Sugar mimics: An artificial receptor for cholera toxin

Article abstract of DOI:10.1021/ja983567c

The paper describes the pseudosugar 2 [Galβ1-3GalNAcβ1-4(NeuAcα2- 3)DCCHD], a high affinity binder of cholera toxin (CT). The molecule was designed using molecular modeling techniques to mimic the natural CT membrane receptor, ganglioside GM1. The central residue of GM1, a 3,4-disubstituted galactose unit, was recognized as the ganglioside scaffold element and substituted with a conformationally locked cyclohexanediol (DCCHD). DCCHD was synthesized in enantiopure form using enantioselective Diels Alder methodology and regioselectively α-sialylation at the equatorial position. Glycosylation with a Galβ(1-3)-GalNAc donor completed the synthesis of 2. The solution structure of 2 and its binding ability to CT were found to be analogous to those of the GM1 oligosaccharide.

Full text of DOI:10.1021/ja983567c

2032
J. Am. Chem. Soc. 1999, 121, 2032-2036
Sugar Mimics: An Artificial Receptor for Cholera Toxin
Anna Bernardi,*,‡ Anna Checchia,‡ Paola Brocca,§ Sandro Sonnino,§ and Fabio Zuccotto|
Contribution from the Dipartimento di Chimica Organica e Industriale, UniVersita` di Milano,
Via Venezian 21, 20133 Milano, Italy, Dipartimento di Chimica e Biochimica Medica,
UniVersita´ di Milano, Italy, and Welsh School of Pharmacy, UniVersity of Cardiff, Cardiff, Wales, UK
ReceiVed October 9, 1998
Abstract: The paper describes the pseudosugar 2 [Galâ1-3GalNAcâ1-4(NeuAcR2-3)DCCHD], a high affinity
binder of cholera toxin (CT). The molecule was designed using molecular modeling techniques to mimic the
natural CT membrane receptor, ganglioside GM1. The central residue of GM1, a 3,4-disubstituted galactose
unit, was recognized as the ganglioside scaffold element and substituted with a conformationally locked
cyclohexanediol (DCCHD). DCCHD was synthesized in enantiopure form using enantioselective Diels Alder
methodology and regioselectively R-sialylation at the equatorial position. Glycosylation with a Galâ(1-3)-
GalNAc donor completed the synthesis of 2. The solution structure of 2 and its binding ability to CT were
found to be analogous to those of the GM1 oligosaccharide.
Introduction
The inhibition of carbohydrate-protein interactions is an
attractive strategy to control a number of important biological
phenomena which are initiated by formation of a sugar-protein
complex. Although carbohydrate mimetics have been used to
this end,1 their rational design has rarely been addressed.1,2 The
complex formed by cholera toxin (CT), a hexameric protein,
and its membrane receptor, the membrane glycolipid ganglioside
GM1 1, is regarded as a paradigm in the study of sugar-protein
interactions. In this paper we report on the use of computational
tools for the rational design of a pseudosugar mimic of GM1.
The pseudosugar 2, which contains the novel conformationally
constrained diol 3 as a 3,4-disubstituted galactose mimic, was
found to exhibit the same three-dimensional structure of GM1
oligosaccaharide and to be a potent inhibitor of the interaction
of GM1 with CT. The methodology presented here, as well as
diol 3, may have wider applications in the study of protein-
sugar interactions.
The sialic acid-containing glycolipids of cell membranes are
implicated in a number of cellular signaling events, and function
as binding sites for various toxins, hormones, and viruses.3 The
specific membrane receptor of the bacterial toxins CT and heat-
labile toxin of Escherichia coli (LT) is the pentasaccharide
portion of ganglioside GM1 [Galâ1-3GalNAcâ1-4(NeuAcR2-
3)Galâ1-4Glcâ1-1Cer] 1. LT and CT are 80% homologous
AB5 hexameric proteins: actual cell intoxication is carried out
by a catalytic fragment of the A subunit, while the B5 pentamer
performs recognition of and anchoring to the cell membrane.
The interaction of GM1 with LT and CT has been fully
characterized.4-6 Biochemical5 and structural6 data indicate that
Figure 1. Structures of ganglioside GM1, 1, its designed mimic 2,
and the conformationally restricted dicarboxy cyclohexanediol (DC-
CHD) 3. The binding determinants of 1 are highlighted in the boxes.
the two sugars at the nonreducing end of GM1, galactose (Gal-
IV) and sialic acid (NeuAc) (see Figure 1), are essential for
binding. In particular, the structure of the CT:GM1 complex6a,b
shows that the GM1 pentasaccharide binds with a two-finger
‡ Dipartimento di Chimica Organica e Industriale.
