An iron reservoir model based on ferrichrome: Iron(III)-binding and metal(III)-exchange properties of tripodal monotopic and ditopic hydroxamate ligands with an L-alanyl-L-alanyl-N-hydroxy-β-alanyl sequence

Article abstract of DOI:10.1021/ja003251g

To gain knowledge about biological iron mobilization, tripodal monotopic and ditopic hydroxamate ligands (1 and 2) are prepared, and their iron-chelating properties are investigated. Ligands 1 and 2 contain three Ala-Ala-β-(HO)Ala units and three [Ala-Ala

Full text of DOI:10.1021/ja003251g

J. Am. Chem. Soc. 2001, 123, 7247-7256
7247
An Iron Reservoir Model Based on Ferrichrome: Iron(III)-Binding
and Metal(III)-Exchange Properties of Tripodal Monotopic and
Ditopic Hydroxamate Ligands with an
L-Alanyl-L-alanyl-N-hydroxy-â-alanyl Sequence
Yukihiro Hara and Masayasu Akiyama*
Contribution from the Department of Applied Chemistry, Tokyo UniVersity of Agriculture and Technology,
Koganei, Tokyo 184-8588, Japan
ReceiVed September 1, 2000
Abstract: To gain knowledge about biological iron mobilization, tripodal monotopic and ditopic hydroxamate
ligands (1 and 2) are prepared, and their iron-chelating properties are investigated. Ligands 1 and 2 contain
three Ala-Ala-â-(HO)Ala units and three [Ala-Ala-â-(HO)Ala]2 units connected with tris(alanylaminoethyl)-
amine, respectively, and form six-coordinate octahedral complexes with iron(III) in aqueous solution. Ligand
1 and 1 equiv of iron give Fe-1, and ligand 2 and 1 or 2 equiv of iron produce Fe1-2, or Fe2-2. These complexes
exhibit absorptions at λmax 425 nm of ꢀ 2800-3000/Fe, characteristic of tris(hydroxamato)iron(III) complexes,
and preferentially assume the ∆-cis configuration. Loading of Fe(III) on 1, 2, and M(III)-loaded ligands (M-1
and M1-2, M ) Al, Ga, In) with ammonium ferric oxalate at pH 5.4 is performed, and the second-order rate
constants of loading with respect to Fe(III) and the ligand or M(III)-loaded ligands are determined. The rates
of loading of Fe(III) on M-1 increase in the order Al-1 < Ga-1 < In-1, and those on M1-2 in the order Al1-2
< Ga1-2 < Fe1-2 < In1-2, indicating that the dissociation tendency of M(III) ions from the hydroxamate
ligand is an important factor. The iron complexes formed with 2 are subjected to an iron removal reaction
with excess EDTA in aqueous pH 5.4 solution at 25.0 °C, and the collected data are analyzed by curve-fitting
using appropriate first-order kinetic equations, providing the rate constants for the upper site and the lower
site of 2. Similar analysis for FeM-2 affords removal rate constants for Feup-2, Mup-2, and Felow-2, and the
iron residence probability at each site. The protonation constants of the hydroxamate groups for 1 and 2 (pK1,
pK2, pK3, and pK1, pK2 ..., pK6) are determined, and the proton-independent stability constants for Fe-1, the
upper site of Fe2-2, and the lower site of Fe1-2 are 1028, 1029, and 1028.5, respectively.
Introduction
iron transport and storage processes requires a profound know-
ledge of iron mobilization by ligand-metal exchange. In this
respect, the design and synthesis of artificial iron(III)-storing
compounds, in particular, multitopic iron-binding compounds,
and evaluation of their iron-binding properties would help us
to better understand the intricate processes of biological iron
mobilization.1
We chose ferrichrome as a prototype for designing iron-
storing compounds. Ferrichrome is the iron(III) complex of a
representative exocyclic tripodal hydroxamate siderophore,9,10
and its characteristic properties have been elucidated through
studies of microbial iron transport,9,10 X-ray analysis,13,14 NMR
Iron is an essential element for almost all living organisms,
but its availability is limited due to the formation of insoluble
ferric oxide polymers.1 In response, organisms have developed
efficient systems for iron capture, storage, and release.1-3 In
mammals, iron is absorbed from digested food and carried by
transferrin to various locations for utilization4,5 or to ferritin
for storage.6-8 Microorganisms use siderophores (low-molecular
weight, chelating compounds) to sequester iron from the en-
vironment, and siderophore-iron complexes are transported into
cells, where the iron is released.9-12 Description of biological
(1) (a) Crichton, R. R. Inorganic Biochemistry of Iron Metabolism; Ellis
Horwood: Chichester, 1991. (b) Crichton, R. R.; Ward, R. J. Analyst 1995,
120, 693-697.
(2) Fatemi, S. J. A.; Kadir, F. H. A.; Williamson, D. J.; Moore, G. R.
AdV. Inorg. Chem. 1991, 36, pp 409-448.
(9) (a) Neilands, J. B. In Microbial Iron Metabolism; Neilands, J. B.,
Ed.; Academic Press: New York, 1974; pp 3-34. (b) Bagg, A.; Neilands,
J. B. Microbiol. ReV. 1987, 51, 509-518.
(10) (a) Emery, T. In Metal Ions in Biological Systems; Sigel, H., Ed;
Marcel Dekker: New York, 1978; Vol. 7, pp 77-126. (b) Emery, T. In
Iron Transport in Microbes, Plants and Animals; Winkelmann, G., van der
Helm, D., Neilands, J. B., Eds.; VCH Publishers: Weinheim, 1987; pp 235-
250.
(11) (a) Raymond, K. N.; Mu¨ller, G.; Matzanke, B. F. Top. Curr. Chem.
1984, 123, 49-102. (b) Matzanke, B. F.; Mu¨ller-Matzanke, G.; Raymond,
K. N. In Iron Carriers and Iron Proteins; Loehr, T. M., Ed.; Physical
Bioinorganic Chemistry Series, Vol. 5; VCH Publishers: New York, 1989;
pp 3-121.
(12) Matzanke, B. F. In Encyclopedia of Inorganic Chemistry; King, R.
B., Ed.; John Wiley & Sons: Chichester, 1994; Vol. 4, pp 1915-1932.
(13) Zalkin, A.; Forrester, J. D.; Templeton, D. H. J. Am. Chem. Soc.
1966, 88, 1810-1814.
(3) Kaim, W.; Schwederski, B. Bioinorganic Chemistry: Inorganic
Elements in the Chemistry of Life; John Wiley & Sons: Chichester, 1994.
(4) Baker, E. N.; Lindley, P. F. J. Inorg. Biochem. 1992, 47, 147-160.
(5) Aisen, P. In Metal Ions in Biological Systems; Sigel, A., Sigel, H.,
Eds.; Marcel Dekker: New York, 1998; Vol. 35, pp 585-632.
(6) Theil, E. C.; Aisen, P. In Iron Transport in Microbes, Plants and
Animals; Winkelmann, G., van der Helm, D., Neilands, J. B., Eds.; VCH
Publishers: Weinheim, 1987; pp 491-520.
(7) Harrison, P. M.; Andrews, S. C.; Artymiuk, P. J.; Ford, G. C.; Guest,
J. R.; Hirzmann, J.; Lowson, D. M.; Livingstone, J. C.; Smith, J. M. A.;
Treffry, A.; Yewdall, S. J. AdV. Inorg. Chem. 1991, 36, 449-486.
(8) Chasteen, N. D. In Metal Ions in Biological Systems; Sigel, A., Sigel,
H., Eds.; Marcel Dekker: New York, 1998; Vol. 35, pp 479-514.
10.1021/ja003251g CCC: $20.00 © 2001 American Chemical Society
Published on Web 07/10/2001

7248 J. Am. Chem. Soc., Vol. 123, No. 30, 2001
Hara and Akiyama
spectroscopy,15 iron-transport experiments,9a and iron-exchange
kinetics.16 Recently, a ferrichrome family has even been sug-
gested as an iron-storing vehicle for a microorganism.12 Fur-
thermore, a variety of tripodal hydroxamate ligands have been
synthesized as ferrichrome models17 and their biological activi-
ties investigated.18 Moreover, potential applications have been
attempted with certain trihydroxamates, for example, in su-
pramolecular chemistry.19-22
Chart 1
Although there have been extensive studies of iron(III)
complexation with monotopic hydroxamate ligands,11,23,24 only
limited information is available about multitopic iron complex-
ation.19,21 Dihydroxamic acids, represented by siderophore
rhodotorulic acid, are known to form binuclear iron complexes,25
but their manner of fragmentary decomplexation makes them
unsuitable for iron-storage experiments.26,27
We have synthesized a number of amino acid- and peptide-
based monotopic hydroxamate ligands to study iron complex-
ation.28-31 Multitopic hydroxamate ligands can be synthesized
by the extension of the framework of these monotopic ligands.
