Identification of optimal anion spacing for anti-HIV activity in a series of cosalane tetracarboxylates

Article abstract of DOI:10.1016/S0960-894X(00)00417-0

The binding of the anti-HIV agent cosalane to CD4 is thought to involve ionic interactions of negatively charged carboxylates of the ligand with positively charged residues on the surface of the protein. An investigation of the optimal anion distances for anti-HIV activity in a series of cosalane tetracarboxylate analogues has been completed, and maximal activity results when the two proximal and the two distal carboxylates are separated by eight atoms. (C) 2000 Elsevier Science Ltd.

Full text of DOI:10.1016/S0960-894X(00)00417-0

Bioorganic & Medicinal Chemistry Letters 10 (2000) 2149±2152
Identi®cation of Optimal Anion Spacing for Anti-HIV Activity
in a Series of Cosalane Tetracarboxylates
a
b
b
b
Gitendra C. Paul, Erik De Clercq, Christophe Pannecouque, Myriam Witvrouw,
a,
c
c
c
Tracy L. Loftus, Jim A. Turpin, Robert W. Buckheit, Jr. and Mark Cushman *
aDepartment of Medicinal Chemistry and Molecular Pharmacology, School of Pharmacy and Pharmacal Sciences,
Purdue University, West Lafayette, IN 47907, USA
b
Rega Institute for Medical Research, Katholieke Universiteit Leuven, Minderbroedersstraat 10, B-3000 Leuven, Belgium
cVirology Research Group, Southern Research Institute, 431 Aviation Way, Frederick, MD 21701, USA
Received 5 June 2000; accepted 21 July 2000
AbstractÐThe binding of the anti-HIV agent cosalane to CD4 is thought to involve ionic interactions of negatively charged carbox-
ylates of the ligand with positively charged residues on the surface of the protein. An investigation of the optimal anion distances for
anti-HIV activity in a series of cosalane tetracarboxylate analogues has been completed, and maximal activity results when the two
proximal and the two distal carboxylates are separated by eight atoms. # 2000 Elsevier Science Ltd. All rights reserved.
Although early and aggressive intervention with combi-
nation chemotherapy has signi®cantly slowed the rate of
disease progression in AIDS, major problems still exist
of 2±4 in which the two distal carboxyl groups were
placed ortho, meta, and para relative to the linkage to the
cosalane skeleton. Maximal activity was found in com-
pound 4, in which the two carboxyl groups were para,
followed by 3 and 2. In order to precisely de®ne how far
the two distal carboxyl groups could be placed from the
two proximal carboxyl groups and still increase activity,
additional compounds in the series have now been syn-
thesized in which the distance is further extended.
1�4
in AIDS treatment, including resistance and toxicity.
Interest therefore remains in the development of new
anti-AIDS agents that would interact with viral proteins
that are not targeted by the currently available anti-
AIDS agents.
The anti-HIV agent cosalane (1) was designed by
attaching a dichlorinated disalicylmethane fragment of
aurintricarboxylic acid to C-3 of cholestane through an
alkenyl linker chain.5 The anti-HIV activity of cosa-
lane (1) results from inhibition of the binding of gp120
to CD4, as well as from the inhibition of an unidenti®ed
,6
7
postattachment event prior to reverse transcription. A
hypothetical model (Fig. 1) has been proposed for the
binding of the cosalane disalicylmethane `pharmaco-
8
phore' to CD4. The model involves the binding of the
We replaced each of the two distal carboxylic acid
groups of 4 with phenyl substituents in order to provide
two platforms for the attachment of carboxylic acid
groups farther away from the two proximal carboxylic
acid groups. The synthesis of the desired compounds
was accomplished by the alkylation of the two phenox-
ide anions derived from cosalane (1) di(methyl ester)
with the benzyl bromides 15, 16, and 17. As shown in
Scheme 1, coupling 4-methylbenzeneboronic acid (5)
with the o-, m-, and p-bromobenzoic acids 6, 7, and 8 in
the presence of palladium hydroxide and aqueous
sodium hydroxide a€orded the o-, m-, and p-tolylbenzoic
two cosalane carboxylates to the positively charged
Arg58 and Arg59 residues. Inspection of the surround-
ing region of the protein reveals several additional
positively charged residues that could be targeted by
additional carboxylates of appropriately designed cosa-
lane analogues, provided they are strategically spaced.8,9
Accordingly, we designed a series of cosalane analogues
*
6
Corresponding author. Tel.: +1-765-494-1459; fax: +1-765-494-
790; e-mail: cushman@pharmacy.purdue.edu
0960-894X/00/$ - see front matter # 2000 Elsevier Science Ltd. All rights reserved.
PII: S0960-894X(00)00417-0

