Mechanism of the stepwise dissociation of FeIII complexes with tripodal ligands as siderophore models

Article abstract of DOI:10.1002/ejic.200300396

The acid hydrolysis and formation kinetics of 1:1 ferric complexes with several tripodal hexadentate ligands were investigated over the [H⁺] range of 0.02-1.0 M at 25 °C and at an ionic strength of 2.0 M (NaClO ₄/HClO₄). These ligands are based on a tris(2-aminoethyl)amine (TREN) spacer connected through a tris(amide) moiety to identical pendant bidentate chelating group arms such as catecholate (TREN-CAMS, L³), 8-hydroxyquinolinate (O-TRENSOX, L⁰) or pyridinophenolate (TRENPYPOLS, L⁵). Mixed ligands with chelating arms such as catecholate and 8-hydroxyquinolinate (TRENSOX2CAMS, L¹ and TRENSOXCAMS2, L²) were also investigated. An analogue of L ⁰ (C-SOX, L⁴) with a C-pivot scaffold instead of TREN was studied for comparison. The acid hydrolysis reaction was found to proceed in four kinetically distinguishable stages. A common mechanism in the series was followed when the ligand structure was varied. The first stage is very fast and can be attributed to the dissociation of one arm of the tripodal ligand. The second and third stages involve protonation of the complex which induces a change of coordination giving the corresponding tetracoordinated bis(salicylate) complex (coordination with carbonyl and o-hydroxy oxygen atoms). The final dissociation of the ligand proceeds in the fourth stage by one or two kinetically detectable steps. For the last step an important structural effect may be observed from 1.2 × 10^-5 to 2.2 × 10_-2 s^-1 for the proton-independent step and from 1.2 × 10 ^-6 to 0.18 M^-1S^-1 for the proton-dependent step. Differences in the rates and mechanism between the different tripodal ligands are discussed in terms of the global electrostatic charge of the ligand and the donor ability of the chelating subunit. The C-N bond rotation of the amide moiety and the importance, depending on the charge, of the H-bond network in the partially uncoordinated complexes have also been considered. The role of electrostatics in complex formation is discussed in relation to the Eigen-Wilkins mechanism. Wiley-VCH Verlag GmbH & Co. KGaA, 69451 Weinheim, Germany, 2004.

Full text of DOI:10.1002/ejic.200300396

FULL PAPER
Mechanism of the Stepwise Dissociation of FeIII Complexes with Tripodal
Ligands as Siderophore Models
[a]
[a]
[a]
[a]
´ ´
´
Guy Serratrice,* Frederic Biaso, Fabrice Thomas, and Claude Beguin
Keywords: Tripodal ligands / Iron / Kinetics and mechanisms / Dissociation / Catechol / 8-hydroxyquinoline
The acid hydrolysis and formation kinetics of 1:1 ferric com-
plexes with several tripodal hexadentate ligands were in-
duces a change of coordination giving the corresponding
tetracoordinated bis(salicylate) complex (coordination with
carbonyl and o-hydroxy oxygen atoms). The final dissoci-
ation of the ligand proceeds in the fourth stage by one or two
vestigated over the [H+] range of 0.02−1.0
M
at 25 °C and at
an ionic strength of 2.0
M (NaClO4/HClO4). These ligands
are based on a tris(2-aminoethyl)amine (TREN) spacer con- kinetically detectable steps. For the last step an important
nected through a tris(amide) moiety to identical pendant bi-
structural effect may be observed from 1.2 × 10−5 to 2.2 ×
dentate chelating group arms such as catecholate (TREN- 10−2 s−1 for the proton-independent step and from 1.2 × 10−6
CAMS, L3), 8-hydroxyquinolinate (O-TRENSOX, L0) or pyri- to 0.18 −1s−1 for the proton-dependent step. Differences in
dinophenolate (TRENPYPOLS, L5). Mixed ligands with che- the rates and mechanism between the different tripodal li-
M
lating arms such as catecholate and 8-hydroxyquinolinate
(TRENSOX2CAMS, L1 and TRENSOXCAMS2, L2) were also
investigated. An analogue of L0 (C-SOX, L4) with a C-pivot
scaffold instead of TREN was studied for comparison. The
acid hydrolysis reaction was found to proceed in four kin-
gands are discussed in terms of the global electrostatic
charge of the ligand and the donor ability of the chelating
subunit. The C−N bond rotation of the amide moiety and the
importance, depending on the charge, of the H-bond net-
work in the partially uncoordinated complexes have also
etically distinguishable stages. A common mechanism in the been considered. The role of electrostatics in complex forma-
series was followed when the ligand structure was varied.
The first stage is very fast and can be attributed to the disso-
tion is discussed in relation to the Eigen-Wilkins mechanism.
ciation of one arm of the tripodal ligand. The second and ( Wiley-VCH Verlag GmbH & Co. KGaA, 69451 Weinheim,
third stages involve protonation of the complex which in-
Germany, 2004)
Introduction
partments of the cell (lysozomes for example), lower pH
levels are present. There have been extensive kinetic studies
of (hydroxamato)FeIII complexes in relation to the predomi-
Microbial iron bio-availability is largely determined by
the coordination chemistry of FeIII with siderophores, a
class of organic compounds having a high and specific chel-
ating affinity for this cation.[1Ϫ3] Natural (siderophores) or
synthetic FeIII chelating agents can be used for therapeutic
or agrochemical applications.[4,5] To gain a better under-
standing of the mechanism of FeIII chelation and transport
in biological systems, quantitative evaluations have been re-
ported through measurements of the kinetics of the FeIII
uptake and release (through acid hydrolysis) in aqueous
solution.[1] A better knowledge of the cleavage of the
ironϪoxygen bonds in iron chelates can help with the de-
sign of new biologically useful ligands, more precisely by
improving iron exchange from ligands to proteins. Studying
the sensitivity of the ironϪoxygen bond to acid hydrolysis
is also interesting since it is well known that, in some com-
nance
of
the
hydroxamate
binding
sites
in
siderophores.[6Ϫ13] The most studied, hexadentate desfer-
rioxamine B (DFB, desferal), has three bidentate hydroxa-
mate groups.[8,9] FeIII complexes of dihydroxamate sider-
ophores and siderophore models have been also investi-
gated[7,10,11] but there are a few kinetic data available for
the dechelation processes of complexes with other types of
ligands.
In particular, numerous ligands having the tripodal struc-
ture have been synthesized and the thermodynamic data of
their FeIII complexes have been determined. The tripodal
structure mimics the backbone of the tris(catecholate) sider-
ophore enterobactin. It is based on tris(2-aminoethyl)amine
(TREN) as a spacer. This central organizing unit is connec-
ted through a tris(amide) moiety to pendant arms each con-
taining a bidentate chelating group such as catecholate
(TRENCAM,[14] TRENCAMS[15]), hydroxamate (TREN-
DROX[16]), 8-hydroxyquinolinate (O-TRENSOX[17]), pyrid-
inophenolate (TRENPYPOLS[18]) or hydroxypyridonate.[19]
These ligands all exhibit strong complexing behavior
towards FeIII. Recently, we synthesized mixed tripodal li-
[a]
´
Institut de Chimie Moleculaire de Grenoble, Laboratoire de
´
´
Chimie Biomimetique, LEDSS, UMR CNRS 5616, Universite
Joseph Fourier,
B. P. 53, 38041 Grenoble Cedex 9, France
E-mail: guy.serratrice@ujf-grenoble.fr
Supporting information for this article is available on the
WWW under http://www.eurjic.org or from the author.
1552
 2004 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
DOI: 10.1002/ejic.200300396
Eur. J. Inorg. Chem. 2004, 1552Ϫ1565