§ Dipartimento di Chimica e Biochimica Medica.
(5) Lanne, B.; Schierbeck, B.; Karlsson, K. A. J. Biochem. 1994, 116,
1269-1274; Fukuta, S.; Magnani, J. L.; Twiddy, E. M.; Holmes, R. K.;
Ginsburg, V. Infect. Immun. 1988, 56, 1748-1753; Ångstrøm, J.; Tenenberg,
S.; Karlsson, K. A. Proc. Natl. Acad. Sci. U.S.A. 1994, 91, 11859-11863;
Schengrund, C.-L.; Ringler, N. J. J. Biol. Chem. 1989, 264, 13233-13237.
(6) (a) Merritt, E. A.; Sarfaty, S.; v. d. Akker, F.; L’Hoir, C.; Martial, J.
A.; Hol, W. G. J. Protein Sci. 1994, 3, 166-175. (b) Merritt, E. A.; Sarfaty,
S.; Jobling, M. G.; Chang, T.; Holmes, R. K.; Hirst, T. R.; Hol, W. G. J.
Protein Sci. 1997, 6, 1516-1528. (c) Merritt, E. A.; Hol, W. G. J. Curr.
Opin. Struct. Biol. 1995, 5, 165-171 and references therein.
| University of Cardiff.
(1) Topics in Current Chemistry; Driguez, H., Thiem, J., Eds.; Springer-
Verlag: Berlin, 1997; Vol. 187; Carbohydrate Mimics: Concepts and
Methods; Chapleur, Y., Ed.; Wiley-VCH: New York, 1998.
(2) Kolb, H. C.; Ernst, B. Chem. Eur. J. 1997, 3, 1571-1578.
(3) Hakomiri, S.-I. Biochem. Soc. Trans. 1993, 21, 583-585.
(4) Spangler, B. D. Microbiol. ReV. 1992, 56, 622-647 and references
therein. Masserini, M.; Freire, E.; Palestini, P.; Calappi, E.; Tettamanti, G.
Biochemistry 1992, 31, 2422-2426.
10.1021/ja983567c CCC: $18.00 © 1999 American Chemical Society
Published on Web 03/02/1999

Artificial Receptor for Cholera Toxin
J. Am. Chem. Soc., Vol. 121, No. 10, 1999 2033
grip: the large majority of interactions between the receptor
and the toxin involve Gal-IV and NeuAc, with a limited
contribution from the N-acetylgalactosamine (GalNAc) residue
(see Figure 1). A computational model of the LT:GM1 complex
which was a fair reproduction of the CT:GM1 X-ray structure
was obtained using a Monte Carlo/energy minimization (MC/
EM) conformational search of the sugar within the toxin binding
pocket.7
NMR data8 and MC/EM calculations9 for GM1 in water
solution have revealed that its pentasaccharide is significantly
conformationally restricted. In particular the core trisaccharide
GalNAcâ1-4(NeuAcR2-3)Galâ was found to exist mainly in
a single conformation, which closely resembles the bound
conformation observed by X-ray crystallography. The Galâ1-
3GalNAc anomeric bond is only slightly more flexible, and the
NOE contacts observed around it can be interpreted as arising
from either two8 or one “average”9 conformations. The overall
picture is that of a highly preorganized receptor for toxin
binding. Very likely, the loss of conformational freedom in GM1
results from the 3,4-branching at Gal-II (Figure 1);10-12 thus
this residue, which does not interact with the protein, appears
to act as a scaffold and hold Gal-IV and NeuAc in the required
position. The GM1 mimic 2 was designed based on this
hypothesis by retaining the ganglioside binding determinants
and replacing the scaffold element with an appropriate diol,
designed to reproduce the topological features of a 3,4-
disubstituted galactose.
Figure 2. (a) Lowest energy conformation of 2. (b) Solution
conformation of GM1 1 (from refs 8 and 9).
their sugar complexes6c and were likewise included in the calculation
of the LT:2 complex.
NMR. ROESY experiments26 were conducted applying a spin lock
pulse of 2.6 kHz strength at one end of the spectrum to avoid scalar
transfer.8,18 For the experiments in D2O, mixing time was varied between
150 and 220 ms and the temperature in the range 303-313 K. For the
experiments in DMSO, mixing time was varied between 180 and 230
ms and the temperature in the range 303-318 K. Spectral assignments
are reported as Supporting Information.
