Immobilization of carbohydrate epitopes for surface plasmon resonance using the Staudinger ligation

Article abstract of DOI:10.1016/j.carres.2010.09.020

The detection of carbohydrate-protein interactions is often performed using techniques that require surface immobilization of the lectin or the glycan. A commonly used assay for lectin binding is surface plasmon resonance (SPR). We describe an implementation of the Staudinger ligation as a method to immobilize carbohydrate epitopes to a biosensor surface. This was accomplished by first introducing an azide functionality to a carboxymethyldextran surface, followed by reaction with a phosphane-modified carbohydrate ligand. The chemistry employed is extremely mild and was easily adapted to a commercial biosensor system. Using this approach, we investigated the binding of jacalin and wheat germ agglutinin (WGA) to galactose, lactose, and N-acetyl-lactosamine. We observed that WGA binding shows evidence of multivalent interaction with the surface. Additionally, we found that jacalin binding was influenced by the presence of a flexible and hydrophobic galactosyl aglycone.

Full text of DOI:10.1016/j.carres.2010.09.020

Carbohydrate Research 345 (2010) 2641–2647
Contents lists available at ScienceDirect
Carbohydrate Research
journal homepage: www.elsevier.com/locate/carres
Immobilization of carbohydrate epitopes for surface plasmon resonance
using the Staudinger ligation
⇑
Ravi S. Loka, Christopher W. Cairo
Alberta Ingenuity Centre for Carbohydrate Science, Department of Chemistry, University of Alberta, Edmonton, Alberta, Canada T6G 2G2
Department of Chemistry, University of Alberta, Edmonton, Alberta, Canada T6G 2G2
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 2 July 2010
Received in revised form 9 September 2010
Accepted 15 September 2010
Available online 21 September 2010
The detection of carbohydrate–protein interactions is often performed using techniques that require sur-
face immobilization of the lectin or the glycan. A commonly used assay for lectin binding is surface plas-
mon resonance (SPR). We describe an implementation of the Staudinger ligation as a method to
immobilize carbohydrate epitopes to a biosensor surface. This was accomplished by first introducing
an azide functionality to a carboxymethyldextran surface, followed by reaction with a phosphane-
modified carbohydrate ligand. The chemistry employed is extremely mild and was easily adapted to a
commercial biosensor system. Using this approach, we investigated the binding of jacalin and wheat
germ agglutinin (WGA) to galactose, lactose, and N-acetyl-lactosamine. We observed that WGA binding
shows evidence of multivalent interaction with the surface. Additionally, we found that jacalin binding
was influenced by the presence of a flexible and hydrophobic galactosyl aglycone.
Keywords:
Lectin
Carbohydrate affinity
Surface plasmon resonance
Multivalent binding
Jacalin
Ó 2010 Elsevier Ltd. All rights reserved.
Wheat germ agglutinin
1. Introduction
SPR has been an important technique for the analysis of lectin–
carbohydrate interactions. Typical immobilization strategies ex-
Lectin–carbohydrate interactions are critical to a variety of bio-
logical recognition events. Among many known processes in hu-
mans, native lectins take part in inflammation, cancer, antigen
recognition, and fertilization.1–3 The measurement of lectin–
carbohydrate interactions can be challenging, and a variety of
methods have been employed including fluorescence spectros-
copy,4 isothermal titration calorimetry (ITC),5 surface plasmon res-
onance (SPR),6 and microarrays.7–9 Solid-phase methods, including
SPR and microarrays, are commonly employed due to their sensi-
tivity and small sample requirements. For lectin–carbohydrate
interactions, surface methods have the added advantage that
surface density can be controlled to mimic multivalent presenta-
tions found in nature.10 However, the application of solid-phase
methods requires suitable surface chemistry that preserves the
lectin–carbohydrate interaction under study. The development of
improved, orthogonal, or more facile immobilization strategies is,
therefore, of continued interest for assay development.
