Hydrophilic and lipophilic iron chelators with the same complexing abilities

Article abstract of DOI:10.1002/1521-3765(20020301)8:5<1091::AID-CHEM1091>3.0.CO;2-Y

A new series of iron chelators with the same coordination sphere as the water-soluble ligand O-trensox, but featuring a variable hydrophilic-lipophilic balance, have been obtained by grafting oxyethylene chains of variable length on a C-pivot tripodal sca

Full text of DOI:10.1002/1521-3765(20020301)8:5<1091::AID-CHEM1091>3.0.CO;2-Y

FULL PAPER
Hydrophilic and Lipophilic Iron Chelators
with the Same Complexing Abilities
Daniel Imbert, Paul Baret, Didier Gaude, Isabelle Gautier-Luneau, Gise¡le Gellon,
Fabrice Thomas, Guy Serratrice, and Jean-Louis Pierre*[a]
Abstract: A new series of iron chelators
with the same coordination sphere as the
water-soluble ligand O-trensox, but fea-
turing a variable hydrophilic lipophilic
balance, have been obtained by grafting
oxyethylene chains of variable length on
a C-pivot tripodal scaffold. The X-ray
structure of a ferric complex exhibiting
tris(8-hydroxyquinolinate) coordination
and solution thermodynamic properties
(pKa of the ligands, stability constants of
the ferric complexes) have been deter-
mined. The complexing ability (pFeIII
values) of the ligands are similar to that
of O-trensox. Partition coefficients be-
tween water and octanol or chloroform
have been measured and transport
across a membrane has been mimicked
(™shuttle process∫). The results of bio-
logical assays (iron chelation with free
ligands or iron nutrition with ferric
complexes) could not be correlated with
the partition coefficients. These results
call into question the role of distribution
coefficients (of the ligands and/or com-
plexes) in the biological activities of iron
chelators.
Keywords: chelates
¥
iron
¥
lipophilicity ¥ partition coefficients ¥
tripodal ligands
Introduction
Iron chelation by natural or abiotic chelators can be applied to
human diseases caused by iron overload, and iron complexes
of similar chelators can be used to alleviate iron deficiency in
plants, preventing and even reversing iron chlorosis.[1] The
ability of the chelator to fulfil these functions is related
primarily to its powerful and selective affinity for the metal,
revealed by the pFeIII value.[2] Water solubility is required, but
it is generally claimed that access to the cell through biological
membranes depends on the lipophilicity of the chelator or of
its iron complex. This lipophilicity is usually measured in
terms of the partition coefficient P between n-octanol and
water or, less often, between chloroform and water.[3] Other
factors affecting a compound×s ability to permeate liquid
membranes freely are molecular weight, size and ionisation.
We have developed the efficient water-soluble chelator
O-trensox[4, 5], which exhibits a strong complexing ability for
ferric iron (pFeIII ˆ 29.5) and a high selectivity towards its
complexation compared with other biological metal ions
(pCuII ˆ 22.8, pZnII ˆ 21.7, pFeII ˆ 17.9, pCaII ˆ 13.6).[6] The
octanol/water partition coefficients of O-trensox (0.02) and of
its iron complex (<0.005) show the insolubility of these
species in octanol.[7] Moreover, the molecular mass of
O-trensox is too high for it to use membrane aquaporin
channels, so O-trensox and its iron complex would not be
expected to exhibit biological activity requiring them to cross
cell membranes. Nevertheless, cellular protection is obtained
with O-trensox through iron mobilisation from rat hepato-
cytes in vitro and in vivo,[8] and the ferric complex of
O-trensox enables plant cells to metabolise iron.[9] These
surprising results led us to imagine a series of iron chelators
with the same coordination sphere as O-trensox (similar pFeIII
values), but a variable hydrophilic lipophilic balance, modu-
lated by a polyoxyethylenic chain of variable length grafted on
a C-pivot tripodal scaffold. Short-chain polyether analogues
of DFO (desferrioxamine) with enhanced lipophilicity, but
still water-soluble, have been evaluated for chelating iron in
rats.[10]
[a] J.-L. Pierre, D. Imbert, P. Baret, D. Gaude, I. Gautier-Luneau,
G. Gellon, F. Thomas, G. Serratrice
¬
Laboratoire de Chimie Biomimetique, LEDSS
¬
UMR CNRS 5616, Universite Joseph Fourier
BP 53, 38041 Grenoble Cedex 9 (France)
Fax : (‡33)4-76-51-48-36
E-mail: jean-louis.pierre@ujf-grenoble.fr
Chem. Eur. J. 2002, 8, No. 5
¹ WILEY-VCH Verlag GmbH, 69451 Weinheim, Germany, 2002
0947-6539/02/0805-1091 $ 17.50+.50/0
1091

J.-L. Pierre et al.
FULL PAPER
Here we describe the synthesis of the ligands and of their
iron complexes, and we try to correlate their physicochemical
properties, including their structural characterisation, solution
thermodynamics and liquid membrane transport, with bio-
logical activities.
LMCT transitions, respectively. These data reveal a tris-
(8-hydroxyquinolate) type of complexation, like that previ-
ously demonstrated for the ferric complex of O-trensox.[5]
X-ray structure of Fe Cox200: Single crystals of the Fe
Cox200 complex [Fe(C50H57N6O11)] ¥ H2O were obtained by
diffusion of diethyl ether into a solution of the complex in
methanol. Its crystal structure reveals a neutral mono-
Results and Discussion
iron(iii) complex with
a distorted octahedral tris(8-hy-
droxyquinolate) coordination geometry for FeIII.
Syntheses: We have described the C-pivot tripodal starting
compound 2 previously.[11] The poly(oxyethylene) (POE)
chains were grafted onto the tripodal template by using the
Williamson×s method, then the cyano functions of 3 were
reduced to give the triamine 4, which in the last step reacted
with 7-carboxy-8-hydroxyquinoline 6 activated by carbon-
yldiimidazole (Scheme 1). The products 6b (Cox750) and 6c
(Cox2000) are polymeric materials, whereas the polyether
chain has a defined length in 6a (Cox200).
The N3O3 iron-coordination sphere is provided by the three
8-hydroxyquinoline arms of the ligand (all the atoms of each
arm connected to the tripodal carbon spacer C14 are labelled
a, b or c) (Figure 1). Each 8-hydroxyquinoline subunit forms a
five-membered chelate ring with iron, leading to the facial
isomer. Selected bond lengths and angles are given in Table 1.
The average five-membered chelate ring O1x-Fe-N1x angle
(x ˆ a, b or c) is 78.68 and is close to that observed for the
Scheme 1. Ligand syntheses.
Iron complexes: The iron(iii) complex of Cox200 was obtained
by adding three equivalents of KOH and one equivalent of
[Fe(acac)3] to a solution of the ligand in degassed methanol/
dichloromethane. The complex was purified by standard
Figure 1. ORTEP plot of Fe Cox200. Hydrogen atoms are omitted for
clarity. Ellipsoids are drawn at the 30% probability level.
‡
procedures. The 1:1 stoichiometry was demonstrated by FAB
Table 1. Selected bond lengths [ä] and angles [8] for Fe Cox200.