Inhibition of the GM1:CT Formation. Binding of 2 to CT was
investigated by measuring the inhibition of GM1:CT formation.19 In
parallel experiments GM1 oligosaccharide was used. Each well of a
96-well microtiter plate (Greiner) was covered with 0.5 µg of GM1
dissolved in 50 µL of 95% ethanol. After overnight drying at room
temperature the wells were incubated for 2.5 h with 250 µL of 1%
albumin in 0.05 M PBS (sodium phosphate buffer, pH 7.4, 9% NaCl).
After removal of the solution, the wells were washed five times with
0.05 M PBS and incubated for 2.5 h with 50 µL of 1% albumin in
0.05 M PBS containing 0.2 µg of Vibrio cholerae toxin B subunit
conjugated to horseradish peroxidase (Sigma) preincubated for 2.5 h
with increasing amounts of GM1 oligosaccharide or 2. The CT solution
was discarded, the wells were washed five times with 250 µL of 0.05
M PBS, and 50 µL of a 0.04% solution of o-phenyldiammine in 0.02
M citrate/phosphate buffer (pH 5) and 17 µL of H2O2 were added.
After 20 min in the dark, 50 µL of H2SO4 were added to block the
reaction, and the color intensity was determined by spectrophotometry.
Materials and Methods
Materials. GM1 ganglioside was prepared13 from the total ganglio-
side mixture extracted from calf brain.14 The penta-oligosaccharide of
GM1 was obtained by ozonolysis of GM1, followed by alkaline
hydrolysis.15
Computational. All calculations were run with MacroModel 4.5,16
using the AMBER* force field with MNDO-derived parameters for
NeuAc9 and following previously established MC/EM protocols.7,9
Twenty thousand MC/EM steps were performed both for the complex
and isolated 2. Bulk water solvation was simulated using MacroModel’s
generalized Born GB/SA continuum solvent model.17 Previous studies7
had shown that under these conditions the crystal structure of the CT:
GM1 complex was best reproduced when five crystallographic water
molecules were retained. Of these, three are located outside the protein
binding site and solvate the carboxy group of Glu-51. The other two
molecules, at crystallographic solvation sites 2 and 3, mediate specific
interactions between the sugar and the protein. These molecules were
consistently found in a set of five different structures of LT, CT, and
Results and Discussion
After the central 3,4-disubstituted Gal-II unit of GM1 was
recognized as the ganglioside scaffold element, the design of a
structural and functional analogue was sought using an ap-
propriate conformationally locked cis-1,2-cyclohexanediol to
replace it. One such molecule is the dicarboxy cyclohexanediol
3 (DCCHD), which possesses the same absolute and relative
configuration of natural galactose and is locked in a single chair
conformation. Indeed, MM3* calculations show that the chair
conformation of DCCHD depicted in Figure 1 is 3.2 kcal mol-1
more stable than the chair which features trans diaxial carboxy
groups.
(7) Bernardi, A.; Raimondi, L.; Zuccotto, F. J. Med. Chem. 1997, 40,
1855-1862.
(8) Acquotti, D.; Poppe, L.; Dabrowski, J.; v. d. Lieth, C.-W.; Sonnino,
S.; Tettamanti, G. J. Am. Chem. Soc. 1990, 112, 7772-7778.
(9) Bernardi, A.; Raimondi, L. J. Org. Chem. 1995, 60, 3370-3377.
(10) Both the unbranched carbohydrates asialo-GM1 (ref 11) and GM4
(NeuAc(R2,3)Gal-âOMe) (ref 12) were shown to be considerably more
flexible than GM1 by NMR spectroscopy.
(11) Scarsdale, N. J.; Prestegard, J. H.; Yu, R. K. Biochemistry 1990,
29, 9843-9855 and references therein.
Based on the above considerations, the pseudo-tetrasaccharide
2, which retains the Gal and NeuAc recognition determinants
and uses 3 as the scaffold element, was devised as an artificial
receptor for LT and CT. Conformational analysis of 2 was
performed both for the isolated molecule and in the binding
pocket of LT, and the predicted three-dimensional structures
were compared to those of GM1.
(12) Poppe, L.; Dabrowski, J.; v. d. Lieth, C.-W.; Numata, M.; Ogawa,
T. Eur. J. Biochem. 1989, 180, 337-342.
(13) Acquotti, D., Cantu´, L., Ragg, E.; Sonnino, S. Eur. J. Biochem.
