On the meaning of affinity: Cluster glycoside effects and concanavalin a

Article abstract of DOI:10.1021/ja991729e

The inhibition of protein - carbohydrate interaction provides a powerful therapeutic strategy for the treatment of myriad human diseases. To date, application of such approaches have been frustrated by the inherent low affinity of carbohydrate ligands for

Full text of DOI:10.1021/ja991729e

10286
J. Am. Chem. Soc. 1999, 121, 10286-10296
On the Meaning of Affinity: Cluster Glycoside Effects and
Concanavalin A
Sarah M. Dimick,† Steven C. Powell,† Stephen A. McMahon,‡ Davina N. Moothoo,‡
James H. Naismith,*,‡ and Eric J. Toone*,†
Contribution from the Department of Chemistry, Duke UniVersity, Durham, North Carolina 27708-0346,
and Center for Biomolecular Science, Purdie Building, The UniVersity,
St. Andrews KY16 9ST,
Scotland, U.K.
ReceiVed May 24, 1999
Abstract: The inhibition of protein-carbohydrate interaction provides a powerful therapeutic strategy for the
treatment of myriad human diseases. To date, application of such approaches have been frustrated by the
inherent low affinity of carbohydrate ligands for their protein receptors. Because lectins typically exist in
multimeric assemblies, a variety of polyvalent saccharide ligands have been prepared in the search for high
affinity. The cluster glycoside effect, or the observation of high affinity derived from multivalency in
oligosaccharide ligands, apparently represents the best strategy for overcoming the “weak binding” problem.
Here we report the synthesis of a series of multivalent dendritic saccharides and a biophysical evaluation of
their interaction with the plant lectin concanavalin A. Although a 30-fold enhancement in affinity on a valence-
corrected basis is observed by agglutination assay, calorimetric titration of soluble protein with a range of
multivalent ligands reveals no enhancement in binding free energies. Rather, IC50 values from agglutination
measurements correlate well with entropies of binding. This observation suggests that hemagglutination measures
a phenomenon distinct from binding that is typified by a large favorable entropy and an unfavorable enthalpy:
this process is almost certainly aggregation. Supporting this assertion, we report crystal structures of multivalent
ligands cross-linking concanavalin A dimers. To the best of our knowledge, these structures are the first reported
of their kind. Our results indicate that hemagglutination assays evaluate the ability of ligands to inhibit the
formation of cross-linked lattices, a process only tangentially related to reversible ligand binding. Cluster
glycoside effects observed in agglutination assays must, therefore, be viewed with caution. Such effects may
or may not be relevant to the design of therapeutically useful saccharides.
Introduction
Despite the intellectual appeal of the methodology, construc-
tion of such compounds is hampered by several fundamental
concerns. Chief among these is the low-affinity binding that
typifies protein-carbohydrate interaction: such binding events
proceed uniformly with millimolar to micromolar dissociation
constants.10 Accordingly, a major focus of contemporary
carbohydrate research surrounds the development of general
strategies for increasing lectin-ligand binding affinities to the
levels required for therapeutic use.
Lectins are seldom found in vivo as monomeric species;
rather, they typically exist as oligomeric structures. This
phenomenon suggests that Nature has dealt with the “tight
binding” problem through multivalency, where multiple simul-
taneous binding events overcome weak individual interaction
free energies. A straightforward corollary of this hypothesis is
that synthetic polyvalent ligands should show high binding
affinities. From this basis, several groups have pursued the goal
of tight lectin-ligand binding through multivalency, and a
number of ligands with valencies from 2 to 20 have been
prepared for binding to a variety of lectins.11-16 Lee and co-
workers have prepared a series of multivalent mannosides as
The initiation of a wide range of human diseases is mediated
by protein-carbohydrate recognition.1-7 In the earliest phases
of infection by a variety of viral, parasitic, mycoplasmal, and
bacterial pathogens, host recognition is achieved through specific
adhesion to cell surface carbohydrate epitopes. Protein-
carbohydrate interaction also plays a major role in the progres-
sion of many human cancers, contributing to both random and
nonrandom metastatic events.8,9 Because of the myriad and
varied roles of protein-carbohydrate interaction in human
disease, the development of small-molecule inhibitors of
pathogenic saccharide-mediated adhesion has been the subject
of intense activity in recent years. The paradigm is an attractive
one since it would provide a non-cytotoxic group of therapeutic
products applicable to a wide range of human diseases.
† Duke University.
‡ The University, St. Andrews.
(1) Hughes, R. C. Curr. Opin. Struct. Biol. 1992, 2, 687.
(2) Drickamer, K.; Taylor, M. E. Annu. ReV. Cell Biol. 1993, 9, 237.
(3) Lee, Y. C.; Lee, R. T. J. Biomed. Sci. 1996, 3, 221.
(4) Dwek, R. A. Chem. ReV. 1996, 96, 683.
(5) Wong, S. Y. C. Curr. Opin. Struct. Biol. 1995, 5, 599.
(6) Cook, G. M. W. Glycobiology 1995, 5, 449.
(7) Kobata, A. Acc. Chem. Res. 1993, 26, 319.
(8) Beuth, J.; K., H. L.; Pulverer, G.; Uhlenbruck, G.; Pichlmaier, H.
Glycoconjugate J. 1995, 12, 1.
(10) Toone, E. J. Curr. Opin. Struct. Biol. 1994, 4, 719.
(11) Roy, R. Polym. News 1996, 21, 226.
(12) Roy, R. Curr. Opin. Struct. Biol. 1996, 6, 692.
(13) Roy, R. Topics Curr. Chem. 1997, 187, 241.
(14) Rini, J. M. Annu. ReV. Biophys. Biomol. Struct. 1995, 24, 551.
(15) Kiessling, L. L.; Pohl, N. L. Chem. Biol. 1996, 3, 71.
(16) Drickamer, K. Struct. Biol. 1995, 2, 437.
(9) Gabius, H.-J.; Kayser, K.; Gabius, S. Naturwissenschaften 1995, 82,
533.
10.1021/ja991729e CCC: $18.00 © 1999 American Chemical Society
Published on Web 10/14/1999

Cluster Glycoside Effects and ConcanaValin A
J. Am. Chem. Soc., Vol. 121, No. 44, 1999 10287
ligands for the mannose binding protein: the tightest-binding
of this group show enhancements of 103-104 over the monova-
lent ligand.17,18 Biessen and co-workers prepared a hexavalent
mannoside for the same protein that showed a valence-corrected
enhancement in binding affinity of 105 over methyl R-D-
mannopyranoside.19 Roy and co-workers have prepared a variety
of polyvalent saccharides for various plant lectins and for the
influenza hemagglutinin.20-25 In general, these ligands show
enhancements of 10-103 over the appropriate monovalent
ligand. Another class of polyvalent glycosidic ligands, the so-
called glycopolymers, provide even more spectacular enhance-
ments in affinity. A series of acrylamide/acrylic acid ester
copolymers show enhancements in affinity to 109, again on a
valence-corrected or per mole of saccharide basis.26-31 The
sometimes spectacular successes of this strategy have led to
general acceptance of the “cluster glycoside” effect, defined as
“an affinity enhancement over and beyond what would be
expected from the concentration increase of the determinant
sugar in a multivalent ligand”.17,32
Despite the phenomenological success of multivalency strate-
gies, a molecular interpretation of the effect is difficult.
Typically, cluster glycoside effects are interpreted in terms of
entropic advantage, either an effective concentration argument
or a chelate effect. However, in many examples the linker region
between recognition epitopes is too short to span two binding
sites. Cluster glycoside effects have been observed for monova-
lent lectins, although the lectins were immobillized on microtiter
plate wells.33-35 On its face, an entropic enhancement to valence-
corrected binding free energies seems unlikely. The binding free
energy for a bivalent ligand to a bivalent receptor is related to
the analogous monovalent binding free energies by the expres-
sion
recognition domains.36 Assuming the linker domain is of
sufficient length and flexibility to allow optimal localization of
the receptor domains in their appropriate binding sites, enthalpic
contributions to ∆Gi will be small. Rather, the dominant
contributions to interaction energies arise from entropic effects.
Specifically, two terms with opposite signs contribute to ∆Gi.
First, tethering of recognition domains reduces the overall
translational and rotational entropy of binding. During complete
binding of a bivalent receptor by monovalent ligand, three
particles are converted to one; during the corresponding binding
of a bivalent ligand, two particles are converted to one. Binding
of a bivalent ligand thus proceeds with an entropic “savings”
equivalent to the translational and rotational entropy of one
monovalent ligand. The magnitude of this effect is difficult to
accurately ascertain: current estimates of the translational and
rotational entropy barrier to bimolecular complex formation in
aqueous solution range from 2.5 to 15 kcal mol-1 near room
temperature.37-40 Countering this effect, the introduction of a
flexible linker region introduces an entropic penalty relative to
monovalent ligand binding as conformational degrees of free-
dom in the linker domain are frozen out during binding. Again,
the exact magnitude of this effect is difficult to estimate,
although the smallest estimates lie in the vicinity of 0.5 kcal
mol-1 per rigid rotor frozen during binding.41-44 Given that
distances between carbohydrate binding sites on most polyvalent
lectins range from 20 to 70 Å, it seems unlikely that a chelate
effect could provide an enhancement to reversible binding free
energies. At this time, then, the molecular basis for the cluster
glycoside effect is not well understood.
Efforts to provide a molecular basis for the cluster glycoside
effect have been frustrated by two issues: (i) polyvalent ligands
are often polydisperse and structurally ill-defined and (ii) assays
typically used to evaluate protein-carbohydrate bindings
especially those based on aggregation of macromoleculess
measure several molecular phenomena including, but not limited
to, reversible thermodynamic protein-carbohydrate association.