2
150
G. C. Paul et al. / Bioorg. Med. Chem. Lett. 10 (2000) 2149±2152
Scheme 1. Reagents and conditions: (a) PdCl
2
, aq NaOH; (b)
, CCl
TMSCHN in hexane, C and MeOH; (c) NBS, (BzO)
6
2
H
6
2
4
.
present series 22±24 contributes three additional analo-
gues in which the spacing is 10±12 atoms. In order to
complete this series, we were therefore interested in
having an analogue in which the proximal and distal
carboxyl groups were spaced by nine atoms. The synth-
esis of the desired compound 27 is outlined in Scheme 3.
The starting material 25 was made by methylation of
commercially available 4-(bromomethyl)phenylacetic
acid with trimethylsilyldiazomethane. Deprotonation
of the two phenolic hydroxyl groups of cosalane di(me-
thyl ester) (18) with potassium carbonate, and alkylation
of the resulting anions with the benzyl bromide 25,
yielded intermediate 26 (44% yield). Hydrolysis of the
four methyl esters with potassium carbonate and a cat-
alytic amount of potassium cyanide a€orded the target
compound 27 (42% yield).
Figure 1. Hypothetical model of the binding of the cosalane `phar-
macophore' to CD4 (programmed for walleyed viewing). Blue dots:
10
acids 9, 10, and 11. These three carboxylic acids were
converted to their corresponding methyl esters 12, 13, and
1
4 with trimethylsilyldiazomethane.10 Free radical bro-
mination of 12, 13, and 14 in the presence of N-bromo-
succinimide and dibenzoyl peroxide yielded the desired
benzyl bromides 15 (90% yield from 9), 16 (90% yield
from 10), and 17 (84% yield from 11). The free radical
bromination of 13 and 14 to yield 16 and 17 was carried
out during irradiation with a 200-W tungsten lamp. Con-
version of 12 to 15 was accomplished without irradiation.
The new cosalane derivatives were evaluated for inhibi-
tion of the cytopathic e€ect of HIV-1RF in CEM-SS cells,
HIV-1IIIB in MT-4 cells, and HIV-2ROD in MT-4 cells.
Cytotoxicities in uninfected CEM-SS cells and MT-4 cells
were also determined. The results are listed in Table 1.
Selective methylation of the two carboxylic acid groups
of cosalane (1) with trimethylsilyldiazomethane a€orded
the di(methyl ester) 18 (100% yield, Scheme 2).10 The
two phenolic hydroxyl groups of 18 were deprotonated
using potassium carbonate as the base in DMF, and
both of the resulting phenoxide anions were alkylated
with the benzyl bromides 15, 16, or 17 to a€ord the
corresponding cosalane derivatives 19 (60% yield), 20
The new cosalane polycarboxylates 22, 23, 24, and 27
synthesized in the present study were generally more
potent against HIV-1RF in CEM-SS cells than they were
versus HIV-1IIIB in MT-4 cells. They were also either
inactive or displayed very low antiviral activity against
HIV-2ROD in MT-4 cells. The low activity of the com-
pounds in the present series versus HIV-2ROD is not
totally unexpected, because the previously reported
polyanions 3 and 4, also having extended polyanionic
(
60% yield), or 21 (46% yield). Hydrolysis of all four
esters present in these compounds yielded the corre-
sponding tetracarboxylic acids 22 (60% yield), 23 (38%
yield), or 24 (42% yield).
pharmacophores, were less active versus HIV-2
9
ROD
than they were versus HIV-1RF or HIV-1IIIB
.
The synthesis of the prior series of compounds 2±4
provided compounds in which the proximal and distal
carboxyl groups were spaced by 6±8 atoms, and the
Movement of the two distal carboxyl groups farther
away from the two proximal carboxyl groups in the