ploit amine–carboxylate crosslinking of the lectin to the surface.11
Alternative approaches have used biotin,12 amine,13 thiol,14 or lipid
tags15 to generate carbohydrate surfaces. There has been a rapid
expansion of bioorthogonal chemistry in recent years,16 with a
large number of groups developing selective strategies for protein
or ligand immobilization.17 An important subset of bioorthogonal
reactions, those with excellent yields and substrate tolerance, is
known as a ‘click’ reaction.18 These reactions can provide useful
alternatives to traditional immobilization strategies. The Stauding-
er ligation has been used to detect glycosyltransferase activity on
microarrays19 and to immobilize azide-labeled carbohydrates or
proteins to a phosphane-derived surface.20,21 A modification of this
strategy, which used a traceless Staudinger ligation, has also been
reported for generating self-assembled monolayer (SAM) and
microarray surfaces.22,23 Native chemical ligation of proteins to
an SPR surface has also been described.24 Microarrays used for lec-
tin-binding studies have been generated using alkyne-modified
carbohydrates and a sulfonyl-azide surface.25 Carbohydrate sur-
faces suitable for SPR lectin-binding studies have been generated
using azide-modified glycans and a SAM containing an alkyne.26,27
Although the Staudinger ligation has been applied for surface
immobilization on SAM and microarray surfaces, there are no re-
ports of adapting this chemistry to commercial SPR systems. Com-
mercial SPR biosensors often use carboxymethyldextran (CMD)
surfaces, and require relatively mild reaction conditions which
Abbreviations: CMD, carboxymethyldextran; DMSO, dimethyl sulfoxide; EDA,
ethylenediamine; EDCI, N-ethyl-N’-(3-diethylaminopropyl)-carbodiimide; ITC, iso-
thermal titration calorimetry; Lac, Galb1,4-Glc; LacNAc, Galb1,4-GlcNAc; NHS,
N-hydroxy-succinimide; SAM, self-assembled monolayer; SPR, surface plasmon
resonance.
⇑
Corresponding author. Tel.: +1 780 492 0377; fax: +1 780 492 8231.
E-mail address: ccairo@ualberta.ca (C.W. Cairo).
0008-6215/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.carres.2010.09.020

2642
R. S. Loka, C. W. Cairo / Carbohydrate Research 345 (2010) 2641–2647
avoid organic solvents. Application of the Staudinger ligation
should be compatible with these requirements. One potential lim-
itation of this strategy is the sensitivity of the phosphane reagents
to oxidation.28 We considered that an azide-modified surface could
be more robust than an immobilized phosphane and would reduce
complications due to oxidation by only capturing intact groups. In
previous work, we employed a synthetic strategy for generating
phosphane-labeled carbohydrate ligands that could be used to
generate glycoprotein conjugates.29 We chose to adapt this strat-
egy as a method to immobilize carbohydrate ligands to a CMD ma-
trix. We found that Staudinger ligation to an azide surface provided
an active and stable biosensor. Using this method, we investigated
the binding of wheat germ agglutinin (WGA) and jacalin binding to
Galb, Galb1,4-Glcb (Lac), and Galb1,4-GlcNAcb (LacNAc).
25 mM NaHCO3, pH 8.5).29,33 Carbohydrate or control surfaces were
then generated by injection of the appropriate phosphane com-
pound (1, 2, 3, or 4) (200 lL 1.25 mM, in 6% DMSO–water). The
change in RU of each surface after treatment with the phosphane
compounds was as follows: 1, 1760; 2, 1970; 3, 1120; and 4, 860 RU.
2.3. Lectin binding
All binding experiments were conducted on a BIAcore 3000 SPR
system in HEPES running buffer (10 mM HEPES, 150 mM NaCl,
1 mM CaCl2, 0.005% surfactant P20, pH 7.4). Experiments were per-
formed at a flow rate of 10 l
L minꢁ1 (kinject). Data obtained from
carbohydrate-modified surfaces were corrected by subtraction of
the response in a control lane modified with compound 1, which
was run in parallel for all injections. Jacalin and WGA were ob-
tained from US Biological (Swampscott, MA) and Sigma–Aldrich
