Effects of C-4 stereochemistry and C-4' hydroxylation on the iron clearing efficiency and toxicity of desferrithiocin analogues

Article abstract of DOI:10.1021/jm990058s

Additional structure-activity studies of desferrithiocin analogues are carried out. The effects of stereochemistry at C-4 on the ligands' iron clearing efficiency are reviewed and assessed using the enantiomers 4,5- dihydro-2-(2,4-dihydroxyphenyl)thiazole-4(R)-carboxylic acid and 4,5-dihydro- 2-(2,4-dihydroxyphenyl)thiazole-4(S)-carboxylic acid. The utility of 4'- hydroxylation as a method of reducing the toxicity of desazadesferrithiocin analogues is also examined further with the synthesis and in vivo comparison of 4,5-dihydro-2-(2-hydroxyphenyl)-4-methylthiazole-4(S)-carboxylic acid, which is the natural product 4-methylaeruginoic acid, and 4,5-dihydro-2- (2,4-dihydroxyphenyl)-4-methylthiazole-4(S)-carboxylic acid. The stereochemistry at C-4 is shown to have a substantial effect on the iron clearing efficiency of desferrithiocin analogues, as does C-4'-hydroxylation on the toxicity profile. All of the compounds are evaluated in a bile-duct- cannulated rodent model to determine iron clearance efficiency and are carried forward to the iron-overloaded primate for iron clearing measurements. On the basis of the results of the present work, although 4,5- dihydro-2-(2,4-dihydroxyphenyl)thiazole-4(S)-carboxylic acid is still the most promising candidate for clinical evaluation, 4,5-dihydro-2-(2,4- dihydroxyphenyl)4-methylthiazole-4(S)-carboxylic acid (4'- hydroxydesazadesferrithiocin) also merits further preclinical assessment.

Full text of DOI:10.1021/jm990058s

2432
J . Med. Chem. 1999, 42, 2432-2440
Effects of C-4 Ster eoch em istr y a n d C-4′ Hyd r oxyla tion on th e Ir on Clea r in g
Efficien cy a n d Toxicity of Desfer r ith iocin An a logu es
Raymond J . Bergeron,* J an Wiegand, J ames S. McManis, Bruce H. McCosar, William R. Weimar,
Gary M. Brittenham,† and Richard E. Smith
Department of Medicinal Chemistry, University of Florida, Gainesville, Florida 32610-0485, and Department of Pediatrics,
Division of Hematology, Columbia University College of Physicians and Surgeons, New York, New York 10032
Received J anuary 25, 1999
Additional structure-activity studies of desferrithiocin analogues are carried out. The effects
of stereochemistry at C-4 on the ligands’ iron clearing efficiency are reviewed and assessed
using the enantiomers 4,5-dihydro-2-(2,4-dihydroxyphenyl)thiazole-4(R)-carboxylic acid and 4,5-
dihydro-2-(2,4-dihydroxyphenyl)thiazole-4(S)-carboxylic acid. The utility of 4′-hydroxylation as
a method of reducing the toxicity of desazadesferrithiocin analogues is also examined further
with the synthesis and in vivo comparison of 4,5-dihydro-2-(2-hydroxyphenyl)-4-methylthiazole-
4(S)-carboxylic acid, which is the natural product 4-methylaeruginoic acid, and 4,5-dihydro-
2-(2,4-dihydroxyphenyl)-4-methylthiazole-4(S)-carboxylic acid. The stereochemistry at C-4 is
shown to have a substantial effect on the iron clearing efficiency of desferrithiocin analogues,
as does C-4′-hydroxylation on the toxicity profile. All of the compounds are evaluated in a bile-
duct-cannulated rodent model to determine iron clearance efficiency and are carried forward
to the iron-overloaded primate for iron clearing measurements. On the basis of the results of
the present work, although 4,5-dihydro-2-(2,4-dihydroxyphenyl)thiazole-4(S)-carboxylic acid
is still the most promising candidate for clinical evaluation, 4,5-dihydro-2-(2,4-dihydroxyphenyl)-
4-methylthiazole-4(S)-carboxylic acid (4′-hydroxydesazadesferrithiocin) also merits further
preclinical assessment.
In tr od u ction
continuous infusion over long periods of time. Patient
compliance can become a major problem over a lifetime
of chelation therapy. Furthermore, the cost of desferri-
oxamine treatment renders the population in areas with
the greatest need untreatable. There are two possible
solutions to these problems: the development of an
orally effective chelator or the identification of more
efficient parenteral chelators. This report focuses on the
former option.
Earlier structure-activity analyses of the desfer-
rithiocin pharmacophore revealed that the three ligating
centerssthe thiazoline nitrogen, the carboxyl group, and
the aromatic hydroxylsare critical to the compounds’
iron clearing capabilities. Modification of some of the
nonchelating fragments, such as removal of the thiazo-
line methyl or the aromatic nitrogen, had little impact
on the deferration properties of the resulting molecules.
However, either replacement of the sulfur with oxygen,
nitrogen, or a methylene or expansion of the five-
membered thiazoline to a six-membered dihydro-∆2-
thiazine resulted in compounds with little iron clearing
activity.12-14 Investigation of benz-fused desazades-
methyldesferrithiocins, i.e., the (R)- and (S)-pairs of
naphthyl analogues 2-(2-hydroxynaphth-1-yl)-∆2-thia-
zoline-4-carboxylic acid and 2-(3-hydroxynaphth-2-yl)-
∆2-thiazoline-4-carboxylic acid, demonstrated that al-
though the former enantiomers were not effective iron
clearing agents after oral administration to the rats,
their positional isomers were.15
Iron is required for life as we know it. Although it is
one of the most abundant elements in the earth’s crust,1
it exists in the biosphere largely as insoluble ferric
hydroxide (Ksp ) 1 × 10-38).2 Microorganisms have
evolved low molecular weight, virtually ferric ion specific
ligands, siderophores, to transport and store iron which
would otherwise be unavailable.3 For example, Strep-
tomyces pilosus assembles and excretes desferrioxamine,
a hydroxamate iron chelator4 that forms a 1:1 hexaco-
ordinate octahedral complex with Fe(III); the formation
constant is 3 × 1030 M-1.5 The microorganism can utilize
this complex as an iron source.
In fact, desferrioxamine B (Desferal) remains the drug
of choice for the treatment of transfusional iron
overload.6-10 Although considerable effort has been
invested in the development of new therapeutics for
managing iron overload diseases, treatments have
remained the same for over a generation. Patients with
primary hemochromatosis are still treated by frequent
phlebotomy; patients suffering from iron overload sec-
ondary to blood transfusions (e.g., Cooley’s anemia)
must be maintained on chelation therapy. Desferriox-
amine’s efficacy and long-term tolerability are well
documented, but there are still problems associated with
its use. If one assumes that the stoichiometry described
above applies when this ligand is administered to an
animal, 5% or less of the calculated iron excretion is
observed.11 This chelator also has a very short half-life
in the body; it must therefore be administered by
In a recent study utilizing (S)-desazadesmethyldes-
ferrithiocin as the parent compound,14 we determined
that introduction of a hydroxyl in the 4-position of the
aromatic ring resulted in a ligand which was orally
* To whom correspondence should be addressed.
† Columbia University College of Physicians and Surgeons.