‡
mass spectroscopy (m/z for [L À 3H‡Fe] ; L ˆ ligand). The
À
À
Fe O1a
1.949(3)
1.935(3)
1.959(3)
78.5(1)
78.9(1)
78.6(1)
95.4(1
98.3(1)
95.6(1)
93.1(1)
95.7(1)
Fe N1a
2.184(4)
2.166(4)
2.143(4)
8-hydroxyquinolate bonding mode (coordination with the
nitrogen and the oxygen atoms) was established unambigu-
ously from the X-ray structural determination of the Cox200
ferric complex (vide infra). The 1:1 stoichiometry of Cox750
and Cox2000 was demonstrated by UV/Vis spectroscopy
(Job×s method) in a pH 7.4 buffered aqueous solution.[12] The
aqueous solutions of the complexes exhibit similar spectra
with main features at 450 (e ˆ 4600mÀ1 cmÀ1) and 590 nm (e ˆ
3900mÀ1 cmÀ1) attributed to phenolate Fe and pyridine Fe
À
À
Fe O1b
Fe N1b
À
À
Fe O1c
Fe N1c
O1a-Fe-N1a
O1b-Fe-N1b
O1c-Fe-N1c
O1b-Fe-O1a
O1b-Fe-O1c
O1a-Fe-O1c
O1a-Fe-N1b
O1c-Fe-N1a
O1a-Fe-N1c
O1b-Fe-N1a
O1c-Fe-N1b
N1b-Fe-N1a
N1c-Fe-N1b
N1c-Fe-N1a
O1b-Fe-N1c
164.2(1)
165.2(1)
171.1(1)
88.0(1)
93.5(1)
87.4(1)
100.0(1)
1092
¹ WILEY-VCH Verlag GmbH, 69451 Weinheim, Germany, 2002
0947-6539/02/0805-1092 $ 17.50+.50/0
Chem. Eur. J. 2002, 8, No. 5

Iron Chelators
1091 1100
tris(8-hydroxyquinolato)iron(iii) complex (79.28),[13] whereas
all cis angles are in the range 87.4(1)8 100.0(1)8 and apical
angles are in the range 164.2(1)8 171.1(1)8. The three
8-hydroxyquinoline groups (O1x, N1x, C1x C9x, x ˆ a, b or
c) are planar with a maximum mean deviation of plane c of
0.018 ä. The dihedral angles between the planes a, b and c are
91.248 (a b), 90.288 (a c) and 71.258 (b c). The three arms
are long enough to allow nonrestricting complexation around
the iron metal. The bond lengths with iron(iii) are in narrow
Figure 2. Potentiometric titration curves for a) ligand Cox2000 (1mm);
b) Cox2000 (1 mm) ‡ FeIII ( 1 mm); m ˆ mol base added per mol ligand,
Tˆ 258C, I ˆ 0.1m (NaClO4).
ranges: between 1.935(3) and 1.959(3) ä for FeIII O and
À
between 2.143(4) and 2.184(4) ä for FeIII N bonds. These are
À
very close to the bond lengths observed in the tris(8-
hydroxyquinolato)iron(iii) complex with mer geometry,[13] in
Hence pKa ˆ 2.89 Æ 0.02, pKa ˆ 4.13 Æ 0.02, pKa ˆ 6.80 Æ
5
4
3
À
which the Fe N bond lengths lie between 2.125(6) and
0.02, pKa ˆ 8.40 Æ 0.01, pKa ˆ 10.38 Æ 0.01, but the lowest
2
1
À
2.172(5) ä, and two of the Fe O bonds are similar in length
pKa values cannot be determined from this analysis.
n
(1.936(5) and 1.956(5) ä), while the third is longer
A spectrophotometric titration was carried out over the pH
range 1 11 (Figure 3). First, the absorbance data of the UV
spectra in the pH range 1.02 4.78 (Figure 3A) were pro-
cessed by the nonlinear least-squares program LETAGROP-
SPEFO[18, 19] (absorbance values at nine wavelengths between
(1.996(5) ä). The difference could be explained by the mer
À
geometry of this complex. The Fe O bond lengths may also
À
be compared with the average Fe O (hydroxyl oxygen) bond
lengths in the tris(salicylate) complexes, 1.94 ä for [Fe(O-
trensox)][14] and 1.92 ä for [Fe(trensam)] and [Fe(tren(3M)-
sam)][15] (trensam ˆ tris[(2-hydroxybenzoyl)-2-aminoethyl]-
amine; tren(3M)sam ˆ tris[(2-benzoxy-3-methoxybenzoyl)-2-
aminoethyl]amine) and in the tris(catecholate), 2.01 ä for
[Fe(trencam)] (trencam ˆ tris[(2,3-dihydroxybenzoyl)-2-ami-
noethyl]amine). To our knowledge this is the first example of
a FeIII complex structure with 1:1 stoichiometry exhibiting fac
tris(8-hydroxyquinolate) coordination. Only the FeIII complex
with O-trensox, obtained in an acidic medium and involving
tris(salicylate)-type coordination, has been described.[7] Fur-
thermore, the structure reveals that amide protons interact
strongly by hydrogen bonding with the quinolinol oxygen
coordinated to iron forming three six-membered rings. The
H2nx ¥¥¥ O1x and the N2x ¥¥¥ O1x distances (x ˆ a, b or c) are
1.83, 1.86, 2.04 ä and 2.677, 2.672, 2.713 ä, respectively.
Hydrogen bonding between the amide hydrogen and the ortho
quinolinol oxygen is an important feature of the structure and
is claimed to enhance the stability of the complex. This
interaction is also observed in iron triscatecholate and other
complexes with this type of ligand. For [Fe(trencam)] the
NH¥¥¥ O distances vary from 1.80 to 1.85 ä.[16]
Figure 3. UV/Vis absorption spectra of Cox2000 as a function of pH:
[Cox2000] ˆ 0.1mm, Tˆ 258C, I ˆ 0.1m (NaClO4). pH values: A) a) 1.02,
b) 4.78; B) c) 6.15, d) 10.88.
320 and 480 nm). The best fit (S(Aexptl À Acalcd)2 ˆ 10À2),
obtained by considering the two deprotonation equilibria
represented by Equations (3) and (4), yielded pKa ˆ 2.43 Æ
6
Thermodynamic solution studies: These studies were per-
formed as described previously for O-trensox.[5] Here we
present the results for the water-soluble ligand Cox2000.
0.15 and pK' ˆ 7.14 Æ 0.09, respectively.
a
5
3‡
2‡
‡
LH6 > LH5 ‡ H
Ka
(3)
(4)
‡
6
K'
2‡
‡
LH5 > LH3 ‡ 2H
a
Ligand deprotonation constants: The fully protonated form of
Cox2000 possesses six deprotonation sites (three pyridinium
nitrogen and three hydroxyl oxygen atoms) and is denoted
LH63‡. The potentiometric titration of the fully protonated
ligand with NaOH in 0.1m NaClO4 at 258C (Figure 2, curve a)
allowed the determination of five deprotonation constants.
Analysis of the potentiometric titration curve over the pH
range 2.5 10.5 by the SUPERQUAD[17] program yielded the
5
The K' value is in good agreement with the sum pK'
a
a
5
5
pK' ˆ 7.02 determined from potentiometric measurements.
a
4
Second, analysis of the absorbance data over the pH range
6.15 10.88 (Figure 3B) by the LETAGROP-SPEFO program
(absorbance values at 10 wavelengths between 340 and
450 nm, S(Aexptl À Acalcd)2 ˆ 10À3) led to the equilibria of
Equations (5) and (6) and yielded the values pKa ˆ 7.11 Æ
3
pKa values (127 points, sfit ˆ 3.7) defined by Equations (1) and
n
0.19 and pK' ˆ 18.75 Æ 0.01, respectively.
a
2
(2).
‡
À
‡
LH3 > LH2 ‡ H
Ka
(5)
(6)
LHn > LHnÀ1 ‡ H
(1)
(
3
LH2À > L3À ‡ 2H
K'
2)
‡
‡
Ka ˆ [LHnÀ1][H ]/[LHn]
n
a
2
Chem. Eur. J. 2002, 8, No. 5
¹ WILEY-VCH Verlag GmbH, 69451 Weinheim, Germany, 2002
0947-6539/02/0805-1093 $ 17.50+.50/0
1093

J.-L. Pierre et al.
FULL PAPER
These values are in good agreement with the pKa (6.80) and
3
the sum pKa ‡ pKa ˆ 18.78 determined from potentiometric
2
1
measurements.
Compared with the pKa values of O-trensox,[5] the three
lowest pKa values are attributed to the pyridinium nitrogen
atoms and the three highest to the hydroxy moieties. The
average deprotonation constants are 3.15 for the pyridinium
nitrogen atoms (pKav ) and 8.53 for the hydroxy groups
(4 6)
(pKav ). The corresponding values for O-trensox, 2.46 and
(1 3)
8.07, respectively, are lower because of the electron-with-
drawing effect of the sulfonate groups. The pKa ranges for the
pyridinium nitrogen atoms (2.43, 2.89 and 4.13) and the
hydroxy groups (6.80, 8.40 and 10.38) differ from the
statistical factor of log3 (i.e., 0.48), indicating that there is
some cooperativity between the three arms of the tripod, in
contrast to the situation in O-trensox.
Stability constants of the ferric complexes: The equilibria of
the metal complexes were studied by means of potentiometric
and spectrophotometric titrations. The potentiometric titra-
tion curve of a 1:1 solution of ferric ion and Cox2000
(Figure 2, curve b) shows a large jump in pH at m ˆ 6,
indicating that all the ligand protons are released when the
ligand binds ferric ion at pH 7.4. However, it was not possible
to determine logb110, since formation of the complex is
complete at low pH values. The spectrophotometric titration
was carried out from pH 0 to pH 11 in order to determine the
deprotonation constants of the ferric complex and the global
constants b11n defined by Equation (7).