Chemical Co. (St Louis, MO), respectively. Stock solutions of each
protein were made in running buffer, and concentrations were
determined by A280 measurements. Equilibrium binding data were
fit to either a single site,
2. Experimental
2.1. Synthetic reagents
2.1.1. Methyl 2-(diphenylphosphino)-4-(2-(2-(5-((3aS,4S,6aR)-2-
oxohexahydro-1H-thieno[3,4-d]imidazol-4-yl)pentanoyloxy)
ethoxy)ethylcarbamoyl)benzoate (1)
Biotin was coupled to a Boc-protected linker, tert-butyl-(2-(2-
hydroxy ethoxy)-ethyl)carbamate S-1, to provide compound S-2.30
Compound S-2 (145 mg, 0.34 mmol) was then stirred in a 3 mL solu-
tion of 1:1 TFA–DCM for 3 h, then concentrated (see Supplementary
data). The resulting amine was added to a solution of DCM (3.5 mL)
Bmax
F
RU ¼
;
ð1Þ
Kd þ F
or two-site model,
Bmax1 Bmax2
F
F
RU ¼
þ
ðsee Ref: 34Þ:
ð2Þ
Kd1 þ F Kd2 þ F
containing Et3N (56.8 lL, 0.41 mmol) and succinimidyl-3-diphenyl-
2.4. Molecular modeling
phosphino-4-methoxycarbonylbenzoate (S-3, 189.1 mg, 0.41 mmol)
and stirred overnight.29,31,32 The solution was concentrated and puri-
fied by column chromatography (9:1 EtOAc–MeOH) to yield com-
pound 1 (153 mg, 67%) as a yellow solid with minor amounts of the
corresponding phosphine oxide (ca. ꢀ4%). 1H NMR (400 MHz, CDCl3)
d 8.06 (dd, J = 8.0, 3.6 Hz, 1H), 7.90 (dd, J = 7.9, 3.3 Hz, 0.1H), 7.81 (dd,
J = 8.1, 1.4 Hz, 1H), 7.38 (dd, J = 3.7, 1.5 Hz, 1H), 7.36–7.26 (m, 10H),
6.91 (s, 1H), 5.85 (s, 1H), 5.27 (s, 1H), 4.63–4.38 (m, 1H), 4.33–4.11
(m, 3H), 3.72 (s, 3H), 3.69–3.43 (m, 6.7H), 3.12–3.07 (m, 1H), 2.87
(dd, J = 12.8, 4.8 Hz, 1H), 2.69 (d, J = 12.8 Hz, 1H), 2.42–2.20 (m, 2H),
1.64–1.61 (m, 4H), 1.47–1.33 (m, 2H); phosphine oxide: d 8.36 (d,
J = 13.7 Hz, 0.1H), 8.15 (d, J = 7.9 Hz, 0.1H), 7.68 (m, 0.5H), 7.52 (m,
0.3H), 7.45 (m, 0.5H), 3.40 (s, 0.3H); 13C NMR (101 MHz, CDCl3) d
173.5, 166.8, 166.76, 166.6, 141.3 (d, JC–P = 29.0 Hz), 137.3, 137.2 (d,
JC-P = 11.0 Hz), 136.7 (d, JC-P = 19.0 Hz), 133.9 (d, JC-P = 21.0 Hz),
132.9, 131.9, 130.8, 128.9, 128.6 (d, JC-P = 7.0 Hz), 126.7, 69.3, 68.9,
62.8, 61.9, 60.1, 55.5, 52.2, 40.5, 39.8, 33.8, 28.2, 28.15, 24.7; 31P
NMR (162 MHz, CDCl3): d 32.90 (s, 0.04P), -2.39(s, 1.00P); HRESIMS:
calcd for C35H40N3O7PSNa (M+Na)+ 700.2217; found 700.2214.
Models of the interaction between jacalin and galactosyl resi-
dues containing a 3,5-dioxaheptyl aglycone were generated using
a reported crystal structure (PDB ID: 1UH1).35,36 The structure
was modified using Macromodel (Schrodinger, Inc.) to incorporate
either an
a- or b-linked aglycone. The carbohydrate and protein
residues within 10 Å of the ligand were then minimized and sub-
jected to molecular dynamics for 10 ps, followed by an additional
minimization (OPLS force field;37 water solvent was modeled by
a continuous dielectric).
3. Results and discussion
3.1. Preparation of the biosensor surface
We developed a surface immobilization protocol suitable for
immobilization of Staudinger–Bertozzi reagents.38 We first func-
tionalized the CMD surface with ethylenediamine (EDA).39 The
amines on the surface could then be easily acetylated with an NHS
ester of azidoacetic acid, 5 (Fig. 1).29 We found that compound 5 re-
quired the inclusion of DMSO in the buffer (10%) for substantial
modification of the surface, and that the compound was subject to
hydrolysis if the solutions were not used rapidly. Once the azide
was immobilized, the surface was stable and could be reacted with
an appropriate Staudinger reagent as desired. We previously re-
ported the synthesis of phosphane reagents that contained galact-
ose (2, Galb), lactose (3, Lac), and N-acetyl-lactosamine (4, LacNAc)
moieties.29 These compounds were used to derivatize the CM5 sur-
face, after treatment as described above, to incorporate the azide
functionality. Separate lanes of a single CM5 chip were then exposed
to compounds 1, 2, 3, and 4 to generate the desired surfaces (Fig. 2).
The biotin-modified surface (compound 1) was used for background
subtraction of the other lanes during binding experiments.
