Chemical determinants of antimalarial activity of reversed siderophores

Article abstract of DOI:10.1128/aac.40.9.2160

Reversed siderophores (RSFs) are artificial hydroxamate-based iron chelators designed after the natural siderophore ferrichrome. The modular molecular design of RSF derivatives allowed the synthesis of various congeners with controlled iron-binding capacities and partition coefficients. These two physicochemical properties were assessed by a novel fluorescent method and were found to be the major determinants of RSF permeation across erythrocyte membranes and scavenging of compartmentalized iron. The partition coefficient apparently conferred upon RSFs two major features: (i) the ability to rapidly access iron pools of in vitro-grown Plasmodium falciparum at all developmental stages and to mobilize intracellular iron and transfer it to the medium and (ii) the ability to suppress parasite growth at all developmental stages. These features of RSFs were assessed by quantitative determination of the structure-activity relationships of the biological activities and partition coefficients spanning a wide range of values. The most effective RSF containing the aromatic group of phenylalanine (RSFm2phe) showed 50% inhibitory concentration of 0.60 ± 0.03 nmol/ml in a 48-h test and a 2-h onset of inhibition of ring development at 5 nmol/ml. The lipophilic compound RSFm2phe and the lipophilic and esterase-cleavable compound RSFm2pee inhibited parasite growth at all developmental stages whether inhibition was assessed in a continuous mode or after discontinuing drug administration. The antimalarial effects of RSFm2phe and clearable RSFm2pee were potentiated in the presence of desferrioxamine (DFO) at concentrations at which DFO alone bad no effect on parasite growth. These studies provide experimental evidence indicating that the effective and persistent antimalarial actions of RSFs are associated with drug access to infected cells and scavenging of iron from intracellular parasites. Moreover, the optimal antimalarial actions of RSFs are apparently also determined by improved accessibility to critical iron pools or by specific interactions with critical parasite targets.

Full text of DOI:10.1128/aac.40.9.2160

ANTIMICROBIAL AGENTS AND CHEMOTHERAPY, Sept. 1996, p. 2160–2166
0066-4804/96/$04.00ϩ0
Copyright ᭧ 1996, American Society for Microbiology
Vol. 40, No. 9
Chemical Determinants of Antimalarial Activity of
Reversed Siderophores
A. TSAFACK,1 J. LIBMAN,2 A. SHANZER,2 AND Z. I. CABANTCHIK1*
Department of Biological Chemistry, Institute of Life Sciences, Hebrew University, Jerusalem, Israel 91904,1
and Department of Organic Chemistry, Weizmann Institute of Science, Rehovot, Israel 716002
Received 2 April 1996/Returned for modification 23 May 1996/Accepted 20 June 1996
Reversed siderophores (RSFs) are artificial hydroxamate-based iron chelators designed after the natural
siderophore ferrichrome. The modular molecular design of RSF derivatives allowed the synthesis of various
congeners with controlled iron-binding capacities and partition coefficients. These two physicochemical prop-
erties were assessed by a novel fluorescent method and were found to be the major determinants of RSF
permeation across erythrocyte membranes and scavenging of compartmentalized iron. The partition coefficient
apparently conferred upon RSFs two major features: (i) the ability to rapidly access iron pools of in vitro-grown
Plasmodium falciparum at all developmental stages and to mobilize intracellular iron and transfer it to the
medium and (ii) the ability to suppress parasite growth at all developmental stages. These features of RSFs
were assessed by quantitative determination of the structure-activity relationships of the biological activities
and partition coefficients spanning a wide range of values. The most effective RSF containing the aromatic
group of phenylalanine (RSFm2phe) showed 50% inhibitory concentration of 0.60 ؎ 0.03 nmol/ml in a 48-h test
and a 2-h onset of inhibition of ring development at 5 nmol/ml. The lipophilic compound RSFm2phe and the
lipophilic and esterase-cleavable compound RSFm2pee inhibited parasite growth at all developmental stages
whether inhibition was assessed in a continuous mode or after discontinuing drug administration. The
antimalarial effects of RSFm2phe and cleavable RSFm2pee were potentiated in the presence of desferrioxam-
ine (DFO) at concentrations at which DFO alone had no effect on parasite growth. These studies provide
experimental evidence indicating that the effective and persistent antimalarial actions of RSFs are associated
with drug access to infected cells and scavenging of iron from intracellular parasites. Moreover, the optimal
antimalarial actions of RSFs are apparently also determined by improved accessibility to critical iron pools or
by specific interactions with critical parasite targets.
Like all living organisms, malaria parasites depend on iron
for growth and replication. This property makes them suscep-
tible to iron deprivation, which can be induced by treatment
with iron chelators (4, 5), as shown in in vitro cultures of
Plasmodium falciparum (17, 23, 26), in animal models of ma-
laria (11, 18, 22), and in human trials (3, 14, 30) with the
clinically approved agent desferrioxamine (DFO). However,
despite its advantages, DFO lacks the requisite speed of action
and therapeutic efficacy to serve as a reliable substitute or
additive for the treatment of severe cases of malaria, in par-
ticular, multidrug-resistant strains (33). Most other synthetic
or naturally occurring chelators showed significant weaknesses
as well, because they were apparently not free of undesirable
side effects (15, 25).
Despite a plethora of studies on iron chelation and malaria,
it is still unclear which factors are responsible for the high
degree of susceptibility of protozoan parasites to chelator-
induced iron deprivation relative to the susceptibility of mam-
malian organisms. Recently, we applied a series of synthetic
chelators of variable lipophilic-hydrophilic balance (28) whose
antimalarial actions were assessed as a function of parasite
stage dependence, time of drug exposure and dosage, and
reversibility of action. New information gained from this ap-
plication allowed us to propose a working model for iron ch-
elator action on parasites (4, 5, 20). In that model, the primary
stage of iron utilization was the trophozoite stage, in which the
metal was mobilized from degraded hemoglobin (5), as re-
cently proposed (1, 12). The iron chelators which were re-
tained within the parasites showed the most persistent inhibi-
tory effects whether they were applied before, during, or after
the stage of major iron utilization (5, 20, 31). An advantage of
the model was its ability to predict the antimalarial potencies
of combinations (5, 13) of chelators with different permeation
properties, which could generate a synergistic mechanism of
action (31).