Figure 4. UV/Vis absorption spectra of FeIII Cox2000 as a function of pH:
[Cox2000] ˆ [Fe] ˆ 0.1mm, I ˆ 0.1m (NaClO4). pH values for A), B) and C):
a) 0, b) 1, c) 1.91, d) 3.48.
increase in lmax from 440 to 450 nm and from 560 to 590 nm
was observed with an increase in the molar extinction
coefficient. The refinement of the absorbance data with the
LETAGROP-SPEFO program (absorbance values at
24 wavelengths between 340 and 800 nm, S(Aexptl À Acalcd)2 ˆ
3  10À2) provided logb110 ˆ 32.12 Æ 0.09, e ˆ 4600mÀ1 cmÀ1
(lmax ˆ 450 nm) and e ˆ 3900 mÀ1 cmÀ1 (lmax ˆ 590 nm) for
the FeL species. No spectral change was observed until pH 9,
beyond which absorbances decreased at all the wavelengths,
suggesting the formation of hydroxo complexes.
The constants and spectral parameters are collected in
Table 2. The value of logb114 is less accurate than that of
logb111 and logb110, because the ionic strength varied over the
pH range 0 1. The distribution of the species according to pH
are shown in Figure 5.
‡
b11n ˆ [FeLHn]/[Fe3‡][L][H ]n
(7)
Between pH 0 and pH 1 (Figure 4A) a charge-transfer band
appears at lmax ˆ 440 nm with a shoulder at 550 nm. The
absorbance data were refined with the program LETAGROP-
SPEFO (absorbance values at 18 wavelengths between 440
and 800 nm). The best refinement (S(Aexptl À Acalcd)2 ˆ 2 Â
10À2) was obtained by considering the formation of the
4‡
FeLH4 species, and provided the values logb114 ˆ 38.98 Æ
0.11 and e ˆ 4600mÀ1 cmÀ1 at lmax ˆ 440 nm. The value
logKFeLH ˆ 3.95 Æ 0.11 was deduced, whereby KFeLH is de-
4
4
fined by Equations (8) and (9).
3‡
4‡
‡
Fe3‡ ‡ LH6 > FeLH4 ‡ 2H
(8)
(
Table 2. Equilibrium constants, logb11n and UV/Vis spectral characteristics
4‡
‡
3‡
KFeLH ˆ [FeLH4 ][H ]2/[Fe3‡][LH6
]
9)
4
of ferric Cox2000 complexes.
Complex
logb11n
pFeIII[a]
lmax [nm]
e [mÀ1 cmÀ1
]
4‡
An increase in the pH from 1 to 1.91 resulted in the
appearance of a charge-transfer band at lmax ˆ 560 nm with
an isosbestic point at 372 nm (Figure 4B), indicating the
presence of only two species with different absorbances.
These absorbance data were also refined with the
LETAGROP-SPEFO program (absorbance values at
24 wavelengths between 340 and 800 nm). The best fit
(S(Aexptl À Acalcd)2 ˆ 9  10À3) showed that deprotonation oc-
curs in a three-proton step yielding logb111 ˆ 34.24 Æ 0.05 and
absorption maxima at lmax ˆ 440 nm (e ˆ 4600mÀ1 cmÀ1) and
FeLH4
FeLH
38.98 Æ 0.11
34.24 Æ 0.05
440
440
560
450
590
4400
4600
2900
4600
3900
‡
FeL
32.12 Æ 0.09
29.1
[a] pFeIII ˆ Àlog[Fe3‡], calculated for [Fe3‡] ˆ 10À3 mm, [L] ˆ 10À2 mm and
pH ˆ 7.4.
The stability constants b110 of the ferric complex FeL were
also determined by spectrophotometric competition experi-
ments with ethylenediaminetetraacetic acid (edta) over the
pH range 2.05 2.49. The competition equilibrium can be
‡
lmax ˆ 560 nm (e ˆ 2900mÀ1 cmÀ1) for the FeLH species.
When the pH was raised from 1.91 to 3.48 (Figure 4C) an
1094
¹ WILEY-VCH Verlag GmbH, 69451 Weinheim, Germany, 2002
0947-6539/02/0805-1094 $ 17.50+.50/0
Chem. Eur. J. 2002, 8, No. 5

Iron Chelators
1091 1100
As a measure of the complexing efficiency of Cox2000 for
iron(iii) at physiological pH, the concentration of free Fe3‡
under standard conditions (pH 7.4, [L]tot ˆ 10À5 m, [Fe]tot
ˆ
10À6 m) expressed as pFeIII (pFeIII ˆ Àlog[Fe3‡]) is 29.1, which
is very close to that of O-trensox (29.5).[5] This indicates that
the substitution of the tren unit N(CH2CH2)3 by the
(POE)C(CH2CH2CH2)3 backbone does not change the com-
plexing efficiency of the ligand significantly.
Figure 5. Distribution curves for FeIII Cox2000 species at pH
[Cox2000] ˆ [Fe3‡] ˆ 0.1mm.
1 8:
Hydrophilic lipophilic balance
Partition coefficients: Two biphasic systems were studied for
the Cox chelators (Cox200, Cox750 and Cox2000). The
partition coefficients between an aqueous phase buffered at
pH 7.4 (Tris buffer) and octanol or chloroform were deter-
mined (Tables 3 and 4, respectively).
expressed by Equations (10) and (11), in which Y denotes
3‡
edta, LH6 the ligand Cox2000, Ka Ka the deprotonation
1
6
constants of LH6 and K' , K' and K' the deprotonation
3‡
a
a
a
3
1
2
[20]
constants of YH3
.
‡
3‡
FeL ‡ YH3 ‡ 3H > FeY‡ LH6
(10)
(11)
Table 3. Octanol water partition coefficients
O-trensox.[a]
P for the Cox chelators and
K ˆ [FeY][LH6]/[FeL][YH3][H ]3
‡
ˆ b110(FeY) K' K' K' /b110(FeL) Ka Ka Ka Ka Ka Ka
Entry Ligand or Fe Ligand
Concentration [%]
in water[b]
in octanol
at 18h at 96h at 18h at 96h at 18h at 96h
P[a] [%]
a
a
a
3
1
2
3
4
5
6
1
2
The concentration of FeL was calculated from the absorb-
ance A at 590 nm according to Equation (12), in which KFeLH
is the deprotonation constant of the FeLH complex.
1
2
3
4
5
6
7
8
Cox2000
Cox750
Cox200
O-trensox[c]
Fe Cox2000
Fe Cox750
Fe Cox200
Fe O-trensox[c]
97
57
6
98
99
53
10
100
94
10
98
3
43
94
2
6
90
2
0.03 0.06
0.78
15.7
0.02
9.0
‡
A ˆ [FeL](eFeL ‡ eFeLH [H ]/KFeLH
)
(12)
1
0.01 0.02
0.90
9.0
47
90
0
The concentrations of the other species in Equations (10)
and (11) were calculated from the mass balance equation and
pH [Eqs. (13) (15), in which each a is the usual Ringbom
coefficient].[21]
< 0.005
[a] P ˆ ratio of the concentrations in water and octanol. [b] With Tris buffer.
[c] Ref. [7].
[Fe]tot ˆ aFeL [FeL] ‡ aFeY [FeY]
[L]tot ˆ aFeL [FeL] ‡ aL [L]
[Y]tot ˆ aFeY [FeY] ‡ aY [Y]
(13)
(14)
(15)
Table 4. Chloroform water partition coefficients P[a] of the Cox chelators
and Fe Cox complexes.
Entry
Ligand or Fe Ligand
Concentration [%]
in water[b]
in chloroform
P[a]
9
10
11
12
13
14
Cox2000
Cox750
Cox200
Fe Cox2000
Fe Cox750
Fe Cox200
1
0
0
1
0
0
99
100
100
99
100
100
99
From the known formation constant of FeY (logb110
ˆ
[b]
25.0),[20] an average formation constant obtained for the ferric
Cox2000 complex (logb110 ˆ 32.3) is in good agreement with
that (32.1) determined from the spectrophotometric titration.