2.1.2. Phosphane compounds 2, 3, and 4
The syntheses of compounds 2, 3, and 4 were carried out as pre-
viously described.29 Briefly, carbohydrate epitopes were generated
with an amine-terminated ethylene glycol linker. The amine was
then reacted with an N-hydroxysuccinimidyl-phosphane reagent,
S-3, to provide the desired compounds.31
2.2. Biosensor surface preparation
Surfaces were prepared using a BIAcore 3000 and CM5 (carb-
oxymethyldextran) sensor chip. Reactions were performed at a flow
rate of 5
equilibrated and then reacted with the following solutions to gen-
erate an azide-modified surface: (1) 200 L of a 1:1 solution of N-
l
L minꢁ1 in order to maximize contact time. The chip was
l
ethyl-N’-(3-diethylaminopropyl)carbodiimide (EDCI) (0.1 M in
water) and N-hydroxysuccinimide (NHS) (0.1 M in water); (2)
To the best of our knowledge, this is the first report of an azide-
modified surface used for SPR detection. However, it is important
to notethat work from the laboratories of Bertozzi and co-workers,19
Waldmann and co-workers,20 and Raines and co-workers23 have
325
200
l
L of ethylenediamine (EDA) (1 M in water, pH 8.5); (3)
lL of azidoacetic acid-NHS ester 5 (50 mM in 10% DMSO, 90%

R. S. Loka, C. W. Cairo / Carbohydrate Research 345 (2010) 2641–2647
2643
O
O
N3
N
NH2
O-
O
O
O
HN
O
5
O
H
N
1) NHS, EDCI
N3
N
H
2)
O
H2N
NH2
O
O
O
R
Ph2P
MeO
O
R
O
N
H
O
N
H
O
H
N
H
N
N
H
compound 1, 2, 3, or 4
O
O
PPh2
O
O
R =
OH
O
HO
HO
O
S
O
O
OH
1
2
3
4
H
NH
H
HN
OH
O
HO
HO
OH
O
O
O
HO
OH
OH
OH
O
HO
HO
OH
O
O
O
HO
OH
NHAc
Figure 1. Conjugation of carbohydrate epitopes to a CMD surface. The preparations of compounds 2, 3, 4, and 5 were carried out as reported.29 Control surfaces were
generated with a biotin-labeled phosphane 1.
25000
20000
15000
10000
5000
0
phosphane wouldfailto reactwiththesurface. By reversingthiscon-
figuration to have the phosphane on the surface, any oxidation
would reduce the number of available surface sites. Therefore, the
azide-surface should provide higher capacity and may be advanta-
geous if the surface requires storage or preparation separate from
the immobilization step. Additionally, in this configuration the pro-
tocol requires less of the costly phosphane reagent.
3.2. Detection of lectin specificity for carbohydrate surfaces
To test the binding of lectins to the carbohydrate-modified sur-
faces, we injected solutions of jacalin (Artocarpus integrifolia agglu-
tinin)40 and wheat germ agglutinin (WGA)41 over all lanes of the
biosensor (Fig. 3). We observed that jacalin bound specifically to
the galactose-containing compound 2, and WGA bound only the
GlcNAc-containing compound 4. Neither lectin recognized the
Lac-modified surface (compound 3), and only bulk-response was
observed for the control lane (compound 1). WGA is known to bind
specifically to GlcNAc mono- and oligosaccharides.41 Jacalin is
a
b
c
d
0
5000
10000
time [sec]
15000
20000
Figure 2. Modification of the biosensor surface with a carbohydrate ligand. A
sensorgram is shown for modification of the sensor surface as described in Figure 1.
The carboxymethyldextran was first activated with (a) EDCI/NHS (200
and then modified with (b) ethylenediamine (325 L, 1 M, pH 8.5). The amine
surface was then labeled with (c) NHS-azidoacetic acid (5, 200 L, 50 mM, 10%
DMSO). The azides on the surface were then allowed to react with (d) the
lL, 0.1 M)
known to bind in preference to aGal; however, only weak binding
l
l
to bGal would be expected (Table 1).42 To explore the binding of
each lectin in more detail, we proceeded to determine the affinity
of the interaction using SPR.
phosphane-modified galactose reagent 2 (200 lL, 1.25 mM, 6% DMSO). The change
in baseline RU was 1970 RU.
3.3. Determination of jacalin and WGA affinity
previously used variations of the Staudinger ligation as a surface
immobilization strategy. In our formulation of the Staudinger liga-
tion, we chose to use an azide surface to avoid any issues with oxida-
tion of the phosphane component.28 In this arrangement, oxidized
One of the advantages of a sensitive SPR assay is that it can be
used to determine the affinity of biomolecular interactions by di-
rect observation of association and dissociation kinetics. However,

2644
R. S. Loka, C. W. Cairo / Carbohydrate Research 345 (2010) 2641–2647
Table 1
200
150
100
50
Lectin affinities determined for jacalin and WGA
Gal
a
Entry Lectin
(ligand)
Model or
Ref.