1994, 225, 271-288
(14) Tettamanti, G.; Bonali, F.; Marchesini, S.; Zambotti, V. Biochim.
Biophys. Acta 1973, 296, 160-170
(15) Wiegandt, H.; Bucking, H. W. Eur. J. Biochem. 1970, 15, 287-
292.
(16) Mohamadi, F.; Richards, N. G. J.; Guida, W. C.; Liskamp, R.;
Lipton, M.; Caufield, C.; Chang, G.; Hendrickson, T.; Still, W. C. J. Comput.
Chem. 1990, 11, 440-467.
(17) Still, W. C.; Tempzyk, A.; Hawley, R.; Hendrickson, T. J. Am.
Chem. Soc. 1990, 112, 6127-6129.
(18) Farmer, H. B. T.; Macura, S.; Brown, L. R. J. Magn. Res. 1987,
72, 347-352.
(19) Wu, G.; Leeden, R. Anal. Biochem. 1988, 173, 368-375; Carpo,
M.; Nobile-Orario, E.; Chigorno, V.; Sonnino, S. Glycoconjugate J. 1995,
12, 729-731.

2034 J. Am. Chem. Soc., Vol. 121, No. 10, 1999
Bernardi et al.
Figure 3. (a) Minimum energy conformation of the LT:2 complex as calculated by MC/EM. LT in blue; 2 in white; the two water molecules in
pink are crystallographic water molecules from the LT crystal structure that are conserved in the calculations (see ref 7). (b) Superimposition of the
calculated LT:2 complex with the X-ray structure of CT:GM1. The picture was obtained by sumperimposing three binding-site residues conserved
in LT and CT (Glu-51, Lys-91, and Trp-88). LT was deleted for clarity. CT in blue; 2 in white; GM1 1 in yellow.
picture is reported in Figure 3b and shows that 2 and GM1 are
expected to adopt a common disposition in the toxin binding
pocket. The root-mean-square deviation (rmsd) for equivalent
residues of the sugar substrates are collected in Table 1. The
buried surface area of the protein binding site in the lowest-
energy conformation of the LT:2 complex was calculated to be
347 Å2, which compares favorably with the experimental value
of 403 ( 10 Å2 for the CT:GM1 complex.20
On this basis, the synthesis of 2 was undertaken and
performed as outlined in Scheme 1.21 Enantiopure 321 was
synthesized using Helmchen’s stereoselective Diels-Alder
methodology,22,23 followed by functional group manipulation
and cis-dihydroxylation (OsCl3, Me3NO) of the cyclohexene
double bond. Regioselective R-sialylation24 of 3 with the
phosphite 425 (TMSOTf, EtCN, -40°, 50% yield, based on
reacted 3) followed by glycosylation with the Galâ(1-3)-
GalNAc donor 5 (TMSOTf, refluxing CH2Cl2, 30% yield)
completed the synthesis of 2.
Figure 4. Map of the interactions of 2 with LT, as calculated in the
global minimum of the MC/EM search.
Two-dimensional 500-MHz ROESY26 spectra of 2 were
obtained in D2O and DMSO-d6 solution. The observed inter-
residual contacts and corresponding NOE distances are collected
in Table 2 and compared with the computational prediction and
the data reported in the literature for GM1 in DMSO-d6.8
The conformational restriction of the core trisaccharide in
GM1 results in a very characteristic set of NOE contacts.
Typically, a strong cross-peak is observed between the H3 of
Gal-II (II-3) and the 3-axial proton of NeuAc (N-3ax) (Table
2, entry 1). The GalNAc anomeric proton (III-1) shows a single
The conformational search of isolated 2 was run simulating
water solvation with the GB/SA model and yielded 10 conform-
ers within 1 kcal mol-1 from the global minimum. They all
shared the common conformation of the branched pseudo-
trisaccharide fragment shown in Figure 2a. Comparison of the
global minimum with the NMR solution structure of GM18
(Figure 2b) clearly shows that DCCHD holds the terminal Gal
and NeuAc in the same relative position as the native galactose.
The analysis of the LT:2 complex was also very promising,
and yielded low energy conformations which featured all the
expected sugar-protein contacts6 (see global minimum in Figure
3a). A map of the interactions of 2 with LT, as seen in the
global minimum of the MC/EM search is reported in Figure 4.
The corresponding maps for GM1:CT and GM1:LT as seen by
X-ray crystallography (from ref 6a) or calculated by molecular
mechanics (from ref 7) are reported as Supporting Information.