FeIII Complexes with Tripodal Ligands as Siderophore Models
FULL PAPER
gands involving both catecholate and hydroxyquinolinate from the coordination with the bidentate unit (8-hydroxy-
chelating subunits (TRENSOX2CAMS and TRENSOX- quinolinate or catecholate) to the so called salicylate-like
CAMS2[20]). They have been shown to be more effective at mode in which protonation of the pyridine nitrogen atoms
pH ϭ 7.4 than the parent homopodal ligands TREN- or meta-phenolic oxygen atoms induce their replacement by
CAMS and O-TRENSOX. Another series of tripodal li- the amide carbonyl oxygen atoms (see for 8-hydroxyquino-
gands bearing a C-pivot scaffold with either 8-hydroxy- line as example in Figure 2). Such a switch is proposed for
quinoline (C-SOX,[21] COX[22]) or catechol subunits (Cac- the mechanism of iron release from enterobactin making
CAM[23]) has also been developed in our laboratory. A C- reduction easier.[24] There are significant structural differ-
pivot allows the grafting of any desired substituent, e.g. a ences (bidentate subunit, spacer) between these tripodal li-
polyoxyethylene chain, in order to modulate the hydrophile- gands which therefore make a comparative study particu-
lipophile balance or the fluorescent group used for labeling. larly interesting in order to elucidate electrostatic and struc-
We previously investigated the dissociation, by acid hy- tural factors affecting the dissociation and the formation of
drolysis, of the FeIII-TRENCAMS[15] and FeIII-O-TREN- complexes. Furthermore, this kinetic investigation provides
SOX[17b] complexes. Acid hydrolysis has been described by data necessary to formulate mechanisms by which the com-
a stepwise mechanism producing the tetracoordinate bis(sa- plexes are formed and dissociated.
licylate) complex and subsequently leading to its complete
dissociation. In contrast to the 8-hydroxyquinoline groups,
the catecholate groups have been found to be very labile. In
this paper, we report an extensive kinetic study on the acid
hydrolysis and formation kinetics of 1:1 ferric complexes
with the ligands TRENSOX2CAMS (L1), TRENSOX-
CAMS2 (L2), CSOX (L4), and TRENPYPOLS (L5). The
structures of the ligands together with those of O-TREN-
SOX (L0) and TRENCAMS (L3) are depicted in Figure 1.
The interesting feature of these ligands is the possibility of
two modes of coordination for each arm of the ligand de-
Figure 2. Salicylate (a) and 8-hydroxyquinolinate (b) mode of coor-
pending on the pH: a lowering of the pH leads to a switch dination of L with FeIII
Figure 1. Chemical formulae of the tripodal ligands
Eur. J. Inorg. Chem. 2004, 1552Ϫ1565
www.eurjic.org
 2004 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1553

´
G. Serratrice, F. Biaso, F. Thomas, C. Beguin
FULL PAPER
Results
shown in Figure S1 of the Supporting Information which
suggests the rate law (3). The UV/Vis spectra were recorded
as a function of the time and the spectrum at the end of
stage 2 (Figure 4, b) was found to be similar to the spectrum
of the [FeL2H4] species (λmax ϭ 570 nm) having a hydroxy-
quinolinate coordination and a salicylate coordination with
one catecholate arm (through carbonyl and hydroxy oxy-
gen atoms).[20]
Acid Hydrolysis Kinetics
General Observations
All the studied FeLn complexes (n ϭ 0Ϫ5 as shown in
Figure 1) have been previously characterized by UV/Vis
spectrophotometry which showed protonated [FeLnHm]
species as the pH was lowered with acidic media.[15,17b,18,20]
The acid hydrolysis kinetics of the complexes were investi-
gated in aqueous solution (ionic strength I ϭ 2 , 25 °C)
by the pH jump method under pseudo-first-order con-
ditions with an excess of [Hϩ] (varying generally over the
range 0.02Ϫ1.0 ) for solutions containing FeIII and the
tripodal ligand in a 1:1 molar ratio at an initial pH ϭ 5.
The kobs rate constants were determined from the ab-
sorbance data versus time at the λmax of the [FeLn] complex.
The absorption spectra were also monitored (λ range
300Ϫ700 nm) using a diode array device in order to provide
additional information on the intermediate species. As the
pH was lowered, the spectral changes observed were the
result of the change of the number ligands at the iron center
and their mode of coordination due to protonation of the
coordination sites. The structures of the protonated com-
plexes which were formed during the acid hydrolysis were
supported by their absorption spectra.[15,17b,18,20] The acid
hydrolysis reaction was found to proceed in three or four
kinetically distinguishable stages depending on the ligand.
All reaction steps in the first stage were found to occur
rapidly and exhibited a loss of absorbance at all the wave-
lengths (20% of the initial absorbance amplitude for all of
the spectrum). This process (denoted stage 1) was too fast
to be monitored by stopped-flow measurements and it can
been attributed to the dissociation of one arm of the trip-
odal ligand. The second and the third stages caused a strik-
ing change in the spectrum. They involve protonation of
the di-coordinated arms which induces a change of coordi-
nation to a salicylate mode giving the bis(salicylate) (tetra-
coordinated) complex which has been clearly identified
from our previous spectrophotometric studies.[15,17b,18,20]
Finally, the dissociation of the ligand occurs in the fourth
stage in one or two kinetically detectable steps. The success-
ive rate constants are noted ki where i ϭ 4 and 3 for stages
two and three, respectively, i ϭ 2 and 1 for the two steps of
the fourth stage (1 is for the last step of the dissociation).
Figure 3. Acid hydrolysis kinetics of the FeIIIL2 complex: ab-
sorbance decay at λ ϭ 530 or 600 nm; [FeIIIL2] ϭ 0.05 m; [Hϩ] ϭ
1 m for stages 1 and 2, [Hϩ] ϭ 10 m for stage 3, [Hϩ] ϭ 0.8 
for stage 4; solvent: water, I ϭ 2  (NaClO4, HClO4), 25 °C
FeIII-TRENSOXCAMS2 (FeL2)
Four first-order kinetic processes were observed in the
[Hϩ] range of 0.001Ϫ1  (Figure 3). The spectrum recorded
after the dead time is similar to that of the complex
[FeL2H3] with a hydroxyquinolinate and catecholate coordi-
nation (λmax ϭ 540 nm) indicating rapid dissociation of a
catechol arm and protonation of the tertiary nitrogen atom
(Figure , a) since the value of its pKa ϭ 3.4.[20] A first-order
absorbance decay was then measured in the time range of
Figure 4. UV/Vis spectra recorded as a function of the time for the
acid hydrolysis kinetics of the FeIIIL2 complex: (a) stage 1, 0Ϫ3 ms
(b) stage 2, 0Ϫ1 s (c) stage 3, 0Ϫ500 ms (d) stage 4, step 1 0Ϫ0.5
s, (e) stage 4, step 2, 0.5Ϫ100 s; experimental conditions: see Fig-
ure 3; the arrow indicates the direction of change as the reaction
proceeds
seconds (stage 2). The corresponding pseudo-first-order
constant k4
This is consistent with the following reactions (charges of
obs
was found to vary linearly with [Hϩ] as the complexes and the ligand omitted) from which the
1554
 2004 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.eurjic.org
Eur. J. Inorg. Chem. 2004, 1552Ϫ1565