[b]
99
[b]
[b]
4‡
The visible spectrum of the FeLH4 species (lmax ˆ 440 nm,
e ˆ 4600mÀ1 cmÀ1 and a shoulder at 550 nm) is similar to that
of the O-trensox bis(salicylate) complex (coordination with
two arms of the ligand through the oxygen atoms of the
carbonyl and hydroxyl groups) but with lower e values. The
[a] P ˆ ratio of the concentrations in water and chloroform after 18 h. [b] In
Tris buffer. [c] High P which could not be measured by the method
employed.
The three Cox ligands and the three corresponding iron
complexes constitute two interesting sets: Cox2000 and its
iron complex are soluble in water and insoluble in octanol. [A
long grafted POE chain can be used here instead of ionised
groups (such as sulfonate in O-trensox) to obtain water
solubility of the free ligand and its iron complex.] Cox200
exhibits the reverse features (water insolubility and octanol
solubility). Cox750 and its iron complex are partitioned
between the two phases. The narrow molecular mass distri-
butions of POE750 and POE2000 (polydispersity indices
<1.2 measured by gel permeation chromatography) imply
that this factor has only a small effect on partition coefficients.
logKFeLH value (3.95) is very close to that for the formation of
4
the O-trensox bis(salicylate) complex (4.21). For the FeL
species the two absorption bands at lmax ˆ 450 nm (e ˆ
4600mÀ1 cmÀ1) and lmax ˆ 590 nm (e ˆ 3900mÀ1 cmÀ1) are char-
acteristic of a tris(8-hydroxyquinolinate) coordination, as
observed with the FeL species of the ferric O-trensox
complex.[5]
However,
the
extinction
coefficient
(3900mÀ1 cmÀ1) at 590 nm is lower than that measured for
O-trensox (5200mÀ1 cmÀ1). The UV/Vis spectrum of the
‡
FeLH species (lmax ˆ 440 nm, e ˆ 4600 mÀ1 cmÀ1 and lmax
ˆ
560 nm, e ˆ 2900 mÀ1 cmÀ1) suggests a monosalicylate, bisoxi-
nate coordination.
Chem. Eur. J. 2002, 8, No. 5
¹ WILEY-VCH Verlag GmbH, 69451 Weinheim, Germany, 2002
0947-6539/02/0805-1095 $ 17.50+.50/0
1095

J.-L. Pierre et al.
FULL PAPER
Furthermore, the results do not vary significantly over time
from 18 to 96 h (Table 3, entries 1, 3, and 5).
The linear relationship in Equation (16), where n is the
stoichiometry of the complex and k is a constant representing
the change in hydrophobicity when the donor groups of the
ligand are moved into the iron coordination sphere of the
complex, is often used to correlate the partition coefficient of
the free ligand with that of the iron complex[22]
.
Figure 6. Percentage of iron transported per mol ligand versus time.
logPcomplex ˆ nlogPfreelig ‡ k
(16)
Water octanol water system: phase A (3 Â 10À4 m):
Cox750 Fe,
*
^
Cox2000 Fe. Water chloroform water system: phase A (3 Â 10À4 m):
*
Cox750 Fe ,
Cox2000 Fe; phase B: [Ferrozine¾] ˆ 10À3 m, [ascorbic
^
The quite similar values of Pcomplex and Pfreelig indicate that
only the POE chain, located outside the metal coordination
sphere, controls the hydrophilic lipophilic balance. O-Tren-
sox and its iron complex were soluble only in water; the three
ionised sulfonate groups prevented solubility in octanol.
The discrepancy between the results from the two biphasic
systems is most surprising: the three ligands and the three iron
complexes are located entirely in the chloroform phase,
despite the water solubility of Cox2000 and of its complex.
This result, explained by the great affinity of the POE chains
for chloroform, highlights the common misuse of partition
coefficients to predict the ability of a compound to cross
biological membranes: in the literature, conclusions are
reached indiscriminately from results on water octanol or
water chloroform systems! The linear relationship used by
Hansch[3] (logPchloroform ˆ 1.12logPoctanol À 1.34) is not of gen-
eral relevance. P varies dramatically with the organic phase:
for example, the percentages of Fe Cox750 in the organic
phase are 4, 5, 13, 47 and 100 for ethyl acetate, cyclohexane,
toluene, octanol and chloroform, respectively.
acid] ˆ 10À1m.
exogenous siderophores. Another shuttle mechanism oper-
ates in the case of Mycobacterium tuberculosis, which
produces two siderophores involving the same iron coordina-
tion sphere: water-soluble exochelin excreted in the external
medium and lipid-soluble mycobactin. Iron is exchanged
between the two siderophores (which differ only in their side
chains, hydrophilic for exochelin and lipophilic for mycobac-
tin) at the membrane surface.[24]
To mimic a shuttle system, the triphasic system described
above has been used, involving an octanol or a chloroform
phase containing Cox200 or Cox750 (10À6 m) in contact with
an aqueous phase A (buffered at pH 7.4) containing Fe edta
(10À3 m) and an aqueous phase B (pH 2.2) containing Ferro-
zine¾ (10À4 m) and ascorbic acid (10À2 m). Formation of the
FeII ferrozine complex was monitored as indicated above.
When the experiments were carried out with octanol, iron
transfer from edta to Cox200 or to Cox750 was slow
(Figure 7): 35% (after 60 h) and 30% (after 120 h) of iron
Water octanol water or water chloroform water triphasic
systems: To mimic a transmembrane transport system, we
have already reported[7] a triphasic system involving an
octanol phase in contact with two separate aqueous phases
A and B. The FeIII complex of the ligand (Cox750 or Cox2000)
lies in the aqueous phase A, buffered at pH 7.4 (standing in
for the external medium). In the second aqueous phase B
(mimicking the cytosol), buffered at pH 2.2, ascorbic acid and
Ferrozine¾ (a strong FeII-complexing agent) were added,
respectively, to reduce FeIII to FeII and to quench FeII. At
pH 2.2, all the FeIII is complexed. The metal migrates
from the source phase A to the receiving phase B across
the octanolic phase and the system is not leaky. Formation
of the FeII ferrozine complex is studied by monitoring
its LMCT transition at 562 nm. The same experiments were
also performed with chloroform instead of octanol (Fig-
ure 6): all the iron is transported across the organic phase,
and faster transport is observed in chloroform than in
octanol.
Figure 7. Percentage of iron transported per mol ligand versus time. Phase
A ( 1À03 m): Fe edta. Organic phase (10À6 m):
Cox750 in octanol,
*
^
*
^
Cox200 Fe in octanol,
Cox750 Fe in chloroform,
Cox200 Fe in
chloroform. Phase B: [Ferrozine¾] ˆ 10À4 m, [ascorbic acid] ˆ 10À2 m.
was transferred to Cox200 and Cox750, respectively. Since
he initial quantity of Ferrozine¾ was calculated in order to
quench only 50% of the iron, an excess of this compound
was added after 60h: iron transfer recommenced and, after
120 h, 60% and 40% of the metal had been transferred
to Cox200 and Cox750, respectively. Similar results were
obtained when chloroform was substituted for octanol
(Figure 7): after 66h 45% and 35% of iron was transferred
to Cox200 and Cox750, respectively and supplementary
addition of Ferrozine¾ led to transfer of 80% and 50% of
iron.
Use of a liquid membrane: mimicking a shuttle process: Some
microorganisms involve ligand exchange for iron acquisition
(a ™shuttle∫ mechanism). In Aeromonas hydrophila, for
example, the exchange of the iron ligand from a ferric
siderophore to an iron-free siderophore bound to the receptor
occurs at the cell surface.[23] This iron acquisition mechanism
provides the bacterium with the ability to steal iron from
1096
¹ WILEY-VCH Verlag GmbH, 69451 Weinheim, Germany, 2002
0947-6539/02/0805-1096 $ 17.50+.50/0
Chem. Eur. J. 2002, 8, No. 5

Iron Chelators
1091 1100
These results feature transfer of iron across a liquid
membrane through a shuttle mechanism, that is, the exchange
of iron between two ligands located in the external aqueous
phase and in the lipophilic phase, respectively. This system is
an oversimplification, since the transport from the organic
phase to the second aqueous phase is driven only by reduction
of iron. The ligands Cox200 and Cox750 exhibited similar
behaviour, but Cox200 had a higher efficiency in accord with
the greater lipophilicity of its iron complex.
