia-
tion (Kd) constants are shown in Table 3. As expected,
GNP-1 and GNP-7–10, displaying glucose and mannose on
their surface showed no binding to the PA-IL channel. The
mannose-GNPs showed strong binding to the BC2L-A sur-
face whilst the galactose-GNPs did not (data not shown).
None of the GNPs tested showed binding to ethanolamine.
It can be seen from Table 3 that the trend in Ka is in
agreement with the trend in inhibitory power from HIA,
with large increases in Ka upon increasing galactose presen-
tation density. By comparing the binding activity of GNP-2
and GNP-5, there is over a 6000-fold difference between a
galactose presentation density of 17 and 90%. However, as
for the HIA experiments, there is a decrease in binding ac-
tivity for the highest presentation density, GNP-6, despite
an increase in GNP valency. The preference of PA-IL for
larger presentation densities could be explained by a combi-
nation of structural preferences and statistical binding. As
the presentation density increases, individual ligands may
self-organise into semi-rigid clusters reducing entropic pen-
alties for multivalent binding. The large increase in Ka ob-
served for GNP-5 is clearly related to a significant decrease
Table 2. Results from hemagglutination inhibition assays for PA-IL. nd=
not determined, no inhibition observed at 2 mgmLÀ1 or below. Errors are
the variation in valency within the population of GNPs.
Ligand Galactose Galactose
Inhibiting
Relative activity
valency valency [%] sugar conc. [mm] per ligand
2SAc
1
45.5
1
GNP-2
GNP-3
GNP-4
GNP-5
GNP-6
GNP-1
12
15
65
57
67
0
17
33
80
90
100
G
ACHTUNGTRENNUNG
ACHTUNGTRENNUNG
ACHTUNGTRENNUNG
ACHTUNGTRENNUNG
(101Æ2)
0
nd
nd
GNP-8 24 (Man)
25 (Man)
nd
Even with low density presentations (GNP-2, galactose
17%), a subtle cluster glycoside effect can be observed for
the ligands presented on the GNP scaffold. A significant im-
provement in inhibition power is observed for low valency
GNP-3 (galactose 33%), with minimum galactoside concen-
trations approaching 1 mm. This inhibition power (or cluster
glycoside effect) is then amplified upon further increasing
galactose presentation density, with minimum galactoside
concentrations in the nm range. An activity factor over 100
times stronger was observed per galactose ligand for GNP-5
(galactose 90%) and GNP-6 (galactose 100%) with respect
to 2SAc. However, an increase in inhibition power between
GNP-4 and GNP-5 indicates that GNP valency is less im-
Chem. Eur. J. 2012, 18, 4264 – 4273
ꢂ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
4267

S. Pꢀrez et al.
The thermograms and titration
curves obtained for 2SAc and
GNP solutions can be seen in
Figure 4.
Lectin concentrations are ex-
pressed as monomer concen-
tration hence the single-site
binding model was used to fit
the data. The concentration of
the nanoparticles was ex-
pressed as a concentration of
galactose residues, confirmed
by a modified phenol–sulfuric
acid method.[61] The reaction
stoichiometry (n) is defined as
the number of monomeric gal-
actose units per monomeric
subunit of the lectin. It should
be noted that the n value for
monovalent 2SAc relative to
the lectin was fixed to the
number of binding sites ob-
served per monomer in the
crystal structure (n=1).[65] Ka,
Kd, thermodynamic parameters
and binding stoichiometry for
Figure 3. A) SPR sensorgram for solutions of GNP-6 passing over a CM5 channel functionalised with 300 RUs
of PA-IL. GNP serial dilutions of 20 mgmLÀ1 to 312.5 ngmLÀ1 were made. B) Plots showing deviations of data
points from the fit (residuals) are also shown. Red points represent the data points, the black lines represent
the theoretical fits.
Table 3. Affinity and dissociation rate constants and calculated Ka as
measured for PA-IL by SPR. Data shown are based on single measure-
ments.
2SAc and GNPs tested are listed in Table 4 and shown in
Figure 5. It was found that GNP-4 and GNP-5, which have
an 80 and 90% coverage of galactose residues, respectively,
were insoluble in buffer solutions hence no ITC data is
given. Injections of buffer solution into GNP solutions re-
sulted in negligible heats of dilution. Injections of PA-IL
into solutions of mannose-functionalised GNP-8 (see the
Supporting Information, Figure 7S) served as a reference
for lectin dilution, which was subtracted from the galactose–
GNP thermograms.
