Ganglioside GM1 mimics: Lipophilic substituents improve affinity for cholera toxin

Article abstract of DOI:10.1016/j.bmcl.2003.07.007

Ganglioside GM1 mimics including (R)-2-hydroxy-3-cyclohexylpropionic acid or (R)-2-hydroxy-3-phenylpropionic acid as replacements for NeuAc are stronger cholera toxin binders than the parent ligand 2, which includes (R)-2-hydroxy-propionic acid.

Full text of DOI:10.1016/j.bmcl.2003.07.007

Bioorganic & Medicinal Chemistry Letters 13 (2003) 3831–3834
Ganglioside GM1 Mimics: Lipophilic Substituents
Improve Affinity for Cholera Toxin
Daniela Arosio, Sergio Baretti, Stefania Cattaldo,
Donatella Potenza and Anna Bernardi*
Dipartimento di Chimica Organica e Industriale, Universita’ di Milano, via Venezian 21, 20133 Milan, Italy
Received 19 May 2003;revised 22 July 2003;accepted 23 July 2003
Abstract—Ganglioside GM1 mimics including (R)-2-hydroxy-3-cyclohexylpropionic acid or (R)-2-hydroxy-3-phenylpropionic acid
as replacements for NeuAc are stronger cholera toxin binders than the parent ligand 2, which includes (R)-2-hydroxy-propionic
acid.
# 2003 Elsevier Ltd. All rights reserved.
The use of sugar mimics to antagonize oligosaccharides
at the protein receptor level is attracting a great deal of
attention as a way to develop drugs with good stability
and synthetic availability.1 Our group has been working
in this area using the cholera toxin/ganglioside GM1
(CT/GM1) recognition pair as a model system.2 In this
context, we have described the rational design and the
synthesis of the pseudo-oligosaccharide 13,4 (Chart 1),
which was found to be as active as the GM1 oligo-
saccharide (o-GM1, Chart 1) in binding to CT.3 More
recently, we have reported three second-generation
mimics5 obtained by replacing the sialic acid (NeuAc)
moiety of 1 with simple a-hydroxyacids, such as (R)-
and (S)-lactic acid and glycolic acid. The most active
compound of this series (Kd=190 mM) was 2, which
includes the (R)-lactic acid [(R)-2-hydroxy propionic
acid] side chain (Chart 1).5,6 NMR data and computer
modeling of the second generation ligands have sug-
gested that they are rather flexible in the hydroxyacid
region.7 A similar analysis carried out on their com-
plexes has shown that the protein selects for binding the
side-chain conformation which reproduces the orienta-
tion of the NeuAc carboxy group in GM1.7,8 The higher
affinity of 2 relative to the other ligands of the series
appears to result from van der Waals interactions
established between its side-chain methyl group and the
toxin cleft. This information suggests that the affinity of
the pseudo-GM1 binders may be improved by adding
appropriate hydrophobic fragments to the framework
of the (R)-lactic acid GM1-mimic 2.
Here, we report that indeed the pseudo-GM1 ligands 3
and 4, which include a cyclohexyl group and a phenyl
group, respectively, do display stronger affinity for CT
than 2. Remarkably, the (R)-2-hydroxy-3-phenyl pro-
pionic acid derivative 4 with a 10-mM dissociation con-
stant is just one order of magnitude less potent than the
natural ligand o-GM19 against the cholera toxin.
The two new ligands were synthesized by minor modifi-
cations of the established sequence,5 as reported in
Scheme 1. Starting from the enantiomerically pure diol
54,10 the monoethers 8 and 9 were synthesized by
Bu2SnO-mediated regioselective alkylation11 using the
triflates 6 and 7,6 respectively. TfOH promoted glyco-
sylation of the axial hydroxy group with the Galb1-
3GalNAc donor 104 gave the protected pseudo-tri-
saccharides 11 and 12. Standard removal of the pro-
tecting groups yielded 3 and 4.12
The interaction of the ligands with the cholera toxin B5
pentamer (CTB) was studied using the intrinsic fluores-
cence of the Trp88 residue in the toxin binding site.
Ligand binding to CTB is known to induce bath-
ochromic shifts and variations in fluorescence intensity
whose extent depends on the structure of the
ligand.9,13,14 The titrations were performed by irradiat-
ing the sample at 280 nm, and collecting the data at the
maximum of the Trp emission curve, at ca. 350 nm. The
normalized changes in the fluorescence emission
*Corresponding author. Tel.: +39-02-5031-4092;fax: +39-02-5031-
4072;e-mail: anna.bernardi@unimi.it
0960-894X/$ - see front matter # 2003 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2003.07.007