Iron mobilisation from iron-overloaded hepatocytes: Despite
its lack of lipophilicity, O-trensox has been found to be able to
mobilise iron from hepatocytes.[8] In a study of the mobilisa-
tion of iron by O-trensox, Cox2000 and Cox750 (chelator
concentration 50 mm), from rat hepatocyte cultures which had
been loaded in vitro with 55Fe dextran,[25] similar results were
observed with both Cox2000 and Cox750, which had mobi-
lised approximately 34% iron after 10 days (28% after two
days, 30% after six days). The hydrophilic lipophilic balance
of the free chelators does not seem to be a criterion of their
ability to mobilise intracellular iron.
Preliminary biological studies: Detailed biological studies will
be published elsewhere. The summary here is to allow a
discussion on the physicochemical data developed in this paper.
Conclusion
Plant cell nutrition: FeIII O-trensox was found to be able to
prevent and to reverse iron chlorosis in several plant species
grown in axenic conditions.[9] The use of radioactive 59Fe
demonstrated the incorporation of iron into ferritin, the
intracellular iron storage protein. Similar studies have been
started with the ferric complexes of Cox750 and Cox2000. The
first results have been obtained with axenic cell cultures of
Arabidopsis thaliana. Ferric complexes of O-trensox, Cox750
and Cox2000 were tested as the single source of iron in
nutritional experiments, in which the growth and greening
(iron incorporation into the chloroplasts, determined by
densitometric measurements) of A. thaliana plant cells were
compared during cell suspension cultures in axenic media. In
one control experiment iron was provided with FeIII edta,
and in another no iron source was added. The ferric
complexes of edta, O-trensox, Cox750 and Cox2000 led to
similar kinetic results for the growth of the cells (Figures 8 and
9). These results reveal that the hydrophilic lipophilic
balance is not a deciding factor for the ability of the ferric
complex to allow incorporation of iron into cells.
Siderophores are low molecular mass (700 1000 Da, and
therefore too large to diffuse through the porin channels),
FeIII-chelating compounds of microbial origin that are ex-
creted into the environment, where they scavenge iron.
Whereas the free siderophores fail to penetrate biological
membranes, the ferric complexes are taken up by microbial
cells by means of specific outer-membrane receptors. They
traverse the membranes in energy-dependent processes and
release bound iron to the cytoplasm. The release may involve
reduction, ligand exchange or decomposition of the side-
rophore to its constituents. Biomimetic siderophores are
usually envisaged as conforming with the following models[26]
:
1) iron-free lipophilic carriers inhibit growth by effectively
penetrating cellular membranes by diffusion, scavenging iron
from intracellular stores and quitting the cells with their iron
loads; 2) iron-loaded hydrophilic carriers act as growth
promoters if they can introduce iron by means of iron-uptake
systems.
The data reported here provide the first example of a series
of iron chelators with the same pFeIII and a modulated
hydrophilic lipophilic balance. We have shown that totally
hydrophilic iron chelators were able to scavenge iron stored in
hepatocytes with the same efficiency as iron chelators
possessing the same complexing abilities but soluble in both
water and organic phases. We have also shown that the ability
to introduce iron into plant cells is not related to the
hydrophilic lipophilic balance of the iron complex. These
results are important with regard to the criteria often claimed
for biological activity of drugs, particularly iron chelators
designed for chelation therapy.[22] Our results show that this
cannot be a general rule and question whether the use of
partition coefficients is pertinent to the prediction of the
activity of iron chelators.
Figure 8. Greening of A. thaliana cells at seven days versus the concen-
tration of the iron source: 258C, 220 rpm, 18 h of light per day.
Experimental Section
Materials and methods: Solvents were purified by the usual techniques. All
other compounds were of reagent grade and were used without further
purification. Iron(iii) stock solutions were prepared by dissolving appro-
priate amounts of ferric perchlorate hydrate (Aldrich) in standardised
solutions of HClO4 or NaClO4. The solutions were standardised spectro-
photometrically for ferric ion (e ˆ 4160mÀ1 cmÀ1 at 240 nm).[27] IR spectra
were collected on a Nicolet Impact 400 spectrometer. UV/Vis absorption
spectra were recorded on a Perkin Elmer Lambda 2 spectrometer with
Figure 9. Greening of A. thaliana cells versus time: [ligand FeIII] ˆ 50mm,
258C, 220 rpm, 18 h of light per day.
Chem. Eur. J. 2002, 8, No. 5
¹ WILEY-VCH Verlag GmbH, 69451 Weinheim, Germany, 2002
0947-6539/02/0805-1097 $ 17.50+.50/0
1097

J.-L. Pierre et al.
FULL PAPER
1.000 cm pathlength quartz cells, connected to an IBM PC 340 micro-
computer. Mass spectra were recorded on a NERMAGR101C mass
spectrometer. Microanalyses were performed by the Central Service of
CNRS, Solaize (France). Melting points were determined with a B¸chi
apparatus and are not corrected. 1H and 13C NMR spectra were obtained in
5 mm tubes at 258C with a Bruker AM 300 or AM 400 spectrometer.
6H; CH2), 3.31 (s, 2H; CH2), 3.39 (s, 3H; CH3), 3.2 3.9 (m, 170H; CH2);
13C NMR (75 MHz, CDCl3): d ˆ 12.8 (CH2), 26.2 (CH2), 40.3 (Cq),
59.4(CH3), 68.4 (CH2), 69.7 70.6 (CH2), 120.0 (CN).
Compound 4a: NaOH (506 mg, 12.64 mmol) and hydrazine (99%, 3 mL)
were added to a solution of 3a (1 g, 2.53 mmol) in 95% ethanol (20 mL) at
08C. Raney nickel (445 mg, 7.6 mmol) was added to this mixture in small
portions over 2 h; the solution was stirred for an additional 6 h at room
temperature. The reaction mixture was heated at reflux for 1 h, filtered hot
and evaporated to give a colourless oil. Addition and evaporation of
toluene were repeated until NaOH was precipitated. Evaporation of the
filtrate gave 4a as a colourless oil, which was used without further
purification (938 mg, 2.3 mmol). 1H NMR (300 MHz, [D6]DMSO): d ˆ
1.10 1.19 (m, 12H; CH2), 2.44 (m, 6H; CH2), 3.08 (s, 2H; CH2), 3.23 (s,
3H; CH3), 3.39 3.53 (m, 16H; CH2); 13C NMR (75 MHz, [D6]DMSO):
d ˆ 27.1 (CH2), 31.6 (CH2), 38.1 (Cq), 42.8 (CH2), 58.1 (CH3), 69.6 71.3
(CH2), 71.3 (CH2), 75.1 (CH2).
Ligand syntheses
Compound 1a: A solution of tetraethylene glycol monomethyl ether (5 g,
24.03 mmol) in THF (7 mL) was added to a solution of NaOH (1.35 g,
33.65 mmol) in water (7 mL). The resulting mixture was cooled to 08C and
a solution of tosyl chloride (4.58 g, 24.03 mmol) in THF (30 mL) was added
over 2 h. The solution was stirred for an additional 2 h at 08C, then
evaporated. The residue was dissolved in water/dichloromethane (50/
100 mL); the organic layer was washed with brine and dried under MgSO4,
and the solvent was removed. The yellow residue was purified by flash
chromatography (silica gel; CH2Cl2). Compound 1a was obtained as a
white oil (8.1 g, 93%) pure enough for the following step. IR (film from
CHCl3): nÄ ˆ 1216, 1102 cmÀ1; 1H NMR (300 MHz, CDCl3): d ˆ 2.45 (s, 3H;
CH3), 3.37 (s, 3H; CH3), 3.52 3.70 (m, 12H; CH2), 3.68 (t, J ˆ 4.9 Hz, 2H;
CH2), 4.15 (t, J ˆ 4.9 Hz, 2H; CH2), 7.36 (d, J ˆ 8.3 Hz, 2H; ArH), 7.79 (d,
J ˆ 8.3 Hz, 2H; ArH); 13C NMR (75 MHz, CDCl3): d ˆ 21.5 (CH), 58.8
(CH3), 68.5 (CH2), 69.1 (CH2), 70.35 71.8 (CH2), 127.8 (CH), 129.7 (CH),
132.9 (Cq), 144.7 (Cq).
Compound 4b: Prepared by the synthetic procedure described for 4a, by
using NaOH (853 mg, 21.35 mmol), 3b (1 g, 1.07 mmol) in 95% ethanol
(40 mL), hydrazine 99% (5 mL) and Raney nickel (188 mg, 3.2 mmol).