FeIII Complexes with Tripodal Ligands as Siderophore Models
FULL PAPER
values k4 ϭ 4440 Ϯ 180 Ϫ1 sϪ1 and kϪ4 ഠ 0 sϪ1 could be the
hydroxamate
ligand
[Equations (9)Ϫ(11)[24] [6]
]
deduced [Equations (1)Ϫ(3)].
FeL2 ϩ3 Hϩ ǟǞ FeL2H3 (fast)
FeL2H3 ϩ Hϩ Ǟǟ FeL2H4 (k4, kϪ4
k4obs ϭ k4 [Hϩ] ϩ kϪ4
{[FeOH]2ϩ for [Fe(OH)(OH2)5]2ϩ, Fe3ϩ for [Fe(OH2)6]3ϩ}.
FeL2H5 ϩ Hϩ Ǟǟ FeL2H6 (k2, kϪ2, K2)
k2obs ϭ k2 [Hϩ] ϩ kϪ2
(7)
(8)
(1)
(2)
(3)
)
FeL2H6 ϩ Hϩ ǟǞ Fe3ϩ ϩ L2H7 (k1, kϪ1, K1)
FeL2H6 ϩ H2O Ǟǟ FeOH2ϩ ϩ L2H7 (k1Ј, kϪ1Ј, K1Ј)
Fe3ϩ ϩ H2O Ǟǟ FeOH2ϩ ϩ Hϩ (KFe ϭ 0.0015 )
(9)
(10)
(11)
The third stage exhibited a single-exponential absorbance
decay measured at λ ϭ 600 nm and was monitored over the
[Hϩ] range 0.01Ϫ0.07 . The spectra collected during this
stage exhibited an isosbestic point at 525 nm and a decrease
of λmax (Figure 4, c) characteristic of the formation of the
By assuming relaxation conditions for the reverse reac-
tions, the experimental rate law can be expressed as Equa-
tion (12).
FeL2H5 species having a bis(salicylate) coordination.[20] The
obs
rate constant k3
did not show a variation with [Hϩ]
(mean value ഠ 60 sϪ1, see Figure S2 in the Supporting In-
formation). By analyzing the absorbance data at the end of
this stage as a function of [Hϩ]n in the form of a Hill plot,
a linear plot was obtained for n ϭ 1 (Figure S3 in the Sup-
porting Information). These data suggest a saturation be-
havior with a fast acid-basic pre-equilibrium preceding the
rate determining step as given by Equations (4)Ϫ(6).
k1obs ϭ k1[Hϩ] ϩ k1Ј ϩ (kϪ1 ϩ kϪ1Ј KFe/(KFe ϩ
(12)
[Hϩ])]([FeIII]e ϩ [L2H7]e)
[FeIII]e and [L2H7ϩ]e are the total equilibrium concen-
trations of ligand and metal under our experimental con-
ditions (calculated from absorbances at equilibrium). The
term in parentheses calculated using the value of kϪ1Ј (2200
FeL2H4 ϩ Hϩ ǟǞ FeL2H5* (K3Ј)
FeL2H5* Ǟǟ FeL2H5 (k3, kϪ3
k3obs ϭ k3 K3Ј [Hϩ]/(1 ϩ K3Ј [Hϩ]) ϩ kϪ3
(4)
(5)
(6)
Ϫ1 Ϫ1) determined from formation kinetics study (vide
s
infra) was found to be negligible (assuming kϪ1 also to be
negligible). The fitted values are k1 ϭ 0.0137 Ϯ 0.0005 Ϫ1
)
sϪ1 and k1Ј ϭ 0.00237 Ϯ 0.00025 sϪ1
.
It was found that K1 K2 ϭ K1Ј K2/KFe ϭ 10Ϫ2.68 is in
reasonable agreement with the value 10Ϫ2.27 determined
from measurements at equilibrium and at an ionic strength
of 0.1 .[20]
obs
Our results imply that K3Ј [Hϩ] ϾϾ 1, so that k3 ഠ k3
ϩ kϪ3. Since [Hϩ] spans the range 0.01Ϫ0.07 , the lower
limit for K3Ј can be evaluated as ഠ103 Ϫ1. Since the pKa
value of the equilibrium FeL2H5/FeL2H4 has been reported
FeIII؊TRENSOX2CAMS (FeL1)
Three distinct first-order kinetic processes could be de-
as 2.2,[20] it can be inferred that k3/kϪ3 Ͻ ഠ10Ϫ0.8 (1/Ka ϭ tected (Figure S5). The spectrum recorded after the dead
K3Јk3/kϪ3) allowing evaluation of an upper limit for k3 as ഠ time suggests a bis-8-hydroxyquinolinate coordination (Fig-
10 sϪ1 and kϪ3 as ഠ 50 sϪ1. The chemical identity of ure S6a in the Supporting Information)[20] indicating rapid
FeL2H5* will be discussed later.
dissociation of a catechol arm and the formation of the
The spectral change observed during stage 4 showed an [FeL1H3] species taking account of the protonation of the
absorbance decrease in two steps at all wavelengths (Fig- tertiary nitrogen. The spectral change observed during stage
ure 4, d and e) in two time ranges which is consistent with 2 reflects the change of coordination to a bis(salicylate)
complete dissociation of the ligand. Two well separated sin- mode with the two hydroxyquinoline arms (Figure S6b) in
gle-exponential absorbance decays (measured at 500 nm) the [FeL1H5] species. This is clearly supported by its ab-
were monitored in the time range of 0.5 s and 100 s over sorbance spectrum.[20] Contrary to the FeL2 complex, only
obs
the [Hϩ] range of 0.2 to 0.65 . The rate constants k2
one absorbance increase was monitored at 435 nm, in the
obs
and k1 exhibited linear variations with [Hϩ] (Figure S4). time range of deciseconds, over the [Hϩ] range of 0.03 to
This result can be interpreted by the successive dissociation 0.9  (Figure S5). It was supposed that formation of the
reactions. The faster involves dissociation of one arm of the [FeL1H4] species, corresponding to the change of coordi-
tripodal ligand (charge omitted for [FeL2Hn]) [Equa- nation to a salicylate mode for one 8-hydroxyquinoline arm,
tions (7) and (8)] with k2 ϭ 33.1 Ϯ 1.9 Ϫ1 sϪ1 and kϪ2
ϭ
was not detectable under our experimental conditions. The
11.2 Ϯ 1.4 sϪ1 and the slower involves dissociation of the absorbance increase allowed determination the rate con-
ligand via two parallel acid-dependent and acid-indepen- stant k3obs. The variation of k3
with [Hϩ] (Figure S7)
obs
dent pathways to give free iron() in agreement with the is linear for [Hϩ] Ͼ 0.4  and exhibits an inverse proton
dissociation pathways proposed for ferric complexes with dependence for [Hϩ] Ͻ 0.4 . This behavior can best be
Eur. J. Inorg. Chem. 2004, 1552Ϫ1565
www.eurjic.org
 2004 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1555

´
G. Serratrice, F. Biaso, F. Thomas, C. Beguin
FULL PAPER
explained by Equations (13)Ϫ(15), by analogy with of 10Ϫ3.21 determined from measurements at equilib-
Equations (9)Ϫ(11), yielding the rate expression according rium.[20]
to Equation (16).
FeIII؊CSOX (FeL4)
FeL1H4 ϩ Hϩ Ǟǟ FeL1H5 (k3, kϪ3
FeL1H4 ǟǞ FeL1H4* (k3Ј, kϪ3Ј)
)
(13)
(14)
(15)
Four distinct stages could be detected by measuring the
absorbance decay at 600 nm (Figures S8 and S9). After the
dead time, a bi-exponential decay of the absorbance was
observed on a 10 s time scale with [Hϩ] in the range of
0.02Ϫ0.1  (stages 2 and 3). This decay could be fitted with
a sum of two exponentials [Equation (18)].
FeL1H4* ϩ Hϩ ǟǞ FeL1H5 (K3ЈЈ, fast)
k3obs ϭ k3 [Hϩ] ϩ k3Ј ϩ (kϪ3K3ЈЈ[Hϩ]) ϩ kϪ3Ј)/(1 ϩ K3ЈЈ[Hϩ]) (16)
A ϭ A ϩ A1exp(Ϫk4obs t) ϩ A2exp(Ϫk3obs t)
(18)
The non-linear fit of these data give the values k3 ϭ 17.9
Ϯ 1.3 Ϫ1 sϪ1, kϪ3Ј ϭ 635 Ϯ 149 sϪ1, K3ЈЈ ϭ 158 Ϯ 43 Ϫ1
(k3Ј and kϪ3 could not be adjusted from the fit). The chemi-
cal identity of FeL1H4* and FeL1H5* will be discussed later.
The spectral change observed during stage 3 showed an
absorbance decrease at all wavelengths (Figure S6c) and is
consistent with dissociation of the ligand from the FeL1H5
species. A mono-exponential absorbance decay (measured
at 435 nm) was monitored on a 1000 s time scale over the
obs
obs
The rate constants k4 and k3 vary apparently with
[Hϩ] with a non-zero intercept (Figure S10). This suggests
two reversible elementary steps [Equations (19) and (20)].
FeL4H2 ϩ Hϩ Ǟǟ FeL4H3 (k4, kϪ4
)
)
(19)
(20)
FeL4H3 ϩ Hϩ Ǟǟ FeL4H4 (k3, kϪ3
obs
[Hϩ] range of 0.2 to 1 . The variation of k1 vs. [Hϩ] is
The matrix formulation of the rate equations allows de-
termination of analytical expressions for the observed rate
constants. The following expressions can thus be deduced
[Equations (21) and (22).
shown in Figure 5. This can be interpreted by Equations (7)
and (9)Ϫ(11) (replacing L2 for L1) and Equation (12) modi-
fied to give Equation (17) by considering Equation (7) as a
fast pre-equilibrium.
k4obs ϩ k3obs ϭ (k4 ϩ k3) [Hϩ] ϩ kϪ4 ϩ kϪ3
(21)
(22)
k1obs ϭ (k1 [Hϩ] ϩ k1Ј) K2 [Hϩ]/(1 ϩ K2 [Hϩ]) ϩ [kϪ1 ϩ kϪ1Ј
(17)
KFe/(KFe ϩ [Hϩ])]([FeIII]e ϩ [L1H7]e)
(k4obs Ϫ k3obs)2 ϭ (k4 [Hϩ] ϩ kϪ4)2 ϩ (k3 [Hϩ] ϩ kϪ3)2 ϩ 2
k
Ϫ4 (k3 [Hϩ] Ϫ kϪ3) Ϫ 2 k4 (k3 [Hϩ] ϩ kϪ3
)
The four microscopic constants were extracted from a
non-linear least-squares refinement of the data according
to Equations (21) and (22): k4 ϭ 189 Ϯ 9 Ϫ1 Ϫ1, kϪ4ϭ
10.0 Ϯ 0.4 sϪ1, k3 ϭ 18.1 Ϯ 0.9 Ϫ1 Ϫ1, kϪ3 ϭ 1.20 Ϯ
0.05 sϪ1
s
s
.
The spectral change observed during stage 4 showed an
absorbance decrease at all wavelengths (Figure S9) and is
consistent with dissociation of the ligand. A mono-ex-
obs
Figure 5. k1 [sϪ1] as a function of [Hϩ] [] in the acid hydrolysis
kinetics of FeL1; [FeL1] ϭ 0.07 m; solvent: water, I ϭ 2  (Na-
ClO4, HClO4), 25 °C
The term in parentheses was calculated using the value of
k
Ϫ1Ј (1320 Ϫ1 sϪ1) determined from the formation kinetics
study (vide infra) and assuming kϪ1 to be negligible. The
fitted values are k1 ϭ 0.0032 Ϯ 0.0004 Ϫ1 sϪ1 k1Ј ഠ 0
s
Ϫ1K2 ϭ 1.70 Ϯ 0.55 Ϫ1
.
Figure 6. k1 [sϪ1] as a function of [Hϩ] [] in the acid hydrolysis
obs
It was found that K1 K2 Յ K1Ј K2/KFe ϭ 10Ϫ3 55
(taking
·
kinetics of FeL4; [FeL4] ϭ 0.1 m; solvent: water, I ϭ 2  (NaClO4,
k1Ј Յ k1/10) which is in reasonable agreement with the value HClO4), 25 °C
1556  2004 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.eurjic.org
Eur. J. Inorg. Chem. 2004, 1552Ϫ1565