Ditopic ligands, for example, present an interesting problem of
iron mobilization within a ligand. It is important to study this
for obtaining information related to intracellular iron migration
behavior which is not yet understood well.1 We describe here
the synthesis of tripodal monotopic and ditopic hydroxamate
ligands (1 and 2) and the formation and kinetic and thermody-
namic properties of their metal-ion complexes. The results are
compared to those reported for linear and cyclic ligands 3 and
428 (Chart 1). In related studies, Shanzer’s group has reported
tripodal, ditopic hydroxamate ligands (for example, 5 in Chart
1) and observed the iron complex formation of a “triple-helical”
structure;19,21 however, further iron-mobilization studies remain
to be done.
To explore the potential of ditopic ligands for an iron reservoir
model, we performed iron-loading and -removal reactions using
even M(III)-preloaded ligands, a study that should help to
elucidate incoming Fe(III) behavior when in competition with
M(III), but which has rarely been carried out.32,33 The group-
13 M(III) ions (M ) Al, Ga, and In) are used as surrogates of
Fe(III). They are diamagnetic and have similar charge/radius
ratios, produce similar octahedral complexes, and exhibit
different affinities for given ligands.15,34-36 It is also convenient
that their hydroxamato complexes do not absorb in the visible
region.
Iron(III) complexes of 1 and 2 were prepared by pH-
dependent transformation from bis(hydroxamato)iron to tris-
(hydroxamato)iron complexes. The latter were characterized by
UV-vis and CD spectroscopy, along with ESIMS. Tris com-
plexes were also formed by iron loading on ligands or on M(III)-
preloaded ligands using ammonium ferric oxalate as a soluble
and exchangeable iron source at near-neutral pH. The iron-
removal reaction of the complexes of 2 with excess EDTA
allowed us to characterize the iron-bound states and distribu-
tions. With the ligand protonation constants and the iron(III)
complex stability constants obtained, we discuss briefly a model
for biological iron mobilization.
(14) van der Helm, D.; Baker, J. R.; Eng-Wilmont, D. L.; Hossain, M.
B.; Loghry, R. A. J. Am. Chem. Soc. 1980, 102, 4224-4231.
(15) (a) Llina´s, M.; Klein, P. M.; Neilands, J. B. J. Mol. Biol. 1970, 52,
399-414. (b) Llina´s, M.; Klein, P. M.; Neilands, J. B. J. Mol. Biol. 1972,
68, 265-284.
(16) Tufano, T. P.; Raymond, K. N. J. Am. Chem. Soc. 1981, 103, 6617-
6624.
(17) (a) Naegeli, H.-U.; Keller-Schierlein, W. HelV. Chim. Acta 1978,
61, 2088-2095. (b) Olsen, R. K.; Ramasamy, K. J. Org. Chem. 1985, 50,
2264-2271. (c) Akiyama, M.; Katoh, A.; Mutoh, T. J. Org. Chem. 1988,
53, 6089-6094.
(18) (a) Lee, B. H.; Miller, M. J. Prody, C. A.; Neilands, J. B. J. Med.
Chem. 1985, 28, 317-323, 323-327. (b) Tor, Y.; Libman, J.; Shanzer, A.
J. Am. Chem. Soc. 1987, 109, 6518-6519. (c) Ng, C. Y.; Rodgers, S. J.;
Raymond, K. N. Inorg. Chem. 1989, 28, 2062-2066. (d) Katoh, A.;
Akiyama, M. J. Chem. Soc., Perkin Trans. 1 1991, 1839-1842. (e) Dayan,
I.; Libman, J.; Agi, Y.; Shanzer, A. Inorg. Chem. 1993, 32, 1467-1475.
(f) Esteves, M. A.; Vaz, M. C. T.; Gonc¸alves, M. L. S. S.; Farkas, E.; Santos,
M. A. J. Chem. Soc., Dalton Trans. 1995, 2565-2573. (g) Ramurthy, S.;
Miller, M. J. J. Org. Chem. 1996, 61, 4120-4124.
(19) Libman, J.; Tor, Y.; Shanzer, A. J. Am. Chem. Soc. 1987, 109,
5880-5881.
(20) Spasojevic´, I.; Batinic´-Haberle, I.; Choo, P. L.; Crumbliss, A. L. J.
Am. Chem. Soc. 1994, 116, 5714-5721.
(21) Blanc, S.; Yakirevitch, P.; Leize, E.; Meyer, M.; Libman, J.; Van
Dorsselaer, A.; Albrecht-Gary, A.-M.; Shanzer, A. J. Am. Chem. Soc. 1997,
119, 4934-4944.
(22) Akiyama, M.; Katoh, A.; Kato, J.; Takahashi, K.; Hattori, K. Chem.
Lett. 1991, 1189-1192.
(23) (a) Crumbliss, A. L. In Handbook of Microbial Iron Chelates;
Winkelmann, G., Ed.; CRC Press: Boca Raton, FL, 1991; pp 177-233.
(b) Albrecht-Gary, A.-M.; Crumbliss, A. L. In Metal Ions in Biological
Systems; Sigel, A., Sigel, H., Eds.; Marcel Dekker: New York, 1998; Vol.
35, pp 239-328.
(24) Birusˇ, M.; Kujundzˇic, N.; Pribanic´, M. Prog. React. Kinet. 1993,
18, 171-271.
Results
(25) Carrano, C. J.; Raymond, K. N. J. Am. Chem. Soc. 1978, 100, 537-
Hydroxamate Ligands. Tris(alanylaminoethyl)amine31 was
combined with a peptide strand containing one or two hydrox-
amate groups to anchor three strands together. Scheme 1
5374.
(26) Barclay, S. J.; Huynh, B. H.; Raymond, K. N. Inorg. Chem. 1984,
23, 2011-2018.
(27) (a) Caudle, M. T.; Cogswell, L. P., III; Crumbliss, A. L. Inorg.
Chem. 1994, 33, 4759-4773. (b) Spasojevic´, I.; Boukhalfa, H.; Stevens,
R. D.; Crumbliss, A. L. Inorg. Chem. 2001, 40, 49-58.
(28) Hara, Y.; Akiyama, M. Inorg. Chem. 1996, 35, 5173-5180.
(29) (a) Shimizu, K.; Hasegawa, M.; Akiyama, M. Bull. Chem. Soc. Jpn.
1984, 57, 495-499. (b) Akiyama, M.; Katoh, A.; Iijima, M.; Takagi, T.;
Natori, K.; Kojima, T. Bull. Chem. Soc. Jpn. 1992, 65, 1356-1361. (c)
Akiyama, M.; Katoh, A.; Mutsui, Y.; Watanabe, Y.; Umemoto, K. Chem.
Lett. 1996, 915-916.
(32) Birusˇ, M.; Bradic´, Z.; Krznaric´, G.; Kujundzˇic, N.; Pribanic´, M.;
Wilkins, P. C.; Wilkins, R. G. Inorg. Chem. 1987, 26, 1000-1005.
(33) Ga(III) in Ga-deferriferrichrome A was replaced by the use of a
large excess of Fe(III)-nitrilotriacetate; Emery, T.; Hoffer, P. B. J. Nucl.
Med. 1980, 21, 935-939.
(34) Anderegg, G.; L’Eplattenier, F.; Schwarzenbach, G. HelV. Chim.
Acta 1963, 46, 1400-1408, 1409-1422.
(35) Borgias, B.; Hugi, A. D.; Raymond, K. N. Inorg. Chem. 1989, 28,
3538-3545.
(36) Evers, A.; Hancock, R. D.; Martell, A. E.; Motekaitis, R. J. Inorg.
Chem. 1989, 28, 2189-2195.
(30) Shimizu, K.; Akiyama, M. J. Chem. Soc., Chem. Commun. 1985,
183-184.
(31) Akiyama, M.; Hara, Y.; Gunji, H. Chem. Lett. 1995, 225-226.