Compound 4b was obtained as a colourless oil (952 mg, 1 mmol). 1H NMR
(300 MHz, [D6]DMSO): d ˆ 1.07 1.23 (m, 12H; CH2), 2.45 (m, 6H; CH2),
3.09 (s, 2H; CH2), 3.29 (s, 3H; CH3), 3.41 3.52 (m, 61H; CH2); 13C NMR
(75 MHz, [D6]DMSO): d ˆ 27.1 (CH2), 31.5 (CH2), 38.1 (Cq), 42.8 (CH2),
58.0 (CH3), 60.2 (CH2), 69.6 71.3 (CH2), 71.6 (CH2), 75.1 (CH2).
Compound 1b: Prepared by the synthetic procedure described for 1a, with
poly(ethylene glycol) methyl ether (Acros; average M ˆ 750, 20 g,
36.7 mmol) in THF (40 mL), NaOH (1.5 g, 37.3 mmol) in water (8 mL),
and tosyl chloride (5.08 g, 26.7 mmol) in THF (60 mL). Compound 1b was
Compound 4c: Prepared by the synthetic procedure described for 4a, by
using NaOH (1.83 g, 45.72 mmol), 3c (2 g, 0.91 mmol) in 95% ethanol
(80 mL), hydrazine (99%, 10 mL), and Raney nickel (160 mg, 2.7 mmol).
Compound 4b was obtained as a colourless oil (1,91 g, 95%). 1H NMR
(300 MHz, [D6]DMSO): d ˆ 1.06 1.23 (m, 12H; CH2), 2.42 (brt, 6H;
CH2), 3.08 (s, 2H; CH2), 3.22 (s, 3H; CH3), 3.34 3.59 (m, 174H; CH2);
13C NMR (75 MHz, [D6]DMSO): d ˆ 27.1 (CH2), 31.7 (CH2), 38.1 (Cq), 42.7
(CH2), 58.0 (CH3), 60.9 (CH2), 69.4 71.2 (CH2), 75.1 (CH2).
1
obtained as a white solid (17.6 g, 73%). H NMR (300 MHz, CDCl3): d ˆ
2.45 (s, 3H; CH3), 3.38 (s, 3H; CH3), 3.58 3.71 (m, 56H; CH2), 4.16 (t, J ˆ
4.8, 2H; CH2), 7.34 (t, J ˆ 8.2 Hz, 2H; ArH), 7.79 (t, J ˆ 8.2 Hz, 2H; ArH);
13C NMR (75 MHz, CDCl3): d ˆ 21.5 (CH3), 58.9 (CH3), 67.3 (CH2), 68.6
(CH2), 69.1 70.6 (CH2), 71.8 (CH2), 72.4 (CH2), 127.9 (CH), 129.7 (CH),
132.9 (Cq), 144.7 (Cq).
Compound 6a (Cox200): Compound 5 (697 mg, 3.68 mmol) was dissolved
in freshly distilled THF (100 mL) under nitrogen and heated at reflux. A
solution of carbonyldiimidazole (656 mg, 4.05 mmol) in dry THF (50 mL)
was added over 30 min. Then a solution of 4a (0.5 g, 1.23 mmol) in dry THF
(50 mL) was added over 30 min. The mixture was stirred overnight and
evaporated. The resulting orange residue was dissolved in chloroform
(150 mL), washed with brine, dried with Na2SO4 and then evaporated to
dryness. The residue was applied to a Sephadex LH-20 column (50%
MeOH CH2Cl2). An orange foam (6a) eluted as a single band (1.42 g,
Compound 1c: Prepared by the synthetic procedure described for 1a, with
poly(ethylene glycol) methyl ether (Aldrich; average M ˆ 2000, 30 g,
15 mmol) in THF (60 mL), NaOH (840 mg, 21 mmol) in water (10 mL),
and tosyl chloride (2.86 g, 15 mmol) in THF (40 mL). Compound 1c was
obtained as a white solid (21.97 g, 68%). 1H NMR (300 MHz, CDCl3): d ˆ
2.44 (s, 3H; CH3), 3.37 (s, 3H; CH3), 3.4 3.8 (m, 176H; CH2), 4.16 (brt,
2H; CH2), 7.39 (t, J ˆ 8.3 Hz, 2H; ArH), 7.80 (t, J ˆ 8.3 Hz, 2H; ArH);
13C NMR (75 MHz, CDCl3): d ˆ 21.9 (CH3), 59.4 (CH3), 69.0 (CH2), 69.6
72.3 (CH2), 72.9 (CH2), 128.0 (CH), 130.0 (CH), 132.7 (Cq), 144.8 (Cq).
40%). IR (film from CHCl3): nÄ ˆ 3460 3200, 1648 cmÀ1
;
1H NMR
Compound 3a: Sodium hydride (429 mg, 60% in paraffin) was washed with
dry pentane and DMSO (5 mL) was added. A solution of 1a in DMSO
(10 mL) was added quickly and the mixture was stirred for 1 h at room
(300 MHz, CDCl3): d ˆ 1.35 (m, 6H; CH2), 1.58 (m, 6H; CH2), 3.13 (s,
2H; CH2O), 3.30 (s, 3H; CH3O), 3.35 3.66 (m, 22H; CH2), 7.22 (d, J ˆ
8.7 Hz, 3H; ArH), 7.39 (dd, J ˆ 7.9 and 4.3 Hz, 3H; ArH), 8.03 (d, J ˆ
7.9 Hz, 3H; ArH), 8.10 (d, J ˆ 8.7 Hz, 3H; ArH), 8.14 (brs, 3H; NH), 8.78
(d, J ˆ 4.3Hz, 3H; ArH), 9.67 (brs, 3H; OH); 13C NMR (70 MHz, CDCl3):
d ˆ 22.9 (CH2), 31.4 (CH2), 38.5 (Cq), 40.3 (CH2), 58.7 (CH3O), 70.3 71.7
(CH2), 75.03 (CH2), 113.1 (Cq), 117.1 (CH), 122.9 (CH), 126.1 (CH), 130.13
temperature under
a nitrogen atmosphere. A solution of 2 (3.53 g,
9.74 mmol) in DMSO (10 mL) was added over 15 min and the resulting
mixture was heated to 708C for 18 h. The solution was evaporated and the
resulting yellow oil was dissolved in chloroform, washed with brine, dried
with MgSO4 and then evaporated to dryness. The white residue was
purified by chromatography (silica gel; 2% MeOH in CH2Cl2). Compound
3a was obtained as a colourless oil (2.77 g, 72%) and used without further
ˆ
(Cq), 135.8 (CH), 138.8 (Cq), 148.4 (CH), 154.1 (COH), 167.1 (C O); MS
(FAB, NBA matrix): m/z: 921 [M‡1] ; elemental analysis calcd (%) for
‡
C50H60N6O11 ¥ 0.75MeOH ¥ 0.25CH2Cl2: C 61.74, H 6.68, N 8.31; found: C
61.79, H 6.58, N 8.62.
purification. IR (film from CHCl3): nÄ ˆ 2248 cmÀ1
;
1H NMR (300 MHz,
CDCl3): d ˆ 1.74 (t, J ˆ 7.9 Hz, 6H; CH2), 2.42 (t, J ˆ 7.9 Hz, 6H; CH2), 3.30
(s, 2 H; CH2), 3.38 (s, 3H; CH3), 3.51 3.70 (m, 16H; CH2); 13C NMR
(75 MHz, CDCl3): d ˆ 12.3 (CH2), 26.3 (CH2), 41.0 (Cq), 59.4 (CH3), 68.5
(CH2), 69.1 (CH2), 70.3 70.6 (CH2), 71.5 (CH2), 120.0 (CN).
Compounds 6b (Cox750): Compound 5 (600 mg, 3.16 mmol) was dissolved
in freshly distilled THF (200 mL) under a nitrogen atmosphere and
refluxed. Carbonyldiimidazole (564 mg, 3.48 mmol) in dry THF (20 mL)
was added over 20 min, followed by a solution of 4b (0.5 g, 0.55 mmol) in
dry dichloromethane (200 mL) over 1 h. The mixture was refluxed with
stirring for an additional 48 h and then evaporated. The resulting residue
was applied to a Sephadex LH-20 column (50% MeOH/CH2Cl2). The first
fractions were evaporated and the orange-green residue was applied to a
Compound 3b: Prepared by the synthetic procedure described for 3a, by
using NaH (107 mg, 2.68 mmol), 1b (2 g, 2.2 mmol), 2 (500 mg, 2.43 mmol).