(
l
M)
[RU]
Lac
LacNAc
1
2
3
Jacalin (Gal
Jacalin (Galb-OMe)
Jacalin (Galb1,3-GlcNAc)
a
-OMe)
42
42
35
29
6250
25,000
na
na
na
4
5
6
Jacalin, (2)
One-site Kd 800 70
Two-site Kd1 200 600
Bmax 55
Bmax1 19
1
4
Kd2 1800 2600 Bmax2 40 36
7
8
9
WGA (GlcNAc)
41
1300
200
1700
na
na
na
WGA (GlcNAcb1,4-GlcNAc) 41
WGA (Galb1,6-GlcNAc)
46
10
11
12
WGA, (4)
One-site Kd 190 40
Two-site Kd1 40 80
Bmax 111
Bmax1 70 40
4
Kd2 1300 1800 Bmax2 60 30
0
values were plotted versus the lectin concentration to provide a
binding isotherm (Fig. 5).
The equilibrium binding data were fit to a single (Eq. 1) or two-
site (Eq. 2) binding model to obtain the affinity of each interaction
(Table 1). When the data were plotted in semi-log form or as a
Scatchard plot (Fig. S2), it could be seen that jacalin binding was
well described by a single-site model, giving a dissociation con-
100
200
300
time [sec]
400
500
600
200
150
100
50
M (r2 = 0.98). Fitting the same data to a two-site
M at the first site, and
Gal
stant of 800 70
model gave an affinity for 2 of 200 600
1800 2600
M for the second site (r2 = 0.99). Although this model
l
b
Lac
l
LacNAc
l
provides a slightly improved fit, the lack of a concave Scatchard
plot suggests that the two-site model should not be used. Based
on the single-site affinity, the binding of a bGal derivative appears
to be remarkably high. Previous reports have determined that the
affinity of bGal to jacalin is close to 5 mM, while the aGal anomer
has much higher affinity.42 Hagiwara et al. have found that Galb-
p-nitrophenyl has greater affinity to jacalin than to GalNAc,
suggesting an ability for the lectin to accommodate a hydrophobic
b-linked aglycone.40 Additionally, Smith et al. have reported that
a
Gal derivatives with a hydrophobic aglycone have slightly im-
proved affinity to jacalin.14 Comparison of the single-site affinity
measured here for a bGal derivative falls between the expected
affinity of a GalaOMe and GalbOMe, suggesting that 2 has higher
than expected affinity for the lectin (vide infra).
0
The data obtained for binding of WGA to compound 4 were
partly described by the single-site model (190 40 l
M, r2 = 0.82).
100
200
300
time [sec]
400
500
600
However, a Scatchard plot of these data gave a concave-up curve,
suggesting heterogeneity in the binding data.44,45 Consistent with
this finding, a fit of the data to the two-site model (r2 = 0.88)
showed improvement over the single-site model, and gave values
Figure 3. Specific binding of lectins to a carbohydrate-modified surface. (a) Jacalin
(9.45 M, 10 M, 10
L minꢁ1) and (b) WGA (21.1 L minꢁ1) were injected over a
l
l
l
l
of 40 80 lM and 1300 1800 lM for the two affinity constants.
sensor chip modified with compounds 2, 3, and 4. Jacalin binds specifically to the
galactose-containing compound 2, and WGA binds specifically to N-acetyl-lactos-
amine compound 4. Sensorgrams shown were reference subtracted using a surface
modified with compound 1. Spikes due to subtraction or injection artifacts have
been removed for clarity.
The WGA lectin is known to be specific for GlcNAc-containing sac-
charides; however, its affinity for LacNAc has not been previously
determined. Measurement of WGA affinity for GlcNAc using iso-
thermal titration calorimetry (ITC) found a value of 1.3 mM.41
Larger oligosaccharides of b1,4-linked GlcNAc gave tighter binding;
kinetic determinations are inherently more challenging as a variety
of artifacts may obscure an accurate measurement.6 We injected a
series of six concentrations of each lectin over the surfaces and
analyzed the binding of jacalin to 2 and WGA to 4 (Fig. 4). Attempts
to fit these data to a typical 1:1 binding model failed. More com-
plex models, including bivalent binding and conformational
change models, were unsuccessful, with all tested models giving
poor fits.43 These observations suggested that lectin binding to
both surfaces followed complex kinetics. Therefore, we used equi-
librium binding to determine the affinity of the lectins. A series of
lectin concentrations were injected over the biosensor surface, and
the response at steady-state binding was recorded. The resulting
for example, the dimer gave an affinity of 200 lM. The affinity of
an isomeric disaccharide of LacNAc, Galb1,6-GlcNAc, for WGA has
been determined to be 1.7 mM.46
The binding of WGA to 4 is, therefore, of unusually high affinity.
When the data were analyzed using the single-site binding model,
the observed affinity of 4 is similar to the GlcNAc dimer. However,
when the non-linearity of the Scatchard plot is taken into account,
a more complex model is required. We interpret the non-ideal
Scatchard plot as an indication of heterogeneous binding sites on
the surface. This heterogeneity is likely the result of high affinity
sites that arise from carbohydrate epitopes on the surface that
are close enough to bridge multiple sites of a single WGA tetramer.