10.1021/jm990058s CCC: $18.00 © 1999 American Chemical Society
Published on Web 06/12/1999

Iron Clearing Efficiency of Desferrithiocin Analogues
J ournal of Medicinal Chemistry, 1999, Vol. 42, No. 13 2433
active in both rodents and primates, yet was profoundly
less toxic than the parent compound. The current series
of experiments further explores 4,5-dihydro-2-(2,4-di-
hydroxyphenyl)thiazole-4(S)-carboxylic acid and how
the stereochemistry at C-4 affects its activity. In addi-
tion, the impact of 4′-hydroxylation of the aromatic ring
on desazadesferrithiocin’s toxicity is assessed.
nitrogen alters the toxicity profile from renal to gas-
trointestinal (GI) and then if 4′-hydroxylation [4,5-
dihydro-2-(2,4-dihydroxyphenyl)-4-methylthiazole-4(S)-
carboxylic acid, 8] helps ameliorate the GI problems.
Syn th etic Meth od s
Chelator 6 was synthesized by condensation of 2,4-
dihydroxybenzonitrile (10) with L-cysteine in phosphate
buffer and methanol (Scheme 1). Nitrile 10 was pre-
pared from 2,4-dihydroxybenzaldehyde with nitroethane
in sodium acetate and acetic acid.17
Design Con cep ts
In the course of our recent structure-activity study
focused on the biological effects of changing the redox
potential of the aromatic ring of desferrithiocin by
introducing electron-donating or -withdrawing groups,14
we found that hydroxylation at the 4′-position had a
pronounced effect on the molecule’s toxicity profile. 4,5-
Dihydro-2-(2,4-dihydroxyphenyl)thiazole-4(S)-carboxy-
lic acid was indeed far less toxic than the 4′-dehydroxy
parent chelator. This decreased toxicity was not related
to a difference in octanol-water partition coefficients
as the log P values between the parent and hydroxylated
analogues were similar. The remarkable difference in
toxicity between these two analogues has prompted us
to further examine the utility of 4′-hydroxylation as a
method of reducing the toxicity of other desferrithiocin
analogues.
In the course of most of our previous investigations
we principally explored desferrithiocin analogues in the
(S)-configuration. This was based on earlier evidence
that the (S)-enantiomers were more active iron clearing
agents in primates.16 In the current series of experi-
ments we reviewed and further assessed the effects of
stereochemistry at C-4 on the ligands’ iron clearing
efficiency.
The production of desazadesferrithiocin (7) was ac-
complished by cyclocondensation of 2-cyanophenol (11)
with (S)-R-methyl cysteine (12) in buffered aqueous CH3-
OH (Scheme 2). The high-field NMR spectra of synthetic
7 and the siderophore 4-methylaeruginoic acid, which
has recently been isolated from the Streptomyces species
KCTC 9303, were virtually identical; however, the
stereochemistry at C-4 of the natural product was not
specified.18 Therefore, a circular dichroism measure-
ment (CD) of 7 was run. While wavelengths of the
maximum and two minima of the CD of 7 matched
closely with literature values of the natural chelator and
the molar ellipicities ([θ]) at 320 nm were within 4% of
each other, the [θ] values at 280 and 260 nm of 7 were
approximately 50% of the reported values.18 There is
no plausible explanation for this discrepancy, since (S)-
R-methyl cysteine (12) used to generate 7 is derived
from hydrolysis of desferrithiocin (DFT, 9) and, pos-
sessing no R-proton, cannot racemize. Thus the first
total synthesis of the natural product 4-methylaerugi-
noic acid proves that, like DFT, it possesses the (S)-
configuration.
The preparation of 4′-hydroxydesazadesferrithiocin 8
also employed the unusual amino acid (S)-R-methyl
cysteine (12), prepared from hydrolysis19 of DFT (9) in
6 N HCl (Scheme 3). The mixture of 12 and the
byproduct, 3-hydroxypicolinic acid (13), was reacted
with 2,4-dihydroxybenzonitrile (10) under weakly acidic
conditions. Filtration of the solid and recrystallization
from aqueous ethanol resulted in crystalline tridentate
ligand 8.
To determine the importance of 4′-hydroxylation and
C-4 stereochemistry, three systems were synthesized
and evaluated in both a non-iron-overloaded, bile-duct-
cannulated rat model and an iron-overloaded Cebus
apella monkey model. The desferrithiocin analogues
included 4,5-dihydro-2-(2,4-dihydroxyphenyl)thiazole-
4(R)-carboxylic acid (6), 4,5-dihydro-2-(2-hydroxyphe-
nyl)-4-methylthiazole-4(S)-carboxylic acid (desazades-
ferrithiocin, 7), and 4,5-dihydro-2-(2,4-dihydroxyphenyl)-
4-methylthiazole-4(S)-carboxylic acid (8) (Table 1).
At the drug discovery level, the fact that the (S)-
enantiomers tend to outperform the (R)-enantiomers in
a given set of desferrrithiocin analogues is more than
an interesting observation. The amino acid reagent
required to assemble the (S)-desferrithiocin analogues
(e.g., D-cysteine) is much more expensive than the
naturally occurring L-cysteine. This issue compelled us
to synthesize and assess the iron clearing properties of
4,5-dihydro-2-(2,4-dihydroxyphenyl)thiazole-4(R)-car-
boxylic acid (6), the (S)-enantiomer of which (5) was
already shown to be a very active iron chelator.
The choice of systems to further investigate the effects
of 4′-hydroxylation on the toxicity profile of desferrithio-
cin analogues was also fairly clear. The parent ligand
in this case was 4,5-dihydro-2-(2-hydroxyphenyl)-4-
methylthiazole-4(S)-carboxylic acid (desazadesferrithio-
cin, 7). As with desazadesmethyldesferrithiocin in the
previous study, desazadesferrithiocin does not have a
nitrogen in the aromatic ring. This choice also makes it
possible to determine whether with 7, similar to des-
azadesmethyldesferrithiocin,14 removal of the aromatic
Meta l Com p lex Stoich iom etr y
In an attempt to determine whether there was
anything unusual about the nature of the Fe(III)
complexes of analogues 7 or 8, we ran J ob’s plots on
these ligands. Not surprisingly, like 9, these chelators
formed 2:1 complexes with Fe(III) (Figure 1). We are
currently investigating both the thermodynamics of
formation and the structure of the complexes.
Ra cem iza tion
One of the interesting observations regarding the
significance of stereochemistry in the biological activity
of the desferrithiocin analogues is that the (S)-enanti-
omers are more active in primates than are their (R)-
counterparts. It was thus of interest to evaluate the
potential for racemization of the 4-carboxyl methine
under physiological conditions. A study on the lability
of the C-4 methine of 4,5-dihydro-2-(2,4-dihydroxy-
phenyl)thiazole-4(S)-carboxylic acid (5) was conducted
at pD 7.2 in D2O. Interestingly, whereas there was little,
if any, exchange at H-4 during the 5-h course of the

2434 J ournal of Medicinal Chemistry, 1999, Vol. 42, No. 13
Ta ble 1. Comparison of Efficiencies of the Desferrithiocins
Bergeron et al.
a
In the rats, doses were 150 µmol/kg via the route indicated. The net iron excretion was calculated by subtracting the iron excretion
of control animals from the iron excretion of treated animals. Efficiency of chelation is defined as net iron excretion/total iron-binding
capacity of chelator administered, expressed as a percent. Directly beneath the efficiency calculation is the percentage breakdown of iron
b
excretion in the bile and urine, respectively. In the monkeys, the doses and routes were as shown in the table. The efficiency of each
compound was calculated by averaging the iron output for 4 days before the administration of the drug, subtracting these numbers from
the 2-day iron clearance after the administration of the drug, and then dividing by the theoretical output; the result is expressed as a
percent. Directly beneath the efficiency calculation is the percentage breakdown of iron excretion in the stool and urine, respectively.
d
c The compound was given as a suspension in dH2O rather than in 60% H2O, 40% Cremophor. The comparison was made at the 150
µmol/kg po dose.