Here we describe experiments designed to provide a molecular
basis for the cluster glycoside effect. Below we detail the
preparation and characterization of a series of dendritic ligands
for the plant lectin concanavalin A. We have characterized the
interaction of polyvalent ligands with this lectin through
agglutination assays, titration microcalorimetry, low-angle
dynamic light scattering, and X-ray crystallography. Together
the results of these studies allow a concise and unambiguous
interpretation of multivalency effects in protein-carbohydrate
interaction.
∆G ) 2∆Gm + ∆Gi
where ∆Gm represents the binding free energy of the component
monovalent ligands and ∆Gi is an interaction energy, the
energetic consequence of physical linkage of the monovalent
(17) Lee, Y. C.; Lee, R. T. Acc. Chem. Res. 1995, 28, 321.
(18) Quesenberry, M. S.; Lee, R. T.; Lee, Y. C. Biochemistry 1997, 36,
2724.
(19) Biessen, E. A. L.; Noorman, F.; van Teijlingen, M. E.; Kuipert, J.;
Barrett-Bergshoeff, M.; Bijsterbosch, M. K.; Rijken, D. C.; van Berkel, T.
J. C. J. Biol. Chem. 1996, 271, 28030.
(20) Page´, D.; Zanini, D.; Roy, R. Bioorg. Med. Chem. 1996, 4, 1949.
(21) Meunier, S. J.; Roy, R. Tetrahedron Lett. 1996, 37, 5469.
(22) Page, D.; Aravind, S.; Roy, R. J. Chem. Soc., Chem. Commun. 1996,
1913.
(23) Zanini, D.; Roy, R. Bioconjugate Chem. 1997, 8, 187.
(24) Zanini, D.; Roy, R. J. Org. Chem. 1996, 61, 7348.
(25) Zanini, D.; Park, W. K. C.; Roy, R. Tetrahedron Lett. 1995, 36,
7383.
(26) Sigal, G. B.; Mammen, M.; Dahmann, G.; Whitesides, G. M. J.
Am. Chem. Soc. 1996, 118, 3789.
(27) Mammen, M.; Dahmann, G.; Whitesides, G. M. J. Med. Chem. 1995,
38, 4179.
(28) Matrosovich, M. N.; Mochalova, L. V.; Marinina, V. P.; Byramova,
N. E.; Bovin, N. V. FEBS Lett. 1990, 272, 209.
(29) Roy, R.; Laferriere, C. A. Carbohydr. Res. 1988, 177, C1.
(30) Gamian, A.; Chomik, M.; Laferriere, G. A.; Roy, R. Can. J.
Micorbiol. 1991, 37, 233.
Results and Discussion
Synthesis of Ligands. We have chosen the plant lectin
concanavalin A as a model lectin for our studies. Lectins are
ubiquitous in plants, and a wide range of mannose-specific
lectins have been isolated from the seeds of legumes. The best
known of these proteins is concanavalin A (con A), first isolated
and crystallized some 60 years ago from the seeds of CanaValia
(36) Jencks, W. P. Proc. Natl. Acad. Sci. U.S.A. 1981, 78, 4046.
(37) Jencks, W. P. AdV. Enzymol. 1975, 43, 219.
(38) Murphy, K. P.; Xie, D.; Thompson, K. S.; Amzel, L. M.; Freire, E.
Proteins: Struct., Fumct., Genet. 1993, 18, 63.
(31) Watson, J. L.; Spaltenstein, A.; Kingery-Wood, J. E.; Whitesides,
G. M. J. Med. Chem. 1994, 37, 3419.
(32) Ozaki, K.; Lee, R. T.; Lee, Y. C.; Kawasaki, T. Glycoconjugate J.
1995, 122, 268.
(39) Dunitz, J. D. Science 1994, 264, 670.
(33) DeFrees, S. A.; Gaeta, F. C. A.; Lin, Y.-Ch.; Ichikawa, Y.; Wong,
C.-H. J. Am. Chem. Soc. 1993, 115, 7549.
(34) DeFrees, S. A.; Kosch, W.; Way, W.; Paulson, J. C.; Sabesan, S.;
Halcomb, R. L.; Huang, D.-H.; Ichikawa, Y.; Wong, C.-H. J. Am. Chem.
Soc. 1995, 117, 66.
(35) Lin, C.-H.; Shimazaki, M.; Wong, C.-H.; Koketsu, M.; Junega, L.
R.; Kim, M. Bioorg. Med. Chem 1995, 3, 1625.
(40) Spolar, R. S.; Recond, M. T. Science 1994, 263, 777.
(41) Starkweather, H. W.; Boyd, R. H. J. Phys. Chem. 1960, 64, 410.
(42) Gellman, S. H.; Dado, S. P.; Liang, G. P.; Adams, B. J. Am. Chem.
Soc. 1991, 113, 1164.
(43) Yoshiharu, T.; Nose, T.; Hata, T. Polym. J. 1974, 6, 51.
(44) Page, M. I.; Jencks, W. P. Proc. Natl. Acad. Sci. U.S.A. 1971, 68,
1678.

10288 J. Am. Chem. Soc., Vol. 121, No. 44, 1999
Dimick et al.
ensiformis.46 The con A monomer has a molecular weight of
26 000 Da and associates into higher order aggregates. In the
pH range 5.0-5.6, con A exists exclusively as dimers, while at
higher pH the dimers form tetramers; at pH 7.0, tetramer is the
predominant form. Succinylation provides a form of the protein
that remains dimeric at all pH values. Numerous crystallographic
structures of con A to resolutions of 2.0 Å both in native and
saccharide-bound forms have been reported.47-49
Scheme 1a
Our multivalent ligands were prepared on a dendritic scaf-
folding. The synthetic approach is convergent, rather than
divergent, allowing glycosidic coupling to the growing polymer
at an early stage of the synthesis. We note that our approach
also allows total flexibility with regard to the surface composi-
tion of polymers displaying two types of functionality, although
we have not utilized this facet of the strategy here. The synthesis
is based on a benzene-1,3,5-tricarboxylic acid core similar to
that devised by Neenan and Miller.50 In their original synthesis,
these researchers formed carboxylic esters at each step and
carried a benzylic alcohol as the nucleophile, protected as a
silyl ether. Ultimately, the synthesis was limited by difficulties
in deprotecting the alcohol nucleophile and acid degradation
of the polymer. Our approach replaces ester linkages by amides
and carries an amine nucleophile protected as the azide. The
enhanced stability of the amide linkage coupled with the facile
exposure of the amine nucleophile obviates the limitations of
the earlier approach. In principle, the tradeoff for the enhanced
stability of the amide relative to the ester is decreased solubil-
ity: such limitations were not encountered here.
a Conditions: (a) H2SO4, MeOH, 80 °C; (b) 1.05 equiv of KOH;
(c) BMS; (d) SOCl2; (e) NaN3; (f) 1 M NaOH; (g) SOCl2.
Synthesis of the dendrimer scaffolding proceeds from benzene-
1,3,5-tricarboxylic acid trimethyl ester 1 (Scheme 1). A 76%
yield of the monoacid is achieved by hydrolysis with KOH in
18 h. Exclusive production of the monoacid can be achieved
by treatment of triester 1 with pig liver esterase. This enzyme,
widely utilized as a chiral catalyst in organic synthesis, does
not accept charged substrates and cleanly converts triester 1 to
monoacid diester 2. The low solubility of 1 renders enzymatic
reaction exceptionally slow: conversion of 9 g of triester with
3000 units of enzyme required 14 days to proceed to completion.
As a result, simple base hydrolysis was typically a superior
reaction, despite generation of minor byproducts. Selective
reduction of monoacid diester 2 was effected with BH3‚Me2S,
and conversion of the resulting benzylic alcohol 3 to the required
benzylic azide 5 was completed by sequential treatment with
thionyl chloride and sodium azide. Although all the ligands
utilized in this study display carbohydrate uniformly on the
surface, patterned surfaces displaying two or more classes of
functionality may be desirable in other applications. Preparation
of such species requires differentiation of the arms of the
growing dendrimer, achievable by selective ester cleavage of
compound 5. For this synthesis, both methyl esters were cleaved
by KOH to yield the diacid 6. Finally, the 5-azidomethylben-
zene-1,3 dicarboxylate was converted to the diacid chloride with
thionyl chloride.
roacetimidate (8) was coupled under TMSOTf promotion to
3-azidopropanol to yield 9. Reduction of the azide preceded
coupling to the appropriate acyl chloride in the presence of
triethylamine to produce protected mono-, bi-, and trivalent
ligands 10, 12, and 14 (Scheme 2). Deprotection under Zemplen
conditions and purification by silica chromatography yielded
ligands 11, 13, and 15, respectively. Alternatively, hydrogenation
of protected bivalent ligand 12 followed by coupling to
5-azidomethylbenzene-1,3-dicarboxylate dichloride (7) and ben-
zene-1,3,5-tricarbonyl trichloride produced protected tetra- and
hexavalent ligands 16 and 18. Deacetylation under Zemplen
conditions and purification by gel permeation chromatography
completed the syntheses of the required ligands 17 and 19
(Scheme 3).
Simple HSEA calculations show a single family of low-
energy conformers of the second-generation (tetravalent) ligand.
With the two amides of the central ring oriented in a cis/trans
orientation, the two dimeric halves extend in a parallel fashion
in different planes, minimizing steric interaction. The extended
conformation of the second-generation ligand places distal
carbonyl carbons some 16 Å apart and separates saccharide
residues by 12, 16, and 22 Å (anomeric oxygen to anomeric
oxygen), with a maximum possible spacing of roughly 30 Å.