R. S. Loka, C. W. Cairo / Carbohydrate Research 345 (2010) 2641–2647
2645
140
120
100
80
60
a
a
50
40
30
20
10
0
60
40
single site
two site
20
0
0.000
0.002
0.004
conc [µM]
0.006
0.008
100
200
300
time [sec]
400
500
600
140
140
120
100
80
b
b
120
100
80
60
40
20
0
60
40
single site
two site
20
0
0.000
0.002
0.004
0.006
0.008
0.010
0.012
100
200
300
400
500
600
conc [µM]
time [sec]
Figure 5. Binding isotherms of Jacalin and WGA. The binding of (a) jacalin to the
galactose-modified surface (7.3, 6.1, 4.5, 3.6, 3.0, 2.4, 1.8, 1.5, 1.2, 0.91, 0.76, 0.61,
Figure 4. Binding of lectins to galactose and N-acetyllactosamine surfaces. (a)
Jacalin was injected over the galactose surface (9.5, 5.7, 4.7, 3.5, 2.8, 2.4 M,
10
L minꢁ1). (b) WGA was injected over the N-acetyllactosamine surface (21.2,
12.7, 10.6, 7.9, 6.4, 5.3 M, 10
L minꢁ1). Sensorgrams shown were reference
subtracted using a surface modified with compound 1. Spikes due to subtraction or
injection artifacts have been removed for clarity.
l
0.46, 0.30
l
M) and (b) WGA to the N-acetyl-lactosamine surface (11, 9.1, 6.9, 5.5,
M) are shown.
l
4.6, 3.7, 2.7, 2.3, 1.8, 1.4, 1.1, 0.92, 0.69, 0.46, 0.37, 0.27, 0.19, 0.09
l
l
l
Each point is a steady-state RU value at the end of the association phase. See
Supplementary data for semi-log and Scatchard analysis (Fig. S2).
In support of this model, the lower affinity site should yield the ex-
pected solution affinity of GlcNAc for WGA. This is indeed the case
when the data are fit to the two-site model. This interpretation
suggests an approximately 30-fold enhancement of affinity at sites
that can engage the lectin in a chelated binding mode.47 Non-ideal
effects due to surface density or multivalent binding are well-
known in SPR measurements, and these results allow us to quan-
tify one mechanism of the enhancement of lectin affinity.10,45,48
ITC has found that Galb-OMe has significantly weaker binding than
the Gala-OMe derivative. Structural determinations of galactoside
binding to jacalin have concluded that the key H-bond interactions
between the protein and the glycoside are at O-4, O-6, and O-3. The
methyl aglycone of the
a-galactoside has been shown to project
over the face of Y122, while the aglycone of the b-galactoside re-
sults in unfavorable steric interactions with the same tyrosine res-
idue. This steric interaction resulted in a decrease in affinity of
approximately 100-fold.49 Previous observations by Smith et al.
3.4. Structural model of jacalin binding
and Hagiwara et al. suggest that a- and b-galactosides with a
hydrophobic aglycone retain tight affinity to jacalin.14,40 In light
of these structural data, we were intrigued at our result. We con-
sidered that a possible explanation for these findings was that a
Our determination of the affinity of jacalin for compound 2 gave
an unexpectedly high value for a b-galactoside. Previous work with

2646
R. S. Loka, C. W. Cairo / Carbohydrate Research 345 (2010) 2641–2647
method, we are able to detect the specificity of lectins for carbohy-
drate-modified surfaces. We confirmed that expected lectin–
carbohydrate interactions, such as WGA binding to GlcNAc, could
be observed and quantified. We are also able to quantify an
enhancement of WGA binding due to multivalent binding of the
lectin with a high-density GlcNAc surface.47 Additionally, we ob-
served an unexpected interaction of a b-galactoside with jacalin.
We propose a structural model for the interaction of the galactosyl
aglycone with secondary subsite A of the protein. Our model helps
to rationalize our binding results, as well as previous reports of jac-
alin binding to synthetic substrates.
Acknowledgments
This work was funded by a grant from the Natural Sciences and
Engineering Research Council of Canada (NSERC) and the Alberta
Ingenuity Centre for Carbohydrate Sciences (AICCS). The authors
would like to acknowledge the Canadian Foundation for Innovation
and Western Economic Diversification Canada for infrastructure
support.
Supplementary data
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.carres.2010.09.020.