Iron Clearing Efficiency of Desferrithiocin Analogues
J ournal of Medicinal Chemistry, 1999, Vol. 42, No. 13 2435
Sch em e 1. Synthesis of 4,5-Dihydro-2-
(2,4-dihydroxyphenyl)thiazole-4(R)-carboxylic Acid (6)a
a
Reagents: (a) L-cysteine/CH3OH/phosphate buffer/reflux/6 h,
(b) 1 N HCl.
Sch em e 2. Synthesis of 4,5-Dihydro-2-
(2-hydroxyphenyl)-4-methylthiazole-4(S)-carboxylic
Acid (7)a
F igu r e 1. J ob’s plots of ligands 7 and 8. The plots are
superimposed on each other to show the similarity between
the two graphs. Solutions containing different ligand/Fe(III)
ratios were prepared so that [ligand] + [Fe(III)] ) 1.0 mM.
Compound 7 is indicated by the open circles, and analogue 8
is shown by the filled squares. The data points were fitted to
the mole fractions (1) from 0-0.667 and (2) from 0.667-1.000;
r2 ) 0.993-1.000. Theoretical mole fraction maximum for a
2:1 ligand:Fe complex is 0.667 (arrow).
a
Reagents: (a) phosphate buffer/CH3OH/34 °C/5 days, (b) citric
acid.
Sch em e 3. Synthesis of 4,5-Dihydro-2-
(2,4-dihydroxyphenyl)-4-methylthiazole-4(S)-carboxylic
Acid (8)a
Thus, it appears that racemization of this ligand in its
free or complexed form is unlikely at physiological pH.
Biologica l Eva lu a tion s
The iron clearing properties of the compounds were
tested both in a non-iron-overloaded, bile-duct-cannu-
lated rat model and in an iron-overloaded C. apella
monkey model. The non-iron-overloaded, bile-duct-can-
nulated rat model12,20,21 represents a very rapid, func-
tional initial screen of potential iron chelators. We can
measure the total amount of iron cleared, the relative
biliary versus urinary iron excretion, and the rate at
which various chelators induce iron clearance in the bile
and urine. If the ligand of interest does not induce iron
clearance in the bile or the urine, then additional
experimentation is generally regarded as unnecessary.
Many chelators have appeared promising in rodents, but
these compounds have failed at the clinical level. The
iron-overloaded C. apella monkey serves as an inter-
mediate screen for evaluating iron chelators before
human studies; there are many similarities in this
model to humans.11,22
a
Reagents: (a) 6 N HCl/100 °C/80 min, (b) 10/CH3OH/71 °C/
3.5 days/phosphate buffer/NaHCO3, (c) 1 N HCl.
Ch ela tor -In d u ced Ir on Clea r a n ce in Rod en ts. In
Table 1, the iron clearing efficiencies of the compounds
in rodents are shown, along with the relative fractions
excreted in the bile and in the urine. We have included
historical data (compounds 1-4 and 9)13,16,20 for the sake
of discussion. A comparison of the historical enantio-
meric pairs (1,2; 3,4) suggests that, in general, the issue
of C-4 stereochemistry is not profoundly significant in
the bile-duct-cannulated rat model at a dose of 150 µmol/
kg. However, the (S)-enantiomer of 4,5-dihydro-2-(2,4-
dihydroxyphenyl)thiazole-4-carboxylic acid (5) was more
effective in terms of efficiency, 2.4 ( 0.9% vs e0.5% for
the (R)-enantiomer 6 (P < 0.02, Table 1).
experiment, the H-3′ aromatic proton between the two
hydroxyls was very labile. The rate constant for the
exchange was first-order in chelator with k ) 4.6 ( 0.6
× 10-4 s-1. Identical experiments did not indicate any
deuterium exchange of carbon-bound protons with
either the parent desazadesmethyldesferrithiocin (3) or
2,4-dihydroxybenzoic acid. In a preliminary study with
the Ga(III) complex of 5 at a 2:1 ligand:metal ion ratio,
although the spectra were somewhat more complicated
owing to the likely presence of diastereomeric mixtures,
it was nevertheless possible to identify the H-3′ proton.
Again, the rate constant was first-order in substrate,
but it was about 7 times slower, k ) 7.5 × 10-5 s-1. We
were also able to identify the H-4 methine in the Ga-
(III) complex, and there was no exchange of this proton.
A comparison of (S)-desazadesferrithiocin (7) with the
corresponding 4′-hydroxylated analogue 8 reveals that

2436 J ournal of Medicinal Chemistry, 1999, Vol. 42, No. 13
Bergeron et al.
these two ligands behaved very differently. Desazades-
ferrithiocin 7 was moderately effective in the rodents
with an efficiency of 2.7 ( 0.5% under normal experi-
mental conditions, (i.e., bile and urine were collected for
24 h post drug administration). Under these conditions,
no chelator-induced iron excretion was observed with
ligand 8; the efficiency was calculated to be e0.5%.
However, because of our interest in how C-4′ hydroxy-
lation might affect the toxicity profile of 7, a toxicity
study was initiated in spite of the apparent lack of
activity. After 2 days of treatment with 8 at a dose of
384 µmol/kg/day, the stools of non-iron-overloaded ro-
dents were red, although guaiac tests were negative.
This compelled us to initiate additional experiments
with this drug in primates, shown below. The explana-
tion for the apparent lack of iron clearing activity in
the rodents under normal experimental conditions is not
clear. It is possible that either the higher dose or the
multiple doses used in the toxicity study led to the
greater iron clearance. Neither can it be dismissed that
uninterrupted enterohepatic circulation in these ani-
mals would result in the recirculation of free ligand and
thus delayed and increased iron clearance. These issues
are currently under consideration.
the prolonged residence times of the latter enantiomer.
The most noteworthy finding from the pharmacokinetic
experiments was the marked enantioselectivity of the
renal clearance observed with 1, a clearance that was
3.5 times greater than that of 2, and an Fe(III):(1)2
clearance that was 6.8 times greater than that of Fe-
(III):(2)2. Owing to the death of one of the primates given
2, we were unwilling to perform intravenous pharma-
cokinetic studies. These observations, coupled with the
very low toxicity of the 4′-hydroxy ligand 5,14 encouraged
us to consider an additional enantiomeric set of com-
pounds, 5,6.
(S)-Enantiomer 5 performed well in the primate
model; the efficiency was 4.2 ( 1.4% when given po at
a dose of 150 µmol/kg14 and 5.3 ( 1.7% when given at a
dose of 300 µmol/kg (Table 1). Iron clearance was largely
in the stool at both 150 and 300 µmol/kg doses, 70% and
90%, respectively; the urinary percentages were 30%
and 10%, respectively. These results were encouraging;
in an effort to more closely mimic clinical applications,
the ligand was given as a suspension in H2O both po
and sc. The iron clearing properties were not altered;
the efficiency was 5.6 ( 0.9% at a dose of 150 µmol/kg
sc and 4.8 ( 1.4% at a dose of 300 µmol/kg po. Again,
iron clearance was overwhelmingly in the stool, 92% and
85%, respectively; 8% and 15% of the iron, respectively,
was excreted in the urine. However, (R)-enantiomer 6
was a poor ligand at this dose with an efficiency of 1.7
( 0.8%. Iron clearance was mostly fecal, 76% vs 24% in
the urine.