GNP
ka
kd
[10À6 sÀ1
Ka
[109 mÀ1
Kd
[nM]
[103 mÀ1 sÀ1
]
G
]
G
]
GNP-2
GNP-3
GNP-4
GNP-5
GNP-6
16
69
69
25
25
291
249
41
0.08
783
0.05
0.28
1.67
306
0.03
20
3.6
0.6
0.003
33
From the raw data, strong exothermic sigmoidal curves
can be observed for the addition of PA-IL to the GNP solu-
tions. The gradients of the sigmoid curves also give an indi-
cation of increased binding. Variations in galactose stoichi-
ometry, n, also suggest varying ligand availabilities at differ-
ent GNP presentation densities. The dissociation constant of
141 mm observed for 2SAc is comparable to that previously
observed for galactose monosaccharides and trisacchar-
ides.[66] However, larger enthalpic and entropic contributions
are observed for 2SAc, most likely due to contributions
from the linker molecule. From the dissociation constants
observed for the GNP systems, there is a clear increase in
PA-IL affinity even with a low presentation density on the
GNP surface, as seen previously for HIA and SPR. The Kd
of GNP-2, with a ligand presentation of 17%, was measured
in the low micromolar range, with a binding potency factor,
b (in which b=Kd,2SAc/Kd,GNP),[3] of 24. This improvement
can be seen thermodynamically as a reduction in entropic
penalties, despite a reduction in enthalpy. Upon increasing
the ligand presentation density to 33%, for GNP-3, the dis-
sociation constant falls to the mid-nanomolar range. The in-
in the macroscopic dissociation rate. This has been related
to a decrease in entropic penalties of the ligand leaving the
binding site.[64] At the same time, higher ligand densities will
increase the probability of correct structural overlap of li-
gands and receptor binding sites as well as statistical rebind-
ing. However, at the largest presentation, the restricted
movement of the ligands at the GNP surface would indeed
reduce entropic penalties, but would in turn exaggerate en-
thalpic penalties due to subtle incompatibilities in protein-
receptor structural complementarity.
Binding thermodynamics by isothermal titration calorime-
try: A further investigation of the thermodynamics of PA-
IL-GNP binding was then carried out using isothermal titra-
tion microcalorimetry (ITC). HIA and SPR studies have
demonstrated that presenting multivalent glycoclusters on a
GNP platform results in significant affinity improvements
for PA-IL binding. Due to the therapeutic importance of
PA-IL regarding pathogen invasion, ITC was used to further
characterise the interaction of PA-IL and galactose–GNPs.
4268
www.chemeurj.org
ꢂ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2012, 18, 4264 – 4273

Multivalent Gold Glycoclusters
FULL PAPER
Figure 4. ITC profiles for A) the “standard” titration of PA-IL ([PA-ILmon]=0.047 mm in the cell) by injections of the galactose monomer, 2SAc
([2SAc]=1.7 mm in the syringe). The molar ratio here refers to the number of galactoside monomers per PA-IL binding site. B)–D) “reverse” titrations
of the GNPs (in the cell) by PA-IL (in the syringe). Molar ratios here refer to the number of PA-IL binding sites per galactoside ligand. B) reverse titra-
tion of GNP-2 (galactoside concentration in the cell, [2]=0.048 mm) by injections of PA-IL ([PA-ILmon]=0.36 mm in the syringe), C) the reverse titration
of GNP-3 (galactoside concentration in the cell, [2]=0.031 mm) by PA-IL ([PA-ILmon]=0.35 mm in the syringe) and D) the reverse titration of GNP-6
(galactoside concentration in the cell, [2]=0.07 mm) by PA-IL ([PA-ILmon]=0.24 mm in the syringe). Concentrations of the galactoside, 2, are given as
(monovalent) thiol concentrations. All experiments were carried out at 258C. Solid black squares represent the data points, the black lines represent the
theoretical fits.
Chem. Eur. J. 2012, 18, 4264 – 4273
ꢂ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
4269

S. Pꢀrez et al.
Table 4. Results of the interactions between PA-IL and galactose-functionalised GNPs, as found by ITC.
Errors are the standard deviations over 2 or 3 experiments. b is the potency of binding with reference to the
monomer 2SAc. All stoichiometries refer to the number of galactosides per PA-IL binding site.
Considering the interaction
at the molecular level, due to
the architecture of the lectin
(Figure 6), chelation of the
four binding sites by ligands
presented on the same GNP is
unlikely. However, intermolec-
ular interactions of one PA-IL
tetramer with several different
GNPs would be possible and
may lead to aggregate struc-
Ligand
Ka
[106 mÀ1
Kd
[mM]
n [stoichio
metry]
DG
[kJmolÀ1
DH
A
ÀTDS
b
A
]
E
]
]
N
]
2SAc
(0.007Æ0.001)
(0.17Æ0.03)
141
5.8
0.76
0.05
1[a]
(À22.0Æ0.4)
(À29.9Æ0.4)
N
G
1
24
185
2824
GNP-2
GNP-3
GNP-6
A
N
(0.76Æ0.04)
(À34Æ3)
(20Æ2)
(2.1Æ0.6)
(À41.7Æ0.3)
[a] Stoichiometry was fixed to 1 according to the crystal structure.
tures, as seen by Sisu et al., and their study of LTBh with
multivalent GM1 ligands,[67] particularly at sub-stoichiomet-
ric concentrations of PA-IL. As the experiment proceeds
however and the PA-IL concentration increases, one would
expect the formation of “chelate complexes”, where two
GNPs bind bivalently to opposite faces of the PA-IL tetra-
mer (Figure 6, a more refined diagram can be found in the
Supporting Information), as suggested previously by Cecioni
et al.[33] and Otsuka et al.[27] In previous ITC studies of PA-
IL with multivalent ligands this chelate complex is charac-
terised by favourable enthalpic contributions and reduced
entropic penalties.[68] As the PA-IL concentration ap-
proaches saturation of the galactose ligands, these chelate
complexes would most likely dissociate to only one face of
the PA-IL tetramer binding to a GNP. From both a structur-
al complementarity and statistical binding point of view, bi-
dentate binding of PA-IL by ligands presented on an AuNP
platform will be increasingly favourable with higher presen-
tation densities. From a low density presentation to a mod-
erate density, a greater probability of correct ligand-binding
site overlap will occur allowing for enthalpic enhancement.