3832
D. Arosio et al. / Bioorg. Med. Chem. Lett. 13 (2003) 3831–3834
Scheme 1. General sequence for the synthesis of 3 and 4: (a) Bu2SnO, benzene, reflux, then CsF, DME and 6, or 7 (30% yield);(b) 10 (0.5 equiv) and
TfOH (0.05 equiv) in CH2Cl2, rt to reflux (20%);(c) H 2/Pd/C, MeOH (80%);cat MeONa in MeOH (90%).
intensity of CTB upon titration with 2–4 are collected in
Figure 1. In contrast with GM1 and ps-GM1 1, ligands
2–4 (Fig. 1) do not display cooperative binding beha-
vior. The dissociation constants determined by non-
linear regression analysis are 190 mM for 2,5 45 mM for
3, and 10 mM for 4. Thus, it appears that the inclusion
of a larger, more lipophilic alkyl group on the hydro-
xyacid side chain does indeed improve the affinity of the
pseudoGM1 binders by up to an order of magnitude.
The phenyl group displays a stronger effect than the
cyclohexyl group.
the solution conformation of ligands 3 and 4 were per-
formed by measuring their NOESY and T-ROESY15
spectra in D2O solution. The flexibility and orientation
of the hydroxyacid chain can be established by examin-
ing the NOE interactions for the proton HL, a to the
carboxy group, to H-3, H-4 and H-2eq of the cyclohex-
anediol residue (CHD) (Fig. 2). The results for the (R)-
2-hydroxy-3-cyclohexylpropionic derivative 3 closely
match those already described for the parent compound
2.7 Thus, the HL proton shows a set of comparable
crosspeaks with CHD-H3 and CHD-H4, and a weaker
interaction with CHD-H2eq (Fig. 3). This is consistent
with the presence of two orientations of the carboxy
group relative to the adjacent diol, which can be idea-
lized as in the Newman projections of Figure 2c and d.
One of these, 2d, which mimics the carboxy group
orientation in bound GM1,8 is presumably the bioactive
one.7
In order to rationalize these results, NMR studies on
In contrast, the (R)-2-hydroxy-3-phenylpropionic acid
derivative 4 appears to be significantly less flexible than
3 and 2 in the side-chain region. In fact, in the ROESY
spectrum of 4 only the HL/CHD-H3 and HL/CHD-
H2eq crosspeaks are seen, whereas the HL/CHD-H4
crosspeak is not observed (Figs. 2b and 4a). This is
consistent with the presence of only one major orienta-
Figure 1. Effects of ligand binding on the fluorescence intensity of
Trp88 emission in CTB. CTB (0.5 mM) was titrated with micromolar
amounts of 2 (black circles, dotted line), 3 (empty circles, dashed line)
and 4 (black triangles, solid line). The ligand concentration is plotted
against the normalized, absolute value of the relative fluorescence
intensity (jdIj) measured at the emission maximum of Trp88.
Chart 1. Ganglioside GM1 headgroup, the pseudo-GM1 mimic 1, and
the (R)-hydroxyacid ligands 2–4.