Ch ela tor -In d u ced Ir on Clea r a n ce in P r im a tes.
Inspection of Table 1 reveals some interesting differ-
ences in the iron clearance properties of the enantio-
meric pairs of desferrithiocin analogues. We have
included historical data (compounds 1-4 and 9)13,16,20
for the sake of discussion; the differences between
enantiomers are greater in the primate model. Of the
three enantiomeric pairs shown (1,2; 3,4; 5,6), it seems
clear that the (S)-enantiomer is consistently better
(compounds 1, 3, and 5). The comparisons between
compounds 1,2 and 3,4 were made at the 300 µmol/kg
dose; the comparison between 5 and 6 was made at the
150 µmol/kg dose. Between compounds 1 and 2, the (S)-
enantiomer 1 was about 16 times better than the (R)-
enantiomer 2, P < 0.002. Ligand 1 at 300 µmol/kg had
an efficiency of 8.0 ( 2.5% with 42% of the iron in the
stool and 58% in the urine, whereas 2 had an efficiency
of only 0.5 ( 2.0% with 54% of the iron in the stool and
46% in the urine.16 The (S)-enantiomer 3, with an
efficiency of 12.4 ( 7.6%, generated 90% of the iron in
the stool and 10% in the urine; (R)-enantiomer 4 as its
sodium salt had an efficiency of 8.2 ( 3.2% with 80% of
the iron in the stool and 20% in the urine. However,
this difference is not significant (P > 0.44), possibly due
to the high variability among the animals in the former
set.
In the case of (S)-desazadesferrithiocin 7 and the
corresponding 4′-hydroxylated analogue 8, the iron
clearing efficiencies and modes of iron excretion of both
compounds were identical. The efficiency of 7 at a dose
of 300 µmol/kg was 13.1 ( 4.0%; that of 8 at a dose of
150 µmol/kg was 13.4 ( 5.8%. In both instances, 86%
of the iron was in the stool and 14% in the urine. The
efficiencies of these compounds were comparable, 21.5
( 12% (76% stool, 24% urine) and 17.7 ( 3.9% (79%
stool, 21% urine), respectively, at a dose of 75 µmol/kg
(Table 1). However, as will be discussed below, the
differences in toxicity profiles between these compounds
are remarkable. Furthermore, the efficiency of these
analogues is not substantially less than that of the
natural product 9, with its 16.1 ( 8.5% efficiency and
78% fecal and 22% urinary iron excretion at a dose of
150 µmol/kg.
Toxicity. If the ligands were shown to be effective
(>3% efficiency) in the primates, a 10-day toxicity trial
was initiated in rodents. The drugs were administered
po to normal rats by gavage or sc once a day for 10 days,
or until the animals expired, as indicated in Table 2.
The data for 9, although historical,20 are included in
Table 2 for the reader’s convenience. Desferrithiocin 9,
the natural product, was a very effective iron chelator
when given orally to rats or monkeys. However, the
compound also elicited significant side effects, particu-
larly nephrotoxicity, when administered chronically to
rats. At a dose of 384 µmol/kg/day (equivalent to 100
mg/kg of the sodium salt), five out of five animals were
dead within 10 days (Table 2). Histologically, the kid-
neys presented with vacuolar degeneration and acute,
diffuse epithelial necrosis throughout the cortex; severe
intralobular granular hyaline casts and severe multi-
Interestingly, a pharmacokinetic study of the initial
set (1,2)16 clearly demonstrated that the peak plasma
concentrations over time curves as well as areas under
the curve (AUCs) of 1 and Fe(III):(1)2 were equal to or
lower than those of the corresponding 2 and Fe(III):(2)2
curves. At 8 h posttreatment, the plasma concentrations
of 1 and Fe(III):(1)2 had declined to levels approaching
the limits of detection, whereas substantial plasma
concentrations of 2 and Fe(III):(2)2 still remained. In
every instance Fe(III):(2)2 in the plasma exceeded 25
mg/L (50 µM) for several hours and remained above 10
mg/L (20 µM) at 8 h. The ratios for AUC0-4/AUC0-∞ are
0.89 and 0.81 for 1 and Fe(III):(1)2, respectively, com-
pared to 0.50 and 0.44 for 2 and Fe(III):(2)2, reflecting

Iron Clearing Efficiency of Desferrithiocin Analogues
J ournal of Medicinal Chemistry, 1999, Vol. 42, No. 13 2437
Ta ble 2. Chronic Toxicity Evaluations of Desferrithiocin and
Desazadesferrithiocin Analogues
rithiocin (7) resulted in compounds (5 and 8, respec-
tively) which were remarkably less toxic than the
corresponding parent drugs.
dose/day
compd (µmol/kg) ×
no. of
Whereas in the case of 3, C-4′-hydroxylation did
reduce toxicity (5), it also diminished the compound’s
iron clearing efficiency by about 60%. Nevertheless, with
very low toxicity this ligand is a promising candidate.
However, with 7, C-4′ hydroxylation had little, if any,
impact on the iron clearing properties of the molecule,
in addition to substantially reducing its toxicity (8).
Although 8 elicited some renal toxicity, because of its
impressive iron clearing efficiency even at lower doses,
further toxicity analyses at lower doses and with iron-
overloaded animals are warranted.
no.
no. of daysa route deaths
comments
7
384 × 10
384 × 10
384 × 10
po
sc
3/3
3/3
0/5
all animals dead by day 5:
severe GI toxicity
all animals dead by day 5:
severe GI toxicity
well-tolerated for the 10-day
test period; histopathologies
normal except for mild
nephrotoxicity
8
po
9
384 × 10
po
5/5
all animals dead by day 5:
severe nephrotoxicity20
a
The compounds were given to rats at the doses, test periods,
and routes shown.
At the level of C-4 stereochemistry, the results
obtained from 5 and 6 confirm that overall the (S)-
enantiomers of the desferrithiocin analogues are indeed
more effective iron clearing agents in primates than
their (R)-counterparts. It further seems clear that
racemization at C-4 under physiological conditions is
unlikely.
Given the iron clearing efficacy at low doses, 4,5-
dihydro-2-(2,4-dihydroxyphenyl)-4-methylthiazole-4(S)-
carboxylic acid (8) is certainly worth further preclinical
assessments. However, the encouraging results obtained
with ligand 5 in our earlier study14 and the iron
clearance observed when this analogue was adminis-
tered to the primates in a more clinically applicable
matrix indicate that 4,5-dihydro-2-(2,4-dihydroxyphe-
nyl)thiazole-4(S)-carboxylic acid (5) remains the most
promising desferrithiocin analogue for continued pre-
clinical development as an orally active iron chelator.
focal tubular ectasia were also noted. Abstraction of the
pyridine nitrogen from 9 to yield 7 resulted in a
compound that presented a different toxicity profile; the
primary toxicity was GI in nature. At a daily dose of
384 µmol/kg either po or sc, all of the animals were dead
by day 5. At necropsy, the stomachs of all of the animals
were hemorrhagic and grossly distended with gas and
fluid; the stomach walls were translucent. The intes-
tines were also hemorrhagic, and pressure necrosis of
the spleen, from the grossly distended stomach, was
noted in one of the animals dosed sc.