The hexavalent structure extends this maximum distance
modestly to near 36 Å.
Binding of Concanavalin A to Polyvalent Mannosides. The
evaluation of protein-carbohydrate interaction strengths is a
nontrivial undertaking. Historically, protein-carbohydrate bind-
ing has been evaluated by variations of the Landsteiner hapten
inhibition assay. This protocol measures the ability of a soluble
saccharide to inhibit the aggregation and precipitation of a
polyvalent lectin by a multivalent saccharide ligand. Protein-
carbohydrate binding free energies are then assumed to be
inversely proportional to IC50 values. While the technique offers
the advantage of simplicity, the results are both notoriously
irreproducible from laboratory to laboratory and difficult to
Mannose was coupled to the growing dendrimer through an
aminopropanol spacer. R-(Tetra-O-acetyl)mannopyranosyl trichlo-
(45) Mammen, M.; Sakhnovich, E. I.; Whitesides, G. M. J. Org. Chem.
1998, 63, 3168.
(46) Goldstein, I. J.; Poretz, R. D. In The Lectins: Properties, Functions,
and Applications in Biology and Medicine; Liener, I. E., Sharon, N.,
Goldstein, I. J., Eds.; Academic: New York, 1986.
(47) Naismith, J. H.; Field R. A. J. Biol. Chem. 1996, 271, 972.
(48) Naismith, J. H.; Emmerich, C.; Habash, J.; Harrop, S. J.; Helliwell,
J. R.; Hunter, W. N.; Raftery, J.; Kalb, A. J.; Yariv, J. J. Acta Crystallogr.
1994, D50, 847.
(49) Moothoo, D. N.; Naismith, J. H. Glycobiology 1998, 8, 173
(50) Miller, T. M.; Kwock, E. W.; Neenan, T. X. Macromolecules 1992,
25, 5, 3143.

Cluster Glycoside Effects and ConcanaValin A
J. Am. Chem. Soc., Vol. 121, No. 44, 1999 10289
Scheme 2
Table 1. Binding of Dendtritic Ligands to Concanavalin A
interpret at a molecular level. More recently, titration micro-
calorimetry has been utilized to study protein-carbohydrate
interaction.51-60 In this technique, a soluble protein is titrated
with aliquots of a soluble ligand. The heat evolved during ligand
addition serves as a reporter signal for binding that is decon-
voluted to yield a binding constant, in turn relatable to the free
energy of binding. The technique also evaluates binding
enthalpies directly: this measure, in conjunction with the free
energy of binding, provides an entropy of binding. Evaluation
of ∆H as a function of temperature yields the change in molar
heat capacity accompanying binding, ∆Cp, which in turn allows
dissection of both binding enthalpies and entropies into con-
tributions arising from specific molecular events.61-64
Calorimetrya Agglutination
ligand
Keq (M-1 Nb ∆Gc ∆Hc T∆Sc IC50 (µM) potencyd
)
Me R-Man
7 700
11 820
18 782
1
1
1
-5.3 -6.8 -1.5
-5.5 -6.4 -0.9
-5.8 -8.9 -3.1
520
280
380
630
17
1
11
13
15
17
19
2
1.4
0.8
31
13
3 734 0.8 -4.9 -8.6 -3.7
9 640
7 504
1
1
-5.4 -4.9 +0.5
-5.3 -3.8 +1.5
41
a All calorimetric values are in terms of mannose equivalents.
b Stoichiometry of binding, carbohydrate:protein. c In kcal mol-1 at 298
K. d Values corrected for valency and expressed relative to methyl R-D-
mannopyranoside.
The bindings of mono-, di-, tri-, tetra-, and hexavalent ligands
11, 13, 15, 17, and 19 to dimeric concanavalin A were evaluated
by both titration microcalorimetry and hemagglutination assay
(Table 1). By agglutination assay, tetra- and hexaValent ligands
demonstrate significant cluster glycoside effects. As has been
previously observed, the magnitude of the effect depends
exquisitely on the size and shape of the ligands. Thus, while
no multivalency effect is observed for bi- or trivalent ligands,
tetra- and hexavalent ligands show cluster glycoside effects of
30- and 17-fold, respectively.
(51) Williams, B. A.; Chervenak, M. C.; Toone, E. J. J. Biol. Chem.
1992, 267, 22907.
(52) Weatherman, R. V.; Hortell, K. H.; Chervenak, M. C.; Kiessling,
L. L.; Toone, E. J. Biochemistry 1996, 35, 9-3624.
(53) Chervenak, M. C.; Toone, E. J. Bioorg. Med. Chem. 1996, 4, 1963-
1977.
(54) Gupta, D.; Cho, M.; Cummings, R. D.; Brewer, C. F. Biochemistry
1996, 35, 15236.
(55) Mandal, D. K.; Bhattacharyya, L.; Koenig, S. H.; Brown, R. D.,
III; Oscarson, S.; Brewer, C. F. Biochemistry 1994, 33, 1157.
(56) Sigurskjold, B. W.; Bundle, D. R. J. Biol. Chem. 1992, 267, 8371.
(57) Sigurskjold, B. W.; Altman, E.; Bundle, D. R. Eur. J. Biochem.
1991, 197, 239.
Calorimetric evaluation of binding provides a markedly
different picture. Titration microcalorimetry of all multivalent
(58) Schwarz, F. P.; Puri, K. D.; Bhat, R. G.; Surolia, A. J. Biol. Chem.
1993, 268, 7668.
(59) Schwarz, F. P.; Puri, K.; Surolia, A. J. Biol. Chem. 1991, 266, 24344.
(60) Bains, G.; Lee, R. T.; Freire, E. Biochemistry 1992, 31, 12624.
(61) Chervenak, M. C.; Toone, E. J. J. Am. Chem. Soc, 1994, 116, 10533
(62) Isbister, B. D.; St. Hilaire, P. M.; Toone, E. J. J. Am. Chem. Soc.
1995, 117, 12877.
(63) Oas, T. G.; Toone, E. J. AdV. Biophys. Chem. 1997, 6, 1.
(64) Chervenak, M. C.; Toone, E. J. Biochemistry 1995, 34, 5685.

10290 J. Am. Chem. Soc., Vol. 121, No. 44, 1999
Dimick et al.
Scheme 3
Scheme 4
to curve shape. An irreversible aggregation subsequent to, i.e.,
coupled with, ligand binding would make the apparent protein-
carbohydrate binding constant infinite and result in a square
curve. If aggregation/irreversible precipitation was slow on the
titration time scale, the value c, which in turn determines the
curve shape, would vary continuously during the titration,
producing a curve that could not be deconvoluted for a single
value of Keq. Apparently then, aggregation/precipitation does
not remove a significant amount of protein from solution during
the course of calorimetric titration.
The thermodynamics of multivalent ligand binding provide
insights into the physical processes evaluated by calorimetry
and hemagglutination. Table 1 reports values of binding free
energies, enthalpies, and entropies averaged over the total
number of monosaccharides in each multivalent ligand. The
narrow range of binding free energies provides binding curves
that are well fit by a single site model, despite the wide disparity
in observed binding enthalpies. It is possible to separate the
contributions of individual microscopic binding events of a
multivalent ligand to the reported average value. Consider
binding of the bivalent ligand 13 (Scheme 4). The observed
thermodynamic parameters are the averages of the two micro-
scopic constants ∆J1 and ∆J2, where ∆J represents the change
in enthalpy, entropy, free energy, or heat capacity accompanying
binding. By assuming ∆J1 is equivalent to that of the monovalent
ligand 11, the values of ∆J2 are accessible. Similarly, micro-
scopic thermodynamic constants for the tetra- and hexavalent
ligands can be extracted.65 Here, absent discrete values for ∆J3
Figure 1. Calorimetric data for titration of concanavalin A, 0.51 mM,
with hexavalent ligand 19, 24.9 mM. Both protein and ligand were
dissolved in buffer conisisting of 50 mM dimethylglutarate, 250 mM
NaCl, 1 mM CaCl2, and 1 mM MnCl2 adjusted to pH 5.2. Top, raw
(power vs time) data; bottom, integrated heat vs molar ratio of ligand.
Solid line shows best fit of data using a one-site model: n ) 0.96 (
0.02; K ) 6900 ( 500; ∆H ) -4.0 ( 0.1 kcal/mol; ø2 ) 14. A fit to
a two-site model does not provide a statistically superior fit.
ligands yields curves indicative of simple reversible binding
(Figure 1). This behavior is somewhat surprising, given that
multivalent ligands aggregate and precipitate multivalent lectins.
Indeed, examination of the contents of the calorimeter cell
following all titrations showed evidence of aggregation; typically
solutions were cloudy. This aggregation is apparently not
reported by the calorimetric experiment, at least with respect
(65) Because thermodynamic parameters are state functions, the order
of binding does not affect this analysis; we do not imply any particular
kinetic progression of binding of any of the multivalent ligands.

Cluster Glycoside Effects and ConcanaValin A
J. Am. Chem. Soc., Vol. 121, No. 44, 1999 10291
Table 2. Thermodynamic Parameters for the Binding of
Successive Saccharide Moieties of Multivalent Ligands 11, 13, 15,
17, and 19
binding of ligand
∆G
∆H
T∆S
1
2
3/4
5/6
-5.5
-6.1
-5.0
-5.1
-6.4
-11.4
-0.9
-0.9
-5.3
+4.1
+3.5
-1.6
and ∆J5, we can obtain only average values for ∆J3/∆J4 and
for ∆J5/∆J6. The values for each set of microscopic constants
are shown in Table 2.