An equal degree of uncertainty exists regarding the factors
which contribute to the antimalarial performance of iron che-
lators (4, 5). The single empirical factor which has been as-
sessed systematically in terms of its contribution to antimalar-
ial efficacy was the drug partition coefficient (PC) (4, 19, 25).
Moreover, structure-activity relationships (SARs) were only
established for a group of reversed siderophores (RSFs) (19,
28), while permeation into cells was only studied with a se-
lected group of fluorescent iron chelators (4) and with 59Fe
complexes of RSFs and DFOs (19).
In the present work we expanded the gamut of RSFs with
agents of higher lipophilicities and hydrophilicities, including
hydrophobic RSFs with enzymatically cleavable ester groups
that can generate hydrophilic-impermeant RSFs intracellu-
larly. Our studies indicate that whereas all RSFs were effective
iron scavengers in solution, their PCs determined their ability
to penetrate into cells, scavenge iron, and exert an antimalarial
effect. This was apparent in the SAR schemes. SARs were
assessed over a wide range of drug lipophilicities, as given by
the PC and model membrane permeation constant. The
marked antimalarial effect of RSFm2phe, the most lipophilic
RSF tested, led us to propose that its therapeutic efficacy might
involve either a unique mechanism of drug accessibility to
* Corresponding author. Phone: 972-2-6585420. Fax: 972-2-6586974.
Electronic mail address: ioav@cc.huji.ac.il.
2160

VOL. 40, 1996
REVERSED SIDEROPHORES AS ANTIMALARIAL AGENTS
2161
The hydroxamate H2NCHBnCONOHMe was prepared by dissolving 1.25 g of
hydroxamate in 80 ml of absolute ethanol and hydrogenation for 3 h under
atmospheric pressure at room temperature in the presence of 550 mg of 5%
Pd/C. Filtration and concentration provided 550 mg of amino hydroxamate.
The Tris hydroxamate RSFm2phe was obtained by reacting 1.7 g of the parent
Tris carboxylate active ester (9) dissolved in 40 ml of dry methylene chloride with
a solution of 1.65 g of amino hydroxamate, 200 mg of N-hydroxysuccinimide, and
100 mg of imidazole in 30 ml of methylene chloride at room temperature. The
overnight reaction yielded a mixture which was concentrated and chromato-
graphed through silica (with CHCl3-MeOH as the eluent) to provide the pure
(ϳ99%) Tris hydroxamate.
Chemical properties of chelators. The chelator structures were determined by
infrared, UV-visible, circular dichroism, and nuclear magnetic resonance imag-
ing methods as described elsewhere (8, 34).
(i) PCs. Estimation of the PCs of the various RSFs was based on determina-
tions of the RSFs’ distributions in octanol:water (saline) after overnight incuba-
tion as described previously (28).
critical iron-containing components of the parasite or a specific
interaction with them.
MATERIALS AND METHODS
Materials. The chemical structures of the RSFs used in the study are depicted
in Fig. 1. The agents were synthesized and chemically analyzed by modifications
of previously published methods (8). They comprise the basic compound RSFm2
and the analogs RSFm2pee and RSFm2pa, whereby pee denotes the propionic
acid ethyl ester and pa denotes the propionic acid moieties extending from the
hydroxamate groups of the binding cavity. RSFm2phe carried a two-carbon (m2)
spacer linking the amino acid phenylalanine to the tripodal anchor. RSFm2(H)
denotes the demethylated analog of RSFm2. DFO was obtained from Ciba-
Geigy (Basel, Switzerland). All other chemicals were from Sigma Co. (St. Louis,
Mo.) or were the best available grade. [3H]hypoxanthine was from the Radio-
chemical Centre (Amersham, Little Chalfont, United Kingdom). All solutions
were prepared in deionized, charcoal-filtered water.
Synthesis of chelators. (i) Preparation of Tris hydroxamate RSFm2(H). The
protected Tris hydroxamate was prepared by dissolving 6.0 g of the parent triacid
(9) in 25 ml of dry acetonitrile to which a solution of 7.0 g of O-benzylhydrox-
ylamine in 10 ml of acetonitrile was added; this was followed by the addition of
300 mg of hydroxybenzotriazole and 3.6 ml of diisopropyl carbodiimide under ice
cooling. The resulting mixture was allowed to warm to room temperature and
was left for 2 days under stirring. The resulting crude reaction mixture was
concentrated to dryness and was purified by chromatography on silica gel (the
eluent was CHCl3-iPrOH, where iPr is isopropanol; yield, 2.6 g of protected Tris
hydroxamate).
A sample of 200 mg of protected Tris hydroxamate dissolved in 30 ml of
ethanol was treated with 80 mg of Pd/C (1/10) and was hydrogenated under
atmospheric pressure at room temperature for 3.5 h. Filtration of the mixture
and concentration of the filtrate provided 104 mg of Tris hydroxamate
RSFm2(H).
Preparation of Tris ester RSFm2pee. The parent tris acyl halide was prepared
from 6.0 g of the parent triacid (9), which was dissolved in 30 ml of dry,
ethanol-free chloroform (dried by passing through basic alumina) and which was
treated dropwise with 12 ml of oxalyl chloride and 60 drops of dimethyl form-
amide, whereupon gas evolution was observed. The reaction mixture was stirred
overnight at room temperature and was subsequently concentrated in vacuo to
dryness under conditions of exclusion of moisture. The crude material was
characterized by its infrared absorption at 1,792 cmϪ1 (in dry chloroform). Half
of the crude triacyl halide was dissolved in 15 ml of dry and ethanol-free chlo-
roform and was treated dropwise (under ice cooling) with a solution of 7.5 g of
HNOBnCH2CH2COOEt (where Bn is benzyl and Et is ethyl) (34) in 20 ml of
chloroform and with 5.8 ml of Et3N. The reaction mixture was allowed to warm
up to room temperature and was left under stirring overnight. Completion of the
reaction was determined by infrared analysis (disappearance of the acyl halide
peak at ca. 1,792 cmϪ1). The crude product was concentrated in vacuo; sus-
pended in 300 ml of ethyl acetate; washed twice with water, twice with 1 N
aqueous HCl, twice with water, twice with 1 N aqueous NaHCO3, and again twice
with water; dried with MgSO4; filtered; and concentrated. Chromatography on
silica gel (with chloroform with increasing amounts of ethyl acetate as the eluent)
provided 3.18 g of pure protected Tris ester product.
