Angewandte
Chemie
seven-carbon sugars for screening against conA, and for
quantitative binding experiments. To this end, our fluorous-
tagged heptomannose monosaccharides, dissolved in a sol-
vent mixture, were arrayed by using a standard DNA
microarray spotting robot. The spotted glass slides were
incubated with a solution of fluorescein isothiocyanate-
labeled concanavalin A (FITC-ConA) in phosphate-buffered
saline (PBS) and scanned with a laser at 488 nm to visualize
the carbohydrate–conA binding (Figure 2). As expected, the
in a way that traditional single concentration assays with
microarrays cannot. Interestingly, the values reported for
mannose binding to conA with various surface-based tech-
niques differ widely; comparisons should only be made to
rank order compounds measured with the same protocol.
Kiessling, Corn, and co-workers used an SPR technique with
thiol-modified mannose linked to a gold surface to measure a
Kd value of 200 ꢁ 50 mm with conA.[17] The authors specifically
note that the inverse of the association value (179 nm for
mannose-conA) could not be used to determine the dissoci-
ation directly and that the method should be used to compare
compounds only with the same technique. More recently, a
microarray based on reaction of carbohydrates with an epoxy-
coated glass surface and blocked with a proprietary protein
solution showed an unusually strong dissociation constant of
80 nm for bound mannose and conA.[16] Clearly, the nature of
the linker and surface matters and quantitative comparisons
between ligands binding to a protein should be limited to
systems in which only the ligands themselves are a variable.
These new results also show, for the first, time that
concanavalin A can recognize seven-carbon mannose ana-
logues as ligands. In 1965 Goldstein et al.[4] suggested that the
modifications of any of the hydroxy groups at C-3, C-4, and C-
6 positions of the d-mannopyranose ring resulted in a
complete loss of affinity with conA. However, our study
demonstrates that the substitution of one of the hydrogen
atoms in the C-6 position by a hydroxymethyl group does not
abrogate carbohydrate-protein binding. The presence of the
second hydroxy group could perhaps even stabilize the
monosaccharide in the binding site of the protein. In an X-
ray crystallographic study of conA binding to a-(1!2)
mannobiose, Naismith and co-workers noted the presence
of a water molecule located near the C-6 hydroxy group of the
nonreducing mannose that helps mediate the interaction of
the sugar with the lectin.[18] This “structural” water molecule,
first described with conA in 1996 by Naismith and Field,[19]
links and stabilizes the nonreducing sugar to the protein
(Asp16). In our case, the extra hydroxymethyl group in the
heptomannose could possibly play the role of this water
molecule and even replace it as a stabilizing link between
sugar and protein. The positioning of this C-7 hydroxy group
may also account for the differences in binding affinity seen
between the heptamannose diastereomers.
Figure 2. Left: Fluorescence images of arrayed carbohydrates probed
with FITC-labeled conA. Columns of five spots each of 125 mm
carbohydrates were incubated for 2 h with FITC-ConA with 1% bovine
serum albumin. Man=fluorous-tag-linked mannose. Right: Graph of
average fluorescence intensities.
fluorous-tagged b-d-mannose[14] exhibits a robust fluorescent
response indicating protein binding, in contrast to the
negative control fluorous-tagged b-d-galactose (not
shown),[14] which shows no binding. Surprisingly, both fluo-
rous-tagged heptomannose monosaccharides show binding to
conA. On the basis of fluorescence intensity, the R diaste-
reomer binds comparably to the S diastereomer to conA. For
quantification, the fluorinated sugars were spotted (2 12
spots) on the fluorinated microarray slide at a concentration
of 125 mm and let dry for 30 minutes in a dark humidified
chamber. The slide was then incubated for 2 hours at eight
different concentrations of conA (0.1–2 mm) and washed (five
to seven times) with deionized water. The relative intensities
of the spots were determined with ImaGene software to
calculate the Kd value for each sugar.[16] The fluorous-tagged
b-d-mannose (Figure 3) bound with a value of Kd = 1.9 ꢁ
Fluorous-based carbohydrate microarrays clearly show
that both diastereomers of glycero-d-manno-heptoses found
in bacteria bind to conA and that this well-studied lectin can
accept modifications at the C-6 position of its usual mannose
ligand. These results illustrate the imperative to incorporate
an ever greater range of carbohydrates into screening arrays
and to very cautiously interpret lectin-binding data as proof of
the presence of distinct structures. In addition, attempts to use
glycero-d-manno-heptoses in the construction of antibacterial
vaccines should proceed with caution as proteins that bind to
this motif can clearly cross-react with mannose residues that
are found in a range of human proteins. Even the most-
studied lectin conA has revelations to offer in its binding
affinities. Microarrays promise to uncover more, and more
surprising carbohydrate-binding protein/sugar complexes.[20]
The fluorous tag described herein not only facilitates the
Figure 3. Binding curve of fluorous-tag-linked mannose.
0.4 mm, which is comparable to the seven-carbon sugars (R
and S diastereomer: Kd = 1.1 ꢁ 0.3 mm). As observed previ-
ously,[16] the one-point threshold-based qualitative analysis
(Figure 2) does not necessarily accurately reflect the relative
binding strengths of even closely related substrates.
These experiments are the first demonstration that
microarrays based on noncovalent fluorous interactions can
be used not only for qualitative but also for the relative
quantitative assessment of carbohydrate/protein interactions
Angew. Chem. Int. Ed. 2008, 47, 1707 –1710
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1709