New bicyclam-GalCer analogue conjugates: Synthesis and in vitro anti-HIV activity

Article abstract of DOI:10.1016/j.bmcl.2003.10.036

The synthesis of bipharmacophore anti-HIV compounds which, in a single molecule, combine two ligands, that is, the bicyclam AMD3100 and a GalCer analogue, that might inhibit several steps of the complex virus/cell cascade interactions has been performed. The 'double-drug' Gal-AMD3100 conjugates elicited inhibitory effects on T (or X4)-tropic HIV-1 replication in all CXCR4 expressing cell lines with EC₅₀ values ranging from 0.25 to 6.0 μM which were however ～40- to 125-fold lower than that of AMD3100. Concerning the mechanism of inhibition of the Gal-AMD3100 conjugates, experiments performed with X4 or R5HIV-1 strains and GHOST cells genetically modified to express CD4 and CXCR4 or CCR5 indicated clearly that the conjugates interact with CXCR4 and not with CCR5.

Full text of DOI:10.1016/j.bmcl.2003.10.036

Bioorganic & Medicinal Chemistry Letters 14 (2004) 495–498
New bicyclam–GalCer analogue conjugates: synthesis and in vitro
anti-HIV activity
Jean-Michel Daoudi,a Jacques Greiner,a Anne-Marie Aubertinb and Pierre Vierlinga,*
a
´
Laboratoire de Chimie Bioorganique UMR-CNRS 6001, Universite de Nice-Sophia Antipolis, Parc Valrose,
´
06108 Nice Cedex 2, France
´
b
´ ´
Institut de Virologie, U 544, Faculte de Medecine, 3, rue Koeberle, 67000 Strasbourg, France
Received 24 April 2003; revised 30 May 2003; accepted 20 October 2003
Abstract—The synthesis of bipharmacophore anti-HIV compounds which, in a single molecule, combine two ligands, that is, the
bicyclam AMD3100 and a GalCer analogue, that might inhibit several steps of the complex virus/cell cascade interactions has been
performed. The ‘double-drug’ Gal-AMD3100 conjugates elicited inhibitory effects on T (or X4)-tropic HIV-1 replication in all
CXCR4 expressing cell lines with EC50 values ranging from 0.25 to 6.0 mM which were however ꢀ40- to 125-fold lower than that of
AMD3100. Concerning the mechanism of inhibition of the Gal-AMD3100 conjugates, experiments performed with X4 or R5HIV-1
strains and GHOST cells genetically modified to express CD4 and CXCR4 or CCR5 indicated clearly that the conjugates interact
with CXCR4 and not with CCR5.
# 2003 Elsevier Ltd. All rights reserved.
The human immunodeficiency virus (HIV) infects
mainly CD4(+) lymphoid cells. Their recognition and
infection are mediated by the binding of HIV, through
its envelope glycoprotein gp120, to CD4 and then to a
cellular co-receptor of the chemokine receptor family,
such as mainly CCR5 or CXCR4, according to the cell
type infected and the M (or R5)- or T (or X4)-tropism
of the HIV strains.1 When this ternary gp120/CD4/co-
receptor complex has formed, the viral transmembrane
gp41, through a fusion peptide located at its external N-
terminal domain, anchors into the target cell membrane.
This initiates the virus-cell fusion and the delivery of the
viral nucleocapside into the cell. Intensive efforts were
undertaken for the discovery of anti-HIV drugs that
target two major viral proteins, that is, the reverse
transcriptase and protease.2 Recently, it has also been
pointed out that virus-cell fusion was a major hallmark
of HIV infection. This observation has led to the dis-
covery of novel highly potent agents that interfere with
the viral binding to the cellular co-receptors, such as the
low molecular weight AMD3100 (Fig. 1) which is a
CXCR4 antagonist,3 or that interact with gp41, such as
the most promising 36-amino acid peptide T20.4 HIV
can also infect many CD4(À) cells.5 Their infection has
been shown to occur through gp120binding to galacto-
sylceramide (GalCer, Fig. 1) they express. Very recently,
GalCer was also found to interact with the HIV gp41.6
Further studies suggested the involvement of GalCer-
rich microdomains,6,7 thus providing an attachment
platform for the virus through its gp120and/or gp41
onto the cell. These findings have urged the synthesis of
various galactolipids and the examination of their anti-
HIV activity.8 Unfortunately, the most active deriva-
tives were found to exhibit only a moderate in vitro
anti-HIV activity (EC50 in the 1–20 mM range).8
Keywords: HIV; CXCR4; AMD3100; Galactosylceramide.