FeIII Complexes with Tripodal Ligands as Siderophore Models
FULL PAPER
ponential absorbance decay (measured at 435 nm) was This is consistent with the reaction according to Equa-
monitored on a 10000 s time scale over the [Hϩ] range of tion (23) and and the rate law according to Equation (24)
vs. [Hϩ] is shown in yielding the values k3 ϭ 36 Ϯ 3 Ϫ1 sϪ1 and kϪ3 ϭ18 Ϯ 1
obs
0.1Ϫ1 . The variation of k1
Figure 6.
s
Ϫ1, respectively.
This can be interpreted by Equations (9)Ϫ(11) and Equa-
tion (12) (by replacing FeL2H5 for FeL1H6 and L2H7 for
L1H6). The term in parentheses was calculated using the
value of kϪ1Ј (1750 Ϫ1 sϪ1) determined from the formation
kinetics study (vide infra) and assuming kϪ1 to be negli-
gible. The fitted values are: k1Ј ϭ (2.0 Ϯ 0.2) ϫ 10Ϫ4 sϪ1
FeL5H4 ϩ Hϩ Ǟǟ FeL5H5 (k3, kϪ3, K3)
k3obs ϭ k3 [Hϩ] ϩ kϪ3
(23)
(24)
and k1 ഠ 0 Ϫ1 sϪ1
.
The spectral change observed during stage 3 showed an
absorbance decrease at all wavelengths (Figures S12c and
S12d) and resembles the spectra recorded for the formation
reaction study. This is consistent with complete dissociation
of the ligand. Two well-separated single-exponential ab-
sorbance decays (measured at 470 nm) were monitored in
the time range of 5 s and 100 s over the [Hϩ] range of 0.1Ϫ1
FeIII؊O؊TRENSOX (FeL0)
The acid hydrolysis kinetics have been previously publi-
shed.[17b] Only the last step of the dissociation was not de-
scribed. As for FeIII؊CSOX, release of the ligand is very
slow (20000 s time scale) even at high Hϩ concentrations. It
was observed at 435 nm as a mono-exponential absorbance
obs
. The rate constants k2 exhibited linear variations with
obs
decay. The variation of k1 vs. [Hϩ] is shown in Figure 7.
obs
[Hϩ] (Figure S14). The rate constants k1 exhibited a lin-
The fitted values from Equation (13) are k1Ј ϭ 1.0 ϫ 10Ϫ5
ear proton dependence for [Hϩ] Ͼ 0.2  and an inverse
proton dependence for [Hϩ] Ͻ 0.1 (Figure 8). These suc-
cessive dissociation reactions are consistent with reactions
and rate laws according to Equations (7)Ϫ(12) (by replacing
FeL2H5 with FeL5H5 and L2H7 with L5H7). The rate con-
stants for Equation (7) have been determined as k2 ϭ 4.9 Ϯ
0.2 Ϫ1 sϪ1; kϪ2 ϭ 2.1 Ϯ 0.1 sϪ1 and for Equations (9) and
(10) the following could be evaluated from a non-linear le-
ast-squares fit of Equation (11) by taking for kϪ1Ј the value
of 514 Ϫ1 sϪ1 determined from the formation kinetic stud-
ies (vide infra) and assuming kϪ1 to be negligible: k1ϭ (7.9
Ϯ 1.0 ϫ 10Ϫ6 sϪ1 and k1 ഠ 0 Ϫ1 sϪ1 [the term in parenth-
eses in Equation (12) was calculated using the value of 789
[17b]
Ϫ1 sϪ1 for kϪ1Ј
and assuming kϪ1 to be negligible].
FeIII؊TRENPYPOLS (FeL5)
The absorbance change was measured at 470 nm. Three
stages were observed in the [Hϩ] range of 0.001Ϫ1  (Fig-
ure S11). The spectrum change after the dead time resulted
in a decrease of λmax from 495 to 469 nm (Figure S12a)
and was attributed to the formation of a tetra-coordinated
complex with one arm of the tripodal ligand in the pyri-
dinophenolate coordination mode and the other in the
salicylate coordination mode.[18] This very fast process was
followed by a slight increase in the absorbance at 470 nm
(stage 2) in the time range of 200 ms. The UV/Vis spectra
recorded as a function of the time (Figure S12b) exhibited
an isosbestic point at 540 nm and the spectrum at the end
of this stage is similar to that of the bis(salicylate) complex
Ϯ 1.0) ϫ 10Ϫ3 Ϫ1 Ϫ1; k1Ј ϭ (9.6 Ϯ 5.7) ϫ 10Ϫ4 sϪ1
s .
[FeL5H5].[18] The corresponding pseudo first-order constant
obs
k3
was found to vary linearly with [Hϩ] (Figure S13).
obs
Figure 8. k1 [sϪ1] as a function of [Hϩ] [] in the acid hydrolysis
kinetics of FeL5; [FeL5] ϭ 0.2 m; solvent: water, I ϭ 2  (NaClO4,
HClO4), 25 °C
It was found that K1 K2 ϭ K1Ј K2/KFe ϭ 10Ϫ2.54 which is
in reasonable agreement with the value of 10Ϫ1.79 deter-
mined from measurements at equilibrium.[18] The values of
k1, k1Ј, k2, and kϪ2 for all the complexes are collected in
Table 1.
Figure 7. k1 [sϪ1] as a function of [Hϩ] [] in the acid hydrolysis
kinetics of FeL0; [FeL0] ϭ 0.07 m; solvent: water, I ϭ 2  (Na-
ClO4, HClO4), 25 °C
obs
Eur. J. Inorg. Chem. 2004, 1552Ϫ1565
www.eurjic.org
 2004 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1557

´
G. Serratrice, F. Biaso, F. Thomas, C. Beguin
FULL PAPER
Table 1. Microscopic rate constants for the dissociation of the FeLnH5 (except for L4, FeL4H4) complexes
ki [Ϫ1
k2 (kϪ2
s
Ϫ1] (kϪi [sϪ1])
FeL0
FeL1
FeL2 [a]
FeL3 [b]
FeL4
FeL5
)
Ϫ
Ϫ
3.2
Ͻ 0.32
1.32
33.1 (11.2)
13.7
2.37
fast
184
22
Ϫ
Ͻ 0.02
0.20
4.9 (2.1)
7.9
0.96
103 ϫ k1
Ͻ 0.0012
0.012
0.789
103 ϫ kЈ1 [sϪ1
]
10Ϫ3 ϫ kЈϪ1 [Ϫ1 sϪ1
]
2.24
2.7
1.75
0.51
[a]
Ref.[17b] [b] Ref.[15]
Formation Kinetics
The kinetics of formation of the ferric complex with li-
gands L1, L2, L4 and L5 were investigated by a stopped-
flow spectrophotometric method under pseudo-first-order
conditions: [FeIII] and [Hϩ] ϾϾ [Ln] and at an ionic
strength I ϭ 2  at 25 °C. The absorbance change vs. time
at λ ϭ 435 nm showed a single exponential curve indicating
that the complex formation reaction proceeds through a
single rate-limiting step. It should be noted that the spectra
recorded with time using a diode array device were identical
to those recorded for the final step of acid hydrolysis. It has
been found that the pseudo-first-order rate constants kobs
at a given acidity level ([Hϩ] varied from 0.03 to 1 ) show
a linear variation as a function of [FeIII
]tot with a zero inter-
cept, and at a given concentration of FeIII have a linear
variation as a function of 1/[Hϩ] (Figure 9, a as an ex-
ample). This suggests that the main reaction is the forma-
tion of the complex [FeLnH6] through the hydrolyzed ferric
ion FeOH2ϩ {for [Fe(OH)(OH2)5]2ϩ} according to Equa-
tion (10) (formation is the reverse reaction). This leads to
the observed pseudo-first-order rate constant given by
Equation (25), where [Fe3ϩ] ϭ [Fe3ϩ
]
Ϫ [FeOH]2ϩ Ϫ 2
tot
[Fe2(OH)2]4ϩ and [Hϩ] takes into account the protons in-
volved in formation of FeOH2ϩ [Equation (11)] and the di-
4ϩ
mer Fe2(OH)2
{for [Fe2(OH)2(OH2)8]4ϩ
}
[Equa-
tion (26)].[25]
k
obs ϭ kϪ1Ј KFe [Fe3ϩ]/[Hϩ] ϩ k1Ј
(25)
(26)
Figure 9. Complex formation kinetics of L1 with FeIII: (a) kobs [sϪ1
]
as a function of [Fe3ϩ tot [] at various [Hϩ] []; [L1] ϭ 0.1 m; (b)
]
2 Fe3ϩ Ǟǟ Fe2(OH)24ϩ ϩ2 Hϩ (KDFe ϭ 0.0024 )
kobs [sϪ1] as a function of [Fe3ϩ]/[Hϩ] according to Equation (25);
solvent: water, I ϭ 2  (NaClO4, HClO4), 25 °C
A linear variation of kobs versus [Fe3ϩ]/[Hϩ] was observed
(Figure 9, b for L1, Figure S15 for L2, Figure S16 for L4 and
Figure S17 for L5). Least-squares fits of the data yielded the
values of kϪ1Ј 1320 Ϯ 60, 2240 Ϯ 120, 1750 Ϯ 40 and 514
Ϯ 8 Ϫ1 sϪ1 for L1, L2, L4 and L5, respectively (k1Ј ഠ 0 sϪ1
for all cases). The values of kϪ1Ј are collected in Table 1
together with the values previously determined for the for-
mation of FeL0 and FeL3.
including data already published for OϪTRENSOX[17b]
(L0) and OϪTRENCAMS[15] (L3). For easy comparison of
the kinetic data of stages 1Ϫ3, we have reported in Table 2
the state of protonation of the tertiary nitrogen atom (for
L0, L1, L2, L3 and L5) and the mode of coordination of
FeIII in the different species according to the state of pro-
tonation, the number of arms coordinated to the metallic
cation and the bidentate coordination (8-hydroxyquinolin-
ate or -catecholate or -salicylate). The following discussion
suggests elementary reaction pathways as depicted in Fig-
ure 10 for FeL2, Figure S18 for FeL1 and Figure 14 of
ref.[17b] for FeL0, as examples.
Discussion
Dissociation Kinetics
Rate constant data for the stepwise acid dissociation of
the FeIIILn complexes are summarized in Tables 1 and 2,
1558
 2004 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.eurjic.org
Eur. J. Inorg. Chem. 2004, 1552Ϫ1565

