Promotion of sugar-lectin recognition through the multiple sugar presentation offered by regioselectively addressable functionalized templates (RAFT): A QCM-D and SPR study

Article abstract of DOI:10.1039/b716214f

The investigation of recognition events between carbohydrates and proteins, especially the understanding of how spatial factors and binding avidity are correlated, remains a great interest for glycobiology. In this context we have investigated by nanogravimetry (QCM-D) and surface plasmon resonance (SPR), the kinetics and thermodynamics of the interaction between concanavalin A (Con A) and various neoglycopeptide ligands of low molecular weight. Regioselectively addressable functionalized templates (RAFT) have been used as scaffolds for the design of multivalent neoglycopeptides bearing thiol or biotin functions for their anchoring on transducer surfaces. Although these multivalent neoglycopeptide ligands cannot span multiple binding sites within the same Con A protein, they have increased activities relative to their monovalent counterpart. Our results emphasize that the multivalent RAFT ligands function by clustering several lectins, which leads to enhanced affinities.

Full text of DOI:10.1039/b716214f

View Article Online / Journal Homepage / Table of Contents for this issue
PAPER
www.rsc.org/obc | Organic & Biomolecular Chemistry
Promotion of sugar–lectin recognition through the multiple sugar presentation
offered by regioselectively addressable functionalized templates (RAFT):
a QCM-D and SPR study†
Marie Wilczewski, Ange´line Van der Heyden,* Olivier Renaudet, Pascal Dumy, Liliane Coche-Gue´rente and
Pierre Labbe´*
Received 22nd October 2007, Accepted 18th December 2007
First published as an Advance Article on the web 19th February 2008
DOI: 10.1039/b716214f
The investigation of recognition events between carbohydrates and proteins, especially the
understanding of how spatial factors and binding avidity are correlated, remains a great interest for
glycobiology. In this context we have investigated by nanogravimetry (QCM-D) and surface plasmon
resonance (SPR), the kinetics and thermodynamics of the interaction between concanavalin A (Con A)
and various neoglycopeptide ligands of low molecular weight. Regioselectively addressable
functionalized templates (RAFT) have been used as scaffolds for the design of multivalent
neoglycopeptides bearing thiol or biotin functions for their anchoring on transducer surfaces. Although
these multivalent neoglycopeptide ligands cannot span multiple binding sites within the same Con A
protein, they have increased activities relative to their monovalent counterpart. Our results emphasize
that the multivalent RAFT ligands function by clustering several lectins, which leads to enhanced
affinities.
Introduction
Lectin–carbohydrate interactions play a crucial role in the
biomolecular interactions of various biological processes. These
interactions are involved in inflammation processes, in cellular
recognition including adhesion of infectious agents and the im-
mune response.1,2 Lectins, carbohydrate binding proteins, contain
two or more specific sugar-combining sites and comprise a large
family of recognition molecules, especially in the immune system.
While the affinity between lectin and monosaccharides is weak, KD
in the 0.1–1 mM range, sugar–protein interactions are very efficient
and specific due to multivalent events commonly known as the
’glycoside cluster effect”.3 This effect has previously been defined
as an “affinity enhancement achieved by multivalent ligands over
monovalent ones that is greater than would be expected from
a simple effect of concentration increase”.4 Multivalent carbo-
hydrate derivatives that can simultaneously interact with several
binding sites of a multivalent lectin (chelating effect) are relevant
for medicinal interest.4 In this context, some small multivalent
ligands, in which the distances between their carbohydrate moieties
are too low to enable their binding to multiple sites of the same
lectin, have proved to be also more efficient than monovalent
ones.5–7 This improvement could be attributed either to local
ligand concentration effects, also defined as proximity/statistical
Fig. 1 Schematic representation of the three main effects at the origin of
the “glycoside cluster effect”, illustrated by the interaction of a bivalent
lectin with small multivalent ligands which cannot span to multiple sites
of a single lectin (A) in solution, (B) immobilized on a surface, and (C)
comparison with the interaction of a bivalent lectin with monovalent
ligands immobilized on a surface: “1 : 1 interaction”, “chelating” effect,
“proximity/statistical” effect and “clustering” effect.
effects8,9 or, as rarely demonstrated, to a favorable clustering of
lectins (interaction of one multivalent ligand with several lectins).5
These various possibilities related to the “glycoside cluster effect”
are shown pictorially in Fig. 1. In the last few years, we have
developed new molecular tools that combine recognition and
effector properties for diverse biological applications such as
vectors for neo-vasculature targeting,10 cell surface mimics,11 anti-
tumoral synthetic vaccines12 or tumor imaging.13 Our approach
utilizes cyclodecapeptide templates based on the TASP model
which were previously described for protein de novo design.14
These topological templates (namely regioselectively addressable
functionalized templates, “RAFT”) exhibit two independent and
De´partement de Chimie Mole´culaire - Equipe Inge´nierie et Interactions
BioMole´culaires, UMR-5250, ICMG FR-2607, CNRS, Universite´ Joseph
Fourier - BP 53, 38 041 GRENOBLE cedex 9, France. E-mail: pierre.labbe@
ujf-grenoble.fr, angeline.van-der_heyden@ujf-grenoble.fr; Fax: 0 (33) 4 76
51 42 67; Tel: 0 (33) 4 76 51 47 18
† Electronic Supplementary Information (ESI) available: Characterization
of the grafting of thiolated RAFT ligands onto gold surfaces. Con A
interactions with RAFT-(Man)1 and RAFT-(Man)4 surfaces studied by
QCM-D. Analysis of Con A binding using the rectangular hyperbolic
relationship. See http://dx.doi.org/10.1039/b716214f/
1114 | Org. Biomol. Chem., 2008, 6, 1114–1122
This journal is
The Royal Society of Chemistry 2008
©

View Article Online
chemically addressable domains. This structural feature permits
the sequential and regioselective assembly of biomolecule-based
ligands (“recognition domains”) and biologically functional units
(“effector domains”).15,16
We have demonstrated recently that clusters of carbohydrate-
based ligands presented at the surface of RAFT molecules ensure
the specific recognition and a significantly enhanced affinity
for lectins through multivalent interactions in solution11 and
on solid supports.17,18 It is important to notice that since the
distance between two mannose residues of this low-molecular
˚
weight tetramannosyl glycoconjugate (∼25 A, as estimated from
molecular modeling) is much lower than the distance between two
binding sites of Con A (∼65 A),19 simultaneous binding to multiple
˚
sites on the same lectin (chelating effect, Fig. 1A) can be ruled out.