Column chromatography (silica gel; 10% acetone in acetonitrile) afforded
1
3b as a white oil (620 mg, 30%). H NMR (300 MHz, CDCl3): d ˆ 1.74 (t,
J ˆ 8.1 Hz, 6H; CH2), 2.42 (t, J ˆ 8.1 Hz, 6H; CH2), 3.32 (s, 2H; CH2), 3.38
(s, 3 H; CH3), 3.46 3.82 (m, 68H; CH2); 13C NMR (75 MHz, CDCl3): d ˆ
12.9 (CH2), 26.6 (CH2), 40.6 (Cq), 59.3 (CH3), 68.6 (CH2), 69.2 (CH2), 70.1
70.9 (CH2), 120.0 (CN).
second column. Compound 6b eluted as
a single band (246 mg,
32%).1H NMR (300 MHz, CDCl3): d ˆ 1.32 (m, 6H; CH2), 1.52 (m, 6H;
CH2), 3.01 (m, 6H; CH2), 3.12 (s, 2H; CH2O), 3.29 (s, 3H; CH3O), 3.37
3.60 (m, 68H; CH2), 7.18 (d, J ˆ 8.8 Hz, 3H; ArH), 7.37 (m, 3H; ArH), 8.04
(m, 9H; ArH ‡ NH), 8.73 (d, J ˆ 4.1 Hz, 3H; ArH); 13C NMR (75 MHz,
CDCl3): d ˆ 23.0(CH2), 31.5 (CH2), 38.6 (Cq), 40.3 (CH2), 58.8 (CH3O),
69.8 71.8 (CH2), 75.2 (CH2), 113.4 (Cq), 116.4 (CH), 122.9 (CH), 126.5
(CH), 130.1 (Cq), 135.8 (CH), 138.9 (Cq), 148.4 (CH), 154.5 (COH), 166.8
Compound 3c: Prepared by the synthetic procedure described for 3a, by
using NaH (107 mg, 2.68 mmol), 1c (4.72 g, 2.2 mmol) in DMSO (50 mL),
and 2 (500 mg, 2.43 mmol). Column chromatography (silica gel; 10%
acetone in acetonitrile) afforded 3c as a white oil (1 g, 21%). 1H NMR
(300 MHz, CDCl3): d ˆ 1.68 (t, J ˆ 8.1 Hz, 6H; CH2), 2.26 (t, J ˆ 8.1 Hz,
ˆ
(C O).
1098
¹ WILEY-VCH Verlag GmbH, 69451 Weinheim, Germany, 2002
0947-6539/02/0805-1098 $ 17.50+.50/0
Chem. Eur. J. 2002, 8, No. 5

Iron Chelators
1091 1100
Compound 6c (Cox2000): Prepared according to the synthetic procedure
described for 6b, by using 5 (260 mg, 1.36 mmol) in THF (100 mL),
carbonyldiimidazole (243 mg, 1.5 mmol) in THF (20 mL), and 4c (500 mg,
0.23 mmol) in dichloromethane (200 mL). Compound 6c was obtained as a
brown solid (249 mg, 40%). 1H NMR (300 MHz, CDCl3): d ˆ 1.32 (m, 6H;
CH2), 1.57 (m, 6H; CH2), 3.17 (s, 2H; CH2O), 3.28 (s, 3H; CH3O), 3.4 3.7
(m, 185H; CH2), 7.25 (d, J ˆ 8.8 Hz, 3H; ArH), 7.4 (dd, J ˆ 8.3 and 4.2 Hz,
3H; ArH), 7.82 (brt, 3H; NH), 8.01 (d, J ˆ 8.8 Hz, 3H; ArH), 8.07 (dd, J ˆ
8.3 and 1.3 Hz, 3H; ArH), 8.72 (d, J ˆ 4.2 and 1.3 Hz, 3H; ArH); 13C NMR
(75 MHz, CDCl3): d ˆ 23.4 (CH2), 32.4 (CH2), 39.0 (Cq), 41.2 (CH2), 59.4
(CH3O), 69.6 72.0 (CH2O), 75.4 (CH2), 113.2 (Cq), 117.3 (CH), 123.9
(CH), 125.9 (CH), 131.1 (Cq), 136.3 (CH), 139.8 (Cq), 149.2 (CH), 156.5
Cox2000 and Na2H2edta were prepared. The samples were allowed to
equilibrate for seven days at 258C.
Partition coefficients and triphasic systems: The partition coefficients
between an aqueous phase buffered at pH 7.4 (Tris buffer) and octanol or
chloroform were determined by a known procedure.[30] The triphasic
water octanol water system has been described previously.[7] Phase A
(2.5 mL) buffered at pH 7.4 (Tris buffer) contained a solution (0.3mm) of
the FeIII complex of Cox2000 or Cox750. Phase B (8 mL, pH 2.2) contained
Ferrozine¾ (1mm) and ascorbic acid (0.1m). Phases A and B, which were
separated by the organic phase (6 mL) of octanol or chloroform, were
stirred at a constant rate at 258C. The time dependence of the UV/Vis
spectrum of phase B (formation of the FeII ferrozine complex) was
followed by monitoring the LMCT transition at 562 nm. For the shuttle
process experiments the triphasic water octanol water system was:
ˆ
(COH), 168.0 (C O).
Fe Cox200: KOH (1m, 170 mL in ethanol) was added to a solution of
Cox200 (50 mg, 0.054 mmol) in methanol (10 mL) and dichloromethane
(10 mL). When [Fe(acac)3] (0.054 mmol) was added, the solution instantly
turned green. It was stirred for 30 min, then evaporated. The residue was
applied to a Sephadex LH-20 column (50% MeOH CH2Cl2). A green
solid (after evaporation) eluted as a single band (49 mg, 93%). MS (FAB,
phase A: 2.5 mL, 10À3
m
FeIII edta, pH 7.4; phase B: 11 mL, 10À4
m
m
Ferrozine¾, 10À2 m ascorbic acid, pH 2.2; octanolic phase, 5.5 mL, 10À6
Cox200 or Cox750. After 60 h, 1 mL of a Ferrozine¾ solution (10À1m) was
added. The triphasic water chloroform water system was: phase A,
2 mL, 10À3 m FeIII edta, pH 7.4; phase B, 2 mL of a Ferrozine¾ solution
(10À4 m) and ascorbic acid(10À2 m), pH 2.2; chloroform phase, 4 mL, 10À6
m
‡
NBA matrix): m/z: 974 [M‡1] .
Cox200 or Cox750. After 60 h, 250 mL of a Ferrozine¾ solution (10À1m) was
added. Phases A and B were stirred at a constant rate at 258C. The time
dependence of UV/Vis spectrum of phase B (formation of the FeII fer-
rozine complex) was followed by monitoring the MLCT transition at
562 nm.
X-ray data collection and crystal structure determination: Crystals of X-ray
quality were obtained by vapour diffusion of diethyl oxide into a solution of
the complex in methanol. The crystal of complex Fe Cox200 was mounted
on an Enraf-Nonius CAD4 diffractometer using a graphite crystal mono-
chromator (l(MoKa) ˆ 0.71073 ä) at 293 K. The reflections were corrected
for Lorentz and polarisation effects but not for absorption. The structure
was solved by direct methods and refined using TEXSAN software.[28]
[FeC50H57N6O11] ¥ H2O: M ˆ 991.90, dark-green platelet (0.1  0.2 Â
Plant cell nutrition: Cell culture and growth measurements were carried
out as previously described.[11] The axenic medium (MS) was prepared
according to reference [31]. An iron-free control was cultivated in the iron-
free nutrient solution MS. The assays were performed in MS supplemented
with FeIII Cox750, FeIII Cox2000, FeIII trensox and FeIII edta, respec-
tively, from 5 to 50mm, with six independent replications. Densitometric
measurements were made daily as previously described.[31] The Biorad
Geldoc 1000 video camera system, with ™molecular analyst∫ software, was
used for quantification. Full-frame images were captured in constant light
conditions (that is, white light was adjusted so that just a few saturated
≈
0.3 mm), triclinic, space group P1, a ˆ 10.611(4), b ˆ 13.129(3), c ˆ
18.109(2) ä, a ˆ 93.22(1)8, b ˆ 101.77(2)8, g ˆ 97.51(2)8, Vˆ 2440(1) ä3,
Z ˆ 2, 1 ˆ 1.350 gcmÀ3, m ˆ 1.081 mmÀ1, 14659 reflections collected in the
range 48 ꢀ 2q ꢀ 608. All non-hydrogen atoms were refined with anisotropic
thermal parameters. Hydrogen atoms were generated in idealised positions,
riding on the carrier atoms, with isotropic thermal parameters, except that
the hydrogen atoms of C18 24 (polyoxyethylene chain) were not
generated because thermal parameters are high. Final cycle refinement,
including 622 parameters, converged to R(F) ˆ 0.077 and Rw(F) ˆ 0.094 for
6135 F > 2s(F), S ˆ 1.87, D1max ˆ 0.99 eäÀ3, D1min ˆ À0.46 eäÀ3. The two
highest peaks on the difference Fourier map were localised near C22, which
is in the tail of the POE chain. Crystallographic data (excluding structure
factors) for the structure reported in this paper have been deposited with
the Cambridge Crystallographic Data Centre as supplementary publication
no. CCDC-170954. Copies of the data can be obtained free of charge on
application to CCDC, 12 Union Road, Cambridge CB21EZ, UK (fax:
(‡44)1223-336-033; e-mail: deposit@ccdc.cam.ac.uk).
pixels appeared in red in the empty optical field).
