S. Gibaud et al. / Journal of Organometallic Chemistry 691 (2006) 1081–1084
1083
3.25–3.28 (t, 2H, CH2OH), 2.83–2.87 (dd, 1H, SCH2). 13C
NMR (DMSO-d6): d 150.24 (aromatic para), 141.56 (aro-
matic ortho), 132.59 (aromatics), 131.09 (aromatics),
115.04 (aromatic meta), 84.42 (C-I aromatic ortho), 63.81
(CH2OH), 59.09 (SCH), 42.96 (SCH2). UV (acetonitrile)
kmax 225, 275 nm. IR (KBr) tmax 385, 1066 cmꢀ1. Anal.
Calc. for C9H11AsINOS2: C, 26.04; H, 2.67; As, 18.05; I,
30.57; N, 3.37; O, 3.85; S, 15.45. Found: C, 25.53; H,
2.65; As, 18.01; I, 30.50; N, 3.20; O, 3.49; S, 14.37%.
As tested in vitro on human leukemic cell lines, the cyto-
toxic activity of each derivative was quite similar for both
cell lines but was significantly better (IC50 ranging from
0.08 0.01 to 6.69 0.14 lM) than that of arsenic trioxide
(IC50 = 24.95 1.76 lM for K562 and 16.73 1.54 for
U937; p < 0.001 vs. all derivatives). Despite the fact that
no strong structure–activity relationship was noticed, some
significant informations were obtained in this series of
compounds. The substitution of one hydrogen atom of
the NH2 group of para-arsanilic acid by various residues
(13–19) did not add any significant advantages compared
with the compound with a primary amine (20), except for
the compound 15 (substituted by 2,4-dimethoxy 1,3,5-tri-
azine ring). This compound (15) was also more cytotoxic
than melarsoprol (13) and (2-chloro-4-amino pyrimidine
ring) 16, who both carried the carbon–nitrogen
NH2ACH@NA arrangement, which was related to the try-
panocidal activity of melarsoprol [15], thus indicating that
this arrangement is not critical for antileukemic properties.
However, an additional substitution of the arsanilic acid by
an iodine atom (21) at the 2-position led to a significant
improvement of the antileukemic activity of 20 (IC50 on
U937 cells = 0.34 0.02 vs. 1.61 0.21 lM, respectively;
p < 0.001), with a limited increase of the systemic toxicity
(60 vs. 103 lmole/kg). Interestingly enough, a nitrogen or
an oxygen atom at the para-position does not seem to be
required for the cytotoxicity since the un-substituted com-
pound 24 showed high antileukemic activity. Moreover, no
correlation between the logP and the activity or the toxic-
ity was found.
2.2. Determination of logP
The logP values of the compounds were determined by
HPLC analysis of the corresponding log k0w (C18 column;
5 lm, 4.6 · 25 cm, Macherey-Nagel, Eckbolsheim, France).
The log k0w of each derivative was determined by measuring
its logk0 using several mixtures of water/acetonitrile as the
mobile phase and extrapolating back to 100% of water. A
standard curve logP vs. log k0w was constructed using the
log k0w of the compounds with known logP.
2.3. Cytotoxic activity
The cytotoxic activity of each starting compound and
each final compound was performed on K562 erythroleu-
kemia and U937 myelomonocytic leukemia cell lines.
Briefly, exponential growing cells were seeded into a 96-
well plate at a final density of 4.104/well using different con-
centrations of organoarsenic compounds (0.01 lM–1 mM).
Cells were incubated for 2 or 3 days at 37 ꢂC in a humidi-
fied 5% CO2 atmosphere. Cell viabilities were determined
using the classical MTT test [12].
Since all derivatives were more cytotoxic than As2O3,
the toxicity/activity ratio (therapeutic index T.I., LD50 in
mice/IC50 on cell lines) was used to select the lead com-
pounds. This ratio was significantly improved for all deriv-
atives when compared to As2O3 [16]. Indeed, T.I. ranged
from 6.80 2.44 (16) to 279.17 41.32 (22) vs.
2.88 0.16 (As2O3) for K562 cells and 6.62 1.63 (16)
to 325.00 20.2 (24) vs. 3.41 0.32 (As2O3). Thus, the
compounds 21, 22 and 24 which exhibited a strong
improvement of T.I., were considered as starter structures
for further studies.
2.4. Determination of the LD50
To determine the lethal dose (LD50) of each compound,
six groups of five mice, weighing 22–24 g, were housed in
cages and observed throughout the quarantine-period
experiments. Organoarsenic compounds were dissolved
and diluted in a mixture of propylene glycol, sodium chlo-
ride and DMSO (25:50:25; v/v). The mortality was assessed
at the 96th hour after injecting 0.25 mL of the solution.
The mechanism of the antileukemic properties of AsIII
derivatives (As2O3 and melarsoprol) was partially eluci-
dated during the last few years and seems quite similar
for both compounds. It was attributed to the linkage
between these compounds and the cystine moieties of
numerous proteins, especially apoptosis proteins [17]. Nev-
ertheless, some differences were shown on multiple mye-
loma cells [7]. In contrast to As2O3, melarsoprol only
slightly reduces the plasma cell differentiation of normal
B cell induced by pokeweed mitogen [7]. Both pokeweed
mitogen-induced normal plasma cells and malignant
plasma cells showed a normal nuclear distribution of
PML protein, which was disrupted by As2O3 but not by
melarsoprol, suggesting that these AsIIl organic derivatives
could act by different mechanisms. As expected, the start-
ing compounds (AsV) were not active (IC50 > 1000 lM).
3. Results and discussion
Since previous results showed that chelating the AAs@O
group by B.A.L. in phenyl arsenoso derivatives induced a
very significant increase of their therapeutic index by
reducing their systemic toxicity [13], all tested compounds
included an identical dithiarsolane ring and thus, were
derivatives of (2-phenyl-[1,3,2]dithiarsolan-4-yl)-methanol.
Actually, the compound 20 (LD50 = 103 lmol/kg) was
clearly less toxic than the corresponding As@O derivative,
4-arsenoso-phenylamine (25; LD50 = 33 lmol/kg). This
dithiarsolane ring permitted to stabilize the molecule since
the AsIII atom in the As@O group exhibits a pronounced
tendency to polymerization and oxidation [14].
Gibaud, Stéphane