Thus, only two effects, i.e. proximity/statistical effects and/or
lectin clustering could a priori explain the observed enhanced
affinity. It is important to note that additional phenomena coupled
to chelating and/or clustering effects could also occur. Brewer
et al. have reported that, in solution, Con A binding to multivalent
ligands depends on the degree of occupancy with negative
cooperativity,20,21 while positive forces between lectins adsorbed
on sugar surfaces were evidenced by Kiessling et al.22 These latter
phenomena will not be developed in the following discussion.
On solid supports, binding tests between horseradish peroxidase
(HRP) labelled lectins and beads of resin derivatized with neo-
glycopeptides displaying clustered lactose, N-acetylgalactose and
mannose residues have qualitatively shown specific recognition
and enhanced affinity through multivalent interactions.17 Similar
conclusions could be obtained with glycopeptide–oligonucleotide
conjugates bearing a glycocluster by using microtiter plate bind-
ing assays with HRP-labelled lectins.18 Since presentation at a
solid surface is formally multivalent for any system, even with
immobilized monovalent ligands, the enhancement obtained with
our immobilized multiple carbohydrate-based ligands emphasizes
that it is not only the multiple presentation of the sugar ligands
that is important for improving the interaction, but also that the
control of their local density is crucial. While some papers report
systematic studies concerning the impact of the density of mono-
valent carbohydrate modified surfaces on lectin recognition,22–29
to our knowledge, only a few examples of lectin–carbohydrate
recognition investigated as a function of the grafting ratio of
multivalent carbohydrate ligands can be found in literature.30
In order to get deeper insights into this recognition process
between multivalent immobilized ligands and a model lectin (Con
A), we have designed and synthesized RAFT molecules displaying
both clusters of sugars and anchoring elements for streptavidin
or gold surfaces (Fig. 2). The RAFT molecules bearing four
mannose or four lactose moieties (RAFT-(Man)4 and RAFT-
(Lac)4 respectively) were synthesized by using a convergent and
chemoselective oxime bond strategy.17 To assess the multivalent
effect of such interactions, we also prepared the corresponding
monovalent glycopeptides as controls (RAFT-(Man)1 and RAFT-
(Lac)1 respectively). As schematically depicted in Fig. 1, the
situation is rather complex when ligands are immobilized on
a surface. At a high RAFT-(Man)4 surface density (Fig. 1B),
the chelating effect induced by the surface presentation cannot
be dissociated from the local multivalent effects induced by
the RAFT-(Man)4 ligands (proximity/statistical effects or lectin
clustering effects). At a sufficiently low RAFT-(Man)4 surface
Fig. 2 Synthesis of tetravalent (A) and monovalent (B) glycoclusters.
Reagents and conditions: (a) See reference;18 (b) biotin-OSu, DIEA, DMF;
(c) i: BocCys(NPys)–OSu, DIEA, DMF; ii: TFA–CH2Cl2. In order to
facilitate the reading of the manuscript, molecules 2, 4, 6 are called
RAFT-(Man)4, molecules 8, 10 are named RAFT-(Man)1. In the same
manner, molecules 3, 5, 7 are called RAFT-(Lac)4 and molecules 9, 11 are
named RAFT-(Lac)1.
density, the mean distance between adjacent ligands is expected
to become larger than the distance between two binding sites
of the same lectin, thus allowing suppression of the chelating
effect and simplifying the overall system. Similarly, only at a
sufficiently low RAFT-(Man)1 surface density, is a monovalent 1 :
1 interaction with Con A expected to occur without the complexity
induced by the chelating effect and the statistical/proximity effect
(Fig. 1C). For these reasons, we have investigated the Con A
binding properties at various RAFT-mannose surface densities
by diluting them with RAFT-(Lac)4 ligands which do not interact
with the protein. Two complementary real-time techniques, quartz
crystal microbalance with energy dissipation monitoring (QCM-
D) and surface plasmon resonance (SPR), were employed to
investigate the binding processes.
Results
Synthesis
The RAFT molecule 1 (Fig. 2) was prepared following the
previously reported protocol.31 A linear decapeptide was first
synthesized on solid phase following the standard Fmoc–tBu
strategy using a parallel peptide synthesizer, then cyclized in
solution to give the orthogonally protected cyclodecapeptide 1.
After coupling serine moieties to the upper face of the template,
followed by acidolysis of protecting groups then generation
of aldehyde functions by periodate oxidation, aminooxylated
mannose and lactose32 ligands were introduced as glycoclusters
2 and 3 through oxime linkages.
This journal is
The Royal Society of Chemistry 2008
Org. Biomol. Chem., 2008, 6, 1114–1122 | 1115
©

View Article Online
To ensure the further immobilization of these molecules on
streptavidin coated surfaces, we used an active ester to introduce
biotin on the side chain of the lysine pointing from the lower face
of the molecules 2 and 3. This reaction occurred in DMF at basic
pH to give biotinylated compounds 4 and 5 in good yields after
reverse-phase HPLC purification (RP-HPLC).
As a more direct immobilization strategy, we functionalized
the compounds 2 and 3 with a cysteine residue bearing the
activating group S-3-nitro-2-pyridinesulfenyl (Npys) on the side
chain. BocCys(Npys) was coupled to the free lysine as the succinic
ester in DMF. After removal of the Boc group by treatment with
TFA solution and RP-HPLC purification we obtained 6 and 7
which could be readily immobilised on gold surfaces.
QCM-D experiments
Fig. 3 Typical QCM-D response, normalized frequency Df n/n (bold line)
and Dn variation (thin line) for the third overtones (n = 3), recorded
during Con A adsorption onto a surface prepared by adsorption of
a 100% RAFT-(Man)4 solution. (a) Lectin adsorption (a1) 0.049 lM,
(a2) 0.098 lM, (a3) 0.49 lM, (a4) 0.98 lM, (a5) 4.9 lM and (a6)
9.8 lM. The injection of buffer leads to partial dissociation of Con
A. Surface regeneration is performed by successive injection of (b) 25 mM
mannose solution and (c) 0.05% SDS. The experiments were performed in
QCM-D measures absolute areal mass density (ng cm−2) without
the need for a reference channel and with a typical mass
sensitivity of 3.5 ng cm−2. RAFT self-assembled monolayers
(SAMs) exhibiting variable densities of mannose residues were
prepared outside the measurement chamber, by adsorption from
0.1 mM aqueous solutions of RAFT-(Man)1 (ligand 10) or RAFT-
(Man)4 (ligand 6) diluted with RAFT-(Lac)4 (ligand 7) at RAFT-
mannose : RAFT-(Lac)4 molar ratios of 100, 33, 20, 10, 5 and 2%.