The main difference between the calculated contacts in LT: 2
and those observed in GM1:CT is the absence in the former of
interactions between the NeuAc side-chain and the water
molecule at crystallographic solvent site 2 (W2 in Figure 3a).
A visual comparison of the lowest energy conformation of LT:2
and the X-ray structure of CT:GM1 was obtained by superim-
posing three active-site residues of LT and CT. The resulting
(20) Merritt, E. A.; Sarfaty, S.; Feil, I. K.; Hol W. G. J. Structure 1997,
5, 1485-1499 and references therein.
(21) Analytical data for compounds 2 and 3 are given in the Supporting
Information. Details of the synthetic procedure will be reported elsewhere.
(22) Hartmann, H.; Hady, A. F. A.; Sartor, K.; Weetman, J.; Helmchen,
G. Angew. Chem., Int. Ed. Engl. 1987, 26, 1143-1144. The desired (S,S)-
diacid was obtained as a 6:1 mixture with the (R,R)-enantiomer and purified
via crystallization of its quinine salt, following a published procedure (ref
23).
(23) Walborsky, H. M.; Barash, L.; Davis, T. C. Tetrahedron 1963, 19,
2333-2351.
(24) Okamoto, K.; Goto, T. Tetrahedron 1990, 17, 5835-5857.
(25) Martin, T. J.; Schmidt, R. R. Tetrahedron Lett. 1992, 33, 6123-
6126; Martin, T. J.; Brescello, R.; Toepfer, A.; Schmidt, R. R. Glycoconj.
J. 1993, 10, 16-25.
(26) Bax, A.; Davis, D. G. J. Magn. Res. 1985, 64, 533-535.

Artificial Receptor for Cholera Toxin
Table 1. Superimposition between the X-ray CT:GM1 Structure and the LT:2 Complex. Rms Deviationsa
NeuAc side chain
J. Am. Chem. Soc., Vol. 121, No. 10, 1999 2035
residue
rmsdb (Å)
Gal-IV
0.495
GalNAc
0.815
Gal-II/DCCHD
0.252
NeuAc
0.587
C7
C8
C9
O7
O8
O9
1.129
1.154
1.560
1.117
1.241
1.755
a As obtained by superimposing Glu-51, Lys-91, and Trp-88 side chains. The lowest minimum conformation was used for the LT:2 complex.
b Rms deviations of sugar residues were measured between ring centroids.
Table 2. Experimental and Calculated Inter-residual Contacts for 2 (Å)
entry
contactsa
calcd distanceb
NOE distance (D2O)
NOE distance (DMSO-d6)
NOE distance in GM1c (contact)
1
2
3
4
5
6
7
8
9
N-3ax/CHD-3
N-8/GN-1
N-8/CHD-4
N-OH8/GN-1
N-OH8/CHD-3
GN-1/CHD-4
GN-Ac/CHD-2ax
G-1/GN-2
2.3
2.9
2.6
3.1
5.4
2.3
3.2
4.0
2.4
4.0
3.4
2.5
d
4.5
2.3
d
2.8
3.0
4.1
2.2
e
3.6
2.3
3.6
3.4
2.1 (N-3ax/II-3)
3.1 (N-8/III-1)
e (N-8/II-4)
2.6 (N-OH8/III-1)
e (N-OH8/II-3)
2.2 (III-1/II-4)
3.1 (III-Ac/II-2)
3.5 (IV-1/III-2)
2.5 (IV-1/III-3)
3.5 (IV-1/III-4)
3.5 (IV-1/III-NH)
2.2
d
3.6
2.2
3.5
G-1/GN-3
G-1/GN-4
G-1/GN-NH
10
11
a Abbreviations: N: NeuAc; CHD: DCCHD; GN: GalNAc; G: Gal. DCCHD numbering as reported in Figure 1. b The distances were calculated
as r ) r-6 -1/6 were r-6 is the Boltzmann average of the r-6 of all the individual conformations found within the first 3 kcal mol-1 from the global
minimum (86 conformations). c DMSO-d6 solution, from ref 8. d Not measurable, due to signal overlap. e No cross-peak detected.