* Corresponding author. Tel.: +33-(0)4-9207-6143; fax: +33-(0)4-
9207-6151; e-mail: vierling@unice.fr
Figure 1. Chemical structure of galactosylceramide (GalCer), GalSer,
an analogue of GalCer, and AMD3100.
0960-894X/$ - see front matter # 2003 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2003.10.036

496
J.-M. Daoudi et al. / Bioorg. Med. Chem. Lett. 14 (2004) 495–498
Our interests were to explore the development of
bipharmacophore anti-HIV compounds which, in a sin-
gle molecule, combine two ligands, that is, the bicyclam
AMD3100 and a GalCer analogue, that might inhibit
several steps of the virus/cell complex cascade interac-
tions. It was indeed of great interest to establish whether
such ‘double-drug’ conjugates could interfere synergis-
tically (with respect to AMD3100 and the GalCer ana-
logue) with both the gp120/CD4 and gp120/CD4/
CXCR4 interactions, and/or with gp41, and, conse-
quently, could constitute highly potent inhibitors of the
viral entry into cells.
reported. However, these ‘double-drug’ prodrugs target
two distinct but remote steps of the viral replicative
cycle.
Herein, we report on the synthesis of conjugates 5a,b
(Scheme 1) deriving formally from GalSer (which is a
GalCer analogue, Fig. 1)8a and AMD3100, and on their
in vitro anti-HIV activities that were established with
X4- and dual (X4 and R5)-tropic HIV-1 strains on var-
ious CXCR4 expressing cells. To have a further insight
into their inhibition mechanism, tests were also per-
formed with a R5-tropic HIV-1 strain and CCR5
expressing cells. It should be emphasized that neither
the 3-D structure of native gp120, gp120/gp41, nor that
of the gp120/CD4/co-receptor complexes are known.
Therefore, it is most difficult to rationally design
bifunctional inhibitors that act simultaneously on two
distinct sites and interfere with the most complex cas-
cade HIV-cell adsorption-fusion events. In order to cir-
cumvent these drawbacks, we adopted a modular design
for the chemical structure of the Gal-AMD3100 con-
jugates 5 wherein the length of the spacer arm linking
the quintessential galactosylated and bicyclam units can
be varied. The synthesis of the Gal-AMD3100 con-
jugates 5a,b is mostly divergent (Scheme 1). The two
regioisomers 3a,b which contain the bicyclam moiety
conjugated to the (CH2)4COOH spacer at different
nitrogen positions constitute the key synthons. Their
time consuming syntheses and purifications were per-
formed following already published procedures.9,13
These two 3a,b regioisomers were then coupled using
conventional peptide coupling reagents to the N-tetra-
decanoyl l-lysine methylester derivative 6 obtained
from commercially available Fmoc-l-lysine through a
standard two-step procedure. Hydrolysis of the
obtained methyl esters under basic conditions followed
by the activation of the resulting acids with HOSu/DCC
then action of the O-Ac protected galactosylated serine
derivative 78a led to the Gal- and AMD3100-protected