Model for Biological Iron Mobilization
J. Am. Chem. Soc., Vol. 123, No. 30, 2001 7249
Scheme 1a
Figure 1. Absorption (ꢀ/M-1 cm-1) at 425 nm vs pH in water at
25 °C.
a Reagents and conditions: i. EDC, HOBt, NMM, -10 °C; ii, TFA
in CH2Cl2, 0 °C; iii, AcONSu, NMM, -10 °C; iv, H2, 10% Pd/C in
MeOH; v, BOP reagent, HOBt, Et3N, 0 °C.
(HL)+,39,42 with λmax at 465 nm. The Fe(HL)+ complexes were
transformed to tris(hydroxamato)iron(III) complexes, Fe(L), by
neutralization to pH 7: Fe(L) for 1, and Fe(L-H3L) and Fe2-
(L-L) for 2, each of which is designated as Fe-1, Fe1-2, or Fe2-
2, respectively. Their UV-vis spectra showed λmax at 425 nm
with values of ꢀ 2800 for Fe-1 and 3000 M-1 cm-1/Fe(III) for
Fe1-2 and Fe2-2, typical of Fe(L) complexes.34 Iron in Fe1-2
resides mostly (93%) at a near tren site of 2 (see Table 3, entry
4b).
Plots of the absorbance at 425 nm versus pH shown in Figure
1 exhibit a plateau region, where iron(III) exists essentially as
Fe(L). When an Fe(L) solution was gradually acidified, the
absorption maximum shifted to the red, and its intensity
decreased.39 Isosbestic points were observed for Fe-1 at 447
nm, and for Fe1-2 and Fe2-2 at 465 nm, respectively, which
indicate an equilibrium reaction occurring between Fe(L) and
Fe(HL)+ (eq 2). With Fe2-2, only two species, Fe2(L-L) and
Fe2(HL-L)+, are present in the pH range of 3.1-4.0, judged
from the magnitude of absorbance.
summarizes the synthesis of ligands 1 and 2.37 The protected
form of 1 was synthesized with a Boc-Ala-Ala-â-(BnO)Ala-
OH unit,28 and the protected form of 2 was derived from this
form by use of the same unit. Ligands 1 and 2 were obtained
1
by deprotection and characterized by IR, H NMR, and ele-
mental analysis. They were soluble in DMF, DMSO, and water,
but not in CHCl3 or acetonitrile.
Data for the 1H NMR spectra of 1 and 2 in DMSO-d6
solutions are presented (Table S1). The spectral patterns of 1
and 2 suggest C3 symmetry of the molecules, and individual
signals were assigned using 2D COSY-NOESY techniques.
Each amino acid residue in the ligands and the Ga(III) com-
plexes (vide infra) is numbered as follows:
1: [Ac-Ala1-Ala2-â-(HO)Ala-Ala3-NHCH2CH2]3N
2: [Ac-Ala1-Ala2-â-(HO)Ala1-Ala3-Ala4-â-(HO)Ala2-Ala5-
NHCH2CH2]3N
In particular, the R-protons of Ala2 of 1 and of Ala2 and Ala4
of 2 appeared at about 4.80 ppm with full intensity of 3H and
6H, respectively, which confirmed the presence of the hydrox-
amic acid groups. The temperature coefficients of amide-proton
chemical shifts were found to be in the range of -4 to -5 ×
10-3 ppm deg-1. These are more negative than the value (-3
× 10-3 ppm deg-1) for intramolecularly hydrogen-bonded
amide-protons.15a,38
Fe(L) + H+ h Fe(HL)+, KFe(HL) ) [Fe(HL)+]/[ Fe(L)][H+]
(2)
Aobs ) (AFe(L) - Aobs)/KFe(HL)[H+] + ꢀFe(HL) ct
(3)
The spectral data was analyzed using Schwarzenbach plot
(eq 3),39 as shown for the case of Fe2-2 (Figure 2), and the
KFe(HL) values obtained are listed in Table 4.
Iron(III) Complexes. Iron(III) complex formation of hy-
droxamate groups is expressed as a series of pH-dependent
equilibrium processes (eq 1),39-41 where Fe3+ represents an aqua
iron species and H3L stands for an iron binding site of three
hydroxamate groups. The symbol H3L-H3L is used for 2.
Gallium(III) Complexes. Ga1-2 exhibited an 1H NMR
spectral pattern of C3 symmetry (Table S1). The signals of R-CH
Ala4 shifted 0.12 ppm downfield and of NH Ala5 0.17 ppm
downfield, relative to those protons of 2. Likewise, the signals
of NH Ala4 shifted 0.04 ppm upfield and of R-CH Ala5 0.04
ppm upfield. These protons exhibited cross-peaks with the NH
and â-CH2 protons of the tren moiety by COSY-NOSEY
measurements. The Ga location in Ga1-2 was assigned to a
nearby tren site, namely, the lower binding site. The NH and
R-CH protons of Ala1, Ala2, and Ala3 residues appear little
shifted.
Fe3+ + H3L h Fe(H2L)2+ + H+ h
Fe(HL)+ + 2H+ h Fe(L) + 3H+ (1)
A ferric nitrate solution (1 equiv or 1 or 2 equiv, respectively)
was mixed with 1 or 2, producing an acidic solution which
contained essentially bis(hydroxamato)iron(III) species, Fe-
(37) Abbreviations are as follows: Ala, L-alanine; â-(HO)Ala or â-(BnO)-
Ala, N-hydroxy- or N-(benzyloxy)-â-alanine; Boc, tert-butyloxycarbonyl;
BOP, (benzotriazol-1-yloxy)tris(dimethylamino)phosphonium hexafluoro-
phosphate(V); EDC, 1-ethyl-3-[(3-dimethylamino)propyl]carbodiimide; EDTA,
ethylenediaminetetraacetic acid; HOBt, 1-hydroxybenzotriazole; NMM,
N-methylmorpholine; TFA, trifluoroacetic acid; tren, tris(aminoethyl)amine;
AcOSu, N-acetoxysuccinimide; THF, tetrahydrofuran; DMSO, dimethyl
sulfoxide.
The spectral pattern of Ga-1 was also C3-symmetric. Its
signals for R-CH Ala2 and NH Ala3 shifted 0.11 and 0.14 ppm
downfield, relative to the corresponding protons of 1. This shift
pattern for Ga-1 is similar to that observed for Ga1-2, supporting
the assignment of the location of Ga(III) in 2.
CD Spectra of Iron Complexes. Cotton effects at 365 and
450 nm were observed for the iron(III) complexes of 1 and 2
(Table 1), and the ∆ configuration around the iron center was
assigned to them by comparison with the structure of fer-
richrome, together with the cis configuration by using NMR
data which indicated C3 symmetric structures.13,14 Notable
(38) Ohnishi, M.; Urry, D. W. Biochem. Biophys. Res. Commun. 1967,
36, 194-202.
(39) Schwarzenbach, G.; Schwarzenbach, K. HelV. Chim. Acta 1963, 46,
1390-1400.
(40) Monzyk, B.; Crumbliss, A. L. J. Am. Chem. Soc. 1979, 101, 6203-
6213.
(41) Birusˇ, M.; Bradic´, Z.; Kujundzˇic, N.; Pribanic´, M.; Wilkins, P. C.;
Wilkins, R. G. Inorg. Chem. 1985, 24, 3980-3983.
(42) Caudle, M. T.; Stevens, R. D.; Crumbliss, A. L. Inorg. Chem. 1994,
33, 843-844, 6111-6115.

7250 J. Am. Chem. Soc., Vol. 123, No. 30, 2001
Hara and Akiyama
Table 2. Iron(III) Loading with Tris(oxalato)Iron(III) Complexa
apparent second-order rate constant (k2 app/M-1 s-1
)
2-equiv loadingc
entry ligand
1-equivb loading
first iron
second iron
1a
1b
1c
2a
2b
2c
3a
3b
3c
1
2
1.7 × 103d
1.4 × 103
1.1 × 103e
1.8 × 102
5.4 × 102
3.4 × 103
1.1 × 102
2.6 × 102
4.3 × 103
1.0 × 103
1.0 × 103
Fe1-2
Al-1
Ga-1
In-1
Al1-2
Ga1-2
In1-2
2.1 × 102
2.3 × 102
3.3 × 103
1.5 × 10
8.1 × 10
3.3 × 103
a Iron(III) was loaded in aqueous acetate buffer at 25.0 ( 0.1 °C,
pH 5.4, and ionic strength 0.10 (KCl), using ammonium ferric oxalate.
b The initial concentration of [ligand] and [Fe(III)] ) 6.5 × 10-5 M.