An even cursory examination of the data shows an immediate
and obvious trend: IC50 Values correlate with calorimetrically
deriVed entropies of ligand binding (i.e., negatiVely with
enthalpies of ligand binding) but not free energies of ligand
binding. It is difficult to unambiguously assign a physical basis
for this observation; two processes might reasonably result in
the observed thermodynamic pattern. First, aggregation could
alter the thermodynamics of protein-carbohydrate binding but
in a compensating fashion. A reduction in favorable binding
enthalpy would be accompanied by an increase in favorable
entropy of binding. While such compensating events have been
observed for a wide range of interacting systems and are indeed
almost a hallmark of association events in aqueous solution, a
physical explanation for such compensating behavior is not
obvious.66 Furthermore, a number of previously reported en-
thalpy-entropy compensation events were subsequently shown
to be the result of correlated errors; Grunwald has written on
the dangers of invoking compensation arguments in the absence
of a clear mechanistic rationale for doing so.67 Thus, while such
an explanation cannot be ruled out, the prescription of Occam’s
razor makes a compensation argument unattractive. Alterna-
tively, an aggregation process, either following or concomitant
with ligand binding and typified by a large positive enthalpy,
would produce the observed pattern of enthalpy-entropy
compensation; the enthalpy of aggregation masks a portion of
the enthalpy of protein-carbohydrate binding. Because of the
experimental design, the free energy change for the aggregation
process is not reported, provided it does not alter the free energy
of protein-carbohydrate binding. A simple linear correlation
between IC50 and ∆H/∆S is not immediately apparent because
a quantitative estimate of the extent of aggregation is unavail-
able.
To assess the validity of our hypothesis, dynamic light
scattering experiments were performed. At pH 7.0 in the
presence of tetravalent ligand, 0.2 mM succinyl concanavalin
A showed an apparent molecular weight of 4-6 MDa. In the
presence of hexavalent ligand, the apparent molecular weight
rises to 6-8 MDa; in both instances the profile is broad,
indicative of a range of species. We note parenthetically that
examination of succinyl concanavalin A without added ligand
yielded results similar to those obtained with native unmodified
protein, which shows a hydrodynamic radius of 4 nm and an
apparent molecular weight of 100 kDa. This result is at odds
with other solution studies that the succinylated protein is
dimeric and emphasizes the caution with which solution studies
must be treated. Together, these observations suggest that the
behavior of polyvalent ligands is more complicated than
previously appreciated. Under all circumstances, a range of
oligomeric species is present, and the detailed structure of the
aggregate depends on the precise experimental conditions.
Figure 2. Quality of the data. The Fo - Fc difference electron density
map is calculated with phases from a model that had never included
the bivalent ligand. The bivlaent ligand is shown in stick form in its
final refined position (red, oxygen; blue, nitrogen; yellow, carbon).
Hydrogen atoms are not included in the figure. The map (blue wire) is
contoured at 3σ above the mean.
Assuming the observed enthalpy-entropy compensation
results from an endothermic, entropically driven aggregation
process masking the enthalpy of protein-carbohydrate binding,
the negative correlation between agglutination IC50 values and
the apparent enthalpy of ligand binding suggests that the
agglutination assay does not eValuate the strength of protein-
carbohydrate binding. Rather, this assay measures the ability
of a polyvalent saccharide ligand to drive aggregation processes.
Accordingly, all “binding” constants obtained by hemaggluti-
nation should be regarded cautiously. Interpretation of ag-
glutination results in terms of protein-carbohydrate affinity is
not warranted in light of the results presented here: 30-fold
variation in IC50 values are observed with no change in protein-
carbohydrate affinity. We note that this caution is not to suggest
that IC50 results are not of use in the study of protein-
carbohydrate interaction. Indeed, in many respects hemagglu-
tination studies are a far more relevant measure of actiVity than
are assays designed exclusively to evaluate protein-carbohy-
drate binding. It is clear, however, that attempts to interpret
agglutination data in terms of affinity, in the normal sense of
the term, can be dangerously misleading.
We note additionally that calorimetry is not the only
experimental technique capable of observing the discrepancies
we have noted here; indeed, any technique that exclusively
observes protein-carbohydrate binding uncoupled to subsequent
aggregation and precipitation events should report values
identical to those reported in Table 1. Such techniques, including
NMR and fluorescence titrations, have previously been utilized
for the assay of protein-carbohydrate affinity, although not with
multivalent ligands. Kiessling and co-workers recently reported
evaluation of the binding of a series of multivalent mannosides
to concanavalin A using surface plasmon resonance.68 In these
studies, enhancements in affinity observed for high-valent
ligands in agglutination assays were greatly reduced in precipitin
assays but were still observed. The longest of these ligands are
capable of spanning two sites, although enhancements in activity
are observed for compounds of lesser lengths. Whether the
ligands used in the Kiessling study will continue to show
enhanced affinity in other assays, such as those used here, is
unclear.
(66) Lumry, R.; Rajender, S. Biopolymers 1970, 9, 1125.
(67) Grunwald, E.; Comeford, L. L. In Protein-SolVent Interact;
Gregory, R., Ed.; Dekker: New York, 1995; pp 421-43.
(68) Mann, D. A.; Dannai, M.; Maly, D. J.; Kiessling, L. L. J. Am. Chem.
Soc. 1998, 120, 10370.

10292 J. Am. Chem. Soc., Vol. 121, No. 44, 1999
Dimick et al.
Figure 3. Cross-link between concanavalin A monomers. The monomers are shown as backbone traces (red and blue) and the ligand in a space-
filling representation (atomic color scheme as in Figure 3). Both the extended conformation of the ligand and the cis/trans orientation about the
amide linkages are obvious. Cross-linked protein monomers do not make substantial contact with each other.
Structure of Concanavalin A-Polyvalent Ligand Com-
plexes. Aggregation and precipitation effects uniformly preclude
X-ray analysis of polyvalent protein-carbohydrate complexes.
In an attempt to circumvent this limitation, we cocrystallized
succinyl concanavalin A with ligands 13, 15, 17, and 19.
Although it is well known that succinylation results in dimeric
protein at all pH values, no crystal structure of this derivative
has been reported. Remarkably, cocrystals of all four ligands
were obtained. To date, we have diffracted cocrystals of bivalent
and trivalent ligands. Details of the crystallization, data collec-
tion, and molecular replacement have been reported elsewhere.69
The divalent complex has space group C2221, with cell
dimensions a ) 99.1 Å, b ) 127.4 Å, c ) 118.9 Å, and diffracts
to 2.66 Å. The crystallographic asymmetric unit contains a dimer
(68% solvent). Overall anisotropic thermal parameter and bulk
solvent corrections were applied to the data using standard CNS
protocols for refinement.70 The starting R-free was 32% for all
data between 26 and 2.66 Å. Restrained refinement of positional
and thermal parameters with CNS reduced the R-free to 25%.
Noncrystallographic restrains were applied to thermal and
positional parameters throughout the refinement. At this stage,
clear density was visible for the ligand (Figure 3) and metal
ions (2Mn2+ and 2Ca2+). These were included in the model
and refinement continued.
In both instances, concanavalin A crystallizes in tetrameric
form despite succinylation: apparently, crystal packing forces
compensate for the repulsive interactions introduced by succi-
nylation. Clear density could be seen for a succinyl group on
Lys 101 in one monomer and was included in the model.
Density for Lys 101 in the other monomer was visible but not
clear enough to build a model of the succinyl group. No other
succinyl groups were visible, raising the issue of a molecular
mechanism for inhibition of tetramerization by protein deriva-
tization. Presumably, other lysine residues are incompletely
succinylated: in aggregate these acylations disfavor dimer
dimerization.
ligands (Figure 4). The structure also allows crystallographic
examination of a saccharide-containing dendron, the first such
structure of which we are aware. The ligand geometry was
defined using the Engh and Huber compendium of bond
distances.71 Statistics on the final refined model are given in
Table 3. The absolute value of R-free (20%) is probably
artificially low due to noncrystallographic symmetry; however,
the trend (decrease upon refinement) validates our refinement
protocol. The loop between residues 118 and 122 is disordered
in this structure, as in all other con A structures. As expected,
the core of the dendrimer structure exists in a planar arrange-
ment. The amides are arranged in a cis/trans orientation,
providing a pseudo-C2 axis of symmetry between the two
saccharide residues. Both saccharides occupy concanavalin A
binding sites in identical fashions, essentially as the monosac-
charide (Figure 2). In both instances, the linker aglycon is
oriented in a -60° conformer, in accordance with the general
observation that carbohydrates are bound by lectins at or near
their HSEA minimum energy conformation.10 The final coor-
dinates and structure factors have been deposited with the
Protein Data Bank (PDB coordinate entry 1qgl).72
The trivalent complex has the same space group and very
similar cell dimensions: a ) 99.4 Å, b ) 127.2 Å, c ) 118.7
Å. A full data set to 3.0 Å was collected on the trivalent ligand
con A complex. Molecular replacement and refinement pro-
ceeded in a manner identical to that described for the divalent
complex above. However, no density was visible for the third
sugar, and as all the monosaccharide binding sites of con A in
the crystal are occupied, we concluded that this part of the ligand
is disordered. In conjunction with the observation of a less than
1:1 stoichiometry during calorimetric experiments, this observa-
tion raises the possibility that the soluble aggregates formed in
solution pack in a fashion similar to the insoluble aggregates
observed during diffraction. Binding of only two carbohydrates
of a trivalent ligand would give rise to a binding stoichiometry
of 0.67. The value of 0.8 observed here suggests that the
aggregates are likely of lower order than the crystal form.