conjugates 4a,b. Quantitative N-Boc deprotection using
an excess of TFA, then quantitative O-Ac deprotection
with a Et3N/MeOH/H2O mixture led to the target
compounds 5a,b.14
To our knowledge, such a ‘double-drug’ approach focus-
ing on the close cascade virus/cell adsorption/fusion pro-
cesses has not been investigated. ‘Double-drug’
prototypes associating in a single molecule a nucleoside
reverse transcriptase inhibitor with a CXCR4 antago-
nist (e.g., AMD3100)9,10 or a virus–cell interaction
inhibitor,11 or a protease inhibitor12 have indeed been
The two Gal-AMD3100 conjugates 5a,b, as well as the
different moieties composing these conjugates [that is,
the AMD3100-linker derivative 2a (Scheme 1),15 and
AMD3100], GalSer, and AZT as control, were eval-
uated for their inhibitory effects on T-tropic HIV-1
replication in CXCR4 expressing cells (CEM-SS, MT4)
and on dual-tropic HIV-1 replication in PBMC cells
(Table 1). Under assay conditions, the ‘double-drug’
Gal–AMD3100 conjugates 5a,b elicited anti-HIV activ-
ities in all CXCR4 expressing cell lines with EC50 values
ranging from 0.25 to 6.0 mM. In the same testing con-
ditions and for a given CXCR4 expressing cell line, the
AMD3100 ‘single drug’ and even its N-alkylated deri-
vative 2a inhibited HIV replication much more effi-
ciently (see Table 1). By contrast, the lipidic GalSer
‘single drug’ displayed a lower anti-HIV activity in
CEM-SS cells than the Gal–AMD3100 conjugates 5a,b,
and was found to be inactive in MT4 cells. These test-
ings show that conjugation of the linker to AMD3100
Scheme 1. Synthetic pathway to the bipharmacophore derivatives
5a,b. Reagents and conditions: (a) (1:1) TFA/CH2Cl2, quantitative; (b)
5N NaOH, (1:1) THF/H2O; (c) 6, HOBt, EDC, CH2Cl2, 90%; (d) O-
(2,3,4,6-tetra-O-acetyl-b-d-galactopyranosyl)-l-serine
HOSu, DCC, CH2Cl2, 33% for a and 53% for b; (e) (1:1:2) Et3N/
7 (1 equiv),
MeOH/H2O, quantitative.

J.-M. Daoudi et al. / Bioorg. Med. Chem. Lett. 14 (2004) 495–498
497
Table 1. Antiviral activity and cytotoxicity data of the AMD3100 conjugates determined in various cells and with various HIV strains (X4 strain:
HIV-1 LAI and HIV-1 IIIB; R5 strain: HIV-1 Bal; dual X4/R5 strain: HIV-1 89.6)
a
b
b
b
d
Compd
EC50 in mM
[Relative potency/
CC50
(mM)
EC50c in mM
[Relative potency/
AMD3100e]
CC50
(mM)
EC50c in mM
[Relative potency/
AMD3100e]
CC50
(mM)
IC50 in mM
[Relative potency/
IC50d (mM)
AMD3100]e
AMD3100e]
CEM-SS
(HIV-1 LAI)
MT4
(HIV-1 IIIB)
PBMC
(HIV-1 89.6)
GHOST CXCR4
(HIV-1 LAI)
GHOST CCR5
(HIV-1 Bal)
2a
5a
0.91 [ꢀ7]
>100
>502.8 [
>100
>10
0.69 [ꢀ10]
ꢀ45]
>100
.47>[500
>100
>10
1000f
>100
0.025 [ꢀ7]
>100
0.0071 [ꢀ7]
0.086 [ꢀ90]
0.021 [ꢀ20]
0.00095
>50
>50
>10
>5
5.2 [ꢀ40]
6.0[ ꢀ50]
0.127
ꢀ125]
0.25 [ꢀ65]
0.0038
28
>10
>10
5b
AMD3100
2.9 [45]
0.065
0.010f
>100
0.0099
GalSerg
AZT
10
0.0063
>100
nt
0.024
nt
nt
a EC50: concentration required to inhibit 50% of virus replication by reverse transcriptase activity measurement on CEM-SS cells.