FeIII Complexes with Tripodal Ligands as Siderophore Models
FULL PAPER
Table 2. Recapitulative table of the measured hydrolysis rate constants for FeLn Ǟ FeLnH5 including modes of protonation and coordi-
nation for each stage (charges are omitted)
Ligand
Species[a]
kn [Ϫ1sϪ1
]
kϪn [sϪ1
]
Tertiary N[b]
NH
Mode of coordination[c]
NO/NO/NO
L0
FeL0H
Ref.[17b]
fast
643
5.7
FeL0H3
k4
NH
NO/NO/free
15
FeL0H4
k3
NH
NO/OOsal(ox)/free
1.3
FeL0H5
FeL1
NH
N
OOsal(ox)/OOsal(ox)/free
NO/NO/OO
L1
fast
Ϫ
FeL1H2
Fast
N
NO/NO/free
FeL1H3
k4
NH
NH
NH
NO/NO/free
Ϫ
Ϫ
FeL1H4
k3
NO/OOsal(ox)/free
OOsal(ox)/OOsal(ox)/free
17.9
Ϫ
FeL1H5
K3ЈЈ ϭ 158
FeL1H4*
FeL2H
N
NH
OOsal(ox)/OOsal(ox)/free
NO/OO/OO
L2
FeL2H3
k4
NH
NH
NH
NO/OO/free
4440
Ͻ 1
FeL2H4
KЈ3 Ͼ 1000
FeL2H5*
k3
NO/OOsal(cat)/free
O(ox)/OOsal(cat)/free
fast
Ͻ 10 sϪ1
ca. 50
FeL2H5
FeL3H
NH
NH
OOsal(ox)/OOsal(cat)/free
OO/OO/OO
L3
Ref.[15]
fast
fast
FeL3H3
NH
OO/OO/free
FeL3H5
FeL4
NH
C
OOsal(cat)/OOsal(cat)/free
NO/NO/NO
L4
fast
189
18.1
FeL4H2
k4
C
C
NO/NO/free
10.0
1.2
FeL4H3
k3
NO/OOsal(ox)/free
FeL4H4
FeL5H
C
NH
OOsal(ox)/OOsal(ox)/free
NO*/NO*/NO*
L5
fast
fast
36
Ϫ
Ϫ
18
FeL5H3
k4
NH
NH
NH
NO*/NO*/free
FeL5H4
k3
NO*/OO*sal/free
OO*sal/OO*sal/free
FeL5H5
[a]
[b]
[c]
The different states of protonation during the pH jump (see text). State of protonation of the anchor. The coordination atoms of
the three arms, separated by /, in the reverse order of the complete removal of the nth arm with FeIII: ‘‘free’’ indicates an arm without a
coordination bond to the iron atom, NO (NO*) refers to 8-hydroxyquinolinate (pyridinophenolate) bidentate coordination to FeIII, OO
refers to catecholate bidentate coordination to FeIII, OOsal(ox) indicates salicylate bidentate coordination of a hydroxyquinoline arm to
FeIII, OOsal(cat) refers to salicylate bidentate coordination of a catechol arm to FeIII, OO*sal represents salicylate bidentate coordination
of a pyridinophenol arm to FeIII
.
Stage 1
netic lability of the hexacoordinate complex, the solvent re-
arrangement resulting from dissociation of the first biden-
On the basis of the spectral change, the reaction steps in tate group, as already supported for hydroxamate dis-
this stage have been attributed to the dissociation of one sociation.[6] It can be suggested that the dissociated arm is
arm of the ligand. This process is very fast whatever the maintained in proximity to the first coordination sphere of
type of chelating group. This suggests that the steric strain FeIII through hydrogen bonds between the water molecules
release in the complex may increase the rate of FeϪO and coordinated to FeIII and the free arm. This results in a small
(or) FeϪNpyr cleavage. We can also invoke, for the high ki- motion of the arm as it dissociates from the FeIII center
Eur. J. Inorg. Chem. 2004, 1552Ϫ1565
www.eurjic.org
 2004 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1559

´
G. Serratrice, F. Biaso, F. Thomas, C. Beguin
FULL PAPER
Figure 10. Proposed mechanism for the acid hydrolysis of the FeL2H complex
and a slight solvent rearrangement. The reaction from the tate group of each arm, as clearly evidenced.[22,26] We were
hexadentate to the tetradentate complex may therefore be a able to obtain information on several elementary steps cor-
facile process.
Stages 2 and 3
Overall Scheme
responding to stages 2 and 3 and rate constants k4 and k3,
depending on the structure of the tripodal ligand, homopo-
dand or heteropodand, and on the nature of the chelating
subunit (catecholate, 8-hydroxyquinolinate, pyridinophen-
olate). Interesting correlations of the rate constants for
same type of process with the chemical structures can be
made.
These stages result in the formation of a bis(salicylate)-
coordinated complex FeLnH5 (except FeL4H4) as clearly
indicated by the spectra recorded at the end of stage 3 in
agreement with the equilibrium studies previously pub-
lished. Each stage is related to one arm of the ligand. It
may involve the FeϪNpyr or FeϪO (depending on the type
Stages 2 and 3 for the Homopodate FeLnH3 Complexes
(n ؍ 0, 3, 4, 5)
The values of k4 and k3 clearly indicate the more labile
of bidentate group in Ln) bond cleavage with protonation of character of the catechol arms (fast for the two arms in
N or O leading to an open-ring structure with the bidentate FeL3H3) compared with the pyridinophenol arms (first
moiety, followed by the FeϪO(ϭC) bond closing the salicy- arm: fast, second arm: 36 Ϫ1 sϪ1 in FeL5H3) and the 8-
late ring. It should be noted that this change of coordi- hydroxyquinoline arms (first arm: 643 and 189 Ϫ1 sϪ1, se-
nation requires conformational reorganization of the arm cond arm: 5.7 Ϫ1 sϪ1 and 18.1 Ϫ1 sϪ1 in FeL0H3 and
by rotation about the CϪN amide bond in order to allow FeL4H2, respectively). This reflects (i) the higher donor
the carbonyl group to be in a suitable position favorable for strength of catecholate vs. pyridinophenolate and 8-
salicylate coordination. Indeed, it has been shown that the hydroxyquinolinate which increases the electron density of
carbonyl group is turned outside the FeIII coordination the metal so that the bonds with the coordinated donor
sphere owing to hydrogen bond between the amide hydro- atoms are weakened; (ii) the chelate bulkiness which influ-
gen atom and the ortho-oxygen atom (quinolinol or cat- ences the rate of rotation about the CϪN amide bond (pyri-
echol or phenol) when FeIII is coordinated with the biden- dinophenol vs. catechol); (iii) the easier rotation of the am-
1560
 2004 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.eurjic.org
Eur. J. Inorg. Chem. 2004, 1552Ϫ1565