Nonetheless, the low stoichiometry is most reasonably inter-
The supramolecular structure consists of infinite plates
extending in two dimensions and cross-linked by bivalent
(69) Moothoo, D. N.; McMahon, S. M.; Dimick, S. M.; Toone, E. J.;
Naismith, J. H. Acta Crystallogr. 1998, D54, 1023.
(70) Adams, P. D.; Pannu, N. S.; Read, R. J.; Brunger, A. T. Proc. Natl.
Acad. Sci. U.S.A. 1997, 94, 5018.
(71) Engh, R. A.; Huber, R. Acta Crystallogr. 1991, A47, 392.
(72) Bernstein, F. C.; Koetzle, T. F.; Williams, G. J. B.; Myer, E. F.,
Jr.; Brice, M. D.; Rodgers, J. R.; Kennard, O.; Shimanouchi, T.; Tasumi,
M. J. Biol. Chem. 1977, 112, 535.

Cluster Glycoside Effects and ConcanaValin A
J. Am. Chem. Soc., Vol. 121, No. 44, 1999 10293
Table 3. Crystallographic Data Collection Statistics and
Refinement Statistics
no. of unique reflections
21 182
completeness of data (%) (26.0-2.66 Å/2.75-2.66 Å) 96.4/98.3
Rmergea (I) (%) (26.0-2.66/2.75-2.66)
average data redundancy (26.0-2.66 Å/2.75-2.66 Å)
% of data > 1σ (26.0-2.6 Å/2.75-2.66 Å)
refinement
7.2/20.7
2.5/2.2
94/83
resolution range (Å)
26-2.66
20.0 (25.7)
17.6 (23.9)
0.008
R-freeb % (uncorrected)c
R-factor % (uncorrected)
bond rms deviation (Å)d
angle rms deviation (deg)d
1.70
noncrystallographic symmetry rms deviation
0.10
(CR atoms) (Å)
B-factor bonded atoms rms deviation (Å2)e
Ramachandran core/additional (%)f
average B-factor (Å2)
1.75
86.7/12.8
28.9
no. of ordered protein atoms
no. of solvent molecules
total no. of ordered non-hydrogen atoms
3624
72
3625
a Rmerge(I) ) ShklSi|Ii - I(hkl)|/ShklSiIi(hkl). b R-free is the crystal-
lographic residual calculated on 10% of data excluded during refine-
ment. c Uncorrected is without CNS anisotropic and bulk solvent
corrections. d Root-mean-square deviation from Engh and Huber ideal
values.71 e Calculated with MOLEMAN (G. J. Kleywegt, unpublished
program). f Core and additionally allowed regions as defined by
PROCHECK.70 One disordered residue is in a generously allowed
region, and no residues are in disallowed regions.
benzophenone ketyl. All other chemicals were reagent grade and used
without further purification. Column chromatography was performed
with flash grade silica gel. Thin-layer chromatography plates were Silica
1
Gel 60 F254 (Merck) and visualized with Hanessian stain. H NMR
spectra were recorded on a GE QE-300, Varian 500, or Varian 600
instrument operating at 300.150, 500.137, or 599.892 MHz, respectively,
in either CDCl3 or D2O referenced to TMS or TSP at 0 ppm,
respectively. 13C NMR spectra were recorded on a GE QE-300 or
Varian 600 instrument operating at 75.48 or 150.86 MHz, respectively,
in CDCl3 referenced to CDCl3 at 77.0 ppm. Mass spectrometry was
performed on a JEOL JMS-SX102A instrument.
Synthesis of Ligands. Trimethyl 1,3,5-Benzenetricarboxylate (1).
Benzenetricarboxylic acid (8.0 g, 38 mmol) was dissolved in methanol
(140 mL). Concentrated sulfuric acid (2 mL) was added. The solution
was refluxed for 24 h. Solvent was removed under reduced pressure,
the residue was dissolved in chloroform (150 mL) and washed with
saturated bicarbonate (200 mL), and the solvent was removed under
reduced pressure to give the desired product 1 as a white powder (9.15
g, 95%): mp 144-144.5 °C (ref 145-147 °C). 1H and 13C NMR spectra
were identical to those previously reported.73
Figure 4. Crystal packing arrangement of the complex. Concanavalin
A is represented in a backbone trace. The cross-linking ligands have
been omitted for clarity. The standard con A tetramer is clearly visible
and forms an infinite two-dimensional array via the cross-linking
ligands. The crystal is formed of offset stacks of two-dimensional arrays.
preted in terms of some form of local order in aggregate
formation that precludes binding of the third saccharide of a
trivalent ligand, even in solution.
Dimethyl 1,3,5-Benzenetricarboxylate (2). Trimethyl 1,3,5-ben-
zenetricarboxylate (1.11 g, 4.40 mmol) was dispersed in MeOH (100
mL). Aqueous NaOH (3.93 mL of 1.0 M, 3.93 mmol, 0.9 equiv) was
added. The suspension was stirred vigorously and slowly dissolved
during 8 h. After 18 h, solvent was removed under reduced pressure.
Dichloromethane (75 mL) was added. The organic phase was washed
with saturated NaHCO3 (2 × 75 mL), dried (MgSO4), and evaporated
to give unreacted starting material (260 mg, 23%). The aqueous washes
were acidified (pH 2.0) with concentrated HCl, yielding a milky white
suspension that was washed with ethyl acetate (2 × 75 mL). The
combined organic phases were washed with brine (1 × 30 mL) and
dried (MgSO4), and the solvent removed to give a 2 as a white powder
(800 mg, 76%): 1H NMR (300 MHz, CDCl3) δ 8.91 (s, 2H), 8.90 (s,
1H), 3.99 (s, 6H) ppm; 13C NMR (75 MHz, CDCl3) δ 169.9, 165.3,
135.2, 135.0, 132.4, 130.4, 52.8 ppm; IR (KBr pellet) ν 3284 (broad),
The conclusions of the work presented here are clear. A range
of species is formed when multivalent ligands are bound to
multivalent lectins, provided the lectins orient binding sites in
several directions and are capable of forming cross-links. Those
observations of high “affinity” of multivalent glycoconjugates
rather reflect the propensity of such ligands to form aggregates,
as opposed to any enhancement in actual protein-carbohydrate
affinity. Finally, IC50 values from agglutination assays correlate
with entropiessnot free energiessof ligand binding; agglutina-
tion assays should not be considered as a measure of protein-
carbohydrate affinity. We continue our studies on the role of
multivalency in protein-carbohydrate interaction and will report
our results in due course.
1735, 1697 cm-1
.
Dimethyl 5-Hydroxymethylbenzene-1,3-dicarboxylate (3). 1,3-
Dimethylbenzene-1,3,5-tricarboxylate (3.40 g, 14.3 mmol) was dis-
solved in dry THF (24 mL). Borane-methyl sulfide complex (14.3
mL of 2 M solution in THF, 28.6 mmol) was added slowly,
Experimental Section
General. D-Mannose (lot no. 05115CG), benzenetricarboxylic acid,
and benzene-1,3,5-tricarbonyl trichloride were purchased from Aldrich
Chemical and used without further purification. Dichloromethane was
distilled from calcium hydride. THF was distilled from sodium/
(73) Aldrich Library of Fourier Transform NMR; Aldrich Chemical
Co.: Milwaukee, WI, 1989; Vol. 2, 1282C.

10294 J. Am. Chem. Soc., Vol. 121, No. 44, 1999
Dimick et al.
accompanied by effervescence. After 24 h of stirring at 27 °C, methanol
(50 mL) was added. The solution was stirred at 27 °C for 30 min and
concentrated under reduced pressure. The resulting white solid was
dissolved in ethyl acetate (75 mL) and washed with water (75 mL),
saturated NaHCO3 (75 mL), and brine (50 mL). Solvent was removed
under reduced pressure, and the crude product was purified by silica
chromatography (PE:EA 3:2) to give 3 as a white solid (2.82 g, 88%):
1H), 4.06-4.10 (dd, 1 H, J ) 2.4, 12.2 Hz), 4.22-4.28 (dd, 1H, J )
5.4, 12.2 Hz), 4.79 (d, 1H, H-1, J ) 1.4 Hz), 5.20-5.28 (m, 3H) ppm;
13C NMR (75.48 MHz, CDCl3) δ 20.00 (OCOCH3), 20.19 (OCOCH3),
27.89 (OCH2CH2CH2N3), 47.35 (OCH2CH2CH2N3), 61.75, 64.11
(OCH2CH2CH2N3), 65.35, 67.89, 68.30, 68.74, 96.89 (C-1), 169.04
(OCOCH3), 169.23 (OCOCH3), 169.37 (OCOCH3), 169.95 (OCOCH3)
ppm; IR (neat) ν 2955, 2257, 2098, 1751, 1370, 1239, 913, 739, 648
1
mp 104 °C; H NMR (300 MHz, CDCl3) δ 8.58 (t, 1H, J ) 1.5 Hz),
cm-1
.