b CC50: concentration required to cause 50% death of uninfected cells.
c EC50: concentration required to protect 50% of the virus-infected MT4 or PBMC cells against virus cytopathogenicity.
d See ref 16
.
e EC50 (IC50) ratio of conjugate versus AMD3100.
f From ref 3.
g From ref 8a. nt=not tested.
as in 2a induces a substantial (ꢀ7- to 10-fold) decrease
of the antiviral activity of AMD3100. Unfortunately,
this decrease is not compensated but even enhanced by
the attachment of the anti-HIV galactosylated-serine
moiety as in the target derivatives 5a,b which display a
ꢀ40- to 125-fold lower antiviral activity than
AMD3100. However, when compared with GalSer,
conjugation of GalSer to the highly anti-HIV active
AMD3100 increases the antiviral activity of GalSer,
though moderately, in CEM-SS cells.
Acknowledgements
We are grateful to the CNRS, the Agence Nationale de
Recherche sur le SIDA (ANRS) and Sidaction for
financial support.
References and notes
1. Berger, E. A.; Murphy, P. M.; Farber, J. M. Annu. Rev.
Immunol. 1999, 17, 657.
2. De Clerq, E. Med. Res. Rev. 2002, 22, 531 and references
cited therein.
3. De Clercq, E. Mol. Pharmacol. 2000, 57, 833.
4. (a) Doms, R. W.; Moore, J. P. J. Cell Biol. 2000, 151, F9.
(b) Pozniak, A. J. HIV Ther. 2001, 6, 91. (c) LaBranche,
C. C.; Galasso, G.; Moore, J. P.; Bolognesi, D. P.; Hirsch,
M. S.; Hammer, S. M. Antiv. Res. 2001, 50, 95. (d) see
also Fujii, N.; Nakashima, H.; Tamamura, H. Exp. Opin.
Investig. Drugs 2003, 12, 185 for other CXCR4 antago-
nists.
Concerning the mechanism of inhibition of the Gal–
AMD3100 conjugates 5a,b, the experiments performed
with X4 or R5HIV-1 strains and GHOST cells geneti-
cally modified to express CD4 and CXCR4 or CCR5
indicate clearly that the Gal–AMD3100 conjugates
interact with CXCR4, blocking viral entry. Indeed,
these compounds, as well as AMD3100 and its N-alkyl-
ated derivative 2a, inhibited viral replication of only the
X4HIV-1 strain in the CXCR4 (+) GHOST cells while
they had no effect on viral replication of the R5HIV-1
strain in the CCR5(+) GHOST cells (Table 1). The
Gal–AMD3100 conjugates 5a,b were further found to
be less (ꢀ20- and 90-fold) active than AMD3100 on the
CXCR4(+) GHOST cells, in line with the results
obtained with the other CXCR4(+) cells investigated
here. To further confirm their interactions with the
CXCR4 co-receptors, it would be interesting to investi-
gate by FACS whether these conjugates can abolish the
binding of the natural SDF-1a ligand to CXCR4.
5. Yahi, N.; Sabatier, J. M.; Nickel, P.; Mabrouk, K.; Gon-
zalez-Scarano, F.; Fantini, J. J. Biol. Chem. 1994, 269,
24349 and references cited therein.
6. (a) Alfsen, A.; Iniguez, P.; Bouguyon, E.; Bomsel, M. J.
Immunol. 2001, 166, 6257. (b) Alfsen, A.; Bomsel, M. J.
Biol. Chem. 2002, 277, 25649 and 2003, 278, 5490.
7. (a) Campbell, S. M.; Crowe, S. M.; Mak, J. J. Clin. Virol.
2001, 22, 217. (b) Raulin, J. Prog. Lipid Res. 2002, 41, 27.
(c) Fantini, J.; Garmy, N.; Mahfoud, R.; Yahi, N. Exp.
Rev. Mol. Med. 2002, http://www.expertreviews.org/
02005392 h.htm.