The Tris ester of RSFm2pee was prepared by dissolving 720 mg of the pro-
tected Tris ester in 50 ml of ethanol and treating it with H2 (atmospheric
pressure) and 280 mg of Pd/C (10%) at room temperature. The suspension was
filtered, and the filtrate was concentrated to provide 460 mg of RSFm2pee.
The protected Tris acid of RSFm2pa was prepared by dissolving a sample of
100 mg of the protected Tris ester in 20 ml of methanol and treating the mixture
with 2 ml of 1 N aqueous NaOH for 1 h and subsequently with additional
portions of 2 ml of 1 N aqueous NaOH until all material had been hydrolyzed
(CHCl3-MeOH [where Me is methyl], 9-1; thin-layer chromatography). Then,
the solution was acidified, concentrated, treated again with base (1 N NaHCO3)
to pH 8, washed with ethyl acetate, acidified, and extracted with ethyl acetate.
Washing with water and drying with MgSO4 provided the protected Tris acid.
The Tris acid RSFm2pa was prepared by dissolving 1.19 g of protected Tris
acid in 60 ml of absolute ethanol and reacting it with H2 under atmospheric
pressure for 1.5 h in the presence of 350 mg of Pd/C (10%). The resulting mixture
was filtered and concentrated to provide 825 mg of Tris acid RSFm2pa (ϳ99%
purity).
(ii) Iron-binding affinity of chelators in solution. The iron-binding affinity
method was based on the application of the probe calcein (CA), whose fluores-
cence (488- and 517-nm excitation and emission maxima, respectively) is
quenched by stoichiometric concentrations of iron upon binding (2). Fluores-
cence recovery obtained by the addition of a competing chelator and the rate and
extent of recovery provide a measure of the RSFs’ ability to scavenge iron from
CA-iron complexes. The measurements were done in N-2-hydroxyethylpipera-
zine-NЈ-2-ethanesulfonic acid (HEPES)-buffered saline (HBS; pH 7.2; 20 ␮mol
of HEPES per ml and 150 ␮mol of NaCl per ml). The fluorescence of 0.5 nmol
of CA per ml was prequenched with 1 nmol of a freshly prepared solution of
ferrous ammonium sulfate per ml. The rate of restoration of the fluorescence was
followed by using a concentration of 5 nmol/ml for each chelator. Fluorescence
was measured in either a PTI station (PhotoMed, Wede, Germany) or a Spex
Fluorolog II Spectrometer (Spex Industries, Edison, N.J.).
(iii) Scavenging of iron from resealed erythrocyte ghosts. The CA incorpo-
rated into erythrocyte ghosts encapsulation, as described previously (6), was
induced to lose virtually all its fluorescence by the addition of membrane-
permeant iron(II) added as ferrous ammonium sulfate (FAS). Removal of re-
sidual extracellular iron and CA was done by washing the ghosts with a solution
containing the nonpermeant chelator diethylenetriaminepentaacetic acid and gel
filtration through a Sephadex G-50 coarse column (Pharmacia, Uppsala, Swe-
den). An aliquot of resealed ghosts (107/ml) was diluted in HBS, and fluores-
cence was recorded with time as described above. The rate of fluorescence
recovery induced by a given chelator provided a measure of the permeation of
the drug into the ghost, with maximal fluorescence recovery attained by the
addition of the fast-permeating chelator salicylaldehyde isonicotinoyl hydrazone
(SIH) (23 ␮g/ml). The pseudo-first-order rate constant of fluorescence recovery
(k) was calculated by linear regression of the normalized traces on the basis of
the equation ln([Fϱ-Ft]/[Fϱ-F ]) ϭ Ϫkt, where F0, Ft, and Fϱ represent the
fluorescence intensities at tim0es zero, t, and infinity (after the addition of SIH),
respectively.
Biological properties of RSFs. (i) Antimalarial activity. The cultivation
method used was a modified version of that of Trager and Jensen (29), as
described elsewhere (28). Parasitemia and growth stage distribution were deter-
mined on methanol-fixed and Giemsa-stained smears. The antimalarial activities
of the free RSFs or their respective iron complexes were assayed as described
previously (28, 31). Briefly, the compounds were added from concentrated stock
solutions in dimethyl sulfoxide to microculture plates (24 wells; Costar, Cam-
bridge, Mass.) containing infected erythrocytes (2.5% hematocrit and 2 to 5%
parasitemia). Twenty-four hours after incubation with the indicated drug, the
cells were supplemented directly with 6 ␮Ci of [3H]hypoxanthine per ml or the
drugs were washed off three times with a large volume of RPMI 1640 medium
and were replenished with fresh growth medium before the addition of radiola-
bel. All systems were run in sextuplicate. Parasite growth was assessed for a
further 24 h by harvesting the frozen-thawed lysate of labeled cells onto glass
fiber filters (Tamar Inc., Jerusalem, Israel) on which the macromolecules were
deposited. The incorporation of label into the nucleic acids was measured on a
Beckman scintillation counter. The 50% inhibitory concentrations (IC50s) were
determined by nonlinear least-square fit to sigmoidal functions (10). All data are
given as average Ϯ standard errors (SEs). Inhibitory actions resulting from the
combined actions of two different drugs were assessed by the additive and
independent concept of mode of action as described by Poch (21). The theoret-
ical values for each of these two modes of action of drug combinations were
statistically compared with the experimental ones by t tests by using the Jandel
Scientific software Sigma Plot for Windows.
(ii) Mobilization of cell labile iron. P. falciparum cultures of 15 to 30%
parasitemia were pooled, and the various parasitic stages were separated over a
Percoll gradient as described elsewhere (24) by using alanine instead of sorbitol.