D. Arosio et al. / Bioorg. Med. Chem. Lett. 13 (2003) 3831–3834
3833
Figure 2. Solution solution conformation of the ligands: (a) computer generated 3D structure of 4;(b) structure of 4, showing the NOE contacts
observed for HL;(c) and (d) Newman projections through the hydroxyacid side chain: both (c) and (d) are populated by 3, only (d) is populated by 4
(see text).
Figure 3. T-ROESY spectrum of 3 in D2O NOEs between the
side-chain proton HL and H-4, H-3H-2eq of the ciclohexanediol resi-
due.
tion of the side chain, corresponding to the bioactive
idealized Newman projection of Figure 2d. Therefore,
contrary to 2 and 3, ligand 4 appears to be preorganized
for toxin binding.
Figure 4. T-ROESY spectrum of 4 in D2O: (a) NOE crosspeaks between
the side-chain proton HL and the ciclohexanediol residue;(b) crosspeaks
observed between the aromatic ring and the GalNAc protons.
Although more work is required to understand the
mode of binding of these ligands, the results reported
here suggest that lipophilic groups on the hydroxy acid
side chain can contribute to the interaction with CT and
that the almost 5-fold increase in affinity of compound 4
over 3 may be interpreted as a result of ligand pre-
organization. Further analysis of the NMR data sug-
gests that the preorganization of 4 may stem from
favorable van der Waals interactions between the
phenyl ring and the a-face of the GalNAc residue. A
close proximity of these two groups is in fact reported
by the strong NOE crosspeaks which are detected
between the a-face protons of GalNAc (GN1, GN3 and
GN5) and the aromatic side-chain residue (Fig. 4b).16
Molecular modeling17 of the free ligand 4 yielded the
three-dimensional structure depicted in Figure 2a as the
best agreement with the available experimental data.
Face-to-face stacking interactions between aminoacid
aromatic residues and Gal or GalNAc are frequently
observed in experimental sugar–protein complexes.18
However, to the best of our knowledge, this is the first
time that such interaction is observed to occur intra-
molecularly to stabilize one conformation of a ligand.
The results reported here suggest that aromatic/sugar
interactions may be used as elements of conformatonal
control in the design of glycomimetics.
Further studies are in progress to determine the mode of
binding of 3 and 4 to the cholera toxin.
Acknowledgements
This work was supported by MIUR and CNR, Istituto
di Scienze e Tecnologie Molecolari.
References and Notes
1. Sears, P.;Wong, C. H. Angew. Chem. 1999, 111, 2446
Sears, P.;Wong, C. H. Angew. Chem. Int. Ed. 1999, 38, 2300,
and references therein.
2. For a review, see: Bernardi, A.;Arosio, D.;Sonnino, S.
Neurochem. Res. 2002, 27, 539.
3. Bernardi, A.;Checchia, A.;Brocca, P.;Sonnino, S.;Zuc-
cotto, F. J. Am. Chem. Soc. 1999, 121, 2032.
4. Bernardi, A.;Boschin, G.;Checchia, A.;Lattanzio, M.;
Manzoni, L.;Potenza, D.;Scolastico, C. Eur. J. Org. Chem.
1999, 6, 1311.
5. Bernardi, A.;Carrettoni, L.;Grosso Ciponte, A.;Monti,
D.;Sonnino, S. Bioorg. Med. Chem. Lett. 2000, 10, 2197.
6. For a previous report on the use of hydroxyacids to mimic
sialic acid in the context of sialyl-LewisX, see: Kolb, H. C.;

