Surface plasmon resonance imaging measurements of the inhibition of Shiga-like toxin by synthetic multivalent inhibitors

Article abstract of DOI:10.1021/ac050423p

A variety of new methodologies to pattern biomolecules on surfaces and to detect binding events are currently being developed for high-throughput assay applications. Carbohydrates serve as attachment sites for toxins, bacteria, and viruses. Immobilized ca

Full text of DOI:10.1021/ac050423p

Anal. Chem. 2005, 77, 7497-7504
Articles
Surface Plasmon Resonance Imaging
Measurements of the Inhibition of Shiga-like Toxin
by Synthetic Multivalent Inhibitors
Vishal Kanda, Pavel Kitov, David R. Bundle, and Mark T. McDermott*
Department of Chemistry, University of Alberta, Edmonton, AB, Canada T6G 2G2
is being driven by the role carbohydrate ligands play in mediating
various biological phenomena. Research has unraveled a large,
diverse family of carbohydrate-mediated interactions between cells
and viruses, bacteria, extracellular matrix scaffolds, and other
cells.3 Cell surfaces can interact with other cells through a variety
of interactions involving membrane-bound and extracellular oli-
gosaccharides, glycolipids, and glycoproteins.
Among the challenges in the study of carbohydrates is their
relative structural complexity. The presence of multiple branching
points, stereochemistry, and functional group modifications in
natural systems introduces a higher degree of synthetic difficulty
for oligosaccharides in comparison to oligonucleotides and linear
polypeptides.4,5 Much of the functional activity of carbohydrates
is governed by highly specific recognition with proteins. Increasing
interest in examining the functional characteristics of carbohydrate
ligands as well as the need for throughput and efficiency in the
analysis of carbohydrate interactions is leading to a surge in
microarray research.2,6-10 The sheer variety of structures and
interactions make a high-throughput analysis system such as
microarrays a necessity for expansion of research into glycomics.
Lectins are a class of proteins that function as receptors for
specific carbohydrate moieties.4,11,12 The interactions of individual
A variety of new methodologies to pattern biomolecules
on surfaces and to detect binding events are currently
being developed for high-throughput assay applications.
Carbohydrates serve as attachment sites for toxins, bac-
teria, and viruses. Immobilized carbohydrate units can
thus be used to directly detect these agents or as a
platform for inhibitor assessment. In this work, modified
glycosides were patterned on gold surfaces to monitor the
binding of the homopentameric B5 cell-recognition sub-
unit of the Shiga-like toxin (SLT). Binding was detected
with the label-free method of surface plasmon resonance
(SPR) imaging. Two synthetic multivalent inhibitors were
used in order to effect inhibitory binding, and SPR
imaging is presented as a simple alternative to ELISA for
the study of toxin inhibition. In contrast to existing
methods for the study of carbohydrate-protein interac-
tions, in particular ELISA, the use of micropatterned
sensor surfaces is shown to be advantageous due to a
decrease in complications and manual labor from numer-
ous blocking, washing, and labeling steps. Carbohydrate
receptor density on the sensor surface was optimized in
order to effect the maximum binding of the SLT. The IC50
values determined were in the low-nanomolar range for
each of the two inhibitors studied.
(2) Wang, D.; Liu, S.; Trummer, B. J.; Deng, C.; Wang, A. Nat. Biotechnol. 2002,
20, 275-281.
(3) Mammen, M.; Choi, S.-K.; Whitesides, G. M. Angew. Chem., Int. Ed. 1998,
37, 2754-2794.
(4) Bertozzi, C.; Kiessling, L. L. Science 2001, 291, 2357-2364.
(5) Houseman, B. T.; Mrksich, M. In Host-Guest Chemistry; Penades, S., Ed.;
Springer-Verlag: Berlin, 2002; Vol. 218, pp 1-44.
(6) Smith, E. A.; Thomas, W. D.; Kiessling, L. L.; Corn, R. M. J. Am. Chem.
Soc. 2003, 125, 6140-6148.
(7) Fang, Y.; Frutos, A. G.; Lahiri, J. Langmuir 2003, 19, 1500-1505.
(8) Houseman, B. T.; Mrksich, M. Chem. Biol. 2002, 9, 443-454.