References
1. Dwek, R. A. Chem. Rev. 1996, 96, 683–720.
2. Dube, D. H.; Bertozzi, C. R. Nat. Rev. Drug Disc. 2005, 4, 477–488.
3. Hernandez, J. D.; Baum, L. G. Glycobiology 2002, 12, 127R–136R.
4. Lee, Y. C. J. Biochem. 1997, 121, 818–825.
5. Dam, T. K.; Brewer, C. F. Chem. Rev. 2002, 102, 387–429.
6. Rich, R. L.; Myszka, D. G. Curr. Opin. Biotechnol. 2000, 11, 54–61.
7. Krishnamoorthy, L.; Mahal, L. K. ACS Chem. Biol. 2009, 4, 715–732.
8. Manimala, J. C.; Roach, T. A.; Li, Z. T.; Gildersleeve, J. C. Glycobiology 2007, 17,
17C–23C.
Figure 6. Molecular modeling of jacalin binding. (a) Models of a b-galactoside with
a PEG aglycone (magenta) were generated, and then minimized (see Section 2). The
critical H-bond contacts at O-4 and O-6 are maintained, although the orientation of
the ring was slightly twisted from that of the
may occupy secondary site A, which is a relatively hydrophobic channel formed by
Y78 and Y122. Both the and the b anomers with a flexible aglycon are able to take
a-galactoside (blue). (b) The aglycone
a
advantage of contacts in secondary site A; however, an additional carbohydrate
residue would be too bulky to be accommodated.
9. Kiessling, L. L.; Cairo, C. W. Nat. Biotechnol. 2002, 20, 234–235.
10. Bamdad, C. Biophys. J. 1998, 75, 1989–1996.
11. Hutchinson, A. M. Anal. Biochem. 1994, 220, 303–307.
12. Linman, M. J.; Taylor, J. D.; Yu, H.; Chen, X.; Cheng, Q. Anal. Chem. 2008, 80,
4007–4013.
13. Gestwicki, J. E.; Cairo, C. W.; Mann, D. A.; Owen, R. M.; Kiessling, L. L. Anal.
Biochem. 2002, 305, 149–155.
14. Smith, E. A.; Thomas, W. D.; Kiessling, L. L.; Corn, R. M. J. Am. Chem. Soc. 2003,
125, 6140–6148.
15. Mann, D. A.; Kanai, M.; Maly, D. J.; Kiessling, L. L. J. Am. Chem. Soc. 1998, 120,
10575–10582.
16. Sletten, E. M.; Bertozzi, C. R. Angew. Chem., Int. Ed. 2009, 48, 6974–6998.
17. Camarero, J. A. Pept. Sci. 2008, 90, 450–458.
flexible and hydrophobic linker could impart additional specificity
to the glycoside. To examine this hypothesis in more detail, we
used molecular modeling to examine the binding of
galactosides with a short flexible linker (Fig. 6). We found that a
modified -galactoside could maintain an almost identical orienta-
a- and b-
a
tion of the saccharide in the binding site as compared to the orien-
tation of the free saccharide. The flexible linker of the aglycone
could insert into a short hydrophobic channel formed by Y78 and
Y122 on the protein. This region of the binding site was referred
to as ‘secondary site A’ by Jeyaprakash et al.35 In the case of the
b-galactoside, we found that in order to accommodate the b-linked
aglycone, the orientation of the ring in the active site was slightly
altered. Importantly, the position of the major H-bond contacts at
O-3, O-4, and O-6 was maintained. We also confirmed that the
aglycone of the b-galactoside was able to gain favorable interac-
tions at secondary site A. Thus, our model suggests that additional
hydrophobic interactions at secondary site A can explain previous
results with synthetic galactosides.14,40 Additionally, the model
suggests that tighter binding ligands for jacalin could be developed
by exploiting these contacts.
18. Kolb, H. C.; Finn, M. G.; Sharpless, K. B. Angew. Chem., Int. Ed. 2001, 40, 2004–
2021.
19. Hang, H. C.; Yu, C.; Pratt, M. R.; Bertozzi, C. R. J. Am. Chem. Soc. 2004, 126, 6–
7.
20. Kohn, M.; Wacker, R.; Peters, C.; Schroder, H.; Soulere, L.; Breinbauer, R.;
Niemeyer, C. M.; Waldmann, H. Angew. Chem., Int. Ed. 2003, 42, 5830–5834.
21. Watzke, A.; Kohn, M.; Gutierrez-Rodriguez, M.; Wacker, R.; Schroder, H.;
Breinbauer, R.; Kuhlmann, J.; Alexandrov, K.; Niemeyer, C. M.; Goody, R. S.;
Waldmann, H. Angew. Chem., Int. Ed. 2006, 45, 1408–1412.
22. Grandjean, C.; Boutonnier, A.; Guerreiro, C.; Fournier, J. M.; Mulard, L. A. J. Org.
Chem. 2005, 70, 7123–7132.