The cells were washed and resuspended in RPMI 1640 without phenol red
growth medium, supplemented as described above for normal cultures with
HEPES, glucose, NaHCO3 and heat-inactivated human plasma. The number of
cells was determined by counting in an improved hemocytometer chamber
(Neubauer). The iron mobilized from cells by iron chelators was determined
at different time intervals after mixing the extracellular medium with an
equal volume of acetone, followed by overnight incubation at Ϫ70ЊC, for
Preparation of RSF-Phe. Pentachlorophenolate CbzNHCHBnCOOC6Cl5 was
prepared by dissolving 8.0 g of Cbz-phenylalanine in 150 ml of acetonitrile and
treating the mixture with 7.2 g of pentachlorophenol and 320 mg of dimethyl-
aminopyridine and cooling the mixture to ice temperature (4ЊC) and reacting it
with 4.2 ml of diisopropyl carbodiimide overnight at room temperature. The
mixture was concentrated and filtered through neutral alumina (containing 5%
water) to provide 10.36 g of active ester.
The hydroxamate CbzNHCHBnCONOHMe was prepared by dissolving 4 g of
phenolate dissolved in 20 ml of dry methylene chloride and treating the mixture
overnight with a suspension of 630 mg of N-methylhydroxylamine hydrochloride,
1.1 ml of triethylamine, and 85 mg of N-hydroxysuccinimide in 15 ml of meth-
ylene chloride. Purification by chromatography over silica provided the pure
hydroxamate.

2162
TSAFACK ET AL.
ANTIMICROB. AGENTS CHEMOTHER.
FIG. 1. Chemical structures of RSFs. RSFm2 and RSFm2(H) are the methylated and demethylated forms of members of the RSF tripodal family (8, 28) which form
the hydroxamate iron-binding cavity. RSFm2pee and RSFm2pa are RSFm2 analogs possessing propionic acid ethyl ester and propionic acid residues, respectively, that
extend from the hydroxamate iron-binding cavity. RSFm2phe is a structural congener which carries phenylalanine residues between the anchor and the hydroxamate-
binding cavity.
deproteinization. The labile iron in infected and noninfected cells was assessed
drug, starting from ring stages), which is given in terms of the
by cell lysis in the presence of a protease inhibitor cocktail (17 ␮g of phenyl-
IC50 of the drug.
methylsulfonyl fluoride per ml, 10 ␮g of benzamidine per ml, 3.7 ␮g of tosyl
phenylalanyl chloromethyl ketone per ml, 3.7 ␮g of tosyl lysylchloromethyl ke-
The quantitative SARs of the RSFs as antimalarial agents
assessed in in vitro cultures are given in Fig. 2 in terms of
inhibitory potency versus the agent’s lipophilicity.
tone per ml, 10 ␮g of leupeptin per ml, and 1 ␮g of pepstatin per ml) and the
indicated chelator. Cell lysis was done by three cycles of freezing in liquid
nitrogen and thawing in a water bath. The lysate was cleared from the particulate
material by centrifugation at 10,000 ϫ g and 4ЊC and was deproteinized as
described above. Estimation of the amount of labile or chelated iron was carried
out by the ferrozine method (7) with fresh preparations of FAS as the standard.
Ferrozine [3-(2-pyridyl)-5,6-bis(4-phenylsulfonic acid)-1,2,4-triazine] solution
(247 ␮g/ml) was prepared in 9.2 ␮g of mercaptoacetic acid per ml–5.8 ␮g of
thiourea per ml. The pH was adjusted to 5 in order to minimize the probable
contamination by heme iron. An equal volume of this solution was added to the
test samples for generation of the iron-ferrozine complex. The absorption was
read against the ferrozine solution at 562 nm in a Milton Roy Spectronic 3000
Array spectrophotometer. Samples of cell-free medium were treated and used
for subtraction of the background iron in the medium. Medium from cells which
were not exposed to the respective chelators had essentially the same iron levels
as cell-free medium. Data are given as averages Ϯ SEs.
All the RSFs used in the present study had similar iron-
binding affinities (Table 1), and their iron-bound complexes
had no inhibitory effects on P. falciparum in culture (data not
shown). Hence, the major determining factor in the antima-
larial potency was apparently the agent’s ability to access crit-
ical intracellular iron pools, as determined by the PC. The
SARs held for all RSFs used in previous studies (19, 28) and in
the present one, which included the most extreme examples of
lipophilic and hydrophilic congeners.
An additional indicator of drug potency, which is dictated by
the drug permeation properties, is the time required by a given
drug concentration to exert an inhibitory effect on parasite
RESULTS
Figure 1 depicts the various derivates of RSFs that were
used in the study. All RSFs used in the study have the tripodal
form, possessing either substituents that extend from the hy-
droxamate-binding cavity or an aromatic amino acid (phenyl-
alanine) as a bridge between the triscarboxylate and the hy-
droxamate-binding cavity. The compounds examined included
RSFm2(H), which represents the demethylated form of
RSFm2(CH3) (referred to in this work as RSFm2) and the
RSFm2pee and RSFm2pa analogs.
The chemical properties of the various RSFs, which are
depicted in Table 1, comprised the octanol:water (saline) PCs
and the iron-binding affinities of the chelators relative to that
of DFO. Table 1 also depicts the permeability constants for the
respective RSFs measured in erythrocyte membrane ghosts
and the antimalarial activity (48 h of continuous exposure to
TABLE 1. Properties of RSFs
Octanol:water
PC
Relative iron
bindinga
k
IC50
RSF
(sϪ1 [10Ϫ4])b,c
(nmol/ml)c,d
m2
m2(H)
m2pa
m2pee
m2phe
1.7
0.11
0.12
13.5
40
0.65
0.44
0.58
0.76
1
1.80 Ϯ 0.06
1.10 Ϯ 0.04
0.70 Ϯ 0.05
3.2 Ϯ 0.1
3.0 Ϯ 1.0
92 Ϯ 18
275 Ϯ 25
2.9 Ϯ 0.3
0.6 Ϯ 0.3
3.9 Ϯ 0.8
a Iron binding is expressed relative to a value of 1.16 for DFO (19).
b The permeation rate constant (k) is for entry into resealed erythrocyte
membrane ghosts.
c Data are given as averages Ϯ SE.
d IC50s were determined at 48 h by nonlinear least-square fit to sigmoidal
functions (10).