(9) Park, S.; Lee, M.-R.; Pyo, S.-J.; Shin, I. J. Am. Chem. Soc. 2004, 126, 4812-
4819.
In recent years, much research effort has been directed at the
development of microarray methods for the parallel analysis of
interactions in proteomics and genomics. In addition to these
widely known areas of research, there is a growing interest in
the analysis of biological interactions mediated by carbohydrates,
or “glycomics”.1,2 The expansion of research efforts in this area
* To whom correspondence should be addressed. E-mail: mark.mcdermott@
ualberta.ca. Voice: 780-492-3687. Fax: 780-492-8231.
(1) Feizi, T. Glycoconjugate J. 2000, 17, 553-565.
(10) Disney, M. D.; Seeberger, P. H. Chem. Eur. J 2004, 10, 3308-3314.
(11) Sharon, N.; Lis, H. Science 1989, 246, 227-246.
10.1021/ac050423p CCC: $30.25 © 2005 American Chemical Society
Published on Web 11/02/2005
Analytical Chemistry, Vol. 77, No. 23, December 1, 2005 7497

carbohydrate moieties with proteins are weak, with typical
dissociation constants in the high-micromolar to millimolar
range.13,14 As a consequence, many lectins interact with their
targets via multiple, homologous binding sites in order to in-
crease affinity. A well-studied example is that of the influenza virus,
which binds to cell surface sialic acids via a homotrimeric lectin
group. The initial binding allows other lectin groups on the viral
surface to come into contact with more cell surface sugars,
eventually deforming the cell membrane and leading to endocy-
tosis.3
The Escherichia coli bacterium produces toxins implicated in
a number of gastrointestinal maladies including diarrhea, hemor-
rhagic colitis, and the hemolytic uremic syndrome.15 The vero-
toxins belong to a class of toxins also known as Shiga-like toxins
(SLTs) due to their structural similarity to the Shiga toxin
produced by Shigella dysenteriae type I. Shiga-like toxins consist
of a 38.4-kDa pentameric B5 subunit, which is responsible for the
recognition of cell surface oligosaccharide units, and a 32-kDa
enzyme portion, the A subunit, which damages ribosomal RNA
once internalized by the cell.16 The cell surface recognition is a
necessary step in allowing the toxin to be internalized and fulfill
its enzymatic function. In mammalian cells, the Gb3 glycolipid is
the recognition element responsible for the binding of the Shiga
toxins.17 The carbohydrate portion of the Gb3 glycolipid is the Pk
accuracy of the analytical data from SPR measurements, as well
as the ease of use of commercial instrumentation, have allowed
researchers to ascertain binding constants from analysis of kinetic
data.
Relevant to the present work, SPR imaging has been used to
study protein-carbohydrate interactions in array format.6 Interac-
tions between immobilized monosaccharides and proteins in
noncompetitive and competitive format were analyzed. In this
work, we fabricate arrays of di- and trisaccharides to study the
binding of SLT. These arrays are employed to explore the
inhibitory effect of two synthetic compounds capable of multivalent
binding to SLT. We compare the methodology developed here to
that previously used to examine SLT inhibition.
EXPERIMENTAL SECTION
HS(CH2)11(OCH2CH2)3OCH3 (EG3-OMe) was synthesized in
accordance with known procedures.26,27 The structures of the two
probe glycosides used are shown in Figure 1. The syntheses of
bis{16-[4-O-[4-O-(R-
glucopyranosyloxy]hexadecanyl}disulfide (C16Pk) and bis{16-[4-
O-(2-acetamido-2-deoxy-â- -galactopyranosyl)-â- -glucopyranosyloxy]-
D-galactopyranosyl)-â-D-galactopyranosyl]-â-D-
D
D
hexadecanyl}disulfide (C16AGM2) were published previously.28
The inhibitors studied are shown in Figure 2. The synthesis of
the daisy inhibitor was reported previously,29 and the synthesis
of the starfish-2 inhibitor (SF-2) is presented in the supplementary
information to this paper.
trisaccharide (R-D-Galp-(1-4)-â-D-Galp-(1-4)-â-D-Glcp). It has been
shown that the binding sites of the B5 subunit can be blocked
with synthetic ligands containing multiple copies of the Pk
trisaccharide unit.16 The 5-fold quasi-symmetric ligands can bind
to all five of the strong binding sites on the B5 subunit simulta-
neously, effectively inhibiting the toxin.16,18 This system is used
here to establish the utility of SPR imaging for studies of toxin
binding inhibition.