Experimental errors are estimated to be (10%, except for faster
reactions (k2 app/M-1 s-1 ≈ 103), in which they are (20%. c For 2-equiv
loading, [ligand] ) 6.5 × 10-5 M and [Fe(III)] ) 1.3 × 10-4 M. d A
similar value was obtained with 0.10 M K(NO3). e Data were obtained
by the second 1-equiv loading after completion of the first 1-equiv
reaction.
Figure 2. UV-vis spectra of Fe2-2 showing an isosbestic point. The
concentrations of [ligand] and [Fe(III)] are 1.1 × 10-4 M and 2.2 ×
10-4 M, respectively, in water at 25 °C. The insert is a Schwarzenbach
plot for the spectral data.
Table 1. Summary of CD Spectral Dataa
The increase in absorbance of Fe(L) at 425 nm was monitored
in an aqueous acetate buffer at 25 °C and pH 5.4. With 1 and
2 equiv of ferric oxalate, ligands 1, M-1, or M1-2 and 2 or M1-2
were cleanly transformed to Fe-1 or FeM-2 and Fe2-2, respec-
tively, in a matter of minutes, as evidenced by UV-vis spectra.
For the purpose of comparison, the rates of formation of Fe-1,
FeM-2, and Fe2-2 were followed, and the second-order rate
constants with respect to the ferric oxalate and either a free
ligand or a M(III)-preloaded ligand,43 as expressed by eq 5, were
obtained.44 Table 2 summarizes these rate constants.45
Fe(III) complex
band/nm (∆ꢀ)
type
ref
Fe-1
Fe1-2
Fe2-2
365 (+2.3)
365 (+2.9)
365 (+5.9)
365 (+2.7)
365 (+2.3)
365 (+2.8)
360 (+3.0)
360 (+4.4)
380 (+1.3)
375 (+5.1)
360 (-3.7)
445 (-4.0)
450 (-6.3)
450 (-12.0)
450 (-5.8)
450 (-5.9)
450 (-5.3)
445 (-7.5)
445 (-6.8)
445 (-0.76)
460 (-2.6)
465 (+2.4)
∆
∆
∆
∆
∆
∆
∆
∆
∆
∆
Λ
FeAl-2b
FeGa-2b
FeIn-2b
Fe-3
28
28
21
21
14
Fe-4
Fe1-5
Fe2-5
ferrichrome
rate ) k2 app [Fe(C2O4)33-][H3L, or M(L)]
(5)
a Determined in water at 25 °C and pH 7.0. b Prepared by 1-equiv
iron loading on M1-2, where M ) Al, Ga, or In; Fe(III) resides mostly
(75%-79%) at the lower site: See Table 3 for the data.
Absorbance changes versus time for 2-equiv loading on 2
and M1-2 are shown in Figure 3 A, with second-order rate plots
in the insert. Shown in Figure 3 B are absorbance changes in
the early stages of 1- and 2-equiv iron loading, emphasizing
different rates for different-M(III) preloaded ligands (each
portion is depicted in a different time scale).
The k2 app values are different and increase in the order Al-1
< Ga-1 < In-1 with factors of 3 and then 6 based on the rate
of Al-1, and Al1-2 < Ga1-2 < Fe1-2 < In1-2 with factors of
2.5, 4, and 4 based on the rate of Al1-2, respectively.
Characterization of FeM-2 complexes was made by CD
spectroscopy (Table 1), ESIMS, and UV-vis spectra; the last
spectra were the same as that of Fe1-2.
Electrospray Ionization Mass Spectrometry. The spectra
of Fe2-2 showed peaks due to tri-, di-, and monocation species,
and the spectra of FeAl-2 and FeGa-2 exhibited dication peaks.
With In1-2, however, peaks were assignable to the ligand, but
not to In1-2. Under these conditions, small peaks ascribable to
FeIn-2 were detected in the spectra recorded on the + or -
voltage mode. Unlike the previous ditopic case,21 fragmentation
peaks of M-16 that arise from the deoxygenated hydroxamate
moiety were not detected.
features are as follows. (1) The CD intensity (from the top of
the peak to the bottom of the trough) of Fe1-2 is comparable to
that of Fe-3 or Fe-4, and higher than that of Fe-1, the latter
being comparable to ferrichrome.14 (2) The intensity of Fe2-2
is quite high, almost twice as high as that of Fe1-2. (3)
Complexes Fe1-2 and Fe2-2 exhibited higher CD intensity than
those of previously reported complexes, for example, Fe1-5 and
Fe2-5.19,21
Iron Loading with Ferric Oxalate. Initially, the loading
process directly from a ligand (eq 4a) or from an M(III)-
preloaded ligand, M(L), (eq 4b) was investigated using am-
monium tris(oxalato)iron(III). The reactions are as outlined
below, which show the processes from H3L and M(L) via mono-
and bis(hydroxamato) complexes to the tris(hydroxamato)iron-
-
(III) complex, Fe(L),16,23,39 with the final production of 3HC2O4
3-
(4a) and M(C2O4)3 (4b), respectively.
H3L + Fe(C2O4)33- h (H2L)Fe(C2O4)22- + HC2O4- h
(HL)F(C2O4)1- + 2HC2O4- h Fe(L) + 3HC2O4 (4a)
-
where H3L stands for an iron binding site in 1 and 2.
M(L) + Fe(C2O4)33- h (C2O4)-M(L)Fe(C2O4)22- h
(43) Wilkins, R. G. Kinetics and Mechanism of Reactions of Transition
Metal Complexes, 2nd ed.; VCH: Weinheim, 1991.
(44) The second-order reactions are among the most commonly encoun-
tered ones (ref 43, pp 24 and 68-69).
(45) The rate constants for slow second phases were obtained from the
slopes of the plots. The progress of reaction in individual cases showed a
pattern which is very similar to that illustrated in Figure 3A for 2-equiv
loading on In1-2.
(C2O4)22-M(L)F(C2O4)- h Fe(L) + M(C2O4)3 (4b)
3-
where M(L) represents an M(III) ion residing in 1 and 2.

Model for Biological Iron Mobilization
J. Am. Chem. Soc., Vol. 123, No. 30, 2001 7251
Scheme 2
For the analysis of cases 2 and 4, consecutive first-order
kinetic equations are applied.43 In case 2 (Fe2-2), iron at the
upper site is disposed of, but iron at the lower site can only be
removed after the upper site. This situation is analyzed by
applying a consecutive first-order kinetic model, and the overall
removal process can be described by pseudo-first-order rate
constants for the ferric ions at the upper site and the lower site,
k1up and k1low (Scheme 2). The absorbance (At) at time t for the
species in the reaction mixture is expressed by eq 7. The two
binding sites have equal ꢀ values, and the contribution of
Fe(EDTA)- can be neglected for most of the reaction.
At ) absorbance of the reaction mixture composed of
Fe2-2, Fe1-2, and Fe(EDTA)- ) A0 exp(-k1upt) +
1/2A0{k1up/(k1up - k1low)}{exp(-k1lowt) - exp(-k1upt)} (7)
In case 3 (Fe1-2), iron may reside partly at the lower site and
partly at the upper site, and the extent of reaction is expressed
by a parallel first-order rate law (eq 8).43 This equation also
Figure 3. (A) Plots of absorbance at 425 nm vs time for 2-equiv iron
loading on 2 and on M1-2 which produces Fe2-2. The insert shows
second-order rate plots for the 2-equiv loading. The same rate
throughout the reaction is seen for 2 and In1-2, while a break is noted
for Al1-2 or Ga1-2 at the 1-equiv loaded stage, and the corresponding
rate constants are obtained from that slope. (B) Absorbance changes
vs time at the early stages for loading of both 1- and 2-equiv cases.
Note the different time scales.
up
uses the rate constants k1 and k1low, with iron-residing
probability (P) at the lower site.