A cleaned gold QCM-D sensor was immersed overnight in the
RAFT solution, rinsed by milliQ water and dried by nitrogen.33,34
The surface densities mentioned in the text correspond to the
RAFT-mannose bulk concentrations of the solutions used for
adsorption. Finally, the functionalized gold-coated quartz was
set up in the measurement chamber. 0.005% P20 was added to
the working buffer (WB) to prevent non-specific adhesion. No
adhesion was observed during a 1lM peanut agglutinin (PNA,
a lectin non-specific for mannose) injection onto a 100% RAFT-
(Man)4 SAM, confirming the efficiency of the P20 use. Ex situ
ellipsometry measurements were performed on the gold coated
quartz crystal before and after the RAFT immobilization (see
ESI†) and gave for the dried RAFT layer an average thickness of
2.1 nm, in agreement with the molecular modeling study (2.5 nm).
An in situ kinetic study of the RAFT self-assembled monolayer
formation was performed by QCM-D (see ESI†). Results confirm
that SAM formation is a slow process that needs ∼15 hours. Since
RAFT immobilization on gold coated quartz offers the ability
of surface regeneration, cyclic adsorptions–desorptions of Con A
lectin were performed at increasing concentrations on the different
SAMs. Each protein adsorption–desorption cycle was followed by
two regeneration steps, firstly with a highly concentrated mannose
solution (25 mM) and then with a 0.05% SDS solution. Indeed,
mannose injections, even at a 50 mM mannose concentration, did
not procure a total regeneration of the surface. A representation of
a single experiment conducted on a 100% RAFT-(Man)4 surface
is shown in Fig. 3.
flow mode at 100 lL min−1
.
A adsorption on the different surfaces, maximum mass uptakes
recorded after 9 min of association have been plotted as a function
of Con A concentration (Fig. 4).
For both RAFT-mannose saturated surfaces, similar maximum
areal mass uptakes of about 710 ng cm−2 are obtained upon
Con A binding. It is interesting to compare this value with a
theoretical mass value calculated for a hydrated Con A monolayer.
Assuming a density for the Con A monolayer of ∼1.165 g cm−3
˚
and considering the hydrated Con A as a cube with edge ∼65 A,
35,36 a theoretical mass of ∼770 ng cm−2 is obtained. The maximum
mass uptake extracted from the QCM-D data agrees well with the
theoretical mass. Clearly the mannose surface density exhibited
by both kinds of surfaces is enough to ensure the formation of a
hydrated Con A monolayer and provides enough sugar heads to
complete a multivalent adhesion of the protein.
Differences between the two surfaces are highlighted for higher
dilutions of RAFT-mannose in the SAMs. For equivalent di-
lutions of RAFT-mannose, larger mass uptakes are observed
for RAFT-(Man)4 surfaces compared to RAFT-(Man)1 ones,
meaning that more protein is adsorbed onto RAFT-(Man)4
SAMs. By increasing the RAFT-mannose dilution in the SAM,
distances between ligands on the surface increase and protein
binding via more than one RAFT-mannose should be avoided.
In this case, the chelating effect involved by the surface (Fig. 1)
should be progressively removed to the benefit of the local ligand
concentration effect (clustering or proximity/statistical effects).
However, while a RAFT-(Man)4 molecule benefits more from a
higher proximity/statistical effect when binding to one protein
than does a RAFT-(Man)1 molecule, this effect could not explain
the significantly higher amount of protein adsorbed onto the
RAFT-(Man)4 surfaces. It could then be supposed that even if the
distance between two sugars on the RAFT scaffold is about 2.5 nm,
a RAFT-(Man)4 molecule allows the immobilization of several
different Con A. More quantitative information is difficult to
extract from these QCM-D experiments for the following reasons.
Con A interactions with RAFT-(Man)1 and RAFT-(Man)4
were compared by overlaying part of the curves concerning
each adsorption–desorption cycle of Con A for the same SAM
(Fig. ESI-3†). Since the dissipation factor remains low (less than
10−6) for adsorption of the protein, we considered as a first
approximation a rigid behaviour for the Con A layer so that
the normalized frequency shifts can be linearly related to mass
uptake using the Sauerbrey equation. In order to compare the Con
1116 | Org. Biomol. Chem., 2008, 6, 1114–1122
This journal is
The Royal Society of Chemistry 2008
©

View Article Online
Fig. 4 Maximum mass uptake recorded for Con A binding onto (A) RAFT-(Man)1 and (B) RAFT-(Man)4 adsorbed on gold surfaces. The dotted lines
have been drawn for better visualization. The numbers indicate the RAFT-mannose molar ratio of the solutions used for adsorption. Other experimental
conditions are reported in the caption for Fig. 3.
Firstly the mass uptake measured by QCM-D includes, in addition
to the mass of dry protein, a contribution due to acoustically
coupled water. Both contributions (dry protein and coupled
water) are not linearly linked. Reliable kinetic or thermodynamic
information cannot be deduced since only the amount of adsorbed
protein is relevant for this kind of data treatment. Secondly, the
association curves of Con A (Fig. ESI-3†) present linear portions
that are characteristic of mass transport limitations. Although
the present QCM-D experiments were conducted under flowing
conditions, both the large measuring chamber volume (∼80 lL)
and the low flow rate (100 lL min−1) lead to mass transport
limitations. For these reasons, SPR experiments based on Biacore
technology were realized.
kinetics recorded by the Biacore system. In situ functionalization
of each of the four Biacore channels could not be realized because
it would have needed such a large amount of thiolated RAFT
molecules, which was incompatible with the available quantities.
To overcome these difficulties, the four Biacore channels were
selectively functionalized with biotinylated RAFT molecules by
the way of biotin–streptavidin links. Streptavidin was, firstly,
grafted to a C1 carboxylated surface using a classical amine
coupling. A similar quantity of streptavidin, ∼370 RU, was grafted
onto each flow cell, in order to guarantee the same number of
available biotin binding sites and the anchoring of a reproducible
RAFT surface. The mass uptake DmSPR can be obtained from
relation (1) with CSPR calibrated to ∼0.066 ng cm−2 for protein
adsorption on a flat gold surface.38
eff
SPR experiments
n
− nbuff er
biomolecule
DmSPR = d
= CSPRDRU
(1)
dn/dc
SAMs exhibiting various sugar dilutions were realized by adsorp-
tion from solutions of thiolated RAFT-mannose diluted to the
desired ratio with thiolated RAFT-(Lac)4 (Fig. 2). Cleaned gold
chips are activated by two injections of 70% ethanol solution.