SPR has previously been applied to the elucidation of binding
constants for various carbohydrate-protein interactions,19 espe-
cially those associated with pathogens.3,20 Some noteworthy
examples that have been studied by SPR include the binding of
HIV-1 protein gp120 to potential inhibitors composed of glyco-
dendrimers and sulfated dextran,21 the binding of the toxin ricin
to various synthetic glycolipids,22,23 and the interactions of cholera
toxin with ganglioside receptors.24,25 The sensitivity and perceived
The recombinant SLT was expressed without the enzymatic
A subunit, leaving only the self-assembling B5 subunit containing
the carbohydrate recognition sites and purified according to
published procedures.30 A 1.6 mg/mL concentration of SLT in PBS
with 0.02% NaN3 as preservative was used as a stock solution.
Protein solutions were diluted in phosphate-buffered saline
(PBS: 8.1 mM Na2HPO4, 1.5 mM KH2PO4, 137 mM NaCl, 2.7
mM KCl in 18 MΩ water).
Poly(dimethylsiloxane) (PDMS) microfluidic channels were
fabricated according to established methods.31 Briefly, a relief
pattern of photoresist on a silicon wafer was created photolitho-
graphically. By curing PDMS prepolymer and cross-linker (Syl-
gard 184, Dow Corning; Midland, MI) 10:1 by weight against this
relief structure, a negative of the relief was formed in the PDMS.
The microchannels measured 200 µm wide by 10-15 µm deep.
The device was through-bored at the ends of the channels to allow
fluids access to the channels when the PDMS device was applied
to a surface. Fluid flow was driven by applying vacuum to one
access point on the microchannel while connecting the other
access point to a reservoir of solution.
(12) Lee, Y. C.; Lee, R. T. Acc. Chem. Res. 1995, 28, 321-327.
(13) Lee, R. T.; Lee, Y. C. Glycoconjugate J. 2000, 17, 543-551.
(14) Horan, N.; Yan, L.; Isobe, H.; Whitesides, G. M.; Kahne, D. Proc. Natl. Acad.
Sci. U.S.A. 1999, 96, 11782-11786.
(15) Soltyk, A. M.; MacKenzie, C. R.; Wolski, V. M.; Hirama, T.; Kitov,
P. I.; Bundle, D. R.; Brunton, J. L. J. Biol. Chem. 2002, 277, 5351-
5359.
(16) Kitov, P. I.; Sadowska, J. M.; Mulvey, G.; Armstrong, G. D.; Ling, H.; Pannu,
N. S.; Read, R. J.; Bundle, D. R. Nature 2000, 403, 669-672.
(17) Kitova, E. N.; Kitov, P. I.; Bundle, D. R.; Klassen, J. S. Glycobiology 2001,
11, 605-611.
Arrays were fabricated by placing PDMS channel structures
in conformal contact with gold surfaces. Solutions consisting of
0.8-1 mM of the disulfides in 1:9 H2O/MeOH were injected into
(18) Mulvey, G. L.; Marcato, P.; Kitov, P. I.; Sadowska, J.; Bundle, D. R.;
Armstrong, G. D. J. Infect. Dis. 2003, 187, 640-649.
(19) Duverger, E.; Frison, N.; Roche, A.-C.; Monsigny, M. Biochimie 2003, 85,
167-179.
(20) Schengrund, C.-L. Biochem. Pharmacol. 2003, 65, 699-707.
(21) Kensinger, R. D.; Yowler, B. C.; Benesi, A. J.; Schengrund, C.-L. Bioconjugate
Chem. 2004, 15, 349-358.
(26) Pale-Grosdemange, C.; Simon, E. S.; Prime, K. L.; Whitesides, G. M. J. Am.
Chem. Soc. 1991, 113, 12-20.