Extent of reaction (At/A0) )
P exp(-k1lowt) + (1 - P) exp(-k1upt) (8)
For analysis of case 4 (FeM-2), we modify eq 7 to evaluate
a contribution from M(III) ion (M ) Al, Ga, and In), and eq 9
is derived, in which P refers to the lower site. The rate constant
Iron Removal Rates. Subsequently, the metal-bound states
of the formed complexes were examined by the measurement
of iron-removal rates with EDTA. The removal reaction is of a
ligand-exchange type and proceeds through the proton-assisted
formation of a ternary complex of a ligand, Fe(III), and EDTA
and its subsequent breakdown into products, the ligand and
Fe(EDTA)- (eq 6), as discussed by Tufano and Raymond and
later Albrecht-Gary et al.16,46
k1up-M for M(III) at the upper site is obtained in the process of
low-M
the curve fit, but the rate constant k1
is not evaluated.
At/A0 ) (1- P) exp(-k1upt) + 1/2P{k1up-M/(k1
-
up-M
k1low)}{exp(-k1lowt) - exp(-k1up-Mt)} (9)
(ii) Reaction Rates. The pseudo-first-order rate constant was
determined to be 2.7 × 10-2 s-1 for Fe-1 (a corresponding
second-order rate constant is calculated to be 10.4 M-1 s-1) in
the presence of EDTA (2.6 × 10-3 M). This rate is comparable
to the rate of 1.8 × 10-2 s-1 reported for Fe-3.28
Fe(L) + H+ h Fe(HL)+
Fe(HL)+ + H2EDTA2- h (HL)Fe(H2EDTA)- h
H3L + Fe(EDTA)- (6)
In each of the cases 2-4, one of the equations of 7, 8, and
9 was used to generate curve fits for the plotted absorbance
change with respect to time, and each rate constant was obtained.
An excellent fit between experimental data (absorbance changes
A decrease in absorbance of Fe(L) at 425 nm was monitored
under pseudo-first-order conditions with a 20 M excess of EDTA
at pH 5.4 and 25 °C;16,28,46 the data sets of absorbance change
with respect to time were collected and analyzed using the
models described below.
(i) Reaction Models. Four models of first-order kinetics are
used, depending on the types of complexes: a monotopic
complex, Fe-1 (case 1); a fully loaded ditopic complex, Fe2-2
(case 2); a half-loaded ditopic complex, Fe1-2 (case 3); and
mixed binuclear complexes, FeM-2 (case 4).
up
low
vs time) and a calculated curve based on given k1 and k1
values is shown for the Fe2-2 reaction (Figure 4), where the
three calculated curves for Fe2-2, intermediate monoferric
Fe1-2, and the total reaction mixture represent the progress of
the reaction. Table 3 lists the kinetic data and the iron
distribution probability. Allowable deviations in all cases were
small; they were determined by the same curve-fit, applying
slightly different values of k1up, k1low, or P.
Furthermore, we carried out iron-removal reactions for Fe2-2
using different EDTA concentrations and observed a hyperbolic
For case 1 (Fe-1), the pseudo-first-order kinetic equation is
used.
up
low
dependence of both k1 and k1 on the EDTA concentration
(Figure 5). In this case the kinetic data was applied to a
modification of eq 6 for simplicity, as the back reaction from
(46) Albrecht-Gary, A.-M.; Palanche-Passeron, T.; Rochel, N.; Hennard,
C.; Abdallah, M. A. New J. Chem. 1995, 19, 105-113.

7252 J. Am. Chem. Soc., Vol. 123, No. 30, 2001
Hara and Akiyama
Figure 4. A typical curve for absorbance (at 425 nm) changes vs time
up
for iron removal (here from Fe2-2) by EDTA. The rate constants k1
and k1low were obtained from a best fit (bold line) of the experimental
Figure 5. Hyperbolic dependence of the rates of iron removal from
the upper site and lower site of Fe2-2 upon EDTA concentration.
up
data (dots) to eq 7. The iron at the upper site is removed by the k1
constant (solid line), while the iron at the lower site is once “exposed”
and removed by the k1 constant (solid line).
low
Table 4. Average Ligand Protonation Constants and the ꢀ Values,
Monoprotonation Constants, and Stability Constants of Iron(III)
Complexesa
Table 3. Pseudo-First-Order Rate Constants for Displacement of
Iron(III) from Monotopic and Ditopic Complexes and Iron(III)
Residence Percentagesa
Fe complex Fe-1 Fe1-2 Fe2-2 Fe-3b
Fe-4b
ferrichromec
d
pKav
8.99 8.82 9.14 9.14 8.98
2800 3000 6000 2890 2830
upper site
102k1up/s-1
(mol %)c
lower site
103k1low/s-1
(mol %)c
upper site
10k1up-M/s-1
for M(III)upd
ꢀ/M-1 cm-1
2890
0.031
29.1
metal ion
KFe(HL) × 10-3 0.926 1.33 5.01 1.44 2.24 × 103
entry complexb
log K Fe(L)
28.2 28.5e 29.0f 27.4 28.3
4a
4b
5a
5b
5c
Fe2-2
2.1 ( 0.1
4.0 ( 0.2
Fe1-2e
1.7 ( 0.2 (7 ( 3) 3.8 ( 0.2 (93 ( 3)
a In water at 25.0 ( 0.1 °C with ionic strength 0.10 (KCl). The
equilibrium exchange reaction was performed using an equimolar
amount of EDTA (1.3 × 10-4 M): Fe1-2 remained 22% (eq 12a) and
Keq1 ) 3.82 × 106; Fe2-2 remained 35% (eq 12b) and Keq2 ) 1.01 ×
106. b Reference 28. c References 14 and 34. d The protonation constant
of the tertiary nitrogen (pKN), 5.68 for 1 and 5.35 for 2. e Value
corresponds to the lower site of 2. f Value stands for the upper site of
2.
FeAl-2 2.2 ( 0.4 (25 ( 5) 4.2 ( 0.2 (75 ( 5)
1.4 ( 0.4
1.2 ( 0.3
6 ( 1
FeGa-2 3.0 ( 0.3 (21 ( 3) 3.9 ( 0.2 (79 ( 3)
FeIn-2
3 ( 1 (25 ( 5)
4.2 ( 0.2 (75 ( 5)
a In an acetate buffer at 25 ( 0.1 °C and pH 5.4, with 20 M excess
of EDTA (1.3 × 10-3 M), at ionic strength (KCl) ) 0.1. [Fe1-2], [Fe2-
2], or [FeM-2] ) 6.5 × 10-5 M. Allowable deviations for both rate
constants and distribution percentages were determined through the
curve fits and are given by digits with ( sign. b FeM-2, where M )
Al, Ga, or In, denotes a dinuclear complex. c Iron residence percentage
was determined by curve fit of the kinetic data to eq 8 or 9 (see text).
d M(III)up removal rate was obtained by curve fit to eq 9, and M(III)
residence percentage was taken as the same that for the lower-site iron.
e In another run, values obtained were 102k1up/s-1 ) 2.2 ( 0.2 with (6
( 3) mol % for the upper site, and 103k1low/s-1 ) 3.6 ( 0.2 with (94
( 3) mol % for the lower site.
It is difficult to experimentally perform an equilibrium reac-
tion involving only the two species, H3L-H3L and Fe2(L-L),
because the transformation inevitably involves the intervening
monoferric complex, Fe(L-H3L). The proton-independent stabil-
ity constants for Fe2-2, therefore, are described by eqs 11a and
11b. The equation for Fe-1 is omitted to avoid duplication of
monotopic complexation.
the product to the intermediate ternary complex is negligible,
as reported by others and by us.16,46,47 In this way, the formation
constant of the intermediate ternary complex, Kint, and the
maximal first-order rate constant for breakdown of the inter-
mediate into products, kmax, were determined to be 0.79 mM-1
and 4.5 × 10-2 s-1 for the upper site and 0.72 mM-1 and 8.8
× 10-3 s-1 for the lower site.
Fe3+ + H3L-L3- h Fe(L-H3L), KFe(L-H L)
)
3
[Fe(L-H3L)]/[Fe3+][H3L-L3-] (11a)
Fe3+ + Fe(L-L)3- h Fe2(L-L), KFe (L-L)
)
2
[Fe2(L-L)]/[Fe3+][Fe(L-L)3-] (11b)
Protonation and Stability Constants. The proton dissocia-
tion equilibrium and its protonation constants are defined by
eq 10 for the hydroxamic acid groups of ligands 1 and 2, where
n ) 1, 2, and 3 for 1, and n ) 1, 2, ..., 6 for 2.