Upon further increasing to high density presentations, the
pre-organised ligands at the GNP surface will induce favour-
able entropic contributions whilst small structural incompa-
tibilities will be exaggerated in the form of enthalpic penal-
ties. Nevertheless, the effective concentrations of ligand-
binding sites will increase the probability of statistical re-
binding.
Figure 5. Graphs of ITC results for PA-IL. Thermodynamic parameters
of the binding of PA-IL to galactose-functionalised GNPs; DG (blue),
DH (red) and ÀTDS (green).
crease in PA-IL affinity in this case is dominated by enthalp-
ic contributions. Further increasing the presentation density
to 100% gives a Kd of 50 nm per galactose ligand, nearly a
3000-fold affinity enhancement, characterised by favourable
entropic contributions. Although there is a clear reduction
in enthalpy, this would be enthalpy averaged over all galac-
tose residues. Given that n=2 for GNP-6, only half of the
galactose residues are available for binding to PA-IL. The
total observed change in enthalpy is subsequently averaged
over all ligands, whether available to interact with a PA-IL
binding site or not. From this we can infer that having an
optimum ligand presentation density is important for in-
creasing enthalpic contributions, yet, higher densities are
necessary to have the same “effective valency” with less en-
tropic penalties. Nevertheless, GNP-6 is the highest affinity
ligand identified for PA-IL to date.
Comparison of HIA, SPR and ITC: To gain a general in-
sight into biochemical interactions, a combination of analyti-
cal techniques is required depending on the nature and
properties of the event to be observed. We have shown that
the different bioanalytical assays used in this study are in
agreement with the strong affinity of galactose ligands pre-
sented on a gold nanoparticle. In all experiments the same
general trend has been observed; ligands presented on a
gold nanoparticle exhibit a significant increase in avidity for
their lectin receptors compared to the ligands in free solu-
tion. Low ligand presentation densities give rise to a signifi-
cant increase in ligand affinity. Upon increasing ligand pre-
sentation density, ligand activity increases further still. HIA
and SPR exhibited similar behaviour regarding inhibitory
power, interaction kinetics and presentation density, with an
increase in ligand activity with presentation density followed
by a reduction in ligand activity at 100% ligand presenta-
Figure 6. Schematic of the chelate complex formed between PA-IL and
galactose-functionalised GNP. Diagram not to scale.
4270
www.chemeurj.org
ꢂ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2012, 18, 4264 – 4273

Multivalent Gold Glycoclusters
FULL PAPER
tion. ITC has shown that GNP-6, which shows a 100%
ligand presentation density, exhibits a 2800-fold increase in
ligand activity compared with the galactose ligand in free so-
lution. With a Kd of 50 nm, this is the most effective PA-IL
ligand to date.
prevention of pathogen invasion.[69,70] The platform has the
potential to be simultaneously tailored with several different
ligands. In particular, a GNP functionalised with both galac-
tose and fucose could prove an interesting multifunctional
inhibitor for both PA-IL and PA-IIL lectins as a combined
approach to prevent infection by Pseudomonas aerugino-
sa.[71] Furthermore, different biologically or diagnostically
important molecules (immunogenic peptides or other anti-
gens, fluorescent markers, etc.) can be grafted to the GNP
surface, allowing the preparation of multifunctional struc-
tures for use as carbohydrate-based inhibitors, drug delivery
agents and diagnostics. Concurrently with applications re-
search, further experimental, computational or crystallo-
graphic studies on AuNP and GNP structure and formation
would provide more detailed, fundamental information re-
garding ligand presentation and behaviour. This in turn
would allow for a fuller interpretation and exploitation of
the multivalence/cluster glycoside effect at the molecular
level.
Although the techniques used are in broad agreement,
they do exhibit certain differences and discrepancies. The
reason for such differences is most certainly due to the fact
that the different techniques are measuring the same bind-
ing event but through different phenomena. HIA, for exam-
ple, is concerned with homogeneous phase aggregation be-
haviour, whereas ITC measures homogeneous phase binding
in solution and SPR heterogeneous binding and dissociation
in which one binding partner is immobilised to a surface.
On one hand, immobilising PA-IL to a SPR chip surface
may reduce the effects of translational and rotational entro-
py upon ligand binding. On the other hand when in free so-
lution, as in the case of an ITC experiment, these entropy
effects maybe further exaggerated during ligand binding
when compared with the natural state of the lectin.
The increase in ligand activity at GNPs can be explained
by a combination of structural complementarities and effec-
tive concentration theories regarding multivalence and clus-
ter glycoside effects. Increasing presentation density may
lead to correct ligand—binding-site overlap whilst also pre-
organising ligand presentation at the GNP surface thus de-
creasing entropic penalties and increasing the probability of
statistical rebinding.