23. Kalia, J.; Abbott, N. L.; Raines, R. T. Bioconjugate Chem. 2007, 18, 1064–1069.
24. Helms, B.; van Baal, I.; Merkx, M.; Meijer, E. W. ChemBioChem 2007, 8, 1790–
1794.
25. Govindaraju, T.; Jonkheijm, P.; Gogolin, L.; Schroeder, H.; Becker, C. F. W.;
Niemeyer, C. M.; Waldmann, H. Chem. Commun. (Cambridge) 2008, 3723–3725.
26. Grabosch, C.; Kleinert, M.; Lindhorst, T. K. Synthesis 2010, 828–836.
27. Kleinert, M.; Winkler, T.; Terfort, A.; Lindhorst, T. K. Org. Biomol. Chem. 2008, 6,
2118–2132.
28. Zhang, H. L.; Ma, Y.; Sun, X. L. Chem. Commun. (Cambridge) 2009, 3032–
3034.
29. Loka, R. S.; Romaniuk, N. A.; Sadek, C. M.; Cairo, C. W., Bioconjugate Chem., in
press, doi:10.1021/bc100209r.
4. Conclusions
We describe an implementation of Staudinger reagents for the
immobilization of carbohydrates to a biosensor surface. To avoid
issues of phosphorus oxidation, we have used an azide-derivatized
surface, which is then reacted with a soluble phosphane. Using this
30. Heydari, A.; Khaksar, S.; Tajbakhsh, M. Synthesis 2008, 3126–3130.
31. Wang, C. C. Y.; Seo, T. S.; Li, Z.; Ruparel, H.; Ju, J. Bioconjugate Chem. 2003, 14,
697–701.
32. Kiick, K. L.; Saxon, E.; Tirrell, D. A.; Bertozzi, C. R. Proc. Natl. Acad. Sci. U.S.A.
2002, 99, 19–24.

R. S. Loka, C. W. Cairo / Carbohydrate Research 345 (2010) 2641–2647
2647
33. Yu, H.; Huang, S. S.; Chokhawala, H.; Sun, M. C.; Zheng, H. J.; Chen, X. Angew.
Chem., Int. Ed. 2006, 45, 3938–3944.
34. Attie, A. D.; Raines, R. T. J. Chem. Educ. 1995, 72, 119.
35. Jeyaprakash, A. A.; Katiyar, S.; Swaminathan, C. P.; Sekar, K.; Surolia, A.;
Vijayan, M. J. Mol. Biol. 2003, 332, 217–228.
41. Bains, G.; Lee, R. T.; Lee, Y. C.; Freire, E. Biochemistry 1992, 31, 12624–12628.
42. Mahanta, S. K.; Sastry, M. V. K.; Surolia, A. Biochem. J. 1990, 265, 831–840.
43. Lienemann, M.; Paananen, A.; Boer, H.; de la Fuente, J. M.; Garcia, I.; Penades,
S.; Koivula, A. Glycobiology 2009, 19, 633–643.
44. Klotz, I. M. Science 1982, 217, 1247–1249.
36. Jeyaprakash, A. A.; Jayashree, G.; Mahanta, S. K.; Swaminathan, C. P.; Sekar, K.;
Surolia, A.; Vijayan, M. J. Mol. Biol. 2005, 347, 181–188.
37. Damm, W.; Frontera, A.; TiradoRives, J.; Jorgensen, W. L. J. Comput. Chem. 1997,
18, 1955–1970.
45. Shinohara, Y.; Hasegawa, Y.; Kaku, H.; Shibuya, N. Glycobiology 1997, 7, 1201–
1208.
46. Muraki, M.; Ishimura, M.; Harata, K. Biochim. Biophys. Acta, Gen. Subj. 2002,
1569, 10–20.
38. Saxon, E.; Bertozzi, C. R. Science 2000, 287, 2007–2010.
39. O’Shannessy, D. J.; Brighamburke, M.; Peck, K. Anal. Biochem. 1992, 205, 132–
136.
47. Kiessling, L. L.; Pohl, N. L. Chem. Biol. 1996, 3, 71–77.
48. Tassa, C.; Duffner, J. L.; Lewis, T. A.; Weissleder, R.; Schreiber, S. L.; Koehler, A.
N.; Shaw, S. Y. Bioconjugate Chem. 2010, 21, 14–19.
40. Hagiwara, K.; Colletcassart, D.; Kobayashi, K.; Vaerman, J. P. Mol. Immunol.
1988, 25, 69–83.
49. Sastry, M. V. K.; Banarjee, P.; Patanjali, S. R.; Swamy, M. J.; Swarnalatha, G. V.;
Surolia, A. J. Biol. Chem. 1986, 261, 1726–1733.