(27) Aherne, D.; Rao, S. N.; Fitzmaurice, D. J. Phys. Chem. B 1999, 103, 1821-
1825.
(28) Kitov, P. I.; Railton, C.; Bundle, D. R. Carbohydr. Res. 1998, 307, 361-
369.
(22) Gustafson, I. Colloids Surf. B 2003, 30, 13-24.
(23) Critchley, P.; Clarkson, G. J. Org. Biomol. Chem. 2003, 1, 4148-4159.
(24) Kuziemko, G. M.; Stroh, M.; Stevens, R. C. Biochemistry 1996, 35, 6375-
6384.
(29) Kitov, P. I.; Paszkiewicz, E.; Wakarchuk, W. W.; Bundle, D. R. Methods
Enzymol. 2003, 362, 86-105.
(30) Mulvey, G.; Vanmaele, R.; Mrazek, M.; Cahill, M.; Armstrong, G. D. J.
Microbiol. Methods 1998, 32, 247-252.
(25) MacKenzie, C. R.; Hirama, T.; Lee, K. K.; Altman, E.; Young, N. M. J. Biol.
Chem. 1997, 272, 5533-5538.
(31) Xia, Y.; Whitesides, G. M. Angew. Chem., Int. Ed. Engl. 1998, 37, 550-
575.
7498 Analytical Chemistry, Vol. 77, No. 23, December 1, 2005

Figure 1. Chemical structures of C16 disulfide-modified Pk trisaccharide and Asialo-GM2 disaccharide used to produce arrays for studies of
SLT inhibition.
the channels, and the probe glycosides were allowed to adsorb
for 20 min. The solvent ratio employed allowed for facile solubi-
lization of the reagents and retained sealing of native PDMS
microchannels against the gold sensor surface for the required
time. Following patterning inside the channels, the entire chip
was immersed into a 1 mM ethanolic solution of EG3-OMe
overnight.
expected to have no interaction with the SLT, were synthesized
as hexadecanyl disulfides for easy immobilization to gold. The
structures of the two disulfides are shown in Figure 1. IRRAS
experiments show that immersion of gold substrates into solutions
containing these disulfides results in densely packed monolayers
presenting the glycoside to the interface (data not shown). Studies
have shown that monolayers formed from disulfides are indistin-
guishable from those formed from the corresponding thiol and
are likely adsorbed via thiolate-gold interactions following cleav-
age of the disulfide bond.32 Arrays to study the binding of SLT
were fabricated by patterning C16Pk and C16AGM2 onto SPR
imaging chips using microfluidic networks in PDMS. Self-
assembly of C16Pk and C16AGM2 to the gold substrate inside the
channels resulted in ∼200-µm-wide lines of glycoside-terminated
monolayers. The chip was immersed in an ethanolic solution of
HS(CH2)11(OCH2CH2)3OCH3 (C11EG3-OMe), which forms a protein-
resistant background on the remaining areas of the gold. No
significant displacement of the glycoside monolayers by the EG3-
OMe is observed by IRRAS.
SPR imaging can be used to assess the effectiveness of our
procedure for fabricating arrays of glycosides. Figure 3A is the
SPR image of an initial array surface featuring two lines each of
C16Pk and C16AGM2. The cross-sectional profile in Figure 3B
reveals a pattern of intensity characteristic of glycoside monolayers
formed inside the channels of the PDMS microfluidic network.
The patterned glycoside regions show higher reflected intensity,
implying a greater refractive index relative to the EG3-OMe
background. The reflectivity at the Pk addresses is observably
All SPR imaging utilized sensing films consisting of 45-nm gold
films deposited on SF10 glass (Schott; Toronto, ON, Canada) with
a 1-nm adhesive layer of chromium. Imaging was performed using
the GWC Instruments SPRimager (GWC Instruments; Madison,
WI). Images displayed are averages of 100 individual snapshot
images. All images were collected in toxin-free PBS. Difference
images result from the subtraction of images before and after toxin
binding. All accompanying image profiles are directional averages
over the entire displayed image. Infrared reflection absorption
spectroscopy (IRRAS) was performed on a Mattson Infinity
spectrometer with an externally housed low-noise MCT-A detector
at a resolution of 2 cm-1. Glycoside monolayer samples for IRRAS
analysis were prepared on glass microscope slides coated with
10 nm of Cr and 300 nm of gold. Solutions consisting of total
disulfide concentration of 0.9 mM were pressed between a clean
glass slide and a gold slide for 20 min. This procedure was used
in order to confine the small volume (50-200 µL) of glycoside
solution employed.