Scheme 1. Synthesis of 2a
Figure 5. Inhibition of the GM1:CT binding in ELISA by the GM1
pentasaccharide (b) and 2 (9). Standard deviation of data determined
on 4 experiments was (30%. Over a concentration of 200 µg/mL
a a. TiCl4, butadiene, CH2Cl2, -50 °C; LiOH. b. DMF-di-tert-
inhibition was always 100%, for both 2 and the pentasaccharide. Each
microtiter well was covered with 1 µg of GM1. Peroxidase conjugate
CT B subunit was used at 1.75 µg/mL.
butylacetal, refluxing benzene; OsCl3, Me3NO. c. TMSOTf, EtCN, -40
°C, then 4; d. TMSOTf, CH2Cl2, 40 °C, then 5; cat. MeONa, MeOH,
then H2O.
the N-OH8/GN-1 cross-peak (Table 2, entry 4) indicates
proximity of the NeuAc side chain and the GalNAc residue.
Furthermore, the spectra of 2 in DMSO show a cross-peak
between N-8 and the H4 proton of DCCHD (N-8/CHD-4; Table
2, entry 3), which is, in turn, in proximity of GN-1 (Table 2,
entry 6).27 The main difference in the spectral patterns of 1 and
2 is the III-Ac/II-2 cross-peak observed for 1, which is not
paralleled by an equivalent contact between the acetyl protons
of GalNac and the H2ax proton of DCCHD in 2 (GN-Ac/CHD-
2ax; Table 2, entry 7). However, NMR analysis of 2 shows a
remarkable similarity to GM1, and clearly supports the hypoth-
esis that the two molecules share a common conformation of
the binding determinants.
The computational values of the interproton distances (first
column in Table 2) were calculated as Boltzmann averages over
all of the conformers included in the first 3 kcal mol-1 from
the global minimum (86 conformers). They reproduce the NOE-
derived distances fairly well, thus validating the computational
procedure and the three-dimensional model of 2 reported in
Figure 2a.
transglycosidic contact with the H4 proton of Gal-II (II-4) (Table
2, entry 6), and gives two cross-peaks with the H8 and OH8
protons of NeuAc (N-8 and N-OH8) (Table 2, entries 2 and
4). This pattern of contacts is consistent with a single dominant
conformation for both the glycosidic NeuAcR2-3Gal and
GalNAcâ1-4Gal bonds. The conformation of the Galâ1-
3GalNAc linkage is defined by the NOE contacts observed for
the anomeric proton of Gal-IV (IV-1, Table 2, entries 8-11),
which shows cross-peaks of equal intensity with the H3, the
H4, and the NH protons of GalNAc. The resulting three-
dimensional picture of GM1 is reported in Figure 2b.8 The
ROESY spectra of 2 showed an equivalent pattern of inter-
residual contacts, and the corresponding NOE distances were
found to be very similar to those observed for GM1. The cross-
peaks and the NOE distances observed for the galactose
anomeric proton (G-1) of 2 are almost identical to those seen
in GM1 (Table 2, entries 8-11). A strong cross-peak also
appears between the NeuAc-3ax proton and the H3 of DCCHD
(N-3ax/CHD-3, Table 2, entry 1), equivalent to the N3ax/II-3
contact in GM1. The contact between NeuAc H8 and the
anomeric proton of GalNAc (N-8/GN-1) cannot be ascertained
due to spectral overlap (Table 2, entry 2), but the presence of
(27) The N-8/CHD-4 cross-peak is much weaker in D2O solution, which
may suggest a different distribution of the H-bonding network for the NeuAc
side chain of 2 in the two solvents.

2036 J. Am. Chem. Soc., Vol. 121, No. 10, 1999
Bernardi et al.
Binding of 2 to CT was investigated by inhibiting the GM1:
CT interaction in ELISA and TLC overlay assays, using pre-
viously published procedures.19 The ELISA inhibition profiles
of 2 and of the GM1 pentasaccharide are reported in Figure 5,
and they are clearly overlapping. This result establishes 2 as
the strongest artificial monovalent CT binder reported thus far.20
In conclusion, the computational tools developed for the study
of GM1 and its toxin complexes7,9 were shown to have
predictive value for the design of a potent CT binder. More
generally, the conformationally locked diol 3 was shown to be
an effective mimic of a 3,4-disubstituted galactose and may have
a wider use in the construction of pseudo-oligosaccharides
featuring 3,4-Gal branching.
Acknowledgment. This work was supported by funding
from the Ministero dell’Universita´ e della Ricerca Scientifica e
Tecnologica (MURST) and the Consiglio Nazionale delle
Ricerche (CNR).
Supporting Information Available: Analytical data for
compounds 2 and 3, maps of the sugar toxin contacts in LT:2,
LT:GM1 and CT:GM1 (PDF). This material is available free
of charge via the Internet at http://pubs.acs.org.
JA983567C