8.23 (t, 2H, J ) 0.8 Hz), 4.81 (d, 2H, J ) 5.9 Hz), 3.95 (s, 6H), 2.20
(t, 1H, J ) 5.9 Hz) ppm; 13C NMR (75 MHz, CDCl3) δ 166.2, 141.9,
132.0, 130.8, 129.8, 64.2, 52.2 ppm; IR (KBr pellet) ν 3446 (broad),
3-(Benzylamido)propyl 2,3,4,6-Tetra-O-acetyl-r-D-mannopyra-
noside (10). 3-Azidopropyl 2,3,4,6-tetra-O-acetyl-R-D-mannopyranoside
(95 mg, 0.22 mmol) was dissolved in ethanol (7 mL), and DeGussa
Pd/C (30 mg) was added. H2 was bubbled through the stirred reaction
mixture for 1.5 h. TLC (2:1 light petroleum ether:ethyl acetate) revealed
complete disappearance of starting material. Pd/C was removed by
filtration over Celite, and the filtrate was concentrated to yield
3-aminopropyl 2,3,4,6-tetra-O-acetyl-R-D-mannopyranoside (9b). The
viscous oil was resuspended in THF (8 mL), and Et3N (111 µL, 4 equiv)
was added. The solution was stirred at 25 °C, and benzoyl chloride
(56 mg, 2 equiv) was added slowly. The solution was stirred for 16 h
at 25 °C; TLC (1:1 light petroleum ether:ethyl acetate) showed complete
conversion to a product which was both UV-active and stained by
Hannesian dip. The solution was concentrated and resuspended in CH2-
Cl2 (25 mL). The organic phase was washed with 1.0 M HCl (1 × 20
mL), saturated NaHCO3 (1 × 20 mL), and saturated NaCl (1 × 20
mL), dried (MgSO4), and concentrated. The residue was purified by
flash chromatography (2:1 light petroleum ether:ethyl acetate) to yield
an amber-colored oil (55 mg, 55%): 1H NMR (600 MHz, CDCl3) δ
1.94-1.98 (m, 2 H), 1.98 (s, 3H), 2.02 (s, 3H), 2.05 (s, 3 H), 2.14 (s,
3H), 3.57-3.60 (m, 3H), 3.78-3.84 (m, 1H), 3.97-4.03 (m, 1H), 4.05-
4.09 (dd, 1H, J ) 2.6, 12.2 Hz), 4.23-4.29 (dd, 1H, J ) 5.4, 12.2
Hz), 4.81 (d, 1H, J ) 1.4 Hz), 5.22-5.29 (m, 3H), 6.56 (br s, 1H,
NH), 7.38-7.48 (m, 3H), 7.76-7.79 (m, 2H) ppm; 13C NMR (75.48
MHz, CDCl3) δ 20.54, 20.65, 20.84, 29.16, 37.59, 62.55, 66.09, 66.49,
68.45, 69.03, 69.38, 97.65, 126.83, 128.54, 131.40, 134.27, 167.55,
169.67, 169.87, 170.04, 170.67 ppm; IR (neat) ν 3395, 2936, 2255,
1748, 1650, 1537, 1370, 1227, 1083, 1049, 912, 731 cm-1; HR FAB
MS (pos) calcd for C24H31NO11 (MH+) 510.1975, obsd 510.1982.
3-(Benzylamido)propyl r-D-Mannopyranoside (11). 3-(Benzyla-
mido)propyl 2,3,4,6-tetra-O-acetyl-R-D-mannopyranoside (50 mg, 0.098
mmol) was dissolved in 0.1 M NaOMe (4 mL) and stirred at 25 °C for
3 h. The solution was neutralized with Dowex H+ and concentrated to
yield 11 as an amber film (30 mg, 89%): 1H NMR (600 MHz, D2O)
δ 1.94 (m, 2H), 3.46 (t, 2H, J ) 5.6 Hz), 3.51-3.64 (m, 3H), 3.69-
3.72 (dd, 1H, J ) 4.5, 12.1 Hz), 3.76-3.78 (dd, 1H, J ) 3.1, 8.6 Hz)
3.81-3.85 (m, 3H), 3.90 (d, 1H, J ) 1.6 Hz), 4.84 (s, 1H), 7.48-7.52
(m, 2H), 7.61-7.62 (m, 1H), 7.71-7.74 (m, 2H) ppm; 13C NMR (150
MHz, D2O) δ 31.30, 40.28, 64.06, 68.63, 69.98, 73.23, 73.81, 75.86,
102.96, 129.99, 131.88, 135.07, 137.08, 173.78, ppm; FAB MS (pos)
calcd for C16H23NO7 341.1, found 342.1 (MH+), 364.1 (MNa+).
5-Azidomethyl-N,N′-bis[3-O-(2,3,4,6-tetra-O-acetyl-r-D-mannopy-
ranosyl)propyl]benzene-1,3-dicarboxamide (12). 3-Azidopropyl 2,3,4,6-
tetra-O-acetyl-R-D-mannopyranoside (250 mg, 0.58 mmol) was dis-
solved in ethanol (5 mL), and DeGussa Pd/C (70 mg) was added. H2
was bubbled through the reaction solution for 3 h at 25 °C. The Pd/C
was removed by filtration over Celite and the filtrate concentrated to
yield 3-aminopropyl 2,3,4,6-tetra-O-acetyl-R-D-mannopyranoside (9b).
The crude material was resuspended in THF (12 mL), and Et3N (324
µL, 4 equiv) was added. 5-Azidomethyl-1,3-dicarbonylbenzene dichlo-
ride (58 mg, 0.26 mmol, 0.45 equiv) was added and the resulting
solution stirred for 16 h at 25 °C. The reaction mixture was concentrated
and resuspended in CH2Cl2. The organic phase was washed with 1.0
M HCl (1 × 20 mL), saturated NaHCO3 (1 × 20 mL), and brine (1 ×
20 mL), dried (MgSO4), and concentrated. The crude material was
purified via flash chromatography (2:1 light petroleum ether:ethyl
acetate) to yield 12 as an amber oil (120 mg, 53%): 1H NMR (600
MHz, CDCl3) δ 1.86-1.97 (m, 2H), 1.95 (s, 6H), 2.00 (s, 6H), 2.03
(s, 6H), 2.11 (s, 6H), 3.48-3.58 (m, 4H), 3.75-3.79 (m, 2H), 3.96-
3.98 (m, 2H), 4.05-4.08 (m, 2H), 4.21-4.24 (dd, 2H, J ) 5.4, 12.2
Hz), 4.43 (s, 2H), 4.78 (d, 2H, J ) 1 Hz), 5.18-5.24 (m, 6H), 6.98 (t,
2H, J ) 5.8 Hz), 7.88 (s, 2H), 8.17 (s, 1H) ppm; 13C NMR (150.86
MHz, CDCl3) δ 20.69, 20.74, 20.88, 29.33, 37.60, 54.04, 60.41, 62.71,
1723 cm-1
.
Dimethyl 5-Chloromethyl-1,3-benzenedicarboxylate (4). To
dimethyl 5-hydroxymethylbenzene-1,3-dicarboxylate (2.0 g, 8.9 mmol)
was added SOCl2 (1.3 mL). The solution was refluxed under an argon
atmosphere for 1 h. CHCl3 was added to the solution, and the organic
phase was washed with 0.1 M NaOH (2 × 50 mL) and brine (50 mL).
Solvent was removed under reduced pressure to yield 4 as a white
1
powder (2.0 g, 93%): mp 119-120 °C; H NMR (300 MHz, CDCl3)
δ 8.62 (t, 1H, J ) 1.9 Hz), 8.25 (d, 2H, J ) 1.9 Hz), 4.66 (s, 2H), 3.96
(s, 6H) ppm; 13C NMR (75 MHz, CDCl3) δ 165.6, 138.6, 133.7, 131.1,
129.8, 52.3, 44.8 ppm; IR (KBr pellet) ν 1725 cm-1
.
Dimethyl 5-Azidomethylbenzene-1,3-dicarboxylate (5). Dimethyl
5-chloromethylbenzene-1,3-dicarboxylate (2.0 g, 8.2 mmol) was dis-
solved in acetone (30 mL) and water (10 mL). NaN3 (3.21 g, 49.3
mmol, 6 equiv) was added, and the solution was refluxed for 16 h.
Solvent was removed in vacuo, and the residue was dissolved in CHCl3
(75 mL). The organic phase was washed with water (3 × 75 mL) and
brine (50 mL). Solvent was removed under reduced pressure to yield
1
5 as a pale yellow powder (1.99 g, 97%): mp 74-75 °C; H NMR
(300 MHz, CDCl3) δ 8.65 (t, 1H, J ) 1.5 Hz), 8.20 (d, 2H, J ) 1.5
Hz), 4.49 (s, 2H), 3.97 (s, 6H) ppm; 13C NMR (75 MHz, CDCl3) δ
165.1, 135.9, 132.7, 132.3, 130.5, 53.2, 51.9 ppm; IR (KBr pellet) ν
2107, 1722 cm-1
.
5-Azidomethylbenzene-1,3-dicarboxylate (6). Dimethyl 5-azidom-
ethylbenzene-1,3-dicarboxylate (0.77 g, 3.1 mmol) was dissolved in 1
M methanolic KOH (25 mL), and the resulting solution was refluxed
for 1 h. Solvent was removed under reduced pressure, and the residue
was dissolved in water (75 mL) and acidified with concentrated HCl
to pH 2 to give a milky suspension. The suspension was extracted with
ethyl acetate (3 × 50 mL), and the combined organic washes were
reduced to yield 6 as a white powder (0.68 g, 99%): mp 212-214 °C;
1H NMR (300 MHz, acetone-d6) δ 8.52 (t, 1H, J ) 1.6 Hz), 8.15 (d,
2H, J ) 1.6 Hz), 4.59 (s, 2H) ppm; 13C NMR (75 MHz, acetone-d6) δ
165.9, 137.5, 133.2, 131.8, 130.1, 53.3 ppm; IR (KBr pellet) ν 2105,
1702 cm-1
.