Experimental Section
General procedures: All starting materials were purchased from Sigma–
Aldrich and used without further purification with the exception of tetra-
chloroauric acid monohydrate which was purchased from Strem Chemi-
cals and used without further purification. Reactions were monitored by
thin-layer chromatography (TLC) on Silica Gel 60 F254 aluminium-
backed sheets (Merck) with visualisation under UV (254 nm) and/or by
staining with para-anisaldehyde solution (anisaldehyde (25 mL), H2SO4
(25 mL), EtOH (450 mL) and CH3COOH (1 mL)), phosphomolybdic
acid solution (phosphomolybdic acid (1.3 g), cerium (IV) sulfate monohy-
drate (1 g), concentrated sulfuric acid (6 mL) water (made up to
100 mL)) or Potassium permanganate solution (KMnO4 (2.5 g), K2CO3
(20 g), NaOH (10%), H2O (200 mL)) as stated in the protocol, followed
by heating at over 2008C. Size-exclusion chromatography was performed
on Sephadex LH-20 (Sigma). Flash column chromatography (FCC) was
carried out on Silica Gel 60 (0.063–0.2 mm; E. Merck). All dialyses were
carried out using SnakeSkin pleated dialysis membranes (Pierce, 3500
MWCO). UV/Vis spectra were measured with Varian Cary 50 Bio UV/
Vis spectrometer. Infrared (IR) spectra were recorded from 4000 to
400 cmÀ1 with a PerkinElmer FT-IR spectrometer. Samples were pressed
into KBr pellets. 1H and 13C NMR spectra were acquired on Bruker
AVANCE 400 MHz and Bruker AC 300 MHz spectrometers. Chemical
shifts (d) are given in ppm relative to the residual signal of the solvent
used. Coupling constants (J) are reported in Hz. Splitting patterns are de-
scribed by using the following abbreviations: br, broad; s, singlet; d, dou-
blet; t, triplet; m, multiplet. Mass spectra were recorded on a ZQ Waters
Electrospray LC/MS. For transmission electron microscopy (TEM) ex-
aminations, a single drop (10 mL) of an aqueous solution (ca. 0.1 mgmLÀ1
in Milli-Q water) of the gold glyconanoparticles (GNPs) was placed on
carbon-coated 200 mm mesh copper grids. The grid was covered and left
to dry in air for several hours at room temperature. TEM analysis was
performed with a JEOL 3010 microscope operating at 300 kV to a mag-
nification of 500000. The photographs were taken on Kodak SO163 films
which were then digitalised using a Kodak Mega Plus camera. The diam-
eters of 350–3000 particles were measured for each GNP, using the Scan-
dium 5.0 software (Soft Imaging Systems). The average diameters and
numbers of gold atoms of the GNPs were deduced as described in a pre-
vious study.[57] Laboratory distilled water was further purified using a
Milli-Q water purification system. For clarity, the neoglycoconjugates are
all named and depicted as disulfides.
Conclusion
In summary, we have reported the synthesis and characteri-
sation of galactose-functionalised GNPs. These GNPs exhib-
it several ligand presentation densities ranging from 17–
100% (with respect to an inert glucose conjugate). These
biofunctionalised gold nanoparticles can be used as multiva-
lent platforms for presenting carbohydrate molecules, as
well as for observing the effect of valency, presentation den-
sity, and ligand behaviour on their recognition by protein re-
ceptors. The increase in lectin affinity for nanoparticle-
based carbohydrate ligands has been observed both qualita-
tively by hemagglutination inhibition assays, as well as quan-
titatively using surface plasmon resonance, and the first use
of isothermal titration microcalorimetry for nanoparticle-
scaffolded carbohydrate ligand systems. For the PA-IL
lectin, from Pseudomonas aeruginosa, we have shown that
even low galactoside presentation densities on the GNPs
can increase lectin recognition. At the highest galactose pre-
sentation density, almost a 3000-fold increase in ligand activ-
ity was observed here.
We have confirmed that glyconanoparticle technology
offers a tuneable synthesis for improved ligand presentation.
For the PA-IL lectin, we have demonstrated that the GNP
multivalent ligand system provides the most efficient bind-
ing partner to date with a Kd of 50 nm–another example of
their potential application as potent anti-adhesives for the
Small-angle X-ray scattering: Aqueous nanoparticle solutions of
2 mgmLÀ1 were prepared and centrifuged for 2ꢃ5 min at 13000 rpm.
Chem. Eur. J. 2012, 18, 4264 – 4273
ꢂ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
4271

S. Pꢀrez et al.
SAXS was carried out at the ID02 high brilliance and ID13 microfocus
beamlines of the European Synchrotron Radiation Facililty (ESRF)
using a monochromatic beam of l=1.0597 ꢄ (E=11.7 keV) and l=1 ꢄ
(E=12.39 keV), respectively. Sample–detector distances of 1.5 m and
0.787 m were used giving q ranges of 0.06–4 and 0.2–3 nmÀ1 respectively.
SAXS patterns were recorded by a 16 bit readout FReLoN charge cou-
pled device (CCD) detector with 2048ꢃ2048 pixels of 51ꢃ51 mm2 binned
to 512ꢃ512 pixels. Nanoparticle and water solutions were studied in a
1.7 mm diameter flow-through cell or 200 mm glass capillary. SAXS pat-
terns were obtained after correction and background subtraction. Be-
tween 5 and 10 SAXS patterns were corrected and averaged for fitting.