VOL. 40, 1996
REVERSED SIDEROPHORES AS ANTIMALARIAL AGENTS
2163
FIG. 4. Quantitative SAR of the in vitro antimalarial actions of the RSFs
(speed of action). The correlation between the speed of action of the RSFs
(given as percent inhibition of nucleic acid synthesis after 6 to 8 h of exposure
relative to the IC50 measured at 48 h of exposure) and the PCs is presented. The
line is the regression line for an apparent linear fit (R ϭ 0.77, n ϭ 10, P Ͻ 0.01).
D and L refer to the respective amino acid isomer in the RSF structure (27).
FIG. 2. Quantitative SAR of the in vitro antimalarial actions of RSFs (in-
hibitory potencies). The antimalarial actions of the RSFs, given in terms of the
reciprocals of the IC50s, were correlated with the respective PCs (data taken
from Table 1). The line corresponds to the regression for an apparent linear fit
(R ϭ 0.9, n ϭ 12, P Ͻ 0.0001). D and L refer to the respective amino acid isomer
in the RSF structure (27).
(full cycle of growth). This normalized measure of speed of
action was found to correlate with the PCs for all the agents
which had detectable effects on parasite growth.
development. As depicted in Fig. 3, the onset of inhibition of
nucleic acid synthesis differed for the various chelators used at
equal concentrations. In cultures composed of early rings, the
most lipophilic agents affected parasites within 4 h of exposure.
We have assessed the speed of action of the drug by following
nucleic acid synthesis in the presence of various concentrations
of chelators for 6 h. The normalized speed of action presented
in Fig. 4 was determined as the percent inhibition attained at
6 h divided by the IC50 of the drug obtained in the 48-h test
The relatively high degrees of potency of the lipophilic iron
chelators at early stages of parasite development prompted us
to assess also their modes of action on the parasites, following
the working model presented elsewhere (4, 20). The first pre-
diction is related to the production of persistent inhibitory
effects at the stage of parasite development in which iron
mobilization is limited, i.e., the ring stage (4). Parasites at the
ring stage were exposed to various concentrations of
RSFm2phe and RSFm2pee for 24 h, and nucleic acid synthesis
was assessed in the continued presence of drug or after re-
moval of the drug from the growth medium. The two experi-
mental systems are denoted either continuous treatment or
discontinuous treatment, respectively (Fig. 5). The data indi-
cate that for either agent administered at the ring stage, most
of the inhibitory effect was essentially irreversible. This was
particularly the case for RSFm2pee, which although it was
relatively less potent than RSFm2phe, it had the advantage of
potentially serving in the cells as a precursor of the imper-
meant RSFm2pa. Thus, the putatively cell-generated
RSFm2pa apparently induced persistent effects on parasite
development because of its retention within the cell.
The second predictive feature of the model was the possi-
bility that RSFs may potentiate the inhibitory action of the
poorly permeant and slowly acting DFO (31). We analyzed the
effects of RSFm2pee and RSFm2phe in combination with
DFO at relatively ineffective concentrations (5 and 10 nmol/
ml) on parasite growth using Poch’s analytical method (21) for
distinguishing between additive and independent modes of
action of pairs of drugs (Fig. 6).
The data indicated that for either drug, the combined action
of the drug in combination with DFO used at biologically
ineffective concentrations was statistically more potent than
the predicted additive or independent effects. Such higher than
additive levels of action of RSFs and DFO reinforced the
validity of the phenomenon which was observed with other
RSF congeners (13, 31) and the suggested modes of action of
the drugs (31).
FIG. 3. Time course of inhibition of nucleic acid synthesis by selected RSFs.
The time dependence of nucleic acid synthesis is given relative to that for the
control at 24 h, as followed in synchronized cultures (starting at the ring stage)
in the presence of the indicated RSFs (5 nmol/ml). The control and RSFm2pa
gave essentially indistinguishable results. Datum points are depicted as means Ϯ
standard errors of the means.
The third predictive feature of the model is the commonly

2164
TSAFACK ET AL.
ANTIMICROB. AGENTS CHEMOTHER.
FIG. 6. Combined actions of RSFs and DFO. DFO was used in combination
with various concentrations of either RSFm2phe or RSFm2pee on a full cycle of
parasite growth (starting at the ring stage). The percentage of growth inhibition
(mean Ϯ standard error of the mean; n ϭ 6) was plotted against the drug
concentration. Full lines represent sigmoidal fits through experimental points.
Poch’s analyses (21) for either the additive or the independent actions of the
drugs were considerably shifted to the right of the experimental lines (shown only
for RSFm2pee). The respective IC50s of RSFm2phe (A) were as follows: (i)
experimental, 0.60 Ϯ 0.03 nmol/ml (DFO ϭ 0, nmol/ml), 0.31 Ϯ 0.01 nmol/ml
(DFO ϭ 5 nmol/ml), and 0.07 Ϯ 0.01 nmol/ml (DFO ϭ 10 nmol/ml); (ii)
theoretical-additive (Poch [21], 0.46 Ϯ 0.07 nmol/ml (DFO ϭ 5 nmol/ml) and
0.16 Ϯ 0.01 nmol/ml (DFO ϭ 10 nmol/ml); (iii) theoretical-independent (Poch
[21]), 0.59 Ϯ 0.02 nmol/ml (DFO ϭ 5 nmol/ml) and 0.25 Ϯ 0.04 nmol/ml (DFO ϭ
10 nmol/ml). The theoretical values for the additive and independent models
were significantly different from the experimental data (P Ͻ 0.001). The respec-
tive IC50s of RSFm2pee (B) were as follows: (i) experimental, 2.9 Ϯ 0.3 ␮g/ml
(DFO ϭ 0 nmol/ml) and 1.58 Ϯ 0.25 ␮g/ml) (DFO ϭ 5 nmol/ml); (ii) theoretical-
additive (Poch [21]), 2.38 Ϯ 0.05 ␮g/ml (DFO ϭ 5 nmol/ml); (iii) theoretical-
independent (Poch [21]), 2.60 Ϯ 0.08 ␮g/ml (DFO ϭ 5 nmol/ml). The theoretical
values for the additive and independent models were significantly different from
the experimental data (P Ͻ 0.001).