G. C. Paul et al. / Bioorg. Med. Chem. Lett. 10 (2000) 2149±2152
2151
ꢀ
ꢀ
Scheme 3. Reagents and conditions: (a) K
ꢀ
2
CO
3
, DMF, 23 C (22 h).
O, 80 C (2 h).
(
b) K
2
CO , KCN, aq EtOH, 80 C (14 h), and then H
3
2
Table 1. Anti-HIV activity of cosalane analoguesa
EC50 (mM)
CC50 (mM)
Scheme 2. Reagents and conditions: (a) TMSCHN
zene and methanol, 25 C (25 min); (b) 15, 16, or 17, K
2
in hexane, ben-
CO , DMF; (c)
for 22: aq NaOH, MeOH and EtOH, re¯ux (14 h); for 23: KCN,
Compound HIV-1RF HIV-1IIIB HIV-2ROD CEM-SS MT-4
ꢀ
2
3
cells
cells
ꢀ
ꢀ
K
2
CO , aq EtOH, 95 C (20 h) followed by H
3
2
O, 80 C (2 h); for 24:
ꢀ
1
2
3
4
2
2
2
5.1
39.8
5.7
3.0
>37
2.2
1.7
12.1
14.5
4.0
>37
>29
22
>125
>95
>80
>200
>316
>316
72
>200
>200
>100
133
>200
37
29
ꢀ
KCN, K
2
CO
3
, aq EtOH, 90 C (20 h) followed by H
2
O, 80 C (1 h).
0.5
88
2
3
4
5.17
6.11
>100
1.78
>125
95
80
biphenyl series 22, 23, and 24 resulted in a progressive
decrease in antiviral activity against both HIV-1 strains.
This e€ect is the reverse of what has been seen in the
benzyl series 2, 3, and 4. The most active of the new
compounds was 27, which was slightly more active than
>80
4.40
27
124.1
>125
aThe antiviral EC50 values are the concentrations required to reduce
the cytopathic e€ect of the virus by 50%, while the CC50 values are the
concentrations required for 50% reduction in cellular viability in
uninfected cells. Antiviral testing versus HIV-1RF was performed in
22, but less active than 4. A plot of the antiviral activity
versus HIV-1RF as a function of the number of atoms
separating the proximal and distal carboxyl groups is
shown in Figure 2.
CEM-SS cells, while antiviral testing versus HIV-1IIIB and HIV-2ROD
was carried out in MT-4 cells.1
least two determinations.
1,12
The values are the averages of at
The results indicate an optimal spacing of the proximal
and distal carboxyl groups that is approximated by the
distances in the cosalane analogue 4. This may re¯ect a
requirement for optimal interaction of the anionic
polycarboxylates with cationic amino acid side chains
present on the surface of CD4. A hypothetical model of
the binding of 4 with CD4, involving ionic bonding
interactions of three of the carboxylates with the side
chains of the Arg58, Arg59, and Lys72 residues is dis-
played in Figure 3.9
as CD4. Both compounds also inhibit the binding of
gp120 to CD4, as well as a postbinding event prior to
reverse transcription.7 A detailed analysis of 4 indicated
that inhibition of fusion occurs at lower concentrations
,9
9
than inhibition of attachment. Like cosalane (1), the
cosalane analogue 4 binds to both gp120 as well as
CD4, and a direct interaction with the HIV co-receptor
CXCR4 was not observed. It is possible that the inter-
action of these compounds with gp120 and CD4 inhibits
the attainment of the proper orientation of the complex
that is required for binding to CXCR4. We assume that
the new compounds reported in the present study are
Mechanism of action studies on both cosalane (1) and
the most active of the tetracarboxylates 4 have revealed
that both of these antiviral agents bind to gp120 as well