8. (a) Faroux-Corlay, B.; Clary, L.; Gadras, C.; Hammache,
D.; Greiner, J.; Santaella, C.; Aubertin, A. M.; Vierling,
P.; Fantini, J. Carbohydr. Res. 2000, 327, 223, and refer-
ences therein. (b) Conboy, J. C.; McReynolds, K. D.;
Gervay-Hague, J.; Saavedra, S. S. Angew. Chem., Int. Ed.
2000, 39, 2882. (c) Mahfoud, R.; Mylvaganam, M.; Ling-
wood, C. A.; Fantini, J. J. Lipid Res. 2002, 43, 1670. (d)
Villard, R.; Hammache, D.; Delapierre, G.; Fotiadu, F.;
Buono, G.; Fantini, J. Chem. Bio. Chem. 2002, 3, 517. (e)
Gu, Y.; LaBell, R.; O’Brien, D. F.; Saavedra, S. S. Angew.
Chem., Int. Ed. 2001, 40, 2320and 2947. (f) McReynolds,
Although the remainder of our initial goal (i.e., devel-
opment of highly potent HIV inhibitors that block viral
entry into cells) was not reached, Gal–AMD3100 con-
jugates such as 5a,b constitute nevertheless ‘double-
drug’ type molecules worth to be extended. The synth-
esis of conjugates with longer spacers between the two
pharmacophore units is currently under progress in our
laboratory. These efforts are intended to show if the
length of the spacer between these two units is indeed
critical for a synergistic antiviral effect.

498
J.-M. Daoudi et al. / Bioorg. Med. Chem. Lett. 14 (2004) 495–498
K. D.; Bhat, A.; Conboy, J. C.; Saavedra, S. S.;
as its hydrochloride salt: ESI-MS (M+Na)+: 653.4 (calcd
for C35H66N8O2+Na=653.4).
Gervay-Hague, J. Bioorg. Med. Chem. 2002, 10, 625. (g)
Conboy, J. C.; McReynolds, K. D.; Gervay-Hague, J.;
Saavedra, S. S. J. Am. Chem. Soc. 2002, 124, 968.
15. The in vitro antiviral activity and cytotoxicity (as esti-
mated by the EC50 and CC50, respectively) of the com-
pounds in CEMS-SS, MT4 and PBMC cells were
determined according to published procedures.16 For the
CEM-SS and PBMC cells, the growing of HIV-1 (LAI
and 89.6 strain, respectively) was evaluated by measuring
the reverse transcriptase (RT). The growing of HIV-1
[HTLV-I (IIIB)] in MT4 cells was followed by the cyto-
pathogenic effect induced by the virus. The EC50 and
CC50 are defined in Table 1. GHOST cells (human osteo-
genic sarcoma) transformed to express human CD4,
CXCR4 or CCR5, and the green fluorescent protein
(GFP) under the control of an HIV-2 LTR, were cultured
in Dulbecco medium supplemented with 10% FCS. Cells
were distributed in 24-well plates at a density of 2.5 105
cells per well in 500 mL medium. Twenty four hours later,
the medium was eliminated and cells were pretreated with
different concentrations of the compounds diluted in
medium for 1 h at 37 ꢁC, then infected in presence of the
drugs with HIV-1 Lai (GHOST-CXCR4) or HIV-1 Bal
(GHOST-CCR5). Untreated cells were infected in paral-
lel. After 2 h adsorption, cells were washed and cultured
in the presence of the same concentrations of drugs for 48
h. At the end of the incubation period, cells were treated
with trypsin, washed, suspended in 700 mL 1.5% paraf-
ormaldehyde and analyzed by flow cytometry to detect
the presence of GFP resulting from LTR transactivation
mediated by Tat which is produced upon cell infection by
HIV. The percentage of infected cells expressing GFP in
the absence or in the presence of different concentrations of
antivirals was measured and the 50% inhibitory concen-
tration of cell infection (IC50) was derived from the com-
puter-generated median effect plot of the dose–effect data.