Then various 0.1 mM RAFT solutions in WB are injected during
20 minutes at 5 lL min−1.37 A 100% RAFT-(Lac)4 SAM is used
as a reference. Kinetics curves recorded on this surface reveal
a non-specific adhesion of Con A (from 24.5 nM to 39.2 lM)
even in the presence of 0.05% of P20 in the working buffer. An
increase in the concentration of the thiolated RAFT solution
to 1 mM, an increase of contact time between the gold surface
and the RAFT solutions to 1 hour, as well as a bovine serum
albumin injection prior to the lectin injection were also tested
unsuccessfully. Since this problem was not observed with QCM-
D experiments, an explanation could be that the contact time
between the thiolated molecules and the gold surface (20 min)
during a functionalization directly in the SPR sensor chamber
is too short compared to the ex situ immobilization (15 hours)
performed for the QCM-D experiments. As a consequence, only a
partial coverage of the gold surface by thiolated RAFT molecules
is obtained whereas the remaining free gold surface is expected
to induce a strong non-specific Con A adhesion. It could be
remarked that while a 15 hour ex situ immobilization of thiolated
RAFT on a gold sensor chip surface is possible, it will lead to
the same functionalization of the four Biacore microchannels. In
consequence, no reference surface will be available to exploit the
The amount of grafted streptavidin was thus estimated to be
∼24.5 ng cm−2, which is about 10% of a monolayer.39 This low
grafting level was chosen in order to minimize mass transport
limitation during SPR kinetic experiments and to dilute RAFT-
mannose on the surface. Anchoring of biotinylated RAFT was
then performed by 3 min injection of 5 lM RAFT solutions in
WB, at 5 lL min−1. Surfaces exhibiting various RAFT-(Man)1
(ligand 8) or RAFT-(Man)4 (ligand 4) densities were prepared by
adsorption of RAFT-mannose–RAFT-(Lac)4 (ligand 5) mixtures
obtained by dilution of RAFT-mannose solutions with RAFT-
(Lac)4 solution to the desired molar ratio (100%, 25%, 10% and
5%). As for the QCM-D study, the surface densities mentioned in
the text correspond to the RAFT–mannose bulk concentrations
of the solutions used for adsorption. A surface coated only with
RAFT-(Lac)4 was tested and used as a reference. After Con A
binding, the RAFT-mannose surfaces could be regenerated by
injecting a 50 mM mannose solution. Similar sensorgrams were
recorded at various flow rates, which demonstrates the absence of
mass transfer limitations (not shown). Fig. 5 compares the binding
curves of Con A, recorded at 25 lL min−1, onto surfaces exhibiting
various surface densities of RAFT-(Man)4 and RAFT-(Man)1. As
already observed by QCM-D, the amount of bound Con A is
systematically higher on the RAFT-(Man)4 surfaces, whatever the
RAFT-mannose surface density is. Another important difference
occurs in the rate of the Con A association–dissociation process,
This journal is
The Royal Society of Chemistry 2008
Org. Biomol. Chem., 2008, 6, 1114–1122 | 1117
©

View Article Online
Fig. 5 Comparison of the sensorgrams recorded at 25 lL min−1 on RAFT-(Man)4 (right) and RAFT-(Man)1 (left) surfaces. The reference surface is a
100% RAFT-(Lac)4. The RAFT-mannose surfaces were prepared by adsorption of mixed solutions of RAFT-mannose–RAFT-(Lac)4 on streptavidin
surfaces. The RAFT-mannose molar ratios of the adsorption solutions are (A) 100% (B) 25% (C) 10% and (D) 5%. At 5% dilution, only the signal
recorded on the RAFT-(Man)4 surface is detectable.
which is much more rapid on the RAFT-(Man)1 surfaces than on
the RAFT-(Man)4 ones. The responses obtained at equilibrium
(Req) have been considered first. At the lowest RAFT-(Man)1
surface density (10%), a linear Scatchard plot (Req/C as a function
of Req, with C = [Con A]) is obtained (Fig. 6A). It can thus
be concluded that at this dilution, the mean distance between
immobilized RAFT-(Man)1 is large enough to avoid a multivalent
interaction of Con A with the surface immobilized mannose
(chelating effect, Fig. 1C) and that a true monovalent 1 : 1
interaction is observed as expected. From this linear Scatchard
plot, a dissociation constant KD of 110 lM was extracted, which
is in good agreement with the values reported in the literature for
the interaction of monomeric mannose with Con A.22
At higher RAFT-(Man)1 surface densities (25% and 100%),
non-linear Scatchard plots were obtained (Fig. 6A) meaning that
the interaction cannot be described as a simple 1 : 1 mechanism.
As depicted in Fig. 1C, a multivalent interaction of Con A with the
RAFT-(Man)1 surface (chelating effect) can explain this deviation.
For surfaces modified by RAFT-(Man)4, non-linear Scatchard
plots were systematically obtained (Fig. 6B), meaning that, even
at the lowest dilution used, a simple monovalent 1 : 1 interaction is
never obtained. In this case, a more complex situation is obtained
since both the chelating and the clustering effects (Fig. 1B) can be
at the origin of this behavior.
Discussion
Our first objective was to rationalize how the presentation of
multiple mannose epitopes offered by RAFT scaffolds increases
their Con A binding activities relative to their monovalent coun-
terparts. For this purpose, RAFT ligands have been immobilized
on gold surfaces at various surface densities, and the binding
affinity of Con A towards the resulting surfaces has been assessed
using QCM-D and SPR. Since QCM-D allows absolute areal
mass determination without needing a reference channel, ex situ
functionalization of the gold surface transducer is possible by way
of self-assembled monolayers of thiolated RAFT. This approach
allowed us to prepare surfaces that were fully saturated by RAFT-
(Man)1 or RAFT-(Man)4 ligands. Both kinds of these saturated
surfaces were able to bind a similar monolayer of Con A, meaning
that at high RAFT surface densities, the multiple presentation of
mannose does not improve, as expected, the binding activity. It
is only at lower surface densities that different behaviors appear
between monovalent and multiple ligands. This work has also
1118 | Org. Biomol. Chem., 2008, 6, 1114–1122
This journal is
The Royal Society of Chemistry 2008
©

View Article Online
Fig. 6 Scatchard plots of the interaction of Con A with increasing surface densities of (A) RAFT-(Man)1 and (B) RAFT-(Man)4. The numbers indicate
the RAFT-mannose molar ratio of the solutions used for adsorption. Insert: part of graph (B) enlarged. Experimental conditions as in Fig. 5.