2
152
G. C. Paul et al. / Bioorg. Med. Chem. Lett. 10 (2000) 2149±2152
acting similarly to cosalane (1) and its more potent
analogue 4.
Acknowledgements
This research was made possible by NIH grant 36624,
and by grants from the Fonds voor Wetenschappelijk
Onderzoek (FWO) Vlaanderen, the Belgian Geconcer-
teerde Onderzoekacties and the Belgian Fonds voor
Geneeskundig Wetenschappelijk Onderzoek (FGWO).
We would like to thank Kristien Erven for excellent
technical assistance.
References and Notes
Figure 2. The relationship between anti-HIV activity (vs HIV-1RF in
CEM-SS cells) and the number of atoms separating the proximal and
distal carboxylates in the series of compounds 2 (6 atoms), 3 (7 atoms), 4
1. Barry, M.; Mulcahy, F.; Back, D. J. Br. J. Clin. Pharmacol.
1998, 45, 221.
2
3
1
4
5
. Fischl, M. A. AIDS 1999, 13, S49.
. Vajpayee, M.; Dar, L. Indian J. Pharmacol. 1999, 31, 96±
03.
. Beach, J. W. Clin. Ther. 1998, 20, 2.
. Golebiewski, W. M.; Bader, J. P.; Cushman, M. Bioorg.
(
8 atoms), 27 (9 atoms), 22 (10 atoms), 23 (11 atoms), and 24 (12 atoms).
Med. Chem. Lett. 1993, 3, 1739.
. Reymen, D.; Witvrouw, M.; Este, J. A.; Neyts, J.; Schols,
Â
D.; Andrei, G.; Snoeck, R.; Cushman, M.; Hejchman, E.;
6
DeClercq, E. Antiviral Chem. Chemother. 1996, 7, 142.
7. Cushman, M.; Golebiewski, W. M.; McMahon, J. B.; Buc-
kheit, R. W. J.; Clanton, D. J.; Weislow, O.; Haugwitz, R. D.;
Bader, J.; Graham, L.; Rice, W. G. J. Med. Chem. 1994, 37,
3
8
040.
. Cushman, M.; Insaf, S.; Ruell, J. A.; Schae€er, W. G.;
Rice, W. G. Bioorg. Med. Chem. Lett. 1998, 8, 833.
. Cushman, M.; Insaf, S.; Paul, G.; Ruell, J.; DeClercq, E.;
Schols, D.; Pannecouque, C.; Witvrouw, M.; Schae€er, C. A.;
9
Turpin, J. A.; Williamson, K.; Rice, W. G. J. Med. Chem.
1
999, 42, 1767.
0. Hashimoto, N.; Aoyama, T.; Shioiri, T. Chem. Pharm.
1
Bull. 1981, 29, 1475.
1. Pauwels, R.; Balzarini, J.; Baba, M.; Snoeck, R.; Schols,
1
D.; Herdewijn, P.; Desmyter, J.; DeClercq, E. J. Virol. Meth-
ods 1988, 20, 309.
Figure 3. Hypothetical model of the binding of the 4 `pharmacophore'
to CD4 (programmed for walleyed viewing). Blue dots: attraction, red
dots: repulsion.
1
2. Rice, W. G.; Bader, J. P. Adv. Pharmacol. 1995, 33, 389.