FIG. 5. Reversible and irreversible modes of action of RSFs. RSFm2phe (A)
and RSFm2pee (B) were assessed on synchronized ring-stage parasites (in sex-
tuplicate) treated with the indicated concentration of drug for 24 h. After wash-
ing with medium, half of the cultures were incubated with the same concentra-
tion of drug (continuous treatment) and half were incubated with medium alone
(discontinuous treatment). Growth inhibition (mean Ϯ standard error of the
mean) is given in terms of nucleic acid synthesis relative to the value obtained for
control cultures (no drug).
accepted, but hitherto not experimentally shown, correlation
between the inhibitory potency of the drug and its capacity to
extract iron from infected cells. This was carried out in the
present study with various RSFs applied at the different stages
of parasite development. Parasites were exposed for 2 to 4 h to
either RSF and were analyzed for the iron that they extracted
from the cell into the medium. Figures 7 and 8 depict data
obtained for RSFm2phe alone, and Table 2 compiles the data
obtained for the different drugs.
As shown in Figure 6, the amount of iron extracted into the
medium increased with time of exposure to the drug and varied
with the stage of parasite growth. The amount of iron extracted
after 2 h of exposure to chelator was highest at the trophozoite
stage and lowest at the schizont stage. This amount further
increased over the subsequent 2 h (4 h total), particularly in
trophozoites. Since, as shown before, RSFm2phe penetrates
very fast across membranes and scavenges intracellular iron,
the amount of iron extracted between 2 and 4 h provides a
measure of the release of free iron within the parasite. Indeed,
the total amount of chelatable or labile iron present in infected
cells (Fig. 8), that is, the amount of iron which chelators can
extract when added to lysed cells, was also highest for the
trophozoites stage, but it was also significant for the ring stage.
The high value of iron availability and scavenging found at the
trophozoite stage is in line with both the high levels of hemo-
globin degradation and the chemical release of iron induced by
glutathione and other factors, characteristic of that stage of
development (1, 12).
In Table 2 we present the values of iron mobilization by all
the RSFs used in the study and assessed at all stages of parasite
development. Impermeant agents such as RSFm2pa or
RSFm2(H) were virtually ineffective in extracting iron at any
stage of parasite growth. For all other agents, the data indi-
cated that the mobilization of iron into the medium was highly
dependent on the drug’s lipophilicity (Table 1) and iron avail-
ability (labile iron) at the particular stage of parasite develop-
ment (Fig. 7). The relatively higher level of extraction of iron
from parasitized cells by RSFm2 compared with that by
RSFm2pee, despite RSFm2pee’s higher degree of lipophilicity,
could be attributed, in part, to some intracellular formation of
impermeant RSFm2pa from RSFm2pee or to differences in

VOL. 40, 1996
REVERSED SIDEROPHORES AS ANTIMALARIAL AGENTS
2165
TABLE 2. Iron mobilization by RSFs in normal and infected cells
Iron mobilizationa
RSF
NRBCb
Rings
Trophozoites
Schizonts
m2
0.8 Ϯ 0.3
2.2 Ϯ 0.7
1.2 Ϯ 0.4
0
2.6 Ϯ 0.8
3.7 Ϯ 1.8
3.1 Ϯ 1.2
2.1 Ϯ 0.5
Ͻ0.1
3.4 Ϯ 1.9
8.5 Ϯ 2.6
3.0 Ϯ 1.0
1.0 Ϯ 0.3
0
0.7 Ϯ 0.2
3.0 Ϯ 0.7
m2(H)
m2pa
m2pee
m2phe
0
0
0.5 Ϯ 0.2
1.7 Ϯ 0.5
a Data are mean Ϯ standard error of the mean of at least three separate
experiments, given as nanomoles of Fe per 1010 cells, mobilized into the growth
medium after 4 h incubation of the cells with 20 nmol of the indicated chelators
per ml. Statistical comparison by the Student t test showed that the amount of
iron mobilized from trophozoites was significantly (P Ͻ 0.001) greater than that
from normal erythrocytes. Iron mobilization by RSFm2phe from trophozoites
was significantly (P Ͻ 0.005) different from that from rings. Data are given as
average values Ϯ SEs.
FIG. 7. RSF-mediated extraction of iron from parasitized cells. Iron mobili-
zation into the medium by RSFm2phe at 2 and 4 h of exposure of different
developmental stages of P. falciparum is presented. The iron-chelator complex
mobilized into the medium was monitored by the ferrozine method as described
in Materials and Methods. NRBC, normal erythrocytes.
b NRBC, normal erythrocytes.
In the work described here we sought to further improve the
antimalarial performance of RSFs, first, by widening the range
of drug lipophilicity and, second, by designing lipophilic RSFs
which can potentially generate in the cell RSFs of impermeant
character. We have done that on the basis of quantitative SAR
schemes and on the basis of studies of RSFs’ modes of action,
which relied on the chelator’s ability to penetrate into the cells
and extract iron, thus inducing a rapid intracellular iron deficit
(28). This mode of action prevails in P. falciparum-infected
cells because of the stage-specific and metabolically restricted
capacity of the parasite to mobilize iron, primarily from the
degradation of host proteins (16, 26). Chelators also affect
mammalian cells, although most of them, in particular, the
hydroxamate-based ones, act cytostatically, that is, only as long
as they are present in the medium (5). This is because mam-
malian cells apparently have the capacity to replenish the cell
iron stores upon removal of the chelator (5) by receptor-me-
diated endocytosis of transferrin. This differential action con-
stitutes the basis for the selective cytotoxicity of the hydrox-
amate-based iron chelators (4, 5).
We have broadened the range of SARs by incorporating
RSFs of more lipophilic (e.g., RSFm2phe and RSFm2pee) and
more hydrophilic [e.g., RSFm2(H) and RSFm2pa] character
(Fig. 1; Table 1). The chemical substitutions had a relatively
small effect on the RSFs’ rates and affinities of iron binding
(Table 1). However, the antimalarial performance conformed
to the lipophilic character of the new RSFs, as evident from the
drug SAR studies on parasite growth (Fig. 2 and 3) and their
ability to extract iron from infected cells (Table 2). Moreover,
the use of potentially cleavable agents such as RSFm2pee
might be advantageous on the basis of their potential to gen-
erate intracellularly impermeant chelators and lead to more
persistent antimalarial effects.
molecular size. Yet, the antimalarial efficacies of these two
agents were not significantly different.