Table 1 Intrinsic dissociation constant (KDX) and effective concentration (CX*) of affinity binding sites for the interaction of Con A with RAFT-mannose
molecules immobilized at various surface densities. These values are obtained by analyzing the amount of bound Con A on the basis of the modified
rectangular hyperbolic relationship with f = 2
RAFT-(Man)1
K
DX/lM
CX*/lM
r
RAFT-(Man)4
K
DX/lM
CX*/lM
r
100%
25%
10%
45
105
154
0.023
0.018
0.012
0.9993
0.9987
0.9970
100%
25%
10%
5%
6.6
34
68
84
0.067
0.049
0.030
0.017
0.9972
0.9960
0.9970
0.9966
pointed out the limitations of QCM-D for these kinds of binding
studies. Since QCM-D is sensitive not only to the mass of protein
but also to the mass of water that is acoustically coupled with
the protein layer, only qualitative results can be obtained by
this technique. As a consequence of the design of the QCM-D
measuring chamber, mass transport limitations also constitute an
important parameter that should be taken into account for any
kinetic studies.
Since SPR is only sensitive to the local change in refractive
index of a thin layer of liquid in proximity to the transducing
surface, the true amount of binding protein can be determined
independently of the coupled water. The disadvantage lies in the
need to subtract the signal of a reference channel (where binding of
the protein does not occur) in order to abstract the refractive index
change of the protein solution with regards to the pure buffer. As
a consequence, surface functionalization of the different Biacore
measuring microchannels must be realized in situ, which in our
case prohibits the use of thiolated RAFT.
On the other hand, the covalent binding of streptavidin on a
C1 sensorchip provided an excellent platform for assessing the
binding affinity of Con A for various ligands immobilized on the
surface through biotin–streptavidin bridges. In particular, using
a low surface density of immobilized streptavidin (∼10% of a
monolayer39) and a high dilution (10%) of the biotinylated RAFT-
(Man)1 with RAFT-(Lac)4 provided a surface where only 1 : 1
interactions occurred between immobilized RAFT-(Man)1 and
Con A in solution. In agreement with the literature, we found a
weak binding affinity (KD = 110 lM) for the monovalent RAFT-
(Man)1–Con A interaction. When the RAFT-(Man)1 surface
density was increased up to 25% and 100% of the available
streptavidin binding sites, larger equilibrium amounts of bound
Con A and non-linear Scatchard plots were obtained (Fig. 6A).
This was interpreted by multivalent interactions of Con A with the
surface (chelating effect, Fig. 1C). Using a modified rectangular
hyperbolic relationship, Winzor and co-workers have reported a
method for determining an intrinsic dissociation constant (KDX
)
and an effective concentration of binding sites (CX*) on a surface
for the multivalent interaction of a protein having f equivalent
and independent binding sites.40 From the raw SPR data, a plot of
effective bound surface concentration versus the effective injected
concentration of Con A was generated (see ESI†) from which KDX
and CX* could be extracted (Table 1).
In the case of RAFT-(Man)1 surfaces, it is first interesting to
observe that the modified rectangular hyperbolic relationship,
used here for a bivalent interaction of Con A (chelating effect
with f = 2), allows an excellent fitting of the experimental data
with high correlation coefficients. As the RAFT-(Man)1 surface
density increases, the number of available binding sites and the
affinity increase (CX* increases and KDX decreases). Since it
was demonstrated previously that at low RAFT-(Man)1 surface
density (10%), the interaction was monovalent, this bivalent
Scatchard analysis appears meaningful only at higher RAFT-
(Man)1 densities (25% and 100%). In the case of RAFT-(Man)4
surfaces, excellent data fitting is also observed. However in this
case, the bivalent Scatchard analysis does not constitute a perfect
theoretical model, since in addition to the chelating effect (that
is only described by this model), some Con A clustering can
occur as a consequence of the multiple mannose presentation
This journal is
The Royal Society of Chemistry 2008
Org. Biomol. Chem., 2008, 6, 1114–1122 | 1119
©

View Article Online
provided by RAFT-(Man)4 molecules. Indeed we observed that, as
the RAFT-(Man)4 surface density increases, the affinity increases,
but with KDX values that are systematically lower than those
observed with RAFT-(Man)1 (Table 1). Another important point
is that the amount of available binding sites (CX*) is systematically
higher (by a ratio of about 2.5–3) on RAFT-(Man)4 surfaces
(Table 1). Even at a similar mannose surface density but different
spatial organization (comparison of 100% RAFT-(Man)1 and 25%
RAFT-(Man)4) a better affinity is obtained for RAFT-(Man)4
as compared to RAFT-(Man)1 surfaces (KDX = 34 lM and
45 lM respectively) as well as a higher amount of binding sites
(CX* = 0.048 lM and 0.023 lM respectively). This qualitative
analysis strongly suggests that a clustering of Con A can occur
on immobilized RAFT-(Man)4, which greatly improves both the
affinity, by a factor of 3 to 7, and increases the number of binding
sites by a factor of about 3.
To confirm the ability of RAFT-(Man)4 molecules to bind
two or more proteins, a competitive binding experiment was
conducted and followed by SPR. Solutions of Con A (16.33 lM)
mixed with various quantities of RAFT-(Man)1 or RAFT-(Man)4
were injected onto the 100% RAFT-(Man)1 surface prepared
via streptavidin–biotin interaction as described previously. The
responses obtained at the end of the association time as a function
of the RAFT-mannose concentration in solution are presented in
Fig. 7.
surfaces. In this range of concentrations, RAFT-(Man)4 behaves
as a promoter of the Con A–RAFT-(Man)1 surface interaction
since Con A aggregates exhibit a more important multivalent
behavior. For higher RAFT-(Man)4 concentrations, inhibition of
the interaction between Con A and the RAFT-(Man)1 surface is
observed as expected. Whereas Con A presents four sugar-binding
sites, inhibition occurs at a RAFT-(Man)4–Con A molar ratio
closer to 1. Probably the lectins are too aggregated to exhibit free
binding sites to the surface.