Interestingly, when an electron-donating hydroxyl
group was added to position 4 of the aromatic ring of 7
to give 8, the GI toxicity problem found with 7 was
absent; no deaths occurred when this compound was
given at 384 µmol/kg/day po for 10 days, greater than
twice the time period for 7 (Table 2). Gross inspection
of organs 1 day after the final dose revealed no drug-
related abnormalities. However, histopathological analy-
sis of tissues (see Experimental Section for a list)
showed that this compound was associated with some
nephrotoxicity, but nothing comparable to what was
observed with the natural product 9. Histologically, the
kidneys presented with some vacuolar degeneration in
the proximal tubular epithelium, with a pathology
considered to be reversible; however, one animal showed
possible early necrosis.
It is also notable that with 7, 8, and 9 in the primates
the fecal iron clearance figures were similar, 86%, 86%,
and 78% of the total, respectively; in the rodents the
biliary iron clearance figures were 100% and 93%,
respectively, for 7 and 9. Thus, the differences in toxicity
in the rodents cannot be readily attributed to the mode
of iron clearance. Finally, it is critical to point out that
(1) these were not iron-overloaded animals and (2) only
a single dosing schedule, i.e., 384 µmol/kg/day, was
evaluated. Dose-ranging studies with 8 are underway,
along with an investigation of the impact of iron load
in test animals on the drug’s toxicity profile.
Exp er im en ta l Section
Chelators 1-5 were prepared by methodology developed in
these laboratories.12,14 All reagents were purchased from
Aldrich Chemical Co. (Milwaukee, WI) and were used without
further purification. Distilled solvents and glassware that had
been presoaked in 3 N HCl for 15 min were employed for
reactions involving chelators. Fisher Optima grade solvents
were routinely used, and DMF was distilled. Organic extracts
were dried with sodium sulfate. Phosphate buffer was made
up to 0.1 M at a pH of 5.95.23 Lipophilic Sephadex LH-20 from
Sigma Chemical Co. (St. Louis, MO) was used for column
chromatography, and melting points are uncorrected. Proton
NMR spectra were obtained on a Varian Unity 300 (300 MHz)
or Varian Unity 600 (600 MHz) spectrometer in the specified
deuterated organic solvent or in D2O; chemical shifts are
expressed in parts per million downfield from tetramethylsi-
lane or sodium 3-(trimethylsilyl)propionate-2,2,3,3-d4, respec-
tively. Coupling constants (J ) are in hertz. Optical rotations
were run at 589 nm (sodium D line) utilizing a Perkin-Elmer
341 polarimeter with c as grams of compound per 100 mL of
solution. Circular dichroism (CD) spectra were obtained with
a J asco J 500 spectropolarimeter. Elemental analyses were
performed by Atlantic Microlabs (Norcross, GA). Cremophor
RH-40 was obtained from BASF (Parsippany, NJ ). Sprague-
Dawley rats were purchased from Charles River (Wilmington,
MA). Nalgene metabolic cages, rat jackets, and fluid swivels
were purchased from Harvard Bioscience (South Natick, MA).
Intramedic polyethylene tubing (PE 50) was obtained from
Fisher Scientific (Pittsburgh, PA). Rodent bedding (Beta-
Chips) was purchased from Northeastern Products Corp.
(Warrensburg, NY). C. apella monkeys were obtained from
World Wide Primates (Miami, FL). Ultrapure salts were
obtained from J ohnson Matthey Electronics (Royston, U.K.).
Imferon, an iron dextran solution, was obtained from Fisons
(Bedford, MA).
Con clu sion
Our earlier structure-activity studies on the impact
of alteration in the aromatic ring on the pharmacological
properties of desazadesmethyldesferrithiocin were predi-
cated on the idea that changing the redox potential of
this fragment of the molecule would change its meta-
bolic profile and potentially its toxicity. Indeed, this
proved to be the case; introduction of an electron-
donating hydroxyl group at the C-4′ position in both
desazadesmethyldesferrithiocin (3) and desazadesfer-
4,5-Dih yd r o-2-(2,4-d ih yd r oxyp h en yl)th ia zole-4(R)-ca r -
boxylic Acid (6).24 L-Cysteine hydrochloride (4.38 g, 27.8

2438 J ournal of Medicinal Chemistry, 1999, Vol. 42, No. 13
Bergeron et al.
mmol), phosphate buffer (50 mL), and NaHCO3 (2.41 g, 28.7
mmol) were added to a solution of 1014 (2.51 g, 18.5 mmol) in
CH3OH (74 mL). The solution was degassed with nitrogen and
was refluxed with stirring for 6 h under nitrogen. The reaction
mixture was cooled to room temperature and was stirred
overnight. After the mixture was acidified to pH 2 with 1 N
HCl, the dark yellow precipitate was filtered, washed with
distilled water (3 × 30 mL) and cold EtOH (2 × 30 mL), and
dried in vacuo, producing 3.065 g of 6 (69%) as a yellow powder,
mp (sealed capillary) 268-270 °C dec (lit. mp 261-262 °C24):
mp 281-283 °C dec: [R]25 +63.9° (c 1.12, DMF); 1H NMR
D
(DMSO-d6) δ 1.56 (s, 3 H), 3.33 (d, 1 H, J ) 11.4), 3.76 (d, 1 H,
J ) 11.5), 6.29 (d, 1 H, J ) 2.1), 6.37 (dd, 1 H, J ) 8.6, 2.2),
7.24 (d, 1 H, J ) 8.4), 10.2 (s, 1 H), 12.6 (br s, 1 H), 13.1 (br s,
1 H). Anal. (C11H11NO4S): C, H, N.
Det er m in a t ion of St oich iom et r y of Liga n d -F e(III)
Com p lexes. The stoichiometry of ligand-Fe(III) complexes
was determined spectrophotometrically from J ob’s plots for
ligands 7 and 8. Desferrithiocin, known to form a 2:1 ligand:
Fe(III) complex,25 was run as a positive control. Solutions were
monitored at the visible λmax (482 nm and 481 nm, respec-
tively). Solutions containing different ligand/Fe(III) ratios were
prepared by mixing appropriate volumes of 0.5 mM ligand in
100 mM TRIS Cl, pH 7.4, and 0.5 mM Fe(III) nitrilotriacetate
(NTA) in 100 mM TRIS Cl, pH 7.4, so that [ligand] + [Fe(III)]
) 1.0 mM. The Fe(III)-NTA solution was prepared im-
mediately prior to use by dilution of a 50 mM Fe(III)-NTA
stock solution with TRIS buffer. The stock solution was
prepared as described previously,14 and the iron content was
verified by atomic absorption spectroscopy.
1H NMR Sa m p le P r ep a r a tion a n d Exp er im en ta l Con -
d ition s. Ligand 5 and 2,4-dihydroxybenzoic acid were ana-
lyzed by 1H NMR (300 MHz) in DMSO-d6 and in D2O/
phosphate buffer (µ ) 0.25, pD ) 7.2, 25 °C). The sodium salt
of 3 and the Ga(III) chelate of ligand 5 were analyzed in D2O/
phosphate buffer only. The Ga(III) chelate of 5 was prepared
by dissolving 5 in a Ga(NO3)3 suspension in D2O/phosphate
buffer in a 2:1 5:Ga(III) molar ratio. The resulting solution
cleared immediately upon addition of ligand. Solutions pre-
pared for the kinetic runs were allowed to temperature
equilibrate at 25 °C for 10 min after buffer addition; spectra
were then collected at approximately 15-min intervals over
2-3 h. Chemical shifts (δ) in D2O/phosphate buffer are
reported in ppm downfield from an external sodium-3-trim-
ethylsilylpropionate-2,2,3,3-d4 reference.