FeIII Complexes with Tripodal Ligands as Siderophore Models
FULL PAPER
ide bond in the catechol arm compared with in the 8- when measured, between k2 and k1 (or kЈ1) for the release
hydroxyquinoline arm that can be related to the electron- of two identical (salicylate) arms as already observed in re-
withdrawing effect of the 8-hydroxyquinoline group as lated systems.[1]
clearly shown from the pKa values of the carboxylic groups
in the 2,3-dihydroxybenzoic acid (2.7) and in the 7-carboxy- Dependence of k1 and kЈ1 as a Function of Structure
8-hydroxyquinoline (1.9).[27]
(Electrostatic Charge)
The important feature is that the rate constants k1 and
k1Ј span over a range of several orders of magnitude: five
orders for k1 (184 to 0.0012 ϫ 10Ϫ3 Ϫ1 sϪ1) and three and
a half orders for k1Ј (22 to 0.012 ϫ 10Ϫ3 sϪ1). Several trends
in relation to the structure of the complex should be
pointed out. The decrease of the rate constants k1 and k1Ј
(except for [FeL5H6]4ϩ) in the series (Table 1) seems to be
related mainly to the increase of the positive charges on the
complexes evaluated without taking account of the sulfon-
ate substituents, as in ϩ3, ϩ4, ϩ5 and ϩ6 for the iron()
complexes with L3, L2, both L1 and L4 and L0, respectively.
For the proton-dependent pathway (k1) this is partly due to
repulsion of the incoming Hϩ by positive charges, suggest-
ing that the dissociated 8-hydroxyquinoline arms of the li-
gand are maintained in proximity to the first coordination
sphere of FeIII probably through H-bonding between pyrid-
inium nitrogen atoms and coordinated water molecules. It
should be pointed out that k1 for the complexes with the
tris(8-hydroxyquinoline) ligands (L0 and L4) is 2 to 3 orders
of magnitude slower than for the other complexes. This is
probably due to enhanced repulsion for Hϩ by positive
charges suggesting a closer proximity for the dissociated
arms than for the other ligands in [FeL0H6] and [FeL4H6].
Indeed, pyridinium nitrogen atoms have been found to es-
tablish H-bonds with oxygen atoms coordinated to FeIII in
the structure of the tris(salicylate) complex with L0 forming
a tight coordination sphere around the FeIII center.[28] The
value of 0.0079 Ϫ1 sϪ1 for [FeL5H6]4ϩ is high when com-
pared with these highly charged complexes. This suggests a
lower repulsion of the incoming Hϩ by positive charges that
are assumed to be far from the coordination sphere due to
a free rotation of the pyridine ring relative to the phenol
ring, thus disfavoring the formation of hydrogen bonds. For
the proton-independent pathway, the decrease of k1Ј paral-
lels that of k1.
Stages 2 and 3 for the Heteropodate FeLnH3 Complexes
(n ؍ 1, 2)
The reaction scheme developed from the analysis of the
kinetic data is slightly different than for the homopodate
complexes. For L1, only the stage FeL1H4/FeL1H5 was de-
tected and can be described by Equations (13)Ϫ(15). This
scheme, developed from an analysis of the kinetic data, has
been proposed since the pKa values of the tertiary amine
nitrogen atom (1.9) and of the two hydroxyquinoline nitro-
gen atoms (2.4) are close.[20] We assumed a complicated ki-
netic pattern involving an acid-basic micro-equilibrium. A
pictorial view of the proposed scheme is shown in Figure
S18 (see also Table 2 for the state of protonation). The equi-
librium between FeL1H4 and FeL1H4* [Equation (14)] in-
volves proton transfer between the tertiary nitrogen atom
and the quinoline nitrogen atom coordinated to the iron
center, leading to the change of coordination of the 8-
hydroxyquinoline moiety. The reaction according to Equa-
tion (15) involves the protonation of the tertiary amine ni-
trogen atom. The k3 value (17.9 Ϫ1
corresponding values determined for FeL0H4 (5.7 Ϫ1 sϪ1
and FeL4H3 (18.1 Ϫ1 Ϫ1). The value of K3ЈЈ (158 Ϫ1) is
s
Ϫ1) is similar to the
)
s
in agreement with the pKa value (1.9) determined from
static measurements at a 0.1  ionic strength. The kϪ3Ј
value (635 sϪ1) has no equivalent in the kinetic treatment
to be related to the previously described micro-equilibrium.
For FeL2H3 the value of k4 (4440 Ϫ1 Ϫ1) is consistent
s
with the high lability of the catechol arm. For stage 3, we
suggest that the pre-equilibrium [Equation (4)] is related to
the FeϪN bond breaking of the 8-hydroxyquinoline arm
leading to FeL2H5* (Figure 10) and that the last step
[Equation (5)] is the change to the bidentate salicylate coor-
dination of this arm (FeL2H5). The upper limit of 10 sϪ1
for k3 is reasonable by comparison with the rate of rotation
about the amide bond in hydroxamate ligands.[6]
Relationship between the Acid-Dependent and Acid-
Independent Steps
Stage 4
The rate constants of these dissociations having acid-de-
pendent and acid-independent steps vary in the same man-
ner as a function of structure. We believe that this similarity
is related to the structures of the transition states differing
only by the presence of a proton. As the starting species is
the same for these two steps, the transition state theory sug-
gests a linear correlation between ln k1 and ln k1Ј [Equa-
tions (27) and (28)] as already discussed.[29,30]
In the last stage, leading from the tetradentate bis(salicyl-
ate) to the free ligand, at least two steps occur, correspond-
ing to the FeϪO bond breakings of each salicylate arm.
The rate constants for the leaving of the penultimate salicy-
late arms, k2, could be measured only for L2 (33 sϪ1 Ϫ1
)
and L5 (4.9 sϪ1

Ϫ1). The smaller value for FeL5 can be
related to the bulkier access for the proton to the tetraden-
tate complex with the pyridinophenol subunits in relation
to the easy bond rotation between the pyridine and phenol
moieties. The rate constants for the leaving of the last sal-
icylate arm, k1 (proton-dependent) and kЈ1 (proton-inde-
pendent) are well documented for all terms of the series
ln k1 ϭ ln k1Ј ϩ ln Kt
(27)
(Table 2). About three orders of magnitude were observed, Ϫln Kt ϭ ∆Gt#/RT
(28)
Eur. J. Inorg. Chem. 2004, 1552Ϫ1565
www.eurjic.org
 2004 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1561

´
G. Serratrice, F. Biaso, F. Thomas, C. Beguin
FULL PAPER
∆Gt# is the free energy difference between the depro- a large range (5.6Ϫ9.23) in the free ligand. The FeϪOϭC
tonated and the protonated transition states, Kt may be con- bond cleavage can therefore be suggested as the rate-limit-
sidered as the protonation constant for the transition state ing step for dissociation after a fast opening of the chelate
and pKAt (ϭ log k1 Ϫ log kЈ1) may be considered as a ring as shown in Figure 12.
pseudo-pKA of the transition state. A good correlation with
Such a relationship has been observed for the dis-
a slope of 0.96 and an intercept lnKt ϭ 1.83 (pKAt ϭ 0.80) sociation of a large series of FeIII-mono(hydroxamato) com-
was obtained for the complexes with L1, L2, L3 and L5 (Fig- plexes yielding a slope of 0.9(1) in agreement with the
ure 11). The correlation was not attempted for L0 and L4 theoretical value of 1, and an intercept ln K1 ϭ Ϫ0.6
since, for these ligands, k1 values were only obtained as su- (pKAt ϭ Ϫ0.26).[7] The correlation shows also that the tran-
perior limits.
sition state acidity of FeIII (L1 or L2 or L3 or L5) is lower
This correlation indicates that all the complexes follow than that of FeIII(hydroxamic acid), i.e. that Hϩ is more
the same dissociation mechanism from salicylate coordi- tightly bound in our series.
nation. The protonation of the transition state is reasonably
constant for this series. This implies that the hydroxy group Linear Free Energy Relationships between k1Ј and K1Ј
is not implied in the transition state since its pKa varies over
The strong structural effect for the dissociation rate con-
stants and the small dependence for the formation rate con-
stants suggests a linear free energy relationships between
ln k1Ј and lnK1Ј (K1Ј has been calculated from k1Ј/kϪ1Ј)
[Equation (29)], where kϪ1Ј (kϪ1) is the formation rate con-
stant when the active iron species is FeOH2ϩ (Fe3ϩ). The
theoretical basis for this correlation has been presented by
Caudle and Crumbliss.[6]
ln k1Ј ϭ ln K1Ј ϩ ln kϪ1Ј
(29)
The plot lnk1Ј vs. lnK1Ј shown in Figure 13 (a) indicates
a reasonably linear relation for the five complexes studied
with a slope of 1.1 in agreement with the theoretical value
of 1. This high degree of correlation suggests a similar
mechanism in the final step of the dissociation. This implies
a
late transition state where the cleavage of the
Figure 11. lnk1 vs. lnkϪ1Ј for the final step dissociation of FeLn
complexes; data were measured in aqueous 2.0  NaClO4/HClO4
at 25 °C
metalϪligand bonds are largely moved forward. The plot
Figure 12. Transition states for proton-dependent and proton-independent dissociation of bidentate FeϪLn complexes
1562 www.eurjic.org
 2004 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Eur. J. Inorg. Chem. 2004, 1552Ϫ1565