3834
D. Arosio et al. / Bioorg. Med. Chem. Lett. 13 (2003) 3831–3834
Ernst, B. Chem. Eur. J. 1997, 3, 1571 In this case, a (S)-
hydroxyacid was found to be the best replacement for NeuAc.
7. Bernardi, A.;Potenza, D.;Capelli, A. M.;Garcı a-Herrero,
A.;Can ada, F. J.;Jime nez-Barbero, J. Chem. Eur. J. 2002, 8,
4597.
8. 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.
JG ÀG =7.6 Hz, JG ÀG =10 Hz);3.59 (dd, 1H, GalH-3,
2
1
2
3
JG ÀG =10 Hz, JG ÀG =3 Hz);3.65 (m, 5H, GalH-5);3.68 (m,
3
2
3
4
1H, GalH-6);3.7 (m, 2H, GalNacH-6, GalNAcH-6 0);3.72 (m,
1H, GalH-60);3.88 (d, 1H, GalH-4, JG ÀG =3 Hz);3.92 (dd,
4
3
1H, GalNAcH-2, JGN ÀGN =8.6 Hz, JGN ÀGN =11 Hz);4.01
2
1
2
3
(d, 1H, GalNAcH-4, JGN ÀGN =3 Hz);4.07 (bs, 1H, CHDH-
4
3
4);4.17 (dd, 1H, CH L, JHLÀCH a=7.4 Hz, JHLÀCH b=4.5 Hz);
2
2
4.35 (d, 1H, GalH-1, JG ÀG =7.6 Hz);4.47 (d, 1H, GalNAcH-
1
2
9. Schon, A.;Freire, E. Biochem. 1989, 28, 5019.
10. Bernardi, A.;Arosio, D.;Manzoni, L.;Micheli, F.;Pas-
quarello, A.;Seneci, P. J. Org. Chem. 2001, 66, 6209.
11. David, S.;Hanessian, S. Tetrahedron 1985, 41, 643.
1, JGN ÀGN =8.6 Hz);7.28–7.35 (m, 5H). 13C NMR-HETCOR
1
2
(D2O, 400 MHz): 22.3;27.2;28.5;33.1;39.1;40.9;44.9;51.8;
61.0;61.5;68.5;69.0;71.1;72.8;73.2;74.9;75.3;78.6;80.5;
81.2;101.7;105.2;128.8.
1
12. 3: H NMR (D2O, 400 MHz): 0.91 (m, 2H, CyH-2);0.14
(m, 4H, CyH-4);1.38 (s, 9H, COO(CH 3)3);1.39 (s, 9H,
13. Mertz, J. A.;McCann, J. A.;Picking, W. D. Biochem.
Biophys. Res. Commun. 1996, 226, 140.
COO(CH3)3);1.52–1.69 (m, 9H, CHDH-5ax, CyH-5, CyCH ,
2
CHDH-2ax);1.93 (s, 3H, NHCOCH );2.04 (m, 1H, CHDH-
3
14. The cholera toxin B5 pentamer (SIGMA1) was resus-
pended to give the desired concentration in 0.1 M TRIS, 0.4 M
NaCl, 2 mM Na2EDTA, 6 mM NaN3, pH 7.5. Fluorescence
measurements were made at room temperature with a Perkin–
Elmer LS-50 spectrofluorometer. The excitation wavelength
was 280 nm, and the spectrum was recorded between 300 and
450 nm.
15. Hwang, T. L.;Shaka, A. J. J. Am. Chem. Soc. 1992, 114,
3157. Hwang, T. L.;Shaka, A. J. J. Magn. Res. Ser. B 1993,
102, 155.
16. No analogous crosspeaks were detected between the Gal-
NAc a protons and the cyclohexyl ring in 3.
17. A conformational search was performed using the MC/
EM protocol of Macromodel 7.0, the AMBER* force field,
and the GB/SA water solvation model, as described in ref 7.
The parameters developed by Kolb and Ernst (from ref 6)
were used for the hydroxyacid fragment. For a description of
Macromodel, see: Mohamadi, F.;Richards, N. G. J.;Guida,
W. C.;Liskamp, R.;Lipton, M.;Caufield, C.;Chang, G.;
Hendrickson, T.;Still, W. C. J. Comp. Chem. 1990, 11, 440.
18. (a) Weis, W. I.;Drickamer, K. Annu. Rev. Biochem. 1996,
65, 441. (b) Elgavish, S.;Shaanan, B. J. Mol. Biol. 1998, 277,
917. (c) Quiocho, F. A. Annu. Rev. Biochem. 1986, 55, 287. (d)
Vyas, N. K. Curr. Opin. Struct. Biol. 1991, 1, 732. (e) Quiocho,
F. A. Biochem. Soc. Trans. 1993, 21, 442.
2eq) 2.17 (m, 1H, CHDH-5eq, J5eqÀ5ax=10 Hz,
J5eqÀ6 ffi J5eqÀ4=3.8 Hz);2.51 (dt, 1H, CHDH-1,
J1À6 ffi J1À2ax=11.5 Hz, J1À2eq=3.7 Hz);2.73 (dt, 1H, CHDH-
6, J6À1 ffi J6À5ax=13.1 Hz, J6À5eq=3.8 Hz);3.45–3.52 (m, 2H,
CHDH-3, GalH-2);3.56–3.63 (m, 5H, GalH-3,0 GalH-5, Gal-
NacH-5);3.68–3.80 (m, 5H, GalH-6, GalH-6 , GalNAcH-6,
GalNAcH-60, GalNacH-3);3.96 (d, 1H, GalH-4, JG4ÀG3=3.2
Hz);4.03 (m, 2H, GalNAcH-2, CH L);4.14 (d, 1H, GalNAcH-
4, JGN ÀGN =3.1 Hz);4.32 (bs, 1H, CHDH-4);4.43 (d, 1H,
4
3
GalH-1, JG1ÀG2=7.8 Hz);4.97 (d, 1H, GalNAcH-1,
JGN ÀGN =8.5 Hz). 13C NMR-HETCOR (D2O, 400 MHz):
1
2
22.8;26.1;26.2;27.5;28.4;33.0;33.4;34.0;40.4;40.6;44.7;
52.1;61.2;68.6;69.2;71.0;72.5;72.6;72.7;75.3;78.4;79.6;
80.4;101.4;105.5.
1
4: H NMR (D2O, 400 MHz): 1.36 (s, 9H, COO(CH3)3);1.38
(s, 9H, COO(CH3)3);1.43 (m, 1H, CHDH-5ax);1.46 (m, 1H,
CHDH-2ax);1.92 (s, 3H, NHCOCH );2.04 (m, 1H, CHDH-
3
2eq) 2.09 (m, 1H, CHDH-5eq);2.43 (dt, 1H, CHDH-1,
J1À6 ffi J1À2ax=12 Hz, J1À2eq=3.6 Hz);2.63 (dt, 1H, CHDH-6,
J6À1 ffi J6À5ax=12 Hz, J6À5eq=3.7 Hz);2.92 (dd, 1H, PhCH 2-a,
Jgem=14 Hz, JCH a-HL=7 Hz);3.05 (dd, 1H, PhCH 2-b,
2
Jgem=14 Hz, JCH b-HL=4.5 Hz);3.28 (m, 1H, GalNAcH-5);
3.35 (m, 2H, GalNAcH-3, CHDH-3);3.45 (dd, 1H, GalH-2,
2