[R]23 -30.4° (c 1.00, DMF); 1H NMR (DMSO-d6) δ 3.57 (dd, 1
D
H, J ) 11.3, 7.2), 3.65 (dd, 1 H, J ) 11.3, 9.4), 5.37 (dd, 1 H,
J ) 9.4, 7.2), 6.31 (d, 1 H, J ) 2.3), 6.38 (dd, 1 H, J ) 8.5, 2.3),
7.25 (d, 1 H, J ) 8.5), 10.26 (s, 1 H), 12.3-13.6 (m, 2 H); 13C
NMR (75.5 MHz, DMSO-d6) δ 33.2, 76.1, 102.3, 107.9, 108.1,
132.0, 160.5, 162.3, 171.6, 171.8. Anal. (C10H9NO4S): C, H,
N, S.
4,5-Dih ydr o-2-(2-h ydr oxyph en yl)-4-m eth ylth iazole-4(S)-
ca r boxylic Acid (4-Meth yla er u gin oic Acid ) (7). Phosphate
buffer (140 mL) and CH3OH (200 mL) were degassed with
nitrogen, and 12 (5.22 g, 38.6 mmol) and then 11 (4.56 g, 38.3
mmol) were added. The reactants, which dissolved after brief
sonication, were heated at 34 °C for 5 days with stirring under
nitrogen. The reaction mixture was cooled to room tempera-
ture, and the bulk of the CH3OH was removed by rotary
evaporation. The concentrate was cooled in ice water, and citric
acid (10.4 g, 54.1 mmol) was added with swirling. The mixture
was extracted with EtOAc (100 mL, 4 × 50 mL). The combined
organic extracts were washed with water (2 × 50 mL) and
brine (50 mL) and were concentrated in vacuo. Purification of
the residue on a Sephadex LH-20 column, eluting with 2%
EtOH/toluene, gave 2.89 g of 7 (32%) as a pale green solid,
1
mp 133-134 °C: [R]23 +42.0° (c 1.58, CH3OH); 600 MHz H
D
NMR (CD3OD) δ 1.67 (s, 3 H), 3.34 (d, 1 H, J ) 11.4), 3.88 (d,
1 H, J ) 11.5), 6.89 (ddd, 1 H, J ) 7.9, 7.9, 1.0), 6.94 (dd, 1 H,
J ) 8.3, 1.0), 7.37 (ddd, 1 H, J ) 8.3, 7.9, 1.5), 7.44 (dd, 1 H,
J ) 7.9, 1.5); 600 MHz 1H NMR18 (CD3OD) δ 1.65 (s, 3 H),
3.33 (d, 1 H, J ) 12.0), 3.85 (d, 1 H, J ) 12.0), 6.88 (ddd, 1 H,
J ) 8.0, 8.0, 0.8), 6.94 (dd, 1 H, J ) 8.3, 0.8), 7.35 (ddd, 1 H,
1H NMR of 4,5-Dih yd r o-2-(2,4-d ih yd r oxyp h en yl)th ia z-
ole-4(S)-ca r boxylic Acid (5) a n d Ga (III):(5) Ch ela te. The
proton resonances for ligand 5 in DMSO-d6 were assigned as
follows: δ 6.31 (d, 1 H, J ) 2.1, H-3′), 6.38 (dd, 1 H, J ) 8.5,
2.1, H-5′), 7.25 (d, 1 H, J ) 8.5, H-6′), 5.38 (dd, 1 H, J ) 9.6,
7.1, H-4), 3.57 (dd, 1 H, J ) 11.3, 7.1, H-5), 3.65 (dd, 1 H, J )
11.3, 9.6, H-5), 10.26 (s, OH), 12.62 (s, OH), 13.14 (br s, OH).
For 5 in D2O/phosphate buffer, observed resonances were
assigned as δ 6.32 (t, 1 H, J ) 2.0, H-3′), 6.38 (dd, 1 H, J )
9.9, 2.0, H-5′), 7.38 (d, 1 H, J ) 9.9, H-6′), 5.14 (dd, 1 H, J )
9.5, 7.3, H-4), 3.55 (dd, 1 H, J ) 11.2, 7.3, H-5), 3.72 (dd, 1 H,
J ) 11.2, 9.5, H-5). The resonance intensity for H-3′ decreased
with time, relative to the other observed resonances, indicative
of slow exchange. Over the range of 1.0-10.0 mM, plots of
concentration of 5 vs time demonstrated that above ap-
proximately 4 mM (in unexchanged 5) the kinetics appeared
as zero-order. As the concentration of unexchanged substrate
fell below 4 mM, the kinetics were first-order; plots of the
natural log of the concentration vs time were linear, with k )
J ) 8.3, 8.0, 1.5), 7.44 (dd, 1 H, J ) 8.0, 1.5). Anal. (C11H11
NO3S): C, H, N.
-
Circular dichroism measurements were run in CH3OH with
the cell path length of 1 cm at 24 °C. Values could not be
obtained for the UV maxima at 218 and 206 nm since the
instrument was not sensitive below about 255 nm due to the
strong UV absorption of 7 below 260 nm.
7 (c ) 0.014)
lit.18 (c ) 0.01)
λ (nm)
[θ] (deg)
λ (nm)
[θ] (deg)
318
280
258
+2530
-3610
-5330
321
280
260
218
206
+2631
-7416
-10058
+4268
+9326
4.7 ( 0.6 × 10-4 s-1
.
Multiple resonances for each carbon-bound proton were
observed in the spectrum of the Ga(III) chelate of 5 (3.0 mM
in 5, 1.5 mM in Ga(NO3)3, D2O/phosphate buffer). Observed
resonances were assigned as H-3′: 5.84 (d, 0.31 H, J ) 2.1),
6.16 (d, 0.10 H, J ) 2.1), 6.28 (d, 0.53 H, J ) 2.1); H-5′: 6.33
(dd, 0.28 H, J ) 8.7, 2.1), 6.39 (dd, 0.42 H, J ) 9.0, 2.1), 6.41
(dd, 1 H, J ) 8.7, 2.1); H-6′: 7.41 (d, 0.21 H, J ) 8.7), 7.46 (d,
0.60 H, J ) 9.0), 7.47 (d, 0.21 H, J ) 8.7); H-4: 5.35 (dd, 0.60
H, J ) 9.0, 13.5), 5.40 (dd, 0.40 H, J ) 9.0, 13.2); H-5: 3.51-
3.67 (m, 1 H), 3.70-3.86 (m, 1 H). As with free ligand 5, the
intensity of the three doublets assigned to H-3′ decreased with
time relative to the other observed resonances. Plots of the
natural log of the concentration (based on both the 5.84 ppm
and 6.28 ppm resonances) vs time were linear, indicative of
first-order kinetics (k ) 7.5 × 10-5 s-1).