We carried out the ligand-exchange reactions given by eqs
12a and 12b, using equimolar amounts of Fe1-2 and H2EDTA2-
,
and of Fe2-2 and H2EDTA2-, respectively, at 25.0 ( 0.1 °C,
ionic strength 0.10, and pH 7.0.
Hn-1L + H+ h HnL, Kn ) [HnL]/[H+][Hn-1L] (10)
Fe(L-H3L) + H2EDTA2- + H+ h Fe(EDTA)- + H3L-H3L
Potentiometric titration was carried out in water at 25.0 ( 0.1
°C; the constants (pK) obtained are 9.69, 9.03, and 8.26 for 1,
and 9.52, 9.39, 9.02, 8.76, 8.46, and 7.81 for 2. Six pK values
for 2 lie in a narrow range of only 1.7 pK units, and the average
hydroxamate pKav values for 1 and 2 are smaller than that of 3
and 4 (Table 4).
(12a)
Fe2(L-L) + H2EDTA2- + H+ h Fe(EDTA)- + Fe(L-H3L)
(12b)
Equilibrium quotients (Keq) of eqs 12a and 12b are expressed
by eqs 13a and 13b and determined from the stoichiometry of
the reactions performed.
(47) Hara, Y.; Shen, L.; Tsubouchi, A.; Akiyama, M.; Umemoto, K.
Inorg. Chem. 2000, 39, 5074-5082.

Model for Biological Iron Mobilization
Keq1
J. Am. Chem. Soc., Vol. 123, No. 30, 2001 7253
)
[Fe(EDTA)-][H3L-H3L]/[ Fe(L-H3L)][H2EDTA2-][H+]
(13a)
Keq2
)
[Fe(EDTA)-][ Fe(L-H3L)]/[ Fe2(L-L)][H2EDTA2-][H+]
(13b)
Equations 14a and 14b may be derived from eq 11, that for
Fe(EDTA)-, and eq 13.
KFe(L-H L) ) (KFe(EDTA)/Keq1)( Kav3/K1edtaK2edta) (14a)
3
edta
KFe (L-L) ) (KFe(EDTA)/Keq2)(Kav3 /K1edtaK2
) (14b)
2
We used the following relations for derivation: [H3L-L3-
]
) K1K2K3[L-L6-][H+]3; [H3L-H3L] ) K4K5K6[H3L-L3-][H+]3;
and [H2EDTA2-] ) K1edtaK2edta[EDTA4-] × [H+]2. However,
the use of these relations means that we presume the follow-
ing: protonation of the six hydroxamates takes place in the order
of the upper site and then the lower site, and proton dissociation
of Fe(L-H3L) occurs as expressed by [Fe(L-H3L)] ) K1K2K3-
[Fe(L-L)3-][H+]3. In practice, Fe(L-H3L) in eqs 11a, 12a, and
13 needs to be replaced by Fe1-2, in which iron may reside at
either site. Errors due to such ambiguous residency are less than
10%, since a probability of 93% at the lower and 7% at the
upper site was determined using eq 8 above. Accordingly, we
used the average Kav value for each Kn in our calculations,
neglecting small differences between the presumed and the real
protonation situations. The results are summarized in Table 4.
Figure 6. Proposed structure of Fe2-2 shown in the ∆-cis configuration.
Similar structures for FeM-2 are possible by replacing Fe with M.
sites. Steric interaction between the CH3 group of the L-Ala-
N(O-) moiety and the â-CH2 group of the NCH2CH2CO residue
at the binding site forces iron(III) to coordinate in the ∆-cis
configuration. Repetitive extension of the Ala-Ala-â-(HO)Ala
unit appears to result in high CD intensity/Fe(III) of Fe1-2 and
Fe2-2 relative to that of Fe-1, as in that of Fe-3 and Fe-4 (Table
1). The three supporting strands are forced locally to assume a
triple-helical orientation in Fe2-2, and the intervening peptide
backbone links the two binding sites in a helical manner,
resulting in a well-defined ditopic chiral complex. A proposed
structure for the overall geometry of Fe2-2 is depicted, aided
by a molecular model examination (Figure 6); similar structures
are also possible for FeM-2, in which M replaces Fe. The tren
moiety has an extended conformation with the nitrogen electron
pair having an inward direction, unlike that shown for Fe2-5.19
Iron Loading. The sequence of eq 4 consists of several
elementary steps, as shown by representative species, and
appears to be complex; eq 4a involves dissociation of Fe(III)-
oxalate and association of Fe(III)-hydroxamate together with
their stepwise intermediary transformation, and eq 4b includes
additional dissociation of M(III)-hydroxamate and association
of M(III)-oxalate. The fact that exact 1- or 2-equiv amounts
of the ferric oxalate virtually transformed M(L) to Fe(L) is
ascribable to the outstanding stability of Fe(L) among M(III)-
hydroxamate and M(III)-oxalate complexes.49-51 A primary
purpose for the experiments is to detect any difference in iron-
loading rates for the different M(III)-preloaded ligands. The rate
constants obtained by applying the second-order reaction
models, therefore, are useful for discussing the formation process
of Fe(L), even though they are apparent constants, and it is not
known which step is rate-determining among those processes.
Discussion
Ligands and Complexes. The ditopic ligands (represented
by 5 in Chart 1) and their Fe(III) complexes were previously
reported to have intramolecularly hydrogen-bonded networks
in chloroform.19,21 However, their lipophilic nature and deoxy-
genated hydroxamate portions and the different ꢀ values at the
two binding sites rendered it inconvenient to make detailed
kinetic and thermodynamic analyses of iron complexation in
aqueous solution. Besides, their moderate CD intensity suggests
that their iron complexes either have a lower helix content or
contain a minor component of the optical antipode, owing to
the less demanding regulatory effect of the backbone.
The present ligands (1 and 2) and their complexes (Ga-1 and
Ga1-2) do not assume any particular, intramolecularly hydrogen-
bonded structures in a polar solvent like DMSO, despite their
peptide nature. This is indicated by large negative values of
the temperature dependence of amide-proton chemical shifts.
Consequently, in much more polar solvent, water, 1 and 2 are
also not preorganized for complexation,48 and their Fe(III)
complexes are not stabilized by intramolecular hydrogen bond-
ing. It is important, however, to use aqueous systems in
determining the properties of iron complexes because of their
relevance to biological systems as well as aqueous chemistry.
Ligand 2 has two discrete binding sites, and each binding
site contributes to an equal ꢀ value/Fe(III) at λmax. The observed
isosbestic behavior and related Schwarzenbach plots indicate
that iron is accommodated at individual binding sites, ruling
out the possibility of forming complicated polymeric complexes.
The CD intensity observed for Fe1-2 and Fe2-2 suggests a high
local symmetry of the chiral binding site. The 2-fold intensity
of Fe2-2 to Fe1-2 is consistent with two similar, separate binding
(49) The logarithmic stability constants of tris(hydroxamato)metal(III)
complexes, to cite one example, for deferriferrioxamine B () L), are as
follows:34 Al3+(L), 24.14; Ga3+(L), 28.17; In3+(L), 20.60; Fe3+(L),32 30.60.
(50) For L ) oxalato:45 (a) Fe3+(L)3, 18.49: (b) Al3+(L)3, 15.21; Ga3+
-
(L)3, 17.86; In3+(L)3, 14.53: (c) For example, L1 ) acetato:45 In3+(L1)3,
7.9; Fe3+(L1)3, 8.3.
(51) Martell, A. E.; Smith, R. M. Critical Stability Constants; Plenum
Press: New York, 1974-1989; Vol. 1-6.
(48) Cram, D. J. Science 1988, 240, 760-767.

7254 J. Am. Chem. Soc., Vol. 123, No. 30, 2001
Hara and Akiyama
Scheme 3
site is also shown in the distributions of FeM-2 in which a large
portion of iron (75∼79%) resides at the lower site.
The observed iron-removal rates are much faster than that
for ferrichrome (6.1 × 10-4 s-1) under comparable conditions.17c
This trend is similar to that observed for previous peptide
hydroxamate ligands.47
Although iron prefers to reside at the lower site, iron may
migrate from the lower to the upper site. Even when this
migration occurs during an iron-removal reaction, the migration
low
rate is included in the constant k1
.
When we consider a global iron mobilization sequence, the
above-described behavior illustrates that iron is captured by the
ligand, stored mostly at the lower site, and then released from
the storage site by a scavenger, EDTA, only after the upper
M(III) release.