Conclusion
Tetravalent carbohydrate-based ligands presented at the surface
of cyclodecapeptidic RAFT scaffolds cannot span the saccharide
binding sites within the same Con A tetramer. In solution, these
multivalent ligands have been found to promote the formation of
Con A clusters that exhibit enhanced affinity toward carbohydrate-
substituted surfaces. When immobilized on a surface at various
surface densities, these multivalent ligands exhibit affinities to-
wards Con A that are typically 3 to 7 fold better than their
monovalent counterpart. At similar surface densities they can
bind about 3-fold higher amounts of Con A as a consequence
of their ability to form clusters of lectins. This lectin clustering
process could be at the origin of the higher affinity exhibited by
the multivalent ligands, although the presence of a high local con-
centration could also participate through a proximity/statistical
effect. Complementary work is in progress to quantify the relative
influence of these two processes.
Experimental
Synthesis
All chemical reagents and solvents were purchased from Sigma
Aldrich, Acros or Carlo-Erba and were used without further
purification. All protected amino acids were obtained from Ad-
vanced ChemTech Europe (Brussels, Belgium), Bachem Biochimie
SARL (Voivins-Le-Bretonneux, France) and France Biochem S.A.
R
(Meudon, France). Fmoc-Gly-SASRINꢀ resin was purchased
from Bachem and PyBOP from France Biochem. Reverse phase
HPLC analyses were performed on Waters equipment: the ana-
˚
lytical (Nucleosil 120 A 3 lm C18 particles, 30 × 4.6 mm) was
operated at 1.3 mL min−1 and the preparative (Delta-Pak 300 A
˚
Fig. 7 SPR response recorded for the association of Con A (16.33 lM)
with 100% RAFT-(Man)1 surfaces obtained through biotin-streptavidin
bridges in the presence of increasing concentrations of RAFT-(Man)4
(solid line) or RAFT-(Man)1 (dashed line).
15 lm C18 particles, 200 × 25 mm) at 22 mL min−1 with UV
monitoring at 214 nm and 250 nm using a linear A–B gradient
(buffer A: 0.09% CF3CO2H in water; buffer B: 0.09% CF3CO2H
in 90% acetonitrile). Mass spectra were obtained by electron spray
ionization (ES-MS) on a VG Platform II in the positive mode.
No signal variations are observed for injection of Con A in
the presence of RAFT-(Man)1 meaning that this monovalent
molecule is not able, within the range of concentrations explored,
to compete efficiently with the binding affinity of Con A towards
the multivalent mannosylated surface. On the opposite hand, a
large signal enhancement is observed for Con A injected in the
presence of 3 to 6 lM of RAFT-(Man)4, with a maximum response
recorded for 5 lM. Thus, tetravalent RAFT-(Man)4 appears to
effectively bind and promote the clustering of Con A in solution
at concentrations where no inhibition by monovalent RAFT-
(Man)1 is observed. It could be supposed that in the bulk phase,
RAFT-(Man)4 binds at least two or three lectins to form soluble
clusters.5 These clusters could then interact with the RAFT-(Man)1
Synthesis of tetravalent neoglycopeptides 2 and 3
Aminooxy a-D-mannopyranosyl (28 mg, 144 lmol) or b-D-
lactopyranoyl (109 mg, 305 lmol)32 was added to a solution of
RAFT molecule bearing aldehyde functions (18 mg, 14.4 lmol for
mannose; 38 mg, 30.5 lmol for lactose) in 20% aqueous acetic
acid (1 mL). After stirring for 3 hours at room temperature,
compounds 2 (20 mg, 71%) and 3 (46 mg, 59%) were purified
by semi-preparative RP-HPLC (linear gradient: 95 : 5 to 60 : 40
A : B in 30 min, detection: k = 214 and 250 nm). Compound 2:
Rt = 7.3 min (linear gradient: 95 : 5 to 60 : 40 A : B in 15 min);
1120 | Org. Biomol. Chem., 2008, 6, 1114–1122
This journal is
The Royal Society of Chemistry 2008
©

View Article Online
ES-MS (positive mode): calcd for C79H130N19O38 1952.88 [M + H]+,
found: 1953.09; compound 3: Rt = 6.6 min (linear gradient: 95 :
5 to 60 : 40 A : B in 15 min); ES-MS (positive mode): calcd for
C103H169N19O58Na 2623.07 [M + Na]+, found: 2622.86.
Buffer and other chemicals
Concanavalin A (Con A) was purchased from Fluka, streptavidin,
bovine serum albumin and peanut agglutinin from Sigma. All
other reactants are analytical grade, CaCl2 came from Normapur,
anhydrous MnCl2 from Acros organics, sodium dodecyl sulphate
(SDS) from Panreac and N-(2-hydroxyethyl)piperazine-Nꢀ-(2-
ethanesulfonic acid) (HEPES) from Euromedex. P20 was provided
by Biacore.
The working buffer (WB) used for all experiments consisted of
HEPES saline buffer (HBS) (0.1 M HEPES, NaCl 0.1 M) pH 7.2
with 1 mM CaCl2 and 1 mM MnCl2.
Synthesis of biotin functionalized tetravalent neoglycopeptides 4
and 5
Neoglycopeptide 2 (7 mg, 3.6 mmol) and biotin-OSu (2.5 mg, 7.2
mmol) were dissolved in DMF (1 mL) and the pH of the solution
adjusted to 8 with DIEA. After 30 minutes at room temperature
and purification by semi-preparative RP-HPLC (linear gradient:
95 : 5 to 60 : 40 A : B in 30 min, detection: k = 214 and 250 nm),
compound 4 (5 mg, 59%) was obtained as a white powder: Rt =
8.8 min (linear gradient: 95 : 5 to 60 : 40 A : B in 15 min); ES-
MS (positive mode): calcd for C89H144N21O40S 2178.96 [M + H]+,
found: 2178.65. The same procedure was followed for the synthesis
of 5: Rt = 8.4 min (linear gradient: 95 : 5 to 60 : 40 A : B in 15
min); ES-MS (positive mode): calcd for C113H184N21O60S 2827.17
[M + H]+, found: 2828.56.
Substrate preparation
Gold QCM-D sensors and gold-coated glass slides (SIA Au kit, Bi-
acore, Sweden) were treated by exposure to UV–ozone for 10 min
followed by immersion in ethanol for 15 min. Then, surfaces were
dried with nitrogen and docked in the SPR measurement chamber
or, for QCM-D experiments, immersed in the functionalization
solution.