4,5-Dih ydr o-2-(2,4-dih ydr oxyph en yl)-4-m eth ylth iazole-
4(S)-ca r boxylic Acid (8). Degassed 6 N HCl (500 mL) was
added to 9 (11.50 g, 48.3 mmol), and the mixture was heated
at 100 °C for 80 min under nitrogen.19 The solution was
concentrated under high vacuum, and the residue was dis-
solved in distilled water (70 mL). Evaporation was repeated
to give 18.77 g of 12 and 13 (quantitative) as a white solid, to
which a solution of 10 (5.72 g, 42.3 mmol) in degassed CH3-
OH (250 mL) was added. Degassed phosphate buffer (250 mL)
and NaHCO3 (14.1 g, 0.168 mol) were introduced to the
reaction to maintain a pH of 6. The reactants were heated at
71 °C for 3.5 days with stirring under nitrogen. The reaction
mixture was cooled to room temperature, and the bulk of the
CH3OH was removed by rotary evaporation. Cold 1 N HCl (150
mL) was slowly poured into the residue with stirring. Solid
was filtered, washed with water (2 × 15 mL), recrystallized
from aqueous EtOH, and dried under high vacuum for 1 day
at 111 °C, affording 6.084 g of 8 (57%) as pale tan needles,
1H NMR of Sod iu m 4,5-Dih yd r o-2-(2-h yd r oxyp h en yl)-
th ia zole-4(S)-ca r boxyla te (3, Sod iu m Sa lt). We have previ-
ously reported the 1H NMR spectrum of the sodium salt of 4
in CD3OD.12 In D2O/phosphate buffer, observed resonances

Iron Clearing Efficiency of Desferrithiocin Analogues
J ournal of Medicinal Chemistry, 1999, Vol. 42, No. 13 2439
were assigned as follows: 6.96-7.10 (m, 2 H, H-3′ and H-5′),
7.48 (td, 1 H, J ) 7.9, 1.5, H-4′), 7.57 (dd, 1 H, J ) 7.9, 1.5,
H-6′), 5.27 (dd, 1 H, J ) 9.5, 7.9, H-4), 3.57 (dd, 1 H, J ) 11.2,
7.9, H-5), 3.74 (dd, 1 H, J ) 11.2, 9.5, H-5). No exchange of
carbon-bound protons was observed.
1H NMR of 2,4-Dih yd r oxyben zoic Acid . 1H NMR spectra
of 2,4-dihydroxybenzoic acid was obtained under the same
conditions as ligand 5 for comparison. In DMSO-d6 the
spectrum was assigned as follows: 6.26 (d, 1 H, J ) 2.3, H-3),
6.34 (dd, 1 H, J ) 8.7, 2.3, H-5), 7.62 (d, 1 H, J ) 8.7, H-6),
10.33 (s, OH), 11.40 (br s, OH), 13.3 (br s, OH). In D2O/
phosphate buffer, assignments were as follows: 6.40 (d, 1 H,
J ) 2.4, H-3), 6.46 (dd, 1 H, J ) 2.4, 8.6, H-5), 7.71 (d, 1 H, J
) 8.7, H-6). No exchange of carbon-bound protons was
observed.
Toxicity Eva lu a tion s in Rod en ts. Male Sprague-Dawley
rats averaging 450 g were housed in polycarbonate cages.
Before the first drug administration, the rats were weighed
and evaluated for their general condition. The rats were given
analogue 7 or 8 at a dose of 384 µmol/kg once daily for 10 days
following an overnight fast. The drugs were solubilized in 40%
Cremophor RH-40/water (v/v). Ligand 7 was administered
either po by gavage or sc (n ) 3/group), whereas 8 was given
po by gavage only (n ) 5/group). The rats were fed 3 h after
drug administration and were allowed access to food for 5 h.
The amount of food and water consumed was recorded daily.
In addition, the animals were weighed and evaluated for their
activity level and general appearance on a daily basis. Control
animals were given an equivalent volume of 40% Cremophor/
water (v/v) either po by gavage or sc as appropriate and were
maintained on the same feeding schedule as the drug-treated
animals.
Ca n n u la tion of Bile Du ct in Ra ts. The cannulation has
been described previously.12,15,26 Briefly, male Sprague-Dawley
rats averaging 450 g were housed in Nalgene plastic metabolic
cages during the experimental period and given free access to
water. The animals were anesthetized using sodium pento-
barbital (55 mg/kg) administered ip. The bile duct was can-
nulated using 22-gauge polyethylene tubing. The cannula was
inserted into the duct about 1 cm from the duodenum and tied
snugly in place. After threading through the shoulder, the
cannula was passed from the rat to the swivel inside a metal
torque-transmitting tether, which was attached to a rodent
jacket around the animal’s chest. The cannula was directed
from the rat to a Gilson microfraction collector (Middleton, WI)
by a fluid swivel mounted above the metabolic cage. Bile
samples were collected at 3-h intervals for 24 h. The urine
sample was taken at 24 h. Sample collection and handling were
as previously described.12
If an animal expired during the course of the evaluation, a
necropsy was performed; any gross abnormalities were re-
corded. After a 10-day exposure to the drug, animals were
sacrificed 24 h after the final dose and extensive tissues
including adrenal gland, aorta, bone marrow, brain, eye,
esophagus, stomach, duodenum, jejunum, ileum, cecum, colon,
rectum, testicle, heart, kidney, liver, lung, lymph node,
pancreas, prostate, salivary gland, skeletal muscle, skin,
spleen, thymus, thyroid, trachea, and bladder were sent to an
outside pathologist for histological evaluation.
Ack n ow led gm en t. Financial support was provided
by National Institutes of Health Grant RO1DK49108.
We thank Katie Ratliff-Thompson, Curt Zimmerman,
and Elizabeth Nelson for their technical assistance and
Dr. Eileen Eiler-Hughes for her editorial comments.
Ir on Loa d in g of C. a pella Mon k eys. C. apella monkeys
were given iron dextran iv as described in detail in earlier
publications.11,20,26 Briefly, the iron dextran was administered
to the animals by slow infusion at a dose of 200-300 mg of
iron/kg of body weight over 45-60 min. Infusions were
performed as necessary to provide about 500 mg of iron/kg of
body weight. After administration of iron dextran, we waited
at least 60 days before using any of the animals in experiments
evaluating iron chelating agents.
Refer en ces
(1) Bergeron, R. J . Synthesis and Solution Structures of Microbial
Siderophores. Chem. Rev. 1984, 84, 587-602.
(2) Raymond, K. N.; Carrano, C. J . Coordination Chemistry and
Microbial Iron Transport. Acc. Chem. Res. 1979, 12, 183-190.
(3) Neilands, J . B. Microbial Iron Compounds. Annu. Rev. Biochem
1981, 50, 715-731.
(4) Bickel, H.; Hall, G. E.; Keller-Schierlein, W.; Prelog, V.; Vischer,
E.; Wettstein, A. Metabolic Products of Actinomycetes. XXVII.
Constitutional Formula of Ferrioxamine B. Helv. Chim. Acta
1960, 43, 2129-2138.
(5) Anderegg, G.; L’Eplattenier, F.; Schwarzenbach, G. Hydroxamate
Complexes. III. Iron(III) Exchange Between Sideramines and
Complexones. A Discussion of the Formation Constants of the
Hydroxamate Complexes. Helv. Chim. Acta 1963, 46, 1409-
1422.
(6) Propper, R. D.; Cooper, B.; Rufo, R. R.; Nienhuis, A. W.;
Anderson, W. F.; Bunn, H. F.; Rosenthal, A.; Nathan, D. G.
Continuous Subcutaneous Administration of Deferoxamine in
Patients with Iron Overload. N. Engl. J . Med. 1977, 297, 418-
423.