Data analysis was carried out using the SAXSutilities package
(www.sztucki.de/SAXSutilities).[72] A core-shell model from this package
was used to fit the SAXS patterns. Core radii and polydispersity were set
to those taken from TEM. Shell radii were extracted from the fits.
flow rate of 20 mLminÀ1, and were allowed to dissociate for 3 mins. The
surface was regenerated with 3ꢃ3 min pulses of 100 mm Me-a-d-galac-
tose. Surface regeneration was again confirmed by a repeat injection of a
GNP solution at 1.25 mgmlÀ1 at the end of the concentration series. Bind-
ing was measured as RU (resonance units) over time. Affinity (ka) and
dissociation constant (kd) rates were calculated using the BIA evaluation
software 1.1 (Biacore). Curves were fitted to a single 1:1 binding model,
which gave the best fit as judged by the lowest c2 value and the best dis-
tribution of residuals. Association constants (Ka) were calculated from
the equation: Ka =1/Kd =ka/kd
Microcalorimetry: Titration calorimetry experiments were performed
using a Microcal VP-ITC microcalorimeter. Titrations were carried out in
0.1m Tris/HCl buffer (pH 7.5) containing 3 mm CaCl2, at 258C. 30–40 ali-
quots of 7.5–10 mL of lectin solutions with concentrations of 0.23 mm to
1 mm, were added at 5 min intervals to the GNP solution present in the
calorimeter cell. In the titrations, the GNP concentration varied from
0.46 mgmLÀ1 to 1.12 mgmLÀ1 PA-IL, giving a saccharide concentration
of 0.031 mm to 0.076 mm as found by the phenol-sulfuric acid method
mentioned above. The corresponding monomer molecule, 2SAc, was also
injected into solutions of PA-IL. 2SAc concentrations were 1.7 mm and
PA-IL concentrations 0.05 mm. The temperature of the cell was con-
trolled to (25Æ0.1)8C. Control experiments performed by injection of
buffer into the GNP solution yielded insignificant heats of dilution. Injec-
tions of lectin into control (mannose functionalised) GNPs yielded heats
of dilution, which were subtracted from experimental data during the
data processing phase. Integrated heat effects were analysed by non-
linear regression using a one-site binding model (Origin 7.0, OriginLab
Corp.). Fitted data yielded association constants (Ka) and the enthalpy of
binding (DH). Other thermodynamic parameters, i.e.; changes in free
energy, DG, and entropy, DS, were calculated from the equation: DG=
DHÀTDS=-RTlnKa, in which T is the absolute temperature and R=
8.314 JmolÀ1 KÀ1. Two to three independent titrations were performed for
each lectin–GNP combination.
Recombinant proteins: The lectin PA-IL from Pseudomonas aeruginosa,
was expressed and purified in recombinant form from Escherichia coli as
documented previously.[65]
Monosaccharide analysis: Monosaccharide analysis was carried out using
a variation of the Phenol-sulfuric acid method documented by Saha
et al.[61] Calibration curves were made using solutions of varying concen-
trations (31.3 ngmLÀ1 to 1 mgmLÀ1) of Me-b-d-Glucopyranoside and
Me-b-d-Galactopyranoside corresponding to GNP coverage densities as
determined by NMR, TEM and elemental analysis. To a GNP solution
(50 mL), 5% (v/v) Phenol (aq.) solution (50 mL) was added and mixed.
H2SO4 (250 mL) was added, the mixture was vortexed, and allowed to
stand for 30 mins at room temperature. Readings were taken at 490 nm
against a blank prepared substituting buffer solution for the GNP solu-
tion. Serial dilutions of GNP solutions were also used to confirm carbo-
hydrate concentration and to ensure readings were taken within the
workable range of the spectrophotometer. A Varian Cary 50 Bio spectro-
photometer was used for the absorbance measurements at 490 nm.
Hemagglutination inhibition assay: Rabbit erythrocytes were bought
from Biomerieux and used without further washing. The erythrocytes
were diluted to
a 2% solution in NaCl (150 mm). Lectin solutions
(1 mgmLÀ1) were prepared in Tris/HCl as for the calorimetry studies.
The Hemagglutination unit (HU) was first obtained by the addition of
25 mL of the 2% erythrocyte solution to 25 mL aliquots of sequential
lectin dilutions. The mixture was incubated at 378C for 30 mins followed
by incubation at RT for 30 mins. The HU was taken as the minimum
lectin concentration required to prevent hemagglutination. For the fol-
lowing lectin-inhibition assays, lectin concentrations of four times that of
the hemagglutination unit were used. For PA-IL, this concentration was
found to be 5 mgmLÀ1. Subsequent assays were then carried out by the
addition of 50 mL lectin solution (at the required concentration) to 50 mL
of sequential dilutions of GNPs, monomer molecules and controls. These
solutions were then incubated at 378C for 30 mins followed by 30 mins at
RT. After which, 50 mL of 2% erythrocyte solution was added followed
by a further 30 mins incubation at 378C and 30 mins at RT. The mini-
mum inhibitory concentration for each GNP molecule was recorded.