Synthesis of cysteine functionalized tetravalent neoglycopeptides 6
and 7
QCM-D measurement
Experiments were performed on a quartz crystal microbalance
with an energy dissipation Q-SENSE D300 system equipped
with a radial flow chamber from Q-sense AB (Va¨stra Fro¨lunda,
Sweden). Changes in the resonance frequencies (Df ), related to
attached mass (including coupled water), and its dissipation, DD,
related to frictional (viscous) losses in the adlayer, are measured
simultaneously during the assembling process for the quartz
crystal fundamental resonance frequency (f 1 = 5 MHz) and odd
overtones (n = 3, 5, 7). In the case of homogeneous, quasi-rigid
films with low thickness, the areal adsorbed masses, DC, could
Neoglycopeptide 2 (5.5 mg, 2.8 mmol) and BocCys(Npys)-OSu
(2.6 mg, 5.6 mmol) were dissolved in DMF (1 mL) and the pH of
the solution adjusted to 8 with DIEA. After 30 minutes the solvent
was evaporated and the peptide precipitated in diethyl ether. The
crude yellow powder was then taken up with a solution of TFA–
CH2Cl2 (1 : 1, 10 mL) and purified by semi-preparative RP-HPLC
(linear gradient: 95 : 5 to 60 : 40 A : B in 30 min, detection: k = 214
and 250 nm). Compound 6 (4.5 mg, 72%) was obtained as a yellow
powder: Rt = 9.9 min (linear gradient: 95 : 5 to 60 : 40 A : B in 15
min); ES-MS (positive mode): calcd for C87H137N22O41S2 2209.87
[M + H]+, found: 2210.93. The same procedure was followed for
the synthesis of compound 7: Rt = 9.2 min (linear gradient: 95 :
5 to 60 : 40 A : B in 15 min); ES-MS (positive mode): calcd for
C111H177N22O61S2 2858.08 [M + H]+, found: 2859.50.
be calculated according to the Sauerbrey equation,41 DC = −C Df
n
n
with the mass sensitivity C = 17.7 ng cm−2 Hz−1 for f 1 = 5 MHz.
QCM-D measurements were operated in flow mode
(0.1 mL min−1). The T-loop in the measurement chamber was
bypassed and all solutions were kept in a thermostated bath
(24.5 ◦C) to ensure stable operation at a working temperature
of 24 ◦C. All solutions were previously degassed in order to avoid
bubble formation in the QCM-D measuring chamber.
Synthesis of biotin functionalized monovalent neoglycopeptides 8
and 9
Compounds 8 and 9 were prepared following the procedure
described above for 4 and 5. Compound 8: Rt = 9.2 min (linear
gradient: 95 : 5 to 60 : 40 A : B in 15 min); ES-MS (positive
mode): calcd for C56H89N15O19S 1307.62 [M + H]+, found: 1308.09;
compound 9: Rt = 8.6 min (linear gradient: 95 : 5 to 60 : 40 A :
B in 15 min); ES-MS (positive mode): calcd for C62H100N15O24S
1470.68 [M + H]+, found: 1470.50.
SPR measurement
The SPR measurements were performed with a BIAcore T100
(Biacore AB, Sweden) operated with BIAcore T100 evaluation
software 1.1. All measurements were performed at 25 ◦C, at
5 lL min−1 for ligand immobilization and 25 lL min−1 for kinetic
measurements. Experiments were realized in WB containing 0.05%
P20. The SPR technique detects changes expressed in relative units
(RU) in the interfacial effective refractive index resulting from
the binding of an analyte in solution to a ligand immobilized
on a sensor chip. For all BIAcore experiments, 0.05% P20 in
buffer solution was used to prevent non-specific interactions
between proteins and the surfaces. A non-specific surface was
used as a reference. Curves obtained on the reference surface were
subtracted from the curves recorded on the recognition surfaces,
allowing elimination of refractive index changes due to buffer
effects. The absence of mass transport effects on experiments was
Synthesis of cysteine functionalized monovalent neoglycopeptides
10 and 11
Compound 10 and 11 were prepared following the procedure
described above for 6 and 7. Compound 10: Rt = 10.3 min (linear
gradient: 95 : 5 to 60 : 40 A : B in 15 min); ES-MS (positive mode):
calcd for C54H83N16O20S2 1339.54 [M + H]+, found: 1339.60;
compound 11: Rt = 10.1 min (linear gradient: 95 : 5 to 60 : 40 A :
B in 15 min); ES-MS (positive mode): calcd for C60H93N16O25S2
1501.59 [M + H]+, found: 1501.62.
This journal is
The Royal Society of Chemistry 2008
Org. Biomol. Chem., 2008, 6, 1114–1122 | 1121
©

View Article Online
checked on each surface by separately running one injection of
Con A (9.8 lM) for 1 min at three different flow rates (5, 15
and 75 lL min−1). The curves obtained are able to be overlaid,
confirming the kinetic control of the experiments (not shown).
12 S. Grigalevicius, S. Chierici, O. Renaudet, R. Lo-Man, E. Deriaud, C.
Leclerc and P. Dumy, Bioconjugate Chem., 2005, 16, 1149–1159.
13 J. Razkin, V. Josserand, D. Boturyn, Z.-h. Jin, P. Dumy, M. Favrot,
J.-L. Coll and I. Texier, ChemMedChem, 2006, 1, 1069–1072.
14 M. Mutter, P. Dumy, P. Garrouste, C. Lehmann, M. Mathieu, C.
Peggion, S. Peluso, A. Razaname and G. Tuchscherer, Angew. Chem.,
Int. Ed. Engl., 1996, 35, 1482–1485.
15 O. Renaudet and P. Dumy, Bioorg. Med. Chem. Lett., 2005, 15, 3619–
3622.
16 E. Garanger, D. Boturyn, O. Renaudet, E. Defrancq and P. Dumy,
J. Org. Chem., 2006, 71, 2402–2410.
17 O. Renaudet and P. Dumy, Org. Biomol. Chem., 2006, 4, 2628–2636.
18 Y. Singh, O. Renaudet, E. Defrancq and P. Dumy, Org. Lett., 2005, 7,
1359–1362.
19 U. Bakowsky, W. Rettig, G. Bendas, J. Vogel, H. Bakowsky, C. Harnagea
and U. Rothe, Phys. Chem. Chem. Phys., 2000, 2, 4609–4614.