RESULTS AND DISCUSSION
Array Fabrication/Optimization. The arrays for detection
of the toxin by SPR imaging were fabricated on thin gold films
by patterning probe glycosides. The Pk trisaccharide, which is
known to bind SLT, and an asialo-GM2 disaccharide, which is
(32) Jung, C.; Dannenberger, O.; Xu, Y.; Buck, M.; Grunze, M. Langmuir 1988,
14, 1103-1107.
Analytical Chemistry, Vol. 77, No. 23, December 1, 2005 7499

Figure 2. Chemical structure of starfish 2 inhibitor (FW 8703.8847) and the daisy 1 inhibitor (FW 9660.79). The Pk trisaccharide unit is presented
twice at each end of five radial spacer arms.
higher due to the larger mass of the C16Pk inducing a greater
change in refractive index.
glycoside chip (Figure 3B) are too large to be ascribed only to
mass differences. Although speculative at this time, the interaction
of the buffer solution with the various surface moieties may
contribute to the contrast of the glycoside chip. Despite this
uncertainty, the signals shown in Figure 3C indicate selective
binding and demonstrate that patterned monolayers of C16Pk show
specific affinity for SLT that is detectable by SPR imaging.
Given the fixed locations of the binding sites on the SLT B5
subunit, the effect of the surface density of C16Pk was studied.
This was accomplished by measuring the amount of SLT binding
to mixed monolayers of C16Pk and a diluent species. Considering
the insignificant amount of SLT bound to C16AGM2 shown in
Figure 3D, we conclude that monolayers composed of interfacial
oligosaccharide groups are quite resistant to nonspecific protein
adsorption. While monolayers presenting mannitol groups at the
surface have been shown to be at least as protein-resistant as oligo-
(ethylene glycols),34 the observation has not been generalized to
The differential binding of SLT by the C16Pk/C16AGM2 array
is demonstrated by SPR imaging as shown in Figure 3C and D.
Figure 3C is the difference image following the exposure of the
array to 130 nM SLT for 10 min and rinsing with PBS. The cross-
sectional profile in Figure 3D reveals an increase in signal on the
Pk trisaccharide addresses, with no detectable adsorption on either
the C16AGM2 or the EG3-OMe background. The magnitude of
the signal change observed in Figure 3D for SLT binding (15 RU)
is similar to that observed for the C16Pk addresses (11 RU) on
the original chip, Figure 3B. Based on previous observed signal
changes for IgG antibodies (MW ) ∼150 000) of 50-60 RU,33
a
signal of 15 RU for the binding of the ∼40 000 SLT is reasonable.
Thus, the differences in reflectivity observed on the orginal
(33) Kanda, V.; Kariuki, J.; Harrison, D. J.; McDermott, M. T. Anal. Chem. 2004,
76, 7257-7262.
7500 Analytical Chemistry, Vol. 77, No. 23, December 1, 2005

Figure 4. SPR difference image and cross-sectional profile following
exposure of a glycoside array to 130 nM SLT for 10 min. The array
elements were prepared from solutions containing both C16Pk and
C
16AGM2 (total concentration, 0.9 mM). Each element is labeled with
the mole fraction of C16Pk used in the assembly solution.
solution. Figure 4 contains the difference image following expo-
sure of the array to SLT. The cross-sectional profile shows
negligible SLT binding on the C16AGM2 line (ø ) 0) and a small
amount of nonspecific adsorption to the C11EG3-OMe background
on this chip. The profile also reveals maximal SLT binding at this
array does not occur at single-component monolayers of C16Pk
but at addresses prepared from solutions of ø ) 0.1. This implies
that the surface density of glycoside receptors is important to
optimal toxin binding. This observation is explored in more detail
with IRRAS.