Stability Constants. The values of the constant KFe(HL), which
represents the monoprotonation stability, increase in the order,
ferrichrome < Fe-1 < Fe1-2 < Fe-3 < Fe2-2 < Fe-4 with the
decreasing stability of Fe(L). We interpret this order as reflecting
the increasing strain present in these complexes. When we
compare them on a monoiron basis, the tren-based tripodal
structure appears to produce flexible complexes relative to those
of 3 and 4.
For the first time, the proton-independent stability constants
for Fe2-2 were obtained via the determination of iron residence
probability in Fe1-2. The stability constants at the lower site of
Fe1-2 and the upper site of Fe2-2 are almost the same order as
that of Fe-1 and also the same as those of Fe-3 and Fe-4. All of
these stability constants are close to that of ferrichrome (Table
4). However, iron preference for the lower site is not apparent
in these values; the constant for the upper site of Fe2-2 is only
determinable in the presence of iron at the lower site.
The absorbance change versus time curves clearly show
different rates for the different M(L) complexes under similar
reaction conditions (Figure 3, A and B). The different rates
strongly suggest that dissociation of M(III)-hydroxamate plays
a crucial role. The iron distributions in the 1-equiv loading on
M1-2 (Table 3) indicate that the residing metal ion scrambles
with the incoming iron for a formally vacant site (vide infra).
A single rate constant for 2-equiv loading on 2 or In1-2 suggests
that Fe(III) competes for the two binding sites with Fe(III) or
In(III) throughout the reaction. Worthy of note is that the loading
rate on In1-2 was faster (ca. 3 times) than the loading rate on
the free ligand 2.52 An apparent rate constant is larger for 1-equiv
loading than 2-equiv loading on In1-2 (entry 3c, Table 2). This
is largely due to a different iron concentration, and, in fact, it
is seen that 2-equiv loading was complete faster at 1-equiv iron
stage (at absorbance 0.14 in Figure 3B).
Conclusions
In the loading on 2 (Scheme 3), there are two main pathways
which are considered to be major and minor. A major pathway
(shown by bold-face lines) leading to the product starts with
Mlow-2, as Ga(III) resides at the lower site. One-equivalent
loading on this produces FeM-2, in which Fe(III) enters rather
the lower site (Table 3, 75-79%) and the next 1-equiv loading
affords Fe2-2, expelling M(III) at the same time. A minor
pathway, which is the rest of the pathway, is also open for Al1-2
and Ga1-2, in which Fe(III) first enters the upper site (21-25%),
as also suggested by similar rates for the two, and then ex-
changes with M(III) in the ligand. Upon the second iron coming
in, M(III) dissociates as a key step, and Fe2-2 is produced.
Iron Removal. The hyperbolic dependence of both constants
k1up and k1low on the EDTA concentration is consistent with the
reaction sequence of eq 6,16 where an intermediate ternary
complex is formed, as previously discussed by Albrecht-Gary
et al.46 and recently by us.47 With Fe2-2, for example, a value
The monotopic and ditopic ligands 1 and 2 prepared were
not preorganized for complexation, and the formed iron(III)
complexes were not stabilized by intramolecular hydrogen
bonding in a solvent like DMSO and, of course, not in much
more polar solvent water. Nevertheless, ligands 1 and 2 afforded
well-defined, chiral iron(III) complexes in water, as shown by
CD spectroscopy, ESIMS, and isosbestic behavior on UV-vis
spectra. Iron loading on ligand 2 or on the M(III)-preloaded
ligands with ammonium ferric oxalate was found to proceed
through Fe(III) and M(III) competition, suggesting that dis-
sociation of the M(III)-hydroxamate bond is a crucial factor
for the rate. In general, the rate of iron loading on M(III)-
preloaded ligands increased in the order Al < Ga < Fe < In.
Iron-removal reactions of the formed complexes under the
pseudo-first-order kinetic conditions with EDTA reflect the
states of iron residence in 2. The two rate constants for iron
up
removal from Fe2-2, k1 and k1low, were determined, and the
up
low
of k1 is faster than that of k1 (a factor of 5 in entry 4a in
Table 3, and also compare two kmax values).
iron residence preference for the lower site in Fe1-2 and in
FeM-2 was noted. The processes of iron loading and its
subsequent removal, when viewed as an iron-holding host
ditopic ligand coupled with a competing guest molecule,
represent a complex iron-mobilization system. We envisage iron
mobilization such as this, although ligating functional groups
are different, in intracellular intricate processes which are not
yet well-characterized.1 The present ditopic ligand serves as an
abiotic iron-storing vehicle and provides a means for studying
iron mobilization in aqueous systems.
low
Rate constants k1 obtained for Fe1-2 through the parallel
first-order kinetics and for FeM-2 through the consecutive first-
order kinetics converge at a value of 4.0 × 10-3 s-1 determined
low
for the rate constant k1 of Fe2-2 (entries 4b and 5a∼5c in
Table 3). This coincidence is reasonable, since the same situation
common to all of the complexes is left after removal of the
upper metal ion.
A preference for iron residence at the lower site is exhibited
by a distribution of Fe1-2 (lower/upper; 93/7), which is in line
with the NMR data of Ga1-2. Similar preference for the lower
Experimental Section
General Procedures. IR spectra were obtained on a JASCO model
A-302 spectrophotometer, and UV-vis spectra were recorded on a
(52) We thought this was due to a preorganization effect of the ligand
for Fe(III) loading with In(III), although a reviewer disagreed with our view.

Model for Biological Iron Mobilization
J. Am. Chem. Soc., Vol. 123, No. 30, 2001 7255
Hitachi 320A spectrophotometer. CD spectra were taken with a JASCO
J-720 spectrophotometer. HPLC was carried out on a JASCO 880-PU
apparatus combined with 875-UV and 100-III attachments, using a
column (4.6 mm × 250 mm) of CrestPak C18 T-5. A solvent system
of CH3CN-H2O (3:1 v/v) containing 0.1% phosphoric acid and 5 mM
sodium 1-octanesulfonate was applied at a flow rate of 1 mL/min, and
the retention time (Rt) was determined. Mass spectra were determined
on Micromass QUATTRO II equipment. Optical rotations were
4.23 (qn, J ) 7.3, 3H), 4.30 (qn, J ) 7.3, 3H), 4.79 (qn, J ) 7.3, 3H),
7.79 (t, J ) 5.1, 3H), 7.80 (d, J ) 7.2,3H), 7.99 (d, J ) 7.8, 3H), 8.09
(d, J ) 7.0, 3H), 9.95 (s, 3H). Anal. Calcd for C48H84N16O18‚4H2O:
C, 46.29; H, 7.45; N, 18.00. Found: C, 46.49; H, 7.26; N, 18.33.
[Ac-Ala-{Ala-â-(HO)Ala-Ala}2-NHCH2CH2]3N (2). Compound 8a
(0.51 g, 0.31 mmol) in CH2Cl2 (1.0 mL) was treated with TFA (14.3
mL, 187 mmol) at 0 °C for 3.5 h, and the solvent was evaporated. A
residue (8b) in DMSO (10 mL) was condensed with 6a (0.57 g, 1.3
mmol) in DMF (13 mL) in the presence of NMM (0.39 mL, 3.5 mmol)
for 3 h at 0 °C and for 24 h at room temperature by a BOP reagent54
(0.88 g, 2.0 mmol) accompanied with NMM (0.33 mL, 3.0 mmol).
DMF was evaporated, and CHCl3 (100 mL) was added. The mix-
ture was washed with water (2 × 50 mL), 5% aqueous NaHCO3 (3 ×
100 mL), dried (MgSO4), and purified by a Sephadex LH-20 column
(with MeOH) to give a product: {Boc-Ala-[Ala-â-(BnO)Ala-Ala]2-
NHCH2CH2}3N (9a) in 80% (0.64 g): HPLC Rt 8.0 min; IR (KBr,
1
measured with a Horiba SWPA-2000 polarimeter at 25 ( 0.1 °C. H
NMR spectroscopy was performed in CDCl3 at an ambient temperature
or in DMSO-d6 with a JEOL GX-400 or an A-500 spectrometer using
Me4Si as the standard. The melting points are uncorrected. Double-
distilled water was deionized by passing through an ion-exchange resin
(Dowex 50W-X8).