16. (a) Moog, C.; Wick, A.; Le Ber, P.; Kirn, A.; Aubertin,
A. M. Antivir. Res. 1994, 24, 275. (b) Pauwels, R.; Bal-
zarini, J.; Baba, M.; Snoeck, R.; Schols, D.; Herdewijn,
P.; Desmyter, J.; De Clercq, E. J. Virol. Meth. 1988, 20,
309. (c) Genu-Dellac, C.; Gosselin, G.; Puech, F.; Henry,
J. C.; Aubertin, A. M.; Aubert, G.; Kirn, A.; Imbach, J. L.
Nucleosides Nucleotides 1991, 10, 1345.
9. Dessolin, J.; Galea, P.; Vlieghe, P.; Chermann, J. C.;
Kraus, J. L. J. Med. Chem. 1999, 42, 229 and references
cited therein.
10. (a) Tamamura, H.; Omagari, A.; Hiramatsu, K.; Kana-
moto, T.; Gotoh, K.; Kanbara, K.; Yamamoto, N.;
Nakashima, H.; Otaka, A.; Fujii, N. Bioorg. Med. Chem.
2001, 9, 2179. (b) Tamamura, H.; Ishihara, T.; Oyake, H.;
Imai, M.; Otaka, A.; Ibuka, T.; Arakaki, R.; Nakashima,
H.; Murakami, T.; Waki, M.; Matsumoto, A.; Yamamoto,
N.; Fujii, N. J. Chem. Soc., Perkin Trans. 1 1998, 495.
11. Vlieghe, P.; Clerc, T.; Pannecouque, C.; Witvrouw, M.;
De Clercq, E.; Salles, J. P.; Kraus, J. L. J. Med. Chem.
2002, 45, 1275.
12. (a) Matsumoto, H.; Matsuda, T.; Nakata, S.; Mitoguchi,
T.; Kimura, T.; Hayashi, Y.; Kiso, Y. Bioorg. Med. Chem.
2001, 9, 417. (b) Matsumoto, H.; Kimura, T.; Hamawaki,
T.; Kumagai, A.; Goto, T.; Sano, K.; Hayashi, Y.; Kiso,
Y. Bioorg. Med. Chem. 2001, 9, 1589. (c) Van Calenbergh,
S.; Nair, V. Nucleosides Nucleotides 1999, 18, 737. (d)
Camplo, M.; Niddam, V.; Barthelemy, P.; Faury, P.;
Mourier, N.; Simon, V.; Charvet, A. S.; Trabaud, C.;
Graciet, J. C.; Chermann, J. C.; Kraus, J. L. Eur. J. Med.
Chem. 1995, 30, 789. (e) Mourier, N.; Trabaud, C.; Gra-
ciet, J. C.; Simon, V.; Niddam, V.; Faury, P.; Charvet,
A. S.; Camplo, M.; Chermann, J. C.; Kraus, J. L.
Nucleosides Nucleotides 1995, 14, 1393.
13. (a) Boitrel, B.; Andrioletti, B.; Lachkar, M.; Guilard, R.
Tetrahedron Lett. 1995, 36, 4995. (b) Fabbrizzi, L.; Gatti,
F.; Pallavicini, P.; Zambarbieri, E. Chem.-Eur. J. 1999, 5,
682.
14. The chemical structure of compounds 1a,b, 2a, 3a,b, 4a,b,
and 5a,b, was unambiguously established by 1H, 13C-
DEPT NMR, ESI-MS, and by comparison with the pub-
lished NMR data of the known compounds (i.e., 1a, 3a).
The regioisomer a of compounds 3 to 5 displays identical
NMR data to that of its respective regioisomer b. 5a,b are
isolated as TFA salts: ESI-MS (M+H)+: 1190.9 (calcd
for C62H115N11O11+H: 1190.9). Compound 2a is isolated