FeIII Complexes with Tripodal Ligands as Siderophore Models
FULL PAPER
Figure 13. (a) ln k1Ј vs. ln K1Ј for proton-independent dissociation of bidentate FeϪLn complexes; (b) ln k1 vs. ln K1 for proton-dependent
dissociation of bidentate FeϪLn complexes; data were measured in aqueous 2.0  NaClO4/HClO4 at 25 °C
ln k1 vs. ln K1 (KЈ1 calculated from K1/KFe) also exhibits a coordination shell. The rate constants may be described ac-
reasonable correlation between data points for L1, L2, L3, cording to the Eigen-Wilkins mechanism[31] with the fast
L5 with a slope of 1.1 (Figure 13, b) (a better correlation formation step of an outer-sphere complex (Kos) followed
was obtained for the series L1, L2, L3), as the correlation by a rate-limiting step involving the water substitution of
related to Equation (29) occurs. The proposed transition the monohydroxylated FeIII species (kex) [Equation (30)],
states are described in Figure 12.
where S is a statistical factor to correct for the composition
of the outer solvation shell.[6]
Formation Kinetics
The rate-determining step is the coordination of FeIII
with one arm of the ligand through the salicylate coordi-
nation followed by a rapid coordination with a second arm
of the ligand leading to the [FeLH5]nϩ species. The spectra
k
Ϫ1Ј ϭ S Kos kex
(30)
The rate constant kϪ1Ј was found to vary by a factor of
recorded during the formation of the complex using diode about 5 for the six tripodal ligands studied. This result sug-
array device clearly show that coordination occurs through gests charge effects of the ligand since increasing values for
4ϩ
the 8-hydroxyquinoline arm for the mixed ligands L1 and
k
Ϫ1Ј were observed on going from L5H74ϩ, L0H7
to
2ϩ
ϩ
L2.
L1H73ϩ, L4H73ϩ, L2H7 and L3H7 corresponding to a
Ligand-substitution reactions at an iron center have been decreasing number of positive charges located on the N
assumed to be controlled by water exchange from the inner anchor and pyridine nitrogen atoms. The role of electro-
Eur. J. Inorg. Chem. 2004, 1552Ϫ1565
www.eurjic.org
 2004 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1563

´
G. Serratrice, F. Biaso, F. Thomas, C. Beguin
FULL PAPER
statics in charge effect concerning complexation reactions (three successive bidentate moieties). The dissociation of the
with hydroxamate ligands has been largely discussed by first arm (from a hexadentate to a tetradentate complex)
Crumbliss and al.[6,32] with discussion involving the Fuoss is in all cases observed but too fast to be measured. The
equation and the Debye-Hückel interionic potential involv- dissociation then follows a process going from the tetraden-
ing aЈ, the distance between the center of the positive charge tate complex with the originally di-coordination of the two
of the metallic cation and the center of the charge (positive concerned arms to the tetradentate bis(salicylate) complex.
or negative) on the ligand.[33Ϫ36] (see Supporting Infor- During this process, bond cleavage of FeϪNpyr or FeϪO
mation, S19). The calculated Kos values are, from these con- with protonation of N or O, followed by proton-indepen-
siderations, 0.024 Ϫ1 for L0 and L5 (ϩ2, ϩ4), 0.039 Ϫ1 dent FeϪO(ϭC) bond formation to close the salicylate ring
for L1 and L4 (ϩ2, ϩ3), 0.062 Ϫ1 for L2 (ϩ2, ϩ2) and could be characterized in most cases. The comparison of
0.100 Ϫ1 for L3 (ϩ2, ϩ1) and were found to exhibit a good these reaction rate constants reflects the influence of chelate
correlation with the kϪ1Ј rate constants determined for li- donor strength (catechol or pyridinophenol vs. 8-hydroxy-
4ϩ
gands with an 8-hydroxyquinoline group (for L0H7
,
quinoline) and the rate of rotation of the amide bond be-
L1H73ϩ, L2H72ϩ, kϪ1Ј ϭ 789, 1320, 2240 Ϫ1 Ϫ1, respec- tween the anchor and the arm in question. The rate con-
s
tively) (Figure S20). This indicates that the complex forma- stants for the leaving of the ultimate salicylate arm indicate
tion follows the Eigen-Wilkins model. By taking a water a proton-dependent (k1) and a proton-independent (kЈ1) re-
[37,38]
exchange rate constant of 1.2 ϫ 105 sϪ1
in action. The rate constants k1 and k1Ј span over a range of
Equation (30), an S value of 2/7 can be deduced for several order of magnitude: five orders for k1 (184 to 0.0012
3ϩ
2ϩ
L0H74ϩ, L1H7 and L2H7 which is a reasonable statisti- ϫ 10Ϫ3 Ϫ1
cal correction for solvent shell composition for this series 0.012 ϫ 10Ϫ3
of ligands with similar structures.[1,32]
the dissociation rate of the last step is evidenced for the 8-
s
Ϫ1) and three and a half orders for kЈ1 (22 to
s
Ϫ1). The role of electrostatics in control of
Some comments for the other terms of the series can be hydroxyquinoline moieties. Interesting correlations could be
made. For L3H7ϩ, complex data point of the correlation obtained which yielded information on the transition states
that kϪ1Ј/Kos falls slightly below the line defined by the for the two mechanisms.
L0H74ϩ, L1H73ϩ and L2H72ϩ complexes suggesting that Kos
From thermodynamic and kinetic studies on aqueous
has probably been overestimated. A lower aЈ value is solutions of such FeIII complexes, we believe that analysis
reasonable owing to the smaller catechol ring size in com- of rates of FeϪN or FeϪO bond cleavage of the species
3ϩ
parison with an 8-hydroxyquinoline ring. For L4H7 (C and the proposed structures for the transition states will
anchor) the kϪ1Ј rate constant (1750 Ϫ1 Ϫ1) is intermedi- help in the designing of new ligands for appropriate ex-
s
ate between those for L1H73ϩ and L2H72ϩ. This may be due change of FeIII in biological systems.
3ϩ
to a larger bulkiness of L4H7 leading to a higher value
for aЈ and hence a higher value for Kos in comparison with
L1H73ϩ. The tripodal anchor contains one additional CH2
and does not allow hydrogen bonding with the carbonyl
oxygen atoms similar to ligands with a Tren anchor, thus
Experimental Section
Materials: All commercial reagents were of the highest purity grade
and were used without further purification. Iron() stock solutions
were prepared by dissolving the appropriate amount of ferric per-
chlorate hydrate (Aldrich) in standardized HClO4 solution. The
solutions were standardized for ferric ion concentration spectro-
photometrically by using a molar extinction coefficient of 4160
Ϫ1·cmϪ1 at 240 nm.[39] The ligands were synthesized according to
procedures previously described (L0,[17a] L1,[20] L2,[20] L3,[15] L4,[40]
3ϩ
3ϩ
increasing the bulkiness of L4H7 with respect to L1H7
.
4ϩ
For L5H7 the kϪ1Ј rate constant (514 Ϫ1 Ϫ1) is lower
s
4ϩ
than that for L0H7 (789 Ϫ1 sϪ1). A possible explanation
could be that the entry of this ligand in the second coordi-
nation sphere of FeIII is less favored owing to the flexibility
of the binding cavity that allows rotation of the pyridine
ring relative to the phenol ring and leads consequently to a L5[18]).
smaller S value.
Kinetics Studies: Kinetic measurements were performed with a
KINSPEC UV (BIO-LOGIC Company, Claix, France) stopped-
flow spectrophotometer equipped with a diode array detector (J &
M) and connected to a microcomputer. The reported rate constants
are the average of about 6 repeat determinations (standard devi-
Conclusion
This work is the first report on the acid dissociation (and ation in the range 1Ϫ2%). The kinetic data were treated on-line
with the commercial BIO-KINE program (BIO-LOGIC Company,
Claix, France). The ionic strength was fixed at I ϭ 2  (NaClO4,
HClO4) due to the Hϩ concentrations which were up to 1 . This
allowed comparisons with literature data. Formation kinetics were
carried out under pseudo-first-order conditions at 25 °C with FeIII
in excess with respect to the ligand. The FeIII concentration
spanned the range 5 ϫ 10Ϫ3 to 3 ϫ 10Ϫ3  for each Hϩ concen-
tration which was over the range 0.03 to 0.1 . A solution contain-
ing FeIII and Hϩ and a solution containing the ligand (1 ϫ 10Ϫ4
) at the same ionic strength were mixed on the stopped-flow ap-
formation) kinetics of iron() complexes with a series of
tripodal ligands, four homopodates [with two types of
anchor (tren or C-anchor)], three types of arm (catechol or
8-hydroxyquinoline or pyridinophenol) and two heteropod-
ates (mixing catechol or 8-hydroxyquinoline).
Analysis of kinetic data shows that there is a common
mechanism in all the series when the ligand structure is va-
ried. The mechanism of the dissociation is more complex
than the successive departure of the three arms around the
iron() as observed in the literature for linear ligands[1] paratus. In each case, first order kinetics were observed. The acid
1564
 2004 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.eurjic.org
Eur. J. Inorg. Chem. 2004, 1552Ϫ1565