DISCUSSION
The present study constitutes a new effort in our pursuit of
antimalarial agents with improved biological performance. It
focuses on a family of hydroxamate-based agents whose mo-
lecular structures could be modulated in order to confer on the
molecules defined chemical properties (28). The RSFs were
previously shown to curb malaria parasite growth in vitro in a
manner commensurate with their PCs (20, 28). This chemical
property was shown to contribute to the speed of action, the
efficacy, and the stage specificity of the drug (20, 31). A pre-
vious SAR study with RSFs showed a positive correlation be-
tween IC50 and PC, although the repertoire of RSFs spanned
a limited range of PC values (19, 28). The most hydrophobic
RSFs markedly outperformed the classical hydroxamate DFO
(5, 31) in all the biological parameters mentioned above. How-
ever, despite their relative faster action profiles and wider
stage specificities, RSFs were less effective than DFO in pro-
ducing persistent effects, particularly at the most metabolically
active and drug-sensitive stage of parasite growth, the tropho-
zoite stage (19). In recent studies we have combined the
unique properties of hydrophobic RSFs and DFO and ob-
tained major improvements in their antimalarial actions (13,
31).
Although the SAR studies indicate that RSFs’ antimalarial
efficacies are apparently limited by their ability to access crit-
ical iron pools or components, it is not clear whether mem-
brane permeation per se is the sole determinant of that effi-
cacy. The PC has conventionally been identified with the ability
of molecules to permeate membranes (19). Consistent with
this notion are our data on high-affinity iron-binding RSFs in
erythrocyte membranes (Table 1) and other classes of iron
chelators studied in both erythrocyte ghosts and living cells (32,
35). However, chelators with high PCs and similar iron-binding
affinities which reach transmembrane equilibrium on a com-
parable time scale markedly differ in their iron extraction ca-
pacities and antimalarial efficacies. This observation implicates
FIG. 8. Total labile cell iron and its RSF-mediated extraction from parasit-
ized cells. The amount of iron mobilized into the medium (hatched portion) after
4 h of exposure of the cells to RSFphe is indicated. Data are given as values
relative to that for the total labile cell iron. NRBC, normal erythrocytes.

2166
TSAFACK ET AL.
ANTIMICROB. AGENTS CHEMOTHER.
microb. Agents Chemother. 39:61–65.
PC, but not merely membrane permeation, as being important
for antimalarial efficiency. Indeed, the lipophilic character
might confer upon RSFs not only topographical specificity,
that is, accessibility to membrane-enclosed compartments, but
also chemical specificity, that is, access to particular chemical
components or iron pools. The results of the studies presented
in Table 2 and Fig. 6 and 7 indicate that RSFm2phe, the most
lipophilic RSF used, not only was the most potent antimalarial
agent but also was the most efficient RSF in extracting iron
from infected cells. At present it remains to be established
whether the extra iron extracted by the most lipophilic RSF is
derived from a specific chemical component or a dispersed
pool of the metal which is critical for parasite survival. Irre-
spective of the source of this extracted iron, the results offer
new possibilities for further improvements in the antimalarial
performance of iron chelators.
14. Gordeuk, V. R., G. M. Brittenham, and P. E. Thuma. 1994. Iron chelation as
a chemotherapeutic strategy in the treatment of malaria, p. 111–130. In R. J.
Bergeron and G. M. Brittenham (ed.), The development of iron chelators for
clinical use. CRC Press, Inc., Boca Raton, Fla.
15. Heppner, D. G., P. E. Hallaway, G. J. Kontoghiorghes, and J. W. Eaton.
1988. Antimalarial properties of orally active iron chelators. Blood 72:358–
361.
16. Hershko, C., and T. E. A. Peto. 1988. Deferoxamine inhibition of malaria is
independent of host iron status. J. Exp. Med. 168:375–387.
17. Loyevsky, M., S. D. Lytton, B. Mester, J. Libman, A. Shanzer, and Z. I.
Cabantchik. 1993. The antimalarial action of desferal involves a direct access
route to erythrocytic (Plasmodium falciparum) parasites. J. Clin. Invest.
91:218–224.
18. Lytton, S. D., M. Loyevsky, B. Mester, J. Libman, I. Landau, A. Shanzer, and
Z. I. Cabantchik. 1993. In vivo antimalarial action of a lipophilic iron (III)
chelator: suppression of Plasmodium vinckei infection by reversed sid-
erophore. Am. J. Hematol. 43:217–220.
19. Lytton, S. D., B. Mester, I. Dayan, H. Glickstein, J. Libman, A. Shanzer, and
Z. I. Cabantchik. 1993. Mode of action of iron (III) chelators as antimalari-
als. I. Membrane permeation properties and cytotoxic activity. Blood 81:
214–221.
20. Lytton, S. D., B. Mester, J. Libman, A. Shanzer, and Z. I. Cabantchik. 1994.
Mode of action of iron (III) chelators as antimalarials. II. Evidence for
differential effects on parasite iron-dependent nucleic acid synthesis. Blood
84:910–915.
21. Poch, G. 1993. Combined effects of drugs and toxic agents. Modern evalu-
ation in theory and practice. Springer-Verlag, New York.
22. Pollack, S., R. N. Rossan, D. E. Davidson, and A. Escjadillo. 1987. Desfer-
rioxamine suppresses Plasmodium falciparum in aotus monkeys. Proc. Soc.
Exp. Biol. Med. 184:162–167.
23. Raventos-Suarez, C., S. Pollack, and R. L. Nagel. 1980. Plasmodium falci-
parum: inhibition of in vitro growth by desferrioxamine. Am. J. Trop. Med.
Hyg. 31:919–922.
24. Rivadeneira, E. M., M. Wasserman, and C. Espinal. 1983. Separation and
concentration of schizonts of Plasmodium falciparum by Percoll gradients. J.
Protozool. 30:367–370.