Acknowledgements
The authors wish to thank the GlycoGold Research Training Network, a
part of the sixth research framework of the European Union, contract
number MRTN-CT-2004–005645, for financial support. Financial support
from the CNRS and ESRF are also acknowledged. We thank also the
Spanish Ministry of Science and Innovation (MICINN, grant CTQ2008–
04638) and the Department of Industry of the Basque Country (Etortek
2009). This project was also supported by COST action CM1102. Dr Ber-
trand Blanchard is greatly appreciated for help with lectin production;
Catherine Gautierꢅs help with ITC is also appreciated. The SPR experi-
ments were ran thanks to access to a Biacore T100 at the Plateforme
Nanobio-Grenoble. We would also like to thank Isabelle Paintrand
(CERMAV-CNRS) and Dr Patricia Donnadieu (SIMAP, INP, Grenoble)
for the TEM photographs. SAXS experiments and analysis were carried
out with the help of Dr. Michael Sztucki and Dr. Emanuela Di Cola
(ESRF, Grenoble). Professors Monica Palcic and Johannis Kamerling are
also thanked for their helpful discussions.
Surface plasmon resonance assays: All SPR experiments were carried
out on a Biacore T100 instrument. CM5 sensor chips (Biacore/GE, Up-
psala, Sweden) were equilibrated with HBS (HEPES-buffered saline:
10 mm HEPES and 150 mm NaCl, pH 7.4) containing 0.005% (v/v)
Tween 20 at 258C with a flow rate of 20 mLminÀ1. Following equilibra-
tion, the chips were activated with two 7 min pulses of a 1:1 mixture (v/v)
of 0.1m N-hydroxy-succinimide and 0.1m N-ethyl-N’-(dimethylaminopro-
pyl)carbodiimide, at 258C and flow rate of 5 mLminÀ1. Ethanolamine hy-
drochloride was immobilised on channel one through an injection of
7 min (1.0m, pH 8.5; ꢀ80 RU) to measure the level of non-specific bind-
ing and to serve as a blank for mathematical data treatment. PA-IL was
[1] A. Varki, R. D. Cummings, J. D. Esko, H. H. Freeze, P. Stanley,
C. R. Bertozzi, G. W. Hart, M. E. Etzler, Essentials of Glycobiology,
2nd ed., Cold Spring Harbour, New York, 2009.
[2] H. Lis, N. Sharon, Chem. Rev. 1998, 98, 637–674.
immobilised to Channel 3 through an injection of 41 mins (100 mgmLÀ1
,
[3] M. Mammen, S. K. Choi, G. M. Whitesides, Angew. Chem. 1998,
110, 2908–2953; Angew. Chem. Int. Ed. 1998, 37, 2754–2794.
[4] Y. C. Lee, R. T. Lee, Acc. Chem. Res. 1995, 28, 321–327.
[5] L. L. Kiessling, J. E. Gestwicki, L. E. Strong, Curr. Opin. Chem.
Biol. 2000, 4, 696–703.
Sodium Acetate buffer, pH 4.5; ꢀ300 RU). Remaining N-hydroxy succi-
nimide esters were blocked by a 7 min pulse of 1.0 Ethanolamine hydro-
chloride, pH 8.5.
Affinity measurements: For the surface functionalised with PA-IL, GNP
solutions (20 mgmLÀ1, and dilutions thereof to 300 ngmLÀ1) in HEPES
buffer were flowed across the sensor chip surface surfaces for 3 mins at a
[6] J. J. Lundquist, E. J. Toone, Chem. Rev. 2002, 102, 555–578.
[7] R. H. Kramer, J. W. Karpen, Nature 1998, 395, 710–713.
4272
www.chemeurj.org
ꢂ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2012, 18, 4264 – 4273

Multivalent Gold Glycoclusters
FULL PAPER
[8] J. M. Gargano, T. Ngo, J. Y. Kim, D. W. Acheson, W. J. Lees, J. Am.
Chem. Soc. 2001, 123, 12909–12910.
[9] T. A. Shewmake, F. J. Solis, R. J. Gillies, M. R. Caplan, Biomacro-
molecules 2008, 9, 3057–3064.
[42] L. Sihelnꢆkovꢇ, I. Tvaroska, Chem. Pap. 2007, 61, 237–255.
[43] M. Marradi, M. Martin-Lomas, S. Penadꢀs, Adv. Carbohydr. Chem.
Biochem. 2010, 64, 211–290.
[44] I. Garcia, J. Gallo, M. Marradi, S. Penadꢀs, in Engineered Carbohy-
drate-Based Materials for Biomedical Applications: Polymers, Surfa-
ces, Dendrimers, Nanoparticles and Hydrogels (Ed.: R. Narain),
Wiley, Hoboken, New Jersey, 2011, pp. 213–259.
[45] P. H. Weigel, R. L. Schnaar, M. S. Kuhlenschmidt, E. Schmell, R. T.
Lee, Y. C. Lee, S. Roseman, J. Biol. Chem. 1979, 254, 10830–10838.
[46] B. T. Houseman, M. Mrksich, Angew. Chem. 1999, 111, 876–880;
Angew. Chem. Int. Ed. 1999, 38, 782–785.
[10] R. S. Kane, Langmuir 2010, 26, 8636–8640.
[11] V. M. Krishnamurthy, V. Semetey, P. J. Bracher, N. Shen, G. M.
Whitesides, J. Am. Chem. Soc. 2007, 129, 1312–1320.