The absorbance of the amide II band is thought to correlate
linearly with the amount of adsorbed protein,38,39 making it a good
diagnostic band for SLT adsorption. By tracking the changes in
the amide II band intensity upon SLT binding to mixed C16Pk/
C16AGM2 monolayers, it was possible to determine the optimal
C16Pk solution ø. Figure 5A shows infrared spectra of a monolayer
prepared from a 0.1 mol fraction solution of C16Pk before and after
adsorption of SLT. Measurements of amide II band intensity were
made after subtracting the spectra before and after SLT adsorp-
tion, thus normalizing for variations in the spectra of the initial
glycoside monolayers.
The plot in Figure 5B shows the change in amide II intensity
for SLT binding with C16Pk mole fraction in solution. This plot
shows very little SLT binding at low ø of C16Pk, a maximum in
binding at ø ∼ 0.15, and a slight decrease as the solution
concentration is increased. This observation points to an optimal
surface density of C16Pk for SLT binding that is less than 100%. A
previous SPR imaging study showed that the binding of lectins
reaches a limiting value when the surface density of immobilized
monosaccharide epitopes is below 100%.6 Our results are not
surprising given that SLT contains a series of spatially well-defined
binding sites and achieves maximum binding strength through
5-fold multivalent interactions with Gb3 glycolipids. Figure 5B
suggests that monolayers formed from solutions containing 10-
15% C16Pk provide the optimum interfacial distribution of the Pk
Figure 3. (A) SPR image of the initial C16AGM2 and C16Pk array.
(B) Cross section showing greater initial signal on the Pk regions than
on the AGM2. (C) Difference image after exposure of array to 130
nM SLT for 10 min. (D) Cross section of difference image shows
exclusive binding to Pk; cross indicates the zero signal level for the
difference image.
a broader class of saccharide-terminated films. The property of
protein resistance would come as little surprise given the role of
cell surface saccharides in specific protein recognition. Given this
resistance to SLT nonspecific adsorption, the AGM2 species was
used as a diluent for varying the surface density of the Pk
trisaccharide.
The effect of diluting C16Pk with C16AGM2 on SLT binding is
shown in Figure 4. Four lines were arrayed from solutions
containing varying mole fractions (ø) of C16Pk relative to C16AGM2.
A number of studies have shown that the surface density of one
component of a two-component monolayer can be controlled by
varying the ratio in solution.35-37 Note that the IR spectral
differences between C16Pk and C16AGM2 are minimal and are
insufficient to directly determine the surface density of each
glycoside. We thus describe the array addresses of mixed
glycoside monolayers by the mole fraction of C16Pk in the assembly
(34) Luk, Y.-Y.; Kato, M.; Mrksich, M. Langmuir 2000, 16, 9604-9608.
(35) Chidsey, C. E. D.; Bertozzi, C. R.; Putvinski, T. M.; Mujsce, A. M. J. Am.
Chem. Soc. 1990, 112, 4301-4306.
(38) Jakobsen, R. J.; Brown, L. L.; Winters, S.; Gendreau, R. M. J. Biomed. Mater.
Res. 1983, 16, 199-201.
(36) Folkers, J. P.; Laibinis, P. E.; Whitesides, G. M.; Deutch, J. J. Phys. Chem.
1994, 98, 563-571.
(37) Kwon, Y.; Mrksich, M. J. Am. Chem. Soc. 2002, 124, 806.
(39) Liedberg, B.; Ivarsson, B.; Lundstrom, I. J. Biochem. Biophys. Methods 1984,
9, 233-243.
Analytical Chemistry, Vol. 77, No. 23, December 1, 2005 7501

Figure 6. Inhibition curves for the SF-2 and daisy inhibitors. See
text for details. The inset is a typical SPR image and profile showing
the signal observed at the array for 42 nM SLT solutions.
the maximal SLT binding to the array surface and is analogous
to previous ELISA results.16 Given that SLT operates by binding
to Gb3 receptors on cell walls, we believe this definition is
appropriate. The curves generated from the assays were fit with
eq 1, which is analogous to a four-parameter logistic model,40
∆%Rmax - ∆%Rmin
1 + ([A]/IC50)b
∆%R ) ∆%Rmin
+
(1)
Figure 5. (A) IRRAS spectra of a glycoside monolayer formed from
a solution containing ø ) 0.1 C16Pk before and after incubation in
130 nM SLT. (B) Amide II band (νAII) intensity vs mole fraction C16Pk
for SLT binding to a range of mixed glycoside monolayers. The error
bars correspond to the standard deviation of two to three measure-
ments.