5-Azidomethylbenzene-1,3-dicarbonyl Dichloride (7). 5-Azidom-
ethylbenzene-1,3-dicarboxylate (0.2 g, 0.9 mmol) was dissolved in 1
mL of SOCl2 and refluxed for 1 h. Excess thionyl chloride was removed
under reduced pressure to yield 7 as a yellow oil (0.233 g, 100%): 1H
NMR (300 MHz, CDCl3) δ 8.77 (s, 1H), 8.35 (s, 2H), 4.62 (s, 2H)
ppm; 13C NMR (75 MHz, CDCl3) δ 166.9, 138.6, 135.8, 134.9, 133.2,
53.2 ppm; IR (neat) ν 2108, 1762 cm-1
.
3-Azidopropyl 2,3,4,6-Tetra-O-acetyl-r-D-mannopyranoside (9).
2,3,4,6-Tetra-O-acetyl-R-D-mannopyranosyl trichloroacetimidate 8 (930
mg, 1.9 mmol) was dissolved in CH2Cl2 (10 mL) with 4-Å powdered
molecular sieves (1 g). 3-Azidopropanol (948 mg, 9.5 mmol, 5 equiv)
was added, and the reaction mixture was chilled to -30 °C. After the
mixture was stirred for 30 min, TMSOTf (155 µL, 0.75 mmol, 0.4
equiv) was added during 5 min. The reaction mixture was stirred and
allowed to warm to 0 °C during 4 h, at which time TLC (4:1 light
petroleum ether:ethyl acetate) showed complete disappearance of
starting material. Solid NaHCO3 (250 mg) was added, and the reaction
mixture was stirred for 10 min. The mixture was filtered over Celite
and concentrated. The crude material was purified by flash chroma-
tography over silica (5:1 light petroleum ether:ethyl acetate) to yield 9
as a viscous, colorless oil (680 mg, 81%): 1H NMR (300 MHz, CDCl3)
δ 1.77-1.88 (m, 2H, OCH2CH2CH2N3), 1.96 (s, 3H, -OCOCH3), 2.02
(s, 3H, -OCOCH3), 2.08 (s, 3H, -OCOCH3), 2.13 (s, 3H, -OCOCH3),
3.42 (t, 2H, J ) 6.4 Hz, OCH2CH2CH2N3), 3.48-3.51 (m, 1H, -OCH2-
CH2CH2N3), 3.72-3.79 (m, 1H, -OCH2CH2CH2N3), 3.80-3.93 (m,

Cluster Glycoside Effects and ConcanaValin A
J. Am. Chem. Soc., Vol. 121, No. 44, 1999 10295
66.28, 68.57, 69.21, 69.53, 97.77, 125.11, 129.57, 135.47, 137.03,
166.51, 169.79, 170.12, 170.20, 170.88 ppm; IR (film) ν 3413, 3058,
2959, 2101, 1745, 1651, 1538, 1434, 1370, 1260, 1136, 1084, 1049,
739 cm-1; HR FAB MS (pos) calcd for C43H57N5O22 (MH+) 996.3573,
obsd 996.3583.
resuspended in CH2Cl2. The organic phase was washed with 1.0 M
HCl (1 × 20 mL), saturated NaHCO3 (1 × 20 mL), and brine (1 × 20
mL), dried (MgSO4), and concentrated. The crude material was purified
by silica chromatography (1:1:0.5 light petroleum ether:ethyl acetate:
methanol) to yield 16 as a pale yellow foam (85 mg, 40%): 1H NMR
(500 MHz, CDCl3) δ 1.93 (m, 8H), 1.97 (s, 12 H), 2.03 (s, 12H), 2.06
(s, 12H), 2.13 (s, 12H), 3.51-3.54 (m, 12H), 3.80 (m, 4H), 3.99 (m,
4H), 4.08-4.12 (m, 4H), 4.23-4.26 (m, 4H), 4.41-4.43 (m, 4H), 4.47
(s, 2H), 4.80 (s, 4H), 5.20-5.26 (m, 12H), 6.97 (br s, 2H), 7.37 (br s,
4H), 7.74-7.87 (m, 9H) ppm; 13C NMR (75.48 MHz, CDCl3) δ 20.67,
20.72, 20.87, 29.12, 29.65, 33.84, 37.31, 49.08, 54.06, 62.46, 65.98,
66.02, 66.13, 68.38, 69.10, 69.45, 97.59, 129.08, 134.60, 137.58, 139.12,
156.94, 167.13, 169.69, 170.08, 170.17, 170.20, 170.80 ppm; IR (neat)
ν 3383, 2936, 2254, 2102, 1746, 1650, 1538, 1432, 1370, 1225, 1050,
909, 730, cm-1; FAB MS (pos) calcd for C95H121N9O46 2124.7, obsd
2125.7 (MH+).
5-Azidomethyl-N,N′-bis[N,N′-bis[3-O-(R-D-mannopyranosyl)pro-
pyl]-1-phenylmethyl-3,5-dicarboxamide]benzene-1,3-dicarboxam-
ide (17). 5-Azidomethyl-N,N′-bis[N,N′-bis[3-O-(2,3,4,6-tetra-O-acetyl-
R-D-mannopyranosyl)propyl]-1-phenylmethyl-3,5-dicarboxamide]benzene-
1,3-dicarboxamide (50 mg, 0.023 mmol) was dissolved in 0.1 M
NaOMe (5 mL) and stirred at 25 °C for 24 h. The reaction was
neutralized with Dowex H+ and concentrated. The crude residue was
purified by size-exclusion chromatography (Sephadex G-25 resin) to
yield 17 as clear oil (24 mg, 63%): 1H NMR (600 MHz, D2O) δ1.76-
1.78 (m, 4H), 1.87-1.89 (m, 4H), 3.30-3.33 (m, 8H), 3.39-3.82 (m,
32 H), 4.34-4.57 (m, 6H), 4.77-4.78 (m, 4H), 7.70 (s, 2H), 7.73 (s,
2H), 7.77 (s, 2H), 7.82 (s, 1H), 7.87 (s, 1H), 7.89 (s, 1H) ppm; 13C
NMR (150 MHz, D2O) δ 26.35, 31.28, 40.35, 46.17, 64.06, 68.47,
69.98, 73.27, 73.87, 75.88, 102.97, 127.66, 132.03, 133.09, 137.86,
172.02 ppm; MALDI-TOF MS calcd for C63H89N9O30 1451.6, obsd
1474.5 (MNa+).
5-Azidomethyl-N,N′-bis[3-(O-r-D-mannopyranosyl)propyl]ben-
zene-1,3-dicarboxamide (13). 5-Azidomethyl-N,N′-bis[3-O-(2,3,4,6-
tetra-O-acetyl-R-D-mannopyranosyl)propyl]benzene-1,3-dicarboxam-
ide (120 mg, 0.12 mmol) was dissolved in 0.1 M NaOMe (10 mL) and
stirred at 25 °C for 2 h. The solution was neutralized with Dowex H+
resin and concentrated. The crude material was purified by flash
chromatography (1:1:0.1 ethyl acetate:ethanol:H2O) to yield 13 as a
white foam (63 mg, 79%): 1H NMR (600 MHz, D2O) δ 1.93-1.95
(m, 4 H), 3.48-3.52 (t, 4H, J ) 6.5 Hz), 3.58-3.62 (m, 6H), 3.69-
3.72 (dd, 2H, J ) 4.3, 11.9 Hz), 3.75-3.77 (dd, 2H, J ) 3.3, 9.0 Hz),
3.80-3.84 (m, 3H), 3.90-3.91 (dd, 2H, J ) 1.8, 3.2 Hz), 4.55 (s, 2
H), 4.83 (s, 2 H), 7.88 (s, 2H), 8.03 (s, 1H) ppm; 13C NMR (150 MHz,
D2O) δ 31.26, 40.42, 56.60, 64.08, 68.57, 69.98, 73.26, 73.83, 75.89,
102.97, 128.41, 132.78, 138.22, 140.44, 172.26 ppm; FAB MS (pos)
calcd for C27H41N5O14 659.3, found 660.2 (MH+), 682.2 (MNa+).
N,N′,N′′-Tris[3-O-(2,3,4,6-tetra-O-acetyl-r-D-mannopyranosyl)-
propyl]benzene-1,3,5-tricarboxamide (14). 3-Azidopropyl 2,3,4,6-
tetra-O-acetyl-R-D-mannopyranoside (220 mg, 0.51 mmol) was dis-
solved in ethanol (5 mL). DeGussa Pd/C was added, and H2 was
bubbled through the reaction mixture. Pd/C was removed by filtration
over Celite, and the filtrate was concentrated. Crude 9b was redissolved
in THF, and Et3N (554 µL, 8 equiv) was added. After 20 min 1,3,5-
benzenetricarbonyl trichloride (45 mg, 0.31 equiv) was added, and the
reaction mixture was stirred for 16 h at 25 °C. The solution was
concentrated, resuspended in CH2Cl2, washed with 1.0 M HCl (1 ×
20 mL), saturated NaHCO3 (1 × 20 mL), and brine (1 × 20 mL),
dried (MgSO4), and concentrated. Crude material was purified via flash
chromatography (2:1:0.2 light petroleum ether:ethyl acetate:methanol)
to yield trivalent mannoside 14 as a white foam (75 mg, 32% yield for
two steps): 1H NMR (600 MHz, CDCl3) δ 1.97 (m, 6H), 1.98 (s, 9H),
2.01 (s, 9H), 2.06 (s, 9H), 2.15 (s, 9H), 3.52-3.56 (m, 9H), 3.65-
3.69 (m, 3H), 3.79-3.84 (m, 3H), 4.06-4.08 (m, 3H), 4.12-4.15 (m,
3H), 4.25-4.29 (dd, 3H, J ) 5.4, 12.2 Hz), 4.81 (s, 3H), 5.23-5.28
(m, 9H), 6.97 (br s, 3H), 8.38 (s, 3H) ppm; 13C NMR (75.48 MHz,
CDCl3) δ 20.54, 20.66, 20.74, 20.80, 20.88, 29.32, 37.31, 62.49, 62.64,
65.97, 66.08, 66.19, 68.29, 68.33, 68.72, 69.06, 69.49, 70.00, 97.70,
128.27, 135.03, 165.95, 169.06, 169.74, 169.92, 169.94, 170.14, 170.74,
170.96 ppm; IR (neat) ν 3413 (br), 1739, 1658, 1537, 1434, 1370, 1226,
1083, 1049, 738 cm-1; HR FAB+ MS calcd for C60H81N3O33 (MH+)
1372.4831, obsd 1372.4768.