[12] M. Reynolds, S. Perez, C. R. Chim. 2011, 14, 74–95.
[13] I. J. Krauss, J. G. Joyce, A. C. Finnefrock, H. C. Song, V. Y. Dudkin,
X. Geng, J. D. Warren, M. Chastain, J. W. Shiver, S. J. Danishefsky,
J. Am. Chem. Soc. 2007, 129, 11042–11044.
[14] O. Renaudet, Mini-Rev. Org. Chem. 2008, 5, 274–286.
[15] B. Benaissa-Trouw, D. J. Lefeber, J. P. Kamerling, J. F. Vliegenthart,
K. Kraaijeveld, H. Snippe, Infect. Immun. 2001, 69, 4698–4701.
[16] P. M. Rendle, A. Seger, J. Rodrigues, N. J. Oldham, R. R. Bott, J. B.
Jones, M. M. Cowan, B. G. Davis, J. Am. Chem. Soc. 2004, 126,
4750–4751.
[47] C. W. Cairo, J. E. Gestwicki, M. Kanai, L. L. Kiessling, J. Am.
Chem. Soc. 2002, 124, 1615–1619.
[48] E. K. Woller, E. D. Walter, J. R. Morgan, D. J. Singel, M. J. Cloning-
er, J. Am. Chem. Soc. 2003, 125, 8820–8826.
[49] S. Takae, Y. Akiyama, H. Otsuka, T. Nakamura, Y. Nagasaki, K. Ka-
taoka, Biomacromolecules 2005, 6, 818–824.
[17] T. K. Dam, H. J. Gabius, S. Andre, H. Kaltner, M. Lensch, C. F.
Brewer, Biochemistry 2005, 44, 12564–12571.
[18] K. J. Doores, D. P. Gamblin, B. G. Davis, Chem. Eur. J. 2006, 12,
656–665.
[19] H. Nagai, A. Onoda, T. Matsuo, T. Hayashi, Supramol. Chem. 2010,
22, 57–64.
[20] M. J. Cloninger, Curr. Opin. Chem. Biol. 2002, 6, 742–748.
[21] R. Roy, M. G. Baek, Rev. Mol. Biotechnol. 2002, 90, 291–309.
[22] W. B. Turnbull, J. F. Stoddart, Rev. Mol. Biotechnol. 2002, 90, 231–
255.
[23] C. D. Heidecke, T. K. Lindhorst, Chem. Eur. J. 2007, 13, 9056–9067.
[24] S. K. Wang, P. H. Liang, R. D. Astronomo, T. L. Hsu, S. L. Hsieh,
D. R. Burton, C. H. Wong, Proc. Natl. Acad. Sci. USA 2008, 105,
3690–3695.
[25] R. Narain, S. P. Armes, Biomacromolecules 2003, 4, 1746–1758.
[26] L. Bes, S. Angot, A. Limer, D. M. Haddleton, Macromolecules 2003,
36, 2493–2499.
[50] A. G. Barrientos, J. M. Fuente, M. Jimenez, D. Solis, F. J. Canada,
M. Martin-Lomas, S. Penadꢀs, Carbohydr. Res. 2009, 344, 1474–
1478.
[51] E. Boisselier, D. Astruc, Chem. Soc. Rev. 2009, 38, 1759–1782.
[52] P. Crespo, R. Litrꢇn, T. C. Rojas, M. Multigner, J. M. de La Fuente,
J. C. Sꢇnchez-Lꢁpez, M. A. Garcꢆa, A. Hernando, S. Penadꢀs, A.
Fernꢇndez, Phys. Rev. Lett. 2004, 93, 087204.
[53] J. Glick, N. Garber, J. Gen. Microbiol. 1983, 129, 3085–3090.
[54] J. R. Govan, V. Deretic, Microbiol. Rev. 1996, 60, 539–574.
[55] A. Imberty, M. Wimmerova, E. P. Mitchell, N. Gilboa-Garber, Mi-
crobes Infect. 2004, 6, 221–228.
[56] O. Martꢆnez-ꢈvila, K. Hijazi, M. Marradi, C. Clavel, C. Campion, C.
Kelly, S. Penadꢀs, Chem. Eur. J. 2009, 15, 9874–9888.
[57] M. J. Hostetler, J. E. Wingate, C.-J. Zhong, J. E. Harris, R. W.
Vachet, M. R. Clark, J. D. Londono, S. J. Green, J. J. Stokes, G. D.
Wignall, G. L. Glish, M. D. Porter, N. D. Evans, R. W. Murray, Lang-
muir 1998, 14, 17–30.
[27] I. Otsuka, B. Blanchard, R. Borsali, A. Imberty, T. Kakuchi, Chem-
BioChem 2010, 11, 2399–2408.
[58] A. G. Barrientos, J. M. de La Fuente, T. C. Rojas, A. Fernandez, S.
Penadꢀs, Chem. Eur. J. 2003, 9, 1909–1921.
[28] D. Cade, E. Ramus, M. Rinaudo, R. Auzely-Velty, T. Delair, T. Ha-
maide, Biomacromolecules 2004, 5, 922–927.
[59] R. Ojeda, J. L. de Paz, A. G. Barrientos, M. Martin-Lomas, S. Pe-
nadꢀs, Carbohydr. Res. 2007, 342, 448–459.