25. Scheibel, L. W., and S. Rodriguez. 1989. Antimalarial activity of selected
aromatic chelators. V. Localization of 59Fe in Plasmodium falciparum in the
presence of oxines, p. 119–149. In G. Brewer (ed.), Malaria and the red cell.
Alan R. Liss, Inc., New York.
26. Scott, M. D., A. Ranz, F. A. Kuypers, B. H. Lubin, and S. R. Meshnick. 1990.
Parasite uptake of desferroxamine: a prerequisite for antimalarial activity.
Br. J. Haematol. 75:598–602.
27. Shanzer, A., and J. Libman. 1991. Biomimetic siderophores, p. 309–338. In
G. Winkelmann (ed.), CRC Handbook of Microbial Iron Chelates. CRC
Press, Inc., Boca Raton, Fla.
28. Shanzer, A., J. Libman, S. D. Lytton, H. Glickstein, and Z. I. Cabantchik.
1989. Reversed siderophores act as antimalarial agents. Proc. Natl. Acad.
Sci. USA 88:6585–6589.
29. Trager, W., and J. B. Jensen. 1976. Continuous culture of human malaria
parasites. Science 193:673–675.
30. Traore, O., P. Carnevalle, L. Kaptue-Noche, J. M’Nbede, M. Desfontaine, J.
Elion, D. Labie, and R. Nagel. 1991. Preliminary report on the use of
desferrioxamine in the treatment of Plasmodium falciparum malaria. Am. J.
Hematol. 37:206–208.
31. Tsafack, A., J. Golenser, J. Libman, A. Shanzer, and Z. I. Cabantchik. 1995.
Mode of action of iron(III) chelators as antimalarials. III. Overadditive
effects in the combined action of hydroxamate-based agents on in vitro
growth of Plasmodium falciparum. Mol. Pharmacol. 47:403–409.
32. Tsafack, A., M. Loyevsky, P. Ponka, and Z. I. Cabantchik. 1996. Mode of
action of iron(III) chelators as antimalarials. IV. Potentiation of desferal
action by benzoyl and isonicotinoyl hydrazone derivatives. J. Lab. Clin. Med.
127:574–582.
ACKNOWLEDGMENTS
This work was supported in part by NIH grant AI20342 and the
Israel Science Fund administered by the Israel Academy of Science
and Humanities.
We thank Rachel Lazar (Weizman Institute, Rehovot, Israel) for
assistance in the synthesis of the various RSFs, Prem Ponka (McGill
University, Montreal, Quebec, Canada) for the gift of SIH, and I. N.
Slotki and W. Breuer (Hebrew University, Jerusalem, Israel) for crit-
ical reading of the manuscript.
REFERENCES
1. Atamna, H., and H. Ginsburg. 1995. Heme degradation in the presence of
glutathione. A proposed mechanism to account for the high levels of non-
heme iron found in the membranes of hemoglobinopathic red blood cells.
J. Biol. Chem. 270:24876–24883.
2. Breuer, W., S. Epsztejn, P. Millgram, and Z. I. Cabantchik. 1995. Transport
of iron and other transition metals into cells as revealed by fluorescent probe.
Am. J. Physiol. 268:C1354–C1361.
3. Bunnag, D., A. A. Poltera, A. Viravan, S. Looareesuwan, K. T. Harinasuta,
and C. Schindlery. 1992. Plasmocidal effect of desferrioxamine B in human
vivax or falciparum malaria from thailand. Acta Trop. 52:59–67.
4. Cabantchik, Z. I. 1995. Iron chelators as antimalarials: the biochemical basis
of selective cytotoxicity. Parasitol. Today 11:74–78.
5. Cabantchik, Z. I., H. Glickstein, J. Golenser, M. Loyevsky, and A. Tsafack.
1996. Iron chelators. Mode of action as antimalarials. Acta Haematol. 95:
70–77.
6. Cabantchik, Z. I., H. Glickstein, P. Milgram, and W. Breuer. 1996. A fluo-
rescence assay for assessing chelation of intracellular iron in a membrane
model system and in mammalian cells. Anal. Biochem. 233:221–227.
7. Ceriotti, F., and G. Ceriotti. 1980. Improved direct specific determination of
serum iron and total iron-binding capacity. Clin. Chem. 26:327–331.
8. Dayan, I., J. Libman, Y. Agi, and A. Shanzer. 1993. Chiral siderophore
analogs: ferrichrome. Inorg. Chem. 32:1467–1475.
9. Dayan, I., J. Libman, A. Shanzer, C. E. Felder, and S. Lifson. 1991. Regu-
lation of molecular conformation of chiral, tripodal structures by Ca2ϩ
binding. J. Am. Chem. Soc. 113:3431–3439.
10. Desjardins, R. E., C. J. Canfield, J. D. Haynes, and J. D. Chulay. 1979.
Quantitative assessment of antimalarial activity in vitro by semiautomated
microdilution technique. Antimicrob. Agents Chemother. 16:710–718.
11. Fritsch, G., J. Treumer, D. T. Spira, and A. Jung. 1985. Plasmodium vinckei:
suppression of mouse infections with desferrioxamine B. Exp. Parasitol.
60:171–174.
33. Wyler, D. J. 1992. Bark, weeds, and iron chelators—drugs for malaria.
N. Engl. J. Med. 327:1519–1521.
34. Yakirevitch, P., N. Rochel, A. M. Albrecht-Gary, J. Libman, and A. Shanzer.
1991. Chiral siderophore analogs: ferrioxamines and their iron(III) coordi-
nation properties. Inorg. Chem. 32:3431–3439.
35. Zanninelli, G., H. Glickstein, W. Breuer, P. Milgram, P. Brissot, R. C. Hider,
A. M. Konijn, J. Libman, A. Shanzer, and Z. I. Cabantchik. Submitted for
publication.
12. Gabay, T., and H. Ginsburg. 1993. Hemoglobin denaturation and iron re-
lease in acidified red blood cell lysate: a possible source of iron for in-
traerythrocytic malaria parasites. Exp. Parasitol. 77:261–272.
13. Golenser, J., A. Tsafack, Y. Amichai, J. Libman, A. Shanzer, and Z. I.
Cabantchik. 1995. Antimalarial action of hydroxamate-based iron chelators
and potentiation of desferrioxamine action by reversed siderophores. Anti-