(7) Aksoy, M.; Seyithanoglu, B. Y.; Bozbora, A. Thalassemia and
Desferrioxamine (Desferal) Introductory Remarks, and Clinical
Laboratory Observations. In Hypertension and Iron Chelation
in Thalassemia; Aksoy, A., and Birdwood, G. F. B., Eds.;
Huber: Berne, Switzerland, 1984; p 11.
(8) Graziano, J . H.; Markenson, A.; Miller, D. R.; Chang, H.; Bestak,
M.; Meyers, P.; Pisciotto, P.; Rifkind, A. Chelation Therapy in
Beta-Thalassemia Major. I. Intravenous and Subcutaneous
Deferoxamine. J . Pediatr. 1978, 92, 648-652.
(9) Pippard, M. J .; Callender, S. T. The Management of Iron
Chelation Therapy. Br. J . Haematol. 1983, 54, 503-507.
(10) Olivieri, N. F.; Brittenham, G. M. Iron-Chelating Therapy and
the Treatment of Thalassemia. Blood 1997, 89, 739-761.
(11) Bergeron, R J .; Streiff, R. R.; Wiegand, J .; Vinson, J . R. T.;
Luchetta, G.; Evans, K. M.; Peter, H.; J enny, H.-B. A Compara-
tive Evaluation of Iron Clearance Models. Ann. N. Y. Acad. Sci.
1990, 612, 378-393.
P r im a te F eca l a n d Ur in e Sa m p les. Fecal and urine
samples were collected and processed as accounted previ-
ously.11,20,26 The collections began 4 days prior to the admin-
istration of the test drug and continued for an additional 5
days after the drug was given. Iron concentrations were
determined by flame atomic absorption spectroscopy as de-
scribed in earlier publications.11,15
Dr u g P r ep a r a tion a n d Ad m in istr a tion . The iron chela-
tors were solubilized in 40% Cremophor RH-40/water (v/v) and
given po to the rats at a dose of 150 µmol/kg. In the primates,
the compounds were solubilized in 40% Cremophor RH-40/
water (v/v) and given po at a dose of 75, 150, or 300 µmol/kg
as indicated in Table 1. In addition, analogue 5 was given po
and sc in dH2O at doses of 300 and 150 µmol/kg, respectively.
Ca lcu la tion of Ir on Ch ela tor Efficien cy. The efficiency
of each chelator was calculated on the basis of a 2:1 ligand:
iron complex. The efficiencies in the rodent model were
calculated by subtracting the iron excretion of control animals
from the iron excretion of treated animals. This number was
then divided by the theoretical output; the result is expressed
as a percentage. In the monkeys, the numbers were generated
by averaging the iron output for 4 days before the administra-
tion of the drug, subtracting these numbers from the 2-day
iron clearance after the administration of the drug, and then
dividing by the theoretical output; the result is expressed as
a percentage.
(12) Bergeron, R. J .; Wiegand, J .; Dionis, J . B.; Egli-Karmakka, M.;
Frei, J .; Huxley-Tencer, A.; Peter, H. H. Evaluation of Desfer-
rithiocin and its Synthetic Analogues as Orally Effective Iron
Chelators. J . Med. Chem. 1991, 34, 2072-2078.
(13) Bergeron, R. J .; Liu, C. Z.; McManis, J . S.; Xia, M. X. B.; Algee,
S. E.; Wiegand, J . The Desferrithiocin Pharmacophore. J . Med.
Chem. 1994, 37, 1411-1417.
Sta tistica l An a lysis. Data are presented as the mean (
the standard error of the mean. For comparisons of the means
of two groups, the two-sample t-test (without the assumption
of equality of variances) was used for analyzing the rodent and
primate data. All tests were two-tailed, and a significance level
of P < 0.05 was used.

2440 J ournal of Medicinal Chemistry, 1999, Vol. 42, No. 13
Bergeron et al.
(14) Bergeron, R. J .; Wiegand, J .; Weimar, W. R.; Vinson, J . R. T.;
Bussenius, J .; Yao, G. W.; McManis, J . S. Desazadesmethyldes-
ferrithiocin Analogues as Orally Effective Iron Chelators. J . Med.
Chem. 1999, 42, 95-108.
(15) Bergeron, R. J .; Wiegand, J .; Wollenweber, M.; McManis, J . S.;
Algee, S. E.; Ratliff-Thompson, K. Synthesis and Biological
Evaluation of Naphthyldesferrithiocin Iron Chelators. J . Med.
Chem. 1996, 39, 1575-1581.
(16) Bergeron, R. J .; Wiegand, J .; Ratliff-Thompson, K.; Weimar, W.
R. The Origin of the Differences in (R)- and (S)-Desmethyldes-
ferrithiocin: Iron-Clearing Properties. Ann. N. Y. Acad. Sci.
1998, 850, 202-216.
(17) Karmarkar, S. N.; Kelkar, S. L.; Wadia, M. S. A Simple Unusual
One-Step Conversion of Aromatic Aldehydes into Nitriles.
Synthesis 1985, 510-512.
(18) Ryoo, I.-J .; Song, K.-S.; Kim, J .-P.; Kim, W.-G.; Koshino, H.; Yoo,
I.-D. A New Cytotoxic Phenylthiazoline, 4-Methylaeruginoic
Acid, from Streptomyces sp. KCTC 9303. J . Antibiot. 1997, 50,
256-258.
(19) Naegeli, H.-U.; Za¨hner, H. Metabolites of Microorganisms. Part
193. Ferrithiocin. Helv. Chim. Acta 1980, 63, 1400-1406.
(20) Bergeron, R. J .; Streiff, R. R.; Creary, E. A.; Daniels, R. D., J r.;
King, W.; Luchetta, G.; Wiegand, J .; Moerker, T.; Peter, H. H.
A Comparative Study of the Iron-Clearing Properties of Desfer-
rithiocin Analogues with Desferrioxamine B in a Cebus Monkey
Model. Blood 1993, 81, 2166-2173.
(21) Bergeron, R. J .; Streiff, R. R.; Wiegand, J .; Luchetta, G.; Creary,
E. A.; Peter, H. H. A Comparison of the Iron-Clearing Properties
of 1,2-Dimethyl-3-Hydroxypyrid-4-One, 1,2-Diethyl-3-Hydroxy-
pyrid-4-One, and Deferoxamine. Blood 1992, 79, 1882-1890.
(22) Wolfe, L. C.; Nicolosi, R. J .; Renaud, M. M.; Finger, J .; Hegsted,
M.; Peter, H.; Nathan, D. G. A Non-Human Primate Model for
the Study of Oral Iron Chelators. Br. J . Haematol. 1989, 72,
456-461.
(23) Gomori, G. Preparation of Buffers for Use in Enzyme Studies.
Methods Enzymol. 1955, 1, 138-146.
(24) Kishore, V.; Mathur, K. B.; Dhar, M. M. Synthesis of R-Poly-
[Nꢀ-(2-aryl-∆2-thiazoline-4-carbonyl)-L-lysines] with Antiviral
Activity. Ind. J . Chem. 1977, 15B, 255-257.
(25) Anderegg, G.; Ra¨ber, M. Metal Complex Formation of a New
Siderophore Desferrithiocin and of Three Related Ligands. J .
Chem. Soc., Chem. Commun. 1990, 1194-1196.
(26) Bergeron, R. J .; Wiegand, J .; Brittenham, G. M. HBED:
A
Potential Alternative to Deferoxamine for Iron-Chelating Therapy.
Blood 1998, 91, 1446-1452.
J M990058S