where ∆%Rmax and ∆%Rmin are asymptotic signal values determined
from the regression analysis, [A] is the adsorbate solution
concentration, and IC50 is the concentration at which the inhibition
is half-maximal. ∆%Rmax, ∆%Rmin, IC50, and b are the fitting
parameters. When the data are plotted on a log scale in [A], the
parameter b reflects the slope of the curve at the inflection point.
When b ) 1, the equation reduces to the Langmuir isotherm.
Although proposed interpretations for deviations from b ) 1 are
that b < 1 implies inhomogeneity in binding affinities or negative
cooperativity while b > 1 indicates a positive cooperativity effect,40
any real physical basis for the observed values is somewhat
speculative.41 The iterative least-squares solution of this equation,
also known as the Hill or Sips plot, is widely used to analyze
binding curves in the life sciences.42
trisaccharide to enable multivalent interactions with SLT. We
attribute the slight decrease in observed SLT binding at higher ø
to crowding of surface-bound Pk that may lead to inhibitory
intermolecular interactions or steric effects.
Inhibition Studies. Having determined an optimal surface
density of C16Pk for SLT binding, assays were performed to study
the inhibitory effect of synthetic pentameric ligands. Figure 2
shows the SF-2 and Daisy ligands used to bind the toxin. Each
structure was designed in such a way that the terminal carbohy-
drate fragments would be able to occupy the strong binding sites
of the toxin (there are 10 additional weak binding sites). The
inhibition of SLT activity by multivalent starfish and daisy ligands
has been previously demonstrated by ELISA16,18 and mass
spectrometry17 as well as by in vitro16 and in vivo18 studies. The
principal ELISA methods used required separate steps for labeling,
blocking, binding interaction, and color development for detection.
The inhibition of SLT was studied in order to demonstrate the
efficacy and simplicity of microfluidic patterning and SPR imaging
as an alternative to ELISA.
Inhibition curves for SF-2 and daisy are shown in Figure 6 and
the parameters obtained from the fits are listed in Table 1. Note
that, in Figure 6, the y-axis was converted from ∆%R to % inhibition
where,
∆%R
∆%Rmax - ∆%Rmin
% inhibition ) 100% -
(2)
(
)
∆%Rmax and ∆%Rmin were obtained from the fits of ∆%R versus
Assays were performed by exposing C16Pk arrays to 1.6 µg/
mL (42 nM) SLT in PBS with varying inhibitor concentrations
and then using SPR imaging to detect the bound SLT. The key
parameter in inhibition studies is IC50, which is defined as the
inhibitor concentration at which 50% of the toxin is inhibited. In
our case, IC50 is the concentration of inhibitor that yields 50% of
(40) Rodbard, D.; Hutt, D. M. In Radioimmunoassay and Related Procedures in
Medicine; IAEA: Vienna, 1974; Vol. 1, pp 165-192.
(41) Henderson, P. J. F. In Enzyme Assays: A Practical Approach; Eisenthal, R.,
Danson, M. J., Eds.; Oxford University Press: New York, 1993; pp 277-
316.
(42) Nix, B.; Wild, D. In The Immunoassay Handbook; Wild, D., Ed.; Nature
Publishing: New York, 2001; pp 198-210.