N,N′,N′′-Tris[3-O-(r-D-mannopyranosyl)propyl]benzene-1,3,5-tri-
carboxamide (15). N,N′,N′′-Tris[3-O-(2,3,4,6-tetra-O-acetyl-R-D-man-
nopyranosyl)propyl]benzene-1,3,5-tricarboxamide (80 mg, 0.06 mmol)
was dissolved in 0.1 M NaOMe (3 mL). The solution was stirred at 25
°C for 24 h and then neutralized with Dowex H+ resin. The resin was
removed by gravity filtration, and the solution was concentrated. The
crude material was purified by size-exclusion chromatography (Sepha-
dex G-10 resin) to yield 15 as a white foam (32 mg, 64%): 1H NMR
(600 MHz, D2O) δ 1.96-1.98 (m, 6H), 3.51-3.54 (t, 6H, J ) 6.7
Hz), 3.59-3.64 (m, 9H), 3.70-3.73 (dd, 3H, J ) 4.2, 12.1 Hz), 3.75-
3.77 (dd, 3H, J ) 2.6, 8.1 Hz), 3.81-3.86 (m, 9H), 3.91 (s, 3H), 4.85
(s, 3H), 8.24 (s, 3H) ppm; 13C NMR (150 MHz, D2O) δ 31.27, 40.51,
64.11, 68.60, 70.00, 73.27, 73.84, 75.92, 102.98, 131.59, 138.31, 171.79
ppm; FAB MS (pos) calcd for C36H57N3O21 867.3, found 890.3 (MNa+),
726.3, 562.3.
5-Azidomethyl-N,N′-bis[N,N′-bis[3-O-(2,3,4,6-tetra-O-acetyl-r-D-
mannopyranosyl)propyl]-1-phenylmethyl-3,5-dicarboxamide]ben-
zene-1,3-dicarboxamide (16). Peracetylated bivalent dendron 12 (200
mg, 0.2 mmol) was dissolved in ethanol (10 mL), and Pd/C (35 mg)
was added. H2 was bubbled through the reaction mixture. Upon
complete disappearance of starting material, catalyst was removed by
vacuum filtration over Celite, and the filtrate was concentrated. Reduced
dendron 12b was resuspended in THF, and Et3N (112 µL, 4 equiv)
was added. After 10 min, 5-azidomethyl-1,3-dicarbonylbenzene dichlo-
ride (25 mg, 0.45 equiv) was added. The reaction mixture was stirred
at 25 °C for 16 h. Solvent was removed in vacuo, and the residue was
N,N′,N′′-Tris[N,N′-bis[3-O-(2,3,4,6-tetra-O-acetyl-r-D-mannopy-
ranosyl)propyl]-1-phenylmethyl-3,5-dicarboxamide]benzene-1,3,5-
tricarboxamide (18). Peracetylated bivalent dendron 12 (190 mg, 0.19
mmol) was dissolved in ethanol (6 mL), and DeGussa Pd/C (45 mg)
was added. H2 was bubbled through the reaction mixture during 2 h.
Pd/C was removed by filtration over Celite, and the filtrate was
concentrated. The residual oil 12b (150 mg) was dissolved in THF,
and Et3N (216 µL, 10 equiv) was added. 1,3,5-Benzenetricarbonyl
trichloride (13 mg, 0.065 mmol, 0.33 equiv) was added and the reaction
mixture refluxed for 18 h. The solution was concentrated and redis-
solved in CH2Cl2, washed with 1.0 M HCl (1 × 20 mL), saturated
NaHCO3 (1 × 20 mL), and brine (1 × 20 mL), dried (MgSO4), and
concentrated. The crude product was purified by silica chromatography
(4:1:0.5 ethyl acetate:ethanol:H2O) to yield 18 as a gold oil (85 mg,
54%): 1H NMR (400 MHz, CDCl3) δ 1.98-2.16 (m, 84H), 3.55-
3.61 (m, 18H), 3.82 (m, 6H) 3.97-4.08 (m, 6H), 4.11-4.13 (m, 6H),
4.24-4.28 (m, 6H), 4.82-4.83 (m, 6H), 5.23-5.30 (m, 18H), 7.83-
8.52 (m, 12H) ppm; 13C NMR (75.48 MHz, CDCl3) δ 20.69, 20.87,
29.09, 29.17, 29.23, 29.65, 37.36, 37.41, 37.68, 62.45, 62.51, 62.55,
62.58, 62.62, 62.68, 66.03, 66.03, 66.09, 66.21, 66.24, 68.34, 68.36,
68.75, 69.02, 69.06, 69.16, 69.20, 69.43, 70.00, 97.66, 97.68, 127.13,
128.34, 131.06, 132.47, 134.23, 135.81, 136.19, 169.74, 169.98, 170.13,
170.15, 170.77, 170.79, 170.91 ppm; FAB MS (pos) calcd for
C
138N9O69H177 3064.2, obsd 3066.4 (MH+).
N,N′,N′′-Tris[N,N′-bis[3-O-(r-D-mannopyranosyl)propyl]-1-phe-
nylmethyl-3,5-dicarboxamide]benzene-1,3,5-tricarboxamide (19).
Peracetylated hexavalent mannoside 18 (52 mg, 0.017 mmol) was
dissolved in 0.1 M NaOMe (15 mL) and stirred at 25 °C for 36 h. The
solution was neutralized with Dowex H+ and concentrated. The crude
product was applied to a G-50 Sephadex column and eluted with H2O
to yield 19 as a white foam (26 mg, 75%): 1H NMR (600 MHz, D2O)
δ 1.81-1.82 (m, 6H), 1.92-1.94 (m, 6H), 3.34-3.92 (m, 60H), 4.50-
4.52 (m, 6H), 4.77 (m, 3H), 4.84 (m, 3H), 7.74 (s, 3H), 7.74-7.83 (m,
6H), 8.27-8.37 (m, 3H) ppm; 13C NMR (150 MHz, D2O) δ 26.40,
31.30, 34.10, 40.41, 46.27, 62.49, 64.08, 68.49, 69.98, 73.27, 73.84,
75.88, 102.95, 127.77, 131.99, 137.61, 142.40, 171.52 ppm; MALDI-
TOF MS calcd for C90H129N9O45 2057.8, obsd 2080.1 (MNa+).
Calorimetry. Titration microcalorimetry was performed using the
MicroCal Omega titration microcalorimeter. Details of instrument

10296 J. Am. Chem. Soc., Vol. 121, No. 44, 1999
Dimick et al.
design and data analysis are described elsewhere.74 Briefly, a solution
of concanavalin A (0.28-0.78 mM) in a buffer of 50 mM dimethyl
glutarate, 250 mM NaCl, 1 mM CaCl2, and 1 mM MnCl2 at pH 5.2
was placed in the sample cell. Glycodendrimer solutions ([mannose]
) 11.3-24.9 mM) in a buffer identical to that used for protein solutions
were added in 25-40, 2.0-2.2-µL increments during 30 s, with 3-min
intervals between injections. Each calorimetric titration was performed
at a sample cell temperature between 26 and 30 °C. Protein concentra-
tions were determined spectrophotometrically using an extinction
coefficient of ꢀ280 ) 1.24 for a 1 mg/mL solution. Carbohydrate
concentrations were determined using the phenol-sulfuric acid charts
of DuBois.75
The heat evolved upon each injection was digitally recorded, and
the data were integrated to generate a titration curve upon completion
of the experiment. The stoichiometry of the association, n, binding
constant, Kc, and the change in enthalpy, ∆H, were obtained from a
nonlinear least-squares fit using the Origin software program. All data
are presented on a valency-corrected basis.
Hemeagglutination. Nonspecific binding to microtiter plates was
blocked by addition of a solution of BSA in PBS and BSA (50 µL).
Carbohydrate solution of a known concentration (50 µL) was added to
the first well and serially diluted. Concanavalin A solution (50 µL)
was added to all wells, and the plate was incubated at 37 C for 2 h. A
2% solution of EDTA-anticoagulated porcine red blood cells (50 µL)
was added to each well. Plates were read following a 1 h incubation.
Acknowledgment. We are grateful to Dr. G. Dubay for the
mass spectroscopy. S.M.D. acknowledges the NIH Biological
Chemistry Training Program for funding. E.J.T. acknowledges
the support of the NIH (GM 48653) J.H.N. acknowledges the
U.K. BBSRC (49/B08307) for support. J.H.N. thanks Prof. J.
R. Helliwell and Prof. S. W. Homans for encouragement and
helpful discussions.
Supporting Information Available: Copies of 1H and 13
C
NMR spectra for compounds 11, 13, 15, 17, and 19 (PDF). This
material is available free of charge via the Internet at
http://pubs.acs.org.
(74) Wiseman, T.; Williston, S.; Brandts, J. F.; Lin, L.-N. Anal. Biochem.
1989, 179, 131.
(75) Dubois, M.; Gilles, K. A.; Hamilton, J. K.; Rebers, P. A.; Smith, F.
Anal. Chem. 1956, 28, 350.
JA991729E