20 T. K. Dam, R. Roy, D. Page and C. F. Brewer, Biochemistry, 2002, 41,
1351–1358.
Ellipsometry
Ellipsometric measurements were performed using an imaging el-
lipsometer EP3-SE from Nanofilm Technology GmbH, Germany.
Experiments were performed ex situ under air conditions at a
wavelength of 630.2 nm and at variable angles of incidence ranging
from 50◦ to 80◦. Optical modeling was performed using the
EP3View software from Nanofilm Technology GmbH. A three-
layer model, substrate–layer–ambient air, was used to fit the data.
The optical properties of the bare gold substrate (a QCM-D gold
coated quartz crystal) were previously measured.
21 T. K. Dam, R. Roy, D. Page and C. F. Brewer, Biochemistry, 2002, 41,
1359–1363.
22 E. A. Smith, W. D. Thomas, L. L. Kiessling and R. M. Corn, J. Am.
Chem. Soc., 2003, 125, 6140–6148.
23 J. Shi, T. Yang, S. Kataoka, Y. Zhang, A. J. Diaz and P. S. Cremer,
J. Am. Chem. Soc., 2007, 129, 5954–5961.
Acknowledgements
The authors thank Eva Pebay-Peyroula for giving us access to the
BIAcore 3000 device in the IBS (Institute of Structural Biology,
Grenoble, France). They are indebted to Hughes Lortat-Jacob
for his introduction to the BIAcore measurements and to Anne
Imberty for useful discussions. The authors thank the Institut de
Chimie Mole´culaire de Grenoble (ICMG FR-2607) for financial
support. M.W. acknowledges the Ministe`re de l’Education Na-
tionale, de l’Enseignement Supe´rieur et de la Recherche for her
Ph.D. fellowship. We are grateful to the Nanobio program for the
facilities of the synthesis and surface characterization platforms.
24 P. Critchley and G. J. Clarkson, Org. Biomol. Chem., 2003, 1, 4148–
4159.
25 P. Critchley, M. N. Willand, A. K. Rullay and D. H. G. Crout, Org.
Biomol. Chem., 2003, 1, 928–938.
26 D. A. Mann, M. Kanai, D. J. Maly and L. L. Kiessling, J. Am. Chem.
Soc., 1998, 120, 10575–10582.
27 Y. Shinohara, Y. Hasegawa, H. Kaku and N. Shibuya, Glycobiology,
1997, 7, 1201–1208.
28 N. Horan, L. Yan, H. Isobe, G. M. Whitesides and D. Kahne, Proc.
Natl. Acad. Sci. U. S. A., 1999, 96, 11782–11786.
29 E. Duverger, N. Frison, A.-C. Roche and M. Monsigny, Biochimie,
2003, 85, 167–179.
30 Y. Suda, A. Arano, Y. Fukui, S. Koshida, M. Wakao, T. Nishimura, S.
Kusumoto and M. Sobel, Bioconjugate Chem., 2006, 17, 1125–1135.
31 M.-P. Dubois, C. Gondran, O. Renaudet, P. Dumy, H. Driguez, S. Fort
and S. Cosnier, Chem. Commun., 2005, 4318–4320.
References
32 O. Renaudet and P. Dumy, Tetrahedron Lett., 2001, 42, 7575–7578.
33 J. C. Love, L. A. Estroff, J. K. Kriebel, R. G. Nuzzo and G. M.
Whitesides, Chem. Rev., 2005, 105, 1103–1169.
1 H. Lis and N. Sharon, Chem. Rev., 1998, 98, 637–674.
2 R. A. Dwek, Chem. Rev., 1996, 96, 683–720.
3 Y. C. Lee and R. T. Lee, Acc. Chem. Res., 1995, 28, 321–327.
4 S.-K. Choi, Synthetic Multivalent Molecules: Concepts and Biomedical
Applications, John Wiley & Sons, Inc., Hoboken, New Jersey, 2004.
5 S. D. Burke, Q. Zhao, M. C. Schuster and L. L. Kiessling, J. Am. Chem.
Soc., 2000, 122, 4518–4519.
6 L. L. Kiessling, J. E. Gestwicki and L. E. Strong, Angew. Chem., Int.
Ed., 2006, 45, 2348–2368.
34 A. Van der Heyden, M. Wilczewski, P. Labbe´ and R. Auzely, Chem.
Commun., 2006, 3220–3222.
35 Z. Derewenda, J. Yariv, J. R. Helliwell, A. J. Kalb, E. J. Dodson, M. Z.
Papiz, T. Wan and J. Campbell, EMBO J., 1989, 8, 2189–2193.
36 J. H. Naismith, C. Emmerich, J. Habash, S. J. Harrop, J. R. Helliwell,
W. N. Hunter, J. Raftery, A. J. Kalb and J. Yariv, Acta Crystallogr.,
Sect. D: Biol. Crystallogr., 1994, D50, 847–858.
7 J. J. Lundquist and E. J. Toone, Chem. Rev., 2002, 102, 555–578.
8 M. L. Wolfenden and M. J. Cloninger, Bioconjugate Chem., 2006, 17,
958–966.
9 S. L. Mangold and M. J. Cloninger, Org. Biomol. Chem., 2006, 4, 2458–
2465.
10 D. Boturyn, J.-L. Coll, E. Garanger, M.-C. Favrot and P. Dumy, J. Am.
Chem. Soc., 2004, 126, 5730–5739.
11 O. Renaudet and P. Dumy, Org. Lett., 2003, 5, 243–246.
37 V. Mansuy-Schlick, R. Delage-Mourroux, M. Jouvenot and W.
Boireau, Biosens. Bioelectron., 2006, 21, 1830–1837.
38 S. Lo¨faas, M. Malmqvist, I. Roennberg, E. Stenberg, B. Liedberg and
I. Lundstro¨m, Sens. Actuators, B, 1991, B5, 79–84.
39 C. Larsson, M. Rodahl and F. Ho¨o¨k, Anal. Chem., 2003, 75, 5080–5087.
40 N. L. Kalinin, L. D. Ward and D. J. Winzor, Anal. Biochem., 1995, 228,
238–244.
41 G. Sauerbrey, Z. Phys. A: Hadrons Nucl., 1959, 155, 206–222.
1122 | Org. Biomol. Chem., 2008, 6, 1114–1122
This journal is
The Royal Society of Chemistry 2008
©