7502 Analytical Chemistry, Vol. 77, No. 23, December 1, 2005

In previous ELISA studies of SLT inhibition, two separate
protocols, termed protocols A and B, were used and are illustrated
in Figure 7.16 In protocol A, SLT was coated on a 96-well ELISA
plate, blocked with milk protein (MP), and then exposed to
solutions with 10 µg/mL of a biotinylated BSA-Pk trisaccharide
conjugate and varying concentrations of inhibitor. Analysis was
performed by adding a streptavidin-horseradish peroxidase
(HRP) conjugate and developing with tetramethylbenzidine (TMB)/
hydrogen peroxide and phosphoric acid. The absorbance was then
detected at 450 nm. In protocol B, C16Pk was applied to a 96-well
ELISA plate before blocking with BSA. SLT with varying inhibitor
concentrations was then added. Surface-bound SLT was detected
by first adding a rabbit anti-SLT antibody to bind the SLT and
then adding HRP-conjugated anti-rabbit IgG antibody. The detec-
tion was again performed by developing with TMB/peroxide and
reading the absorbance. A schematic comparison of the interac-
tions employed in the ELISA protocols and the SPR imaging
protocol used in this work is shown in Figure 7. The SPR imaging
method is shown to be advantageous in that far fewer reaction
steps are necessary in order to perform the assay. As a result,
there are fewer steps at which a systematic or random error can
be introduced to the assay. In addition to the time savings, by
which a 2-day procedure essentially becomes a 1-day procedure,
the elimination of many tedious pipetting and washing steps, which
Table 1. Curve Fitting Parameters for SF-2 and Daisy
Inhibition Curves in Figure 6a
inhibitor
IC50 (µM)
∆%Rmin ∆%Rmax
b
R2
SF-2
daisy
1.3(0.2) × 10-3
∼0
∼0
5.6(0.1) 0.62(0.05) 0.991
5.1(0.1) 0.69(0.04) 0.996
2.6(0.3) × 10-3
a Values in parentheses are standard deviations of the indicated
values.
log[inhibitor] plots and are given in Table 1. The inhibition values
plotted reflect the fraction of the maximum signal that was not
observed as a result of inhibitor binding to SLT.
The IC50 values obtained (Table 1) are in the low-nanomolar
range for each inhibitor, in agreement with ELISA work performed
previously. The value of 2.6 nM for Daisy is slightly less than the
reported value of 8.05 nM using ELISA (protocol B, see below).18
The value of 1.3 nM for SF-2 is slightly larger than the values of
0.24 (protocol A) and 0.4 nM (protocol B) for a similar inhibitor.16
The similarity in the IC50 values reported here and those
determined by various methods in the literature16,18 imply that SPR
imaging is a viable approach for inhibition studies. The clear
advantages of SPR for this type of assay are operational simplicity
and, more importantly, a minimization of the number of interac-
tions necessary to produce a signal as discussed below.
Figure 7. Comparison of inhibition assay procedures. Protocol A and protocol B are ELISA-based assays used in previous work.16 The SPR
protocol is used in this work. The top row illustrates the assay format while the bottom row illustrates the detection method. See text for assay
details. BSA, bovine serum albumin; MP, milk protein; SA, streptavidin; HRP, horseradish peroxidase.
Analytical Chemistry, Vol. 77, No. 23, December 1, 2005 7503

require high precision and training, is a major advantage of the
SPR imaging methodology presented. The color development step
for detection is a kinetic process that also requires a high degree
of precision.
array elements was explored here, this methodology can be scaled
to fabricate a higher density of addresses (10-100) for screening
glycoside libraries.
ACKNOWLEDGMENT
This work was supported by the Natural Sciences and
CONCLUSIONS
Engineering Research Council of Canada (NSERC). We thank the
U of A NanoFab for the use of their facility in the fabrication of
masters and PDMS microfluidic networks and the U of A for
support of the Nanofab.
More efficient assays will emerge with the development of
facile and reliable methods to pattern biomolecules on to surfaces
and with new, sensitive, label-free detection schemes. We have
demonstrated that SPR imaging is a viable technique for label-
free screening of toxin inhibition. Patterning of disulfide glycosides
on gold with PDMS microfluidic networks provides an easy and
effective means of fabricating carbohydrate arrays for SPR imag-
ing. In this work, SLT binding was maximized by lowering the
surface receptor density to a level reflecting 15% mol fraction in
solution. The inhibition assays demonstrated here provided similar
IC50 values to ELISA protocols but were shown to be far simpler
and reliant upon fewer interactions. Although a low density of
SUPPORTING INFORMATION AVAILABLE
Additional information as noted in text. This material is
available free of charge via the Internet at http://pubs.acs.org.
Received for review March 10, 2005. Accepted August 18,
2005.
AC050423P
7504 Analytical Chemistry, Vol. 77, No. 23, December 1, 2005