[29] J. E. Gestwicki, L. E. Strong, S. L. Borchardt, C. W. Cairo, A. M.
Schnoes, L. L. Kiessling, Bioorg. Med. Chem. 2001, 9, 2387–2393.
[30] J. F. Nierengarten, J. Iehl, V. Oerthel, M. Holler, B. M. Illescas, A.
Munoz, N. Martin, J. Rojo, M. Sanchez-Navarro, S. Cecioni, S.
Vidal, K. Buffet, M. Durka, S. P. Vincent, Chem. Commun. 2010, 46,
3860–3862.
[31] 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.
[60] N. R. Jana, L. A. Gearheart, S. O. Obare, C. J. Johnson, K. J. Edler,
S. Mann, C. J. Murphy, J. Mater. Chem. 2002, 12, 2909–2912.
[61] S. K. Saha, C. F. Brewer, Carbohydr. Res. 1994, 254, 157–167.
[62] A. M. Blow, G. M. Botham, D. Fisher, A. H. Goodall, C. P. Tilcock,
J. A. Lucy, FEBS Lett. 1978, 94, 305–310.
[63] E. Lameignere, L. Malinovska, M. Slavikova, E. Duchaud, E. P.
Mitchell, A. Varrot, O. Sedo, A. Imberty, M. Wimmerova, Biochem.
J. 2008, 411, 307–318.
[64] I. Papp, C. Sieben, A. L. Sisson, J. Kostka, C. Bottcher, K. Ludwig,
A. Herrmann, R. Haag, ChemBioChem 2011, 12, 887–895.
[65] G. Cioci, E. P. Mitchell, C. Gautier, M. Wimmerova, D. Sudakevitz,
S. Pꢀrez, N. Gilboa-Garber, A. Imberty, FEBS Lett. 2003, 555, 297–
301.
[66] B. Blanchard, A. Nurisso, E. Hollville, C. Tetaud, J. Wiels, M. Pokor-
na, M. Wimmerova, A. Varrot, A. Imberty, J. Mol. Biol. 2008, 383,
837–853.
[67] C. Sisu, A. J. Baron, H. M. Branderhorst, S. D. Connell, C. A. Weij-
ers, R. de Vries, E. D. Hayes, A. V. Pukin, M. Gilbert, R. J. Pieters,
H. Zuilhof, G. M. Visser, W. B. Turnbull, ChemBioChem 2009, 10,
329–337.
[68] T. K. Dam, C. F. Brewer, Chem. Rev. 2002, 102, 387–429.
[69] J. Rojo, V. Diaz, J. M. de La Fuente, I. Segura, A. G. Barrientos,
H. H. Riese, A. Bernad, S. Penadꢀs, ChemBioChem 2004, 5, 291–
297.
[32] L. Baldini, A. Casnati, F. Sansone, R. Ungaro, Chem. Soc. Rev.
2007, 36, 254–266.
[33] S. Cecioni, R. Lalor, B. Blanchard, J. P. Praly, A. Imberty, S. E. Mat-
thews, S. Vidal, Chem. Eur. J. 2009, 15, 13232–13240.
[34] A. Dondoni, A. Marra, Chem. Rev. 2010, 110, 4949–4977.
[35] S. Cecioni, S. Faure, U. Darbost, I. Bonnamour, H. Parrot-Lopez, O.
Roy, C. Taillefumier, M. Wimmerova, J. P. Praly, A. Imberty, S.
Vidal, Chem. Eur. J. 2011, 17, 2146–2159.
[36] A. Robinson, J. M. Fang, P. T. Chou, K. W. Liao, R. M. Chu, S. J.
Lee, ChemBioChem 2005, 6, 1899–1905.
[37] J. D. M. de La Fuente, S. Penades, Tetrahedron: Asymmetr. 2005, 16,
387–391.
[38] K. Niikura, S. Sekiguchi, T. Nishio, T. Masuda, H. Akita, Y. Matsuo,
K. Kogure, H. Harashima, K. Ijiro, ChemBioChem 2008, 9, 2623–
2627.
[39] R. Kikkeri, B. Lepenies, A. Adibekian, P. Laurino, P. H. Seeberger,
J. Am. Chem. Soc. 2009, 131, 2110–2112.
[70] O. Martꢆnez-ꢈvila, L. M. Bedoya, M. Marradi, C. Clavel, J. Alcami,
S. Penadꢀs, ChemBioChem 2009, 10, 1806–1809.
[40] Y. Yang, M. Yu, T. T. Yan, Z. H. Zhao, Y. L. Sha, Z. J. Li, Bioorg.
Med. Chem. 2010, 18, 5234–5240.
[71] I. Deguise, D. Lagnoux, R. Roy, New J. Chem. 2007, 31, 1321–1331.
[72] M. Sztucki, T. Narayanan, J. Appl. Crystallogr. 2007, 40, S459-S462.
[41] J. M. de La Fuente, A. G. Barrientos, T. C. Rojas, J. Rojo, J. Canada,
A. Fernandez, S. Penadꢀs, Angew. Chem. 2001, 113, 2317–2321;
Angew. Chem. Int. Ed. 2001, 40, 2257–2261.
Received: July 1, 2011
Published online: February 23, 2012
Chem. Eur. J. 2012, 18, 4264 – 4273
ꢂ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
4273