A reproducible
relationship was found between the absorbance of the wells in densito-
metric scanning profiles of recorded images and the number of cells in the
wells.
Acknowledgements
The authors thank the members of GDR CNRS1879: R. R. Crichton, C.
¬
¡
Henry (Universite de Louvain, Belgium), J.-P. Laulhere (CEA Grenoble),
Potentiometric measurements: All the measurements were made at 258C
in deionised and twice-distilled water. The ionic strength was kept constant
at 0.1m (sodium perchlorate, PROLABO, Puriss). The potentiometric
titrations were performed by using a DMS Titrino (Metrohm) automatic
titrator system equipped with a combined glass electrode (Metrohm filled
with saturated NaCl solution) and connected to an IBM Aptiva micro-
computer. The electrodes were calibrated to read the pH according to the
classical method[29] (from titration of 0.01m HClO4 with 0.01m NaOH). The
ligand and its iron(iii) complex (1mm) were titrated with standardised
0.02m NaOH. Argon was bubbled through the solutions to exclude CO2
and O2. Carbonate content was checked by Gran×s method. The titration
data were refined by the non-linear least-squares fitting program SUPER-
QUAD[16] to determine the ligand deprotonation constants.
¬
and G. Lescoat and N. Rakba (Universite de Rennes, France) for the
biological studies mentioned in this paper and for fruitful discussions.
[1] ™Iron Transport and Storage in Microorganisms, Plants and Animals∫:
Met. Ions Biol. Syst. 1998, 35, whole volume.
[2] K. N. Raymond, G. M¸ller, B. F. Matzanke, Top. Curr. Chem. 1984,
123, 49 102.
[3] C. Hansch, Acc. Chem. Res. 1969, 2, 232 239.
[4] P. Baret, C. Beguin, H. Boukhalfa, C. Caris, J.-P. Laulhere, J.-L. Pierre,
¬
¡
G. Serratrice, J. Am. Chem. Soc. 1995, 117, 9760 9761.
[5] G. Serratrice, H. Boukhalfa, C. Beguin, P. Baret, C. Caris, J.-L. Pierre,
¬
Spectrophotometric measurements: The temperature was maintained at
258C with a Perkin Elmer PTP-1 variable-temperature unit and the
acquisition was made with UV Winlab software (Perkin Elmer). The ionic
strength was kept constant at 0.1m (sodium perchlorate). pH was measured
with a 713 Metrohm digital pH meter equipped with a microelectrode. The
titration of a solution containing Cox2000 and ferric ion was carried out as
previously described[5] and the absorbance data were analysed using the
non-linear least-squares fitting program LETAGROP-SPEFO.[18, 19] The
calculations were performed by using absorbance values at 10 24 wave-
lengths and the residual square sums of the fit were 10À3 10À4. The
spectrophotometric competition experiments were carried out with Na2-
H2edta over the pH range 2 3. Equimolar (10À4 m) solutions in ferric ions,
Inorg Chem. 1997, 36, 3898 3910.
[6] G. Serratrice, F. Biaso, P. Baret, J.-L. Pierre, J. Inorg. Biochem., in
press.
[7] F. Thomas, P. Baret, D. Imbert, J.-L. Pierre, G. Serratrice, Bioorg. Med.
Chem. Lett. 1999, 9, 3035 3040.
[8] N. Rakba, F. Aouad, C. Henry, C. Caris, I. Morel, P. Baret, J.-L. Pierre,
P. Brissot, R. J. Ward, G. Lescoat, R. R. Crichton, Biochem. Pharma-
col. 1998, 55, 1797 1806.
¬
¡
[9] C. Caris, P. Baret, C. Beguin, G. Serratrice, J.-L. Pierre, J.-P. Laulhere,
Biochem. J. 1995, 312, 879 885.
[10] R. J. Bergeron, J. Wiegand, J. S. Mc Manis, T. Thirumalai Perumal, J.
Med. Chem. 1991, 34, 3182 3187.
Chem. Eur. J. 2002, 8, No. 5
¹ WILEY-VCH Verlag GmbH, 69451 Weinheim, Germany, 2002
0947-6539/02/0805-1099 $ 17.50+.50/0
1099

J.-L. Pierre et al.
FULL PAPER
[11] D. Imbert, F. Thomas, P. Baret, G. Serratrice, D. Gaude, J.-L. Pierre,
[22] J. T. Edwards, Biometals 1998, 11, 203 205.
¡
J.-P. Laulhere, New J. Chem. 2000, 24, 281 288.
[23] A. Stinzi, C. Barnes, J. Xu, K. N. Raymond, Proc. Natl. Acad. Sci. USA
2000, 97, 10691 10696.
[12] P. Job, Ann. Chim. 1927, 9, 113 203.
[13] L. Pech, Y. A. Bankovsky, A. Kemme, J. Lejeijs, Acta Crystallogr.
Sect. C 1997, 53, 1943 1945.
[24] J. M. Roosenberg, Y. M. Lin, Y. Lu, M. Miller, Curr. Med. Chem. 2000,
7, 159 197.
[14] G. Serratrice, P. Baret, H. Boukhalfa, I. Gautier-Luneau, D. Luneau,
J.-L. Pierre, Inorg Chem. 1999, 38, 840 841.
[15] M. S. Cohen, M. Meyer, K. N. Raymond, J. Am. Chem. Soc. 1998, 120,
6277 6286.
[16] T. D. P. Stack, T. B. Karpishin, K. N. Raymond, J. Am. Chem. Soc.
1992, 114, 1512 1514.
[25] Unpublished results.
[26] A. Shanzer, J. Libman, Met. Ions Biol. Syst. 1998, 35, 329 354
[27] R. Bastian, R. Weberling, F. Palilla, Anal. Chem. 1956, 28, 459 465.
[28] TEXSAN: Single-Crystal Structure Analysis Software, V.1.7, Molec-
ular Structure Corporation, 3200 Research Forest Drive, The Wood-
lands, TX 77381 (USA), 1995.
[17] P. Gans, A. Sabatini, A. Vacca, J. Chem. Soc. Dalton Trans. 1985,
1195 1200.
[29] A. E. Martell, R. J. Motekaitis, Determination and Use of Stability
Constants, 2nd ed., VCH, New York, 1992.
[18] L. G. Sillen, B. Warsquist, Ark. Kemi 1969, 31, 377 390.
[19] M. Meloun, J. Havel, E. Hogfeldt, Computation of Solution Equilibria,
Ellis Horwood, Chichester, 1988.
[30] J. E. Bollinger, J. Mague, C. J. O×Connor, W. A. Banks, D. M. Round-
hill, J. Chem. Soc. Dalton Trans. 1995, 1677 1688.
[31] T. Murashige, F. Skoog, Physiol. Plant 1962, 15, 473 479.
[20] A. E. Martell, R. M. Smith, Critical Stability Constants, Plenum, New
York, 1974.
[21] A. Ringbom, Complexation in Analytical Chemistry, Interscience,
New York, 1963.
Received: May 25, 2001
Revised: October 31, 2001 [F3288]
1100
¹ WILEY-VCH Verlag GmbH, 69451 Weinheim, Germany, 2002
0947-6539/02/0805-1100 $ 17.50+.50/0
Chem. Eur. J. 2002, 8, No. 5