FeIII Complexes with Tripodal Ligands as Siderophore Models
FULL PAPER
[17] [17a]
hydrolysis kinetics of the FeIIIϪLn complexes were studied under
pseudo-first-order conditions in the presence an of excess protons
([Hϩ] range 0.01 to 1 ) at 25 °C. The initial pH of the FeIIIϪLn
solution (ca. 1 ϫ 10Ϫ4 ) was ca. 5.
´
`
P. Baret, C. Beguin, H. Boukhalfa, C. Caris, J.-P. Laulhere,
J.-L. Pierre, G. Serratrice, J. Am. Chem. Soc. 1995, 117,
9760Ϫ9761. [17b] G. Serratrice, H. Boukhalfa, C. Beguin, P. Ba-
´
ret, C. Caris, J.-L. Pierre, Inorg. Chem. 1997, 36, 3898Ϫ3910.
P. Baret, C. Beguin, G. Gellon, J.-L. Pierre, G. Serratrice, F.
Thomas, J.-P. Laulhere, E. Saint-Aman, Eur. J. Inorg. Chem.
[18]
[19]
´
`
2000, 1219Ϫ1227.
M. Meyer, J. R. Telford, S. M. Cohen, D. J. White, J. Xu, K.
N. Raymond, J. Am. Chem. Soc. 1997, 119, 10093Ϫ10103; S.
M. Cohen, B. O’Sullivan, K. N. Raymond, Inorg Chem. 2000,
96, 4339Ϫ4346.
A.-M. Albrecht, S. Blanc, F. Biaso, F. Thomas, P. Baret, G.
Gellon, J.-L. Pierre, G. Serratrice, Eur. J. Inorg. Chem. 2003,
2596Ϫ2605.
Acknowledgments
The authors thank Professor J.-L. Pierre (LEDSS) for his interest
in these studies and Professor P. Baret for managing the syntheses
of the ligands Ln.
[20]
[1]
[21]
[22]
´
A.-M. Albrecht-Gary, A. L. Crumbliss, Met. Ions Biol. Syst.
F. Biaso, Thesis, Universite Joseph Fourier, Grenoble, 2002.
1998, 35, 243Ϫ327.
B. F. Matzanke, G. Müller-Matzanke and K. N. Raymond in
D. Imbert, P. Baret, D. Gaude, I. Gautier-Luneau, G. Gellon,
F. Thomas, G. Serratrice, J.-L. Pierre, Chem. Eur. J. 2002, 8,
1091Ϫ1100.
[2]
Physical Bioinorganic Chemistry Series: Iron Carriers and Iron
[23]
[24]
[25]
[26]
Proteins (Ed.: T. M. Loehr), VCH, New York, 1989, p.1.
J. R. Telford, K. N. Raymond in Comprehensive Supramolecu-
D. Imbert, F. Thomas, P. Baret, G. Serratrice, D. Gaude, J.-L.
Pierre, J.-P. Laulhere, New J. Chem. 2000, 24, 281Ϫ288.
C.-W. Lee, D. J. Ecker, K. N. Raymond, J. Am. Chem. Soc.
[3]
`
lar Chemistry (Eds.: J. M. Lehn, G. W. Gokel), Pergamon Press,
London, 1996, p. 245.
R. R. Crichton in Inorganic Biochemistry of Iron Metabolism,
1985, 107, 6920Ϫ6923.
[4]
R. M. Milburn, W. C. Vosburgh, J. Am. Chem. Soc. 1955, 77,
1352Ϫ1355.
A. Shanzer, S. Libman, S. Lifson, Pure Appl. Chem. 1992, 64,
John Wiley & Sons, Chichester, 2001.
[5] [5a]
´
J.-F. Briat, S. Lobreaux in Metals Ions in Biological Systems
(Eds.: A. Sigel, H. Sigel), Marcel Dekker, Inc., New York, 1998,
1421Ϫ1428.
[5b]
[27]
[28]
vol. 35, chapter 14, p. 563;
R. R. Crichton, R. J. Ward,
D. Chapon, G. Serratrice, unpublished results.
G. Serratrice, P. Baret, H. Boukhalfa, I. Gautier-Luneau, D.
Luneau, J.-L. Pierre, Inorg. Chem. 1999, 38, 840Ϫ841.
A. L. Crumbliss, in CRC Handbook of Microbial Iron Chelates
(Ed.: G. Winkelman), CRC Press, Boca Raton, Fl, 1991, chap-
ter 7.
L. E. Asher, E. Deutsch, Inorg. Chem. 1973, 12, 1774Ϫ1778.
M. Eigen, R. G Wilkins, ‘‘Mechanism of Inorganic Reactions’’,
Adv. Chem. Ser. 1965, 55.
J. I. Wirgau, I. Spasojevic, H. Boukhalfa, I. Batinic-Haberle,
A. C. Crumbliss, Inorg. Chem. 2002, 41, 1464Ϫ1473.
R. M. Fuoss, J. Am. Chem. Soc. 1958, 80, 5059Ϫ5061.
M. Z. Eigen, Phys. Chem. 1954, 1, 176Ϫ185.
C.-T. Lin, D. B. Rorabacher, Inorg. Chem. 1973, 12,
2402Ϫ2410.
N. J. Hair, J. K. Beattie, Inorg. Chem. 1977, 16, 245Ϫ250.
M. Grant, R. B. Jordan, Inorg. Chem. 1981, 20, 55Ϫ60.
T. W. Swaddle, A. E. Merbach, Inorg. Chem. 1981, 20,
4212Ϫ4216.
chapter 16, p. 633; [5c] G. S. Tilbrook, R. C. Hider, chapter 18,
p. 691.
[6]
[7]
[8]
[9]
[29]
M. T. Caudle, A. L. Crumbliss, Inorg. Chem. 1994, 33,
4077Ϫ4085.
M. T. Caudle, L. P. Cogswell, A. L. Crumbliss, Inorg. Chem.
1994, 33, 4759Ϫ4773.
B. Monzyk, A. L. Crumbliss, J. Am. Chem. Soc. 1982, 104,
4921Ϫ4929.
M. Birus, Z. Bradic, G. Krznaric, N. Kujundzic, M. Pribanic,
P. C. Wilkins, R. G. Wilkins, Inorg. Chem. 1987, 26,
1000Ϫ1005.
H. Boukhalfa, A. L. Crumbliss, Inorg. Chem. 2000, 39,
4318Ϫ4331.
H. Boukhalfa, T. J. Brickman, S. K. Armstrong, A. L. Crum-
bliss, Inorg. Chem. 2000, 39, 5591Ϫ5602.
Y. Hara, L. Shen, A. Tsubouchi, M. Akiyama, Inorg. Chem.
2000, 39, 5074Ϫ5082.
[30]
[31]
[32]
[33]
[34]
[35]
[10]
[11]
[12]
[36]
[37]
[38]
[13]
[14]
Y. Hara, M. Akiyama, J. Am. Chem. Soc. 2001, 39, 7247Ϫ7256.
S. J. Rodgers, C. W. Lee, C. Y. Ng, K. N. Raymond, Inorg.
Chem. 1987, 26, 1622Ϫ1625.
[39]
[40]
R. Bastian, R. Weberling, F. Palilla, Anal. Chem. 1956, 28,
459Ϫ462.
[15]
[16]
´
´
F. Thomas, C. Beguin, J.-L. Pierre, G. Serratrice, Inorg. Chim.
D. Imbert, Thesis, Universite Joseph Fourier, Grenoble, 2000.
Acta 1999, 291, 148Ϫ157.
C. Y. Ng, S. J. Rodgers, K. N. Raymond, Inorg. Chem. 1989,
28, 2062Ϫ2066.
Received June 25, 2004
Early View Article
Published Online March 5, 2004
Eur. J. Inorg. Chem. 2004, 1552Ϫ1565
www.eurjic.org
 2004 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1565