Synthesis. Boc-Ala-Ala-â-(BnO)Ala-OH (6a) was obtained as
previously reported.28
1
cm-1) 1680, 1640, 1540, 1450, 1370, 750, 700; H NMR (CDCl3) δ
(Boc-Ala-NHCH2CH2)3N (7a). To a mixture of Boc-Ala-OH (6.81
g, 36 mmol), HOBt (6.12 g, 40 mmol) and EDC‚HCl (9.59 g, 50 mmol)
in CH2Cl2 (100 mL) was added tris(2-aminoethyl)amine (1.46 g, 10
mmol) in DMSO (10 mL). The mixture was stirred for 3 h at -10 °C
and for 50 h at room temperature, and the solvent was evaporated,
followed by addition of CHCl3 (100 mL). The resulting mixture was
washed with water (2 × 50 mL) and 5% aqueous NaHCO3 (3 × 100
mL) and dried (MgSO4). Evaporation of the solvent and tritiation of
the residue with CHCl3-EtOAc gave crystals of 7a (3.78 g, 57%):
mp 162-163 °C; IR (KBr, cm-1) 3300, 2900, 1700 (CdO Boc), 1670
(CdO amide), 1535 (NH amide), 1370, 1250, 1175; 1H NMR (CDCl3)
δ 1.36 (d, J ) 7.1, 9H), 1.42 (s, 27H), 2.56 (t, 6H), 3.00-3.50 (m,
6H), 4.40 (m, 3H), 5.60 (br d, 3H), 8.78 (br s, 3H).
1.25 (m, 45H), 1.43 (s, 27H), 2.47 (m, 18H), 2.9-3.3 (m, 6H), 3.8-
4.1 (m, 12H), 4.25 (m, 3H), 4.58 (m, 6H), 4.93 (d, J ) 7.1, 12H), 4.95
(m, 6H), 5.38 (br s, 3H), 7.00 (br s, 3H), 7.37 (m, 30H), 7.41 (br s,
6H), 7.60 (d, J ) 6.6, 3H). Compound 9a (0.597 g, 0.232 mmol) in
CH2Cl2 (2.3 mL) and TFA (10.6 mL, 139 mmol) gave 9b, which was
acetylated in DMSO (1.0 mL)-CHCl3 (15 mL) with N-acetoxysuc-
cinimide (0.182 g, 1.16 mmol) and Et3N (0.54 mL) at -10 °C for 3 h
and for 24 h at room temperature. The residue obtained was purified
by a Sephadex LH-20 column (with MeOH) to give a product (0.35 g,
63%): {Ac-Ala-[Ala-â-(BnO)Ala-Ala]2-NHCH2CH2}3N (9c): HPLC
Rt 7.2 min; IR (KBr, cm-1) 1660, 1640, 1540, 1450, 1370, 750, 700;
1H NMR (400 MHz, CDCl3) δ 1.24 (m, 45H), 1.96 (s, 9H), 2.46 (m,
12H), 2.53 (m, 6H, s), 2.9-3.3 (m, 6H), 3.7-4.2 (m, 12H), 4.57 (m,
9H), 4.88 (s, 12H), 4.93 (m, 6H), 6.80 (br s, 3H), 7.28 (br s, 3H), 7.36
(m, 30H), 7.64 (br s, 12H). Compound 9c (0.15 g, 0.063 mmol) in
MeOH (100 mL) was hydrogenated with H2 in the presence of Pd on
carbon (15 mg, 10%) for 20 h at room temperature. The product (2)
was purified with a Sephadex G-15 column to afford a white solid (60
mg, 52%): HPLC Rt 2.1 min; Optical rotation [R]25D -60 (c 0.3, H2O);
IR (KBr, cm-1) 3260 (N-OH), 1640, 1540, 1450 (CONOH); 1H NMR
(DMSO-d6 at 30 °C) δ 1.17 (d, J ) 6.8, 27H), 1.19 (d, J ) 5.4, 18H),
1.83 (s, 9H), 2.40 (t, J ) 6.5, 18H), 3.10 (m, 6H), 3.69 (m, 12H), 4.27
(qn, J ) 7.0, 3H), 4.29 (qn, J ) 7.3, 6H), 4.78 (qn, J ) 6.6, 6H), 7.79
(t, J ) 6.5, 3H), 7.84 (d, J ) 7.3, 6H), 7.98 (d, J ) 7.3, 3H), 8.09 (d,
J ) 7.3, 6H), 9.91 (s, 6H). Anal. Calcd for C75H129N25O30‚6.5H2O: C,
45.54; H, 7.24; N, 17.70. Found: C, 45.68; H, 6.93; N, 17.38.
[Boc-Ala-Ala-â-(BnO)Ala-Ala-NHCH2CH2]3N (8a). Compound 7a
(4.67 g, 7.09 mmol) was treated in TFA (130 mL, 1.7 mol) for 3.5 h
at 0 °C to give tris(CF3CO2H‚H-Ala-NHCH2CH2)N (7b) as an oil with
inclusion of TFA (9.95 g, 7.09 mmol). To a mixture of 6a (11.7 g,
24.8 mmol), HOBt (6.13 g, 40 mmol), and EDC‚HCl (7.67 g, 40 mmol)
in CHCl3 (100 mL) was added dropwise 7b and NMM (7.14 mL) in
DMSO (30 mL) at -10 °C. The mixture was stirred for 3 h at -10 °C
and for 24 h at room temperature, followed by addition of CHCl3 (100
mL). The resulting mixture was washed with water (2 × 200 mL) and
aqueous 5% NaHCO3 (3 × 200 mL) and dried (MgSO4). After
evaporation of the solvent, the residue was purified by silica gel
chromatography with CHCl3-MeOH (5:1 v/v), yielding a product (8a)
(10,4 g, 91%): HPLC Rt 3.0 min; IR (KBr, cm-1) 3300, 2970, 1650,
1
1530, 750, 700; H NMR (DMSO-d6 at 40 °C) δ 1.15 (m, 27H), 1.38
Iron(III) Complex Formation. Stock solutions of 3.05 × 10-3
M
(s, 27H), 2.45 (m, 12H), 3.08 (m, 6H), 3.72-4.04 (m, 6H), 4.05 (m,
3H), 4.28 (m, 3H), 4.85 (m, 3H), 4.95 (ABq, J ) 9.2, 6H), 6.80 (br s,
3H), 7.41 (m, 15H), 7.70 (t, J ) 6.4, 3H), 7.93 (d, J ) 7.3, 3H), 8.05
(d, J ) 7.1, 3H).
and 1.47 × 10-3 M for ligands 1 and 2 in water were prepared. A
stock solution of ferric nitrate (2.93 × 10-3 M) was prepared by
dissolving Fe(NO3)3‚9H2O in 0.1 M nitric acid solution.
(a) In a 10-mm cell held at a constant temperature of 25.0 ( 0.1 °C
in the cell compartment of the spectrophotometer, a solution of 1 (7.84
× 10-7 mol) and a KCl solution were placed and diluted with water.
The total volume was 3.00 mL with 0.10 M KCl. In the case of 2, a
ligand solution (3.9 × 10-7 mol) was used to make a 3.00 mL solution
with 0.10 M KCl. To these was added an iron(III) solution (7.82 ×
10-7 mol). The pH of the solution was determined (pH 2.1 ( 0.1).
For studies at pH 7.0, the pH of the solutions was adjusted with 0.1
M KOH (ca. 0.3 mL). The CD spectrum was determined for this
solution, and the concentration was corrected for added volume.
(b) For pH titration studies, the pH of the above neutral solutions
was adjusted with small quantities of either 0.1 M HNO3 or 0.1 M
KOH. The UV-vis spectra were recorded, and the concentrations of
the resulted solutions were corrected for added volume.
(c) Schwarzenbach plots were generated from these solutions that
exhibited isosbestic behavior in the UV-vis spectra during the pH
titration studies.
NMR Determinations of Gallium(III) Complexes. A 1.5-equiv or
a 1.0-equiv amount of Ga(OH)3, prepared in situ from Ga(NO3)3 and
aqueous alkali, was combined with 1 or 2 in H2O, respectively, stirred
for 24 h, and filtered, and the solution was evaporated. The solid was
dissolved in DMSO-d6.15 Alternatively, Ga1-2 was prepared in situ by
[Ac-Ala-Ala-â-(BnO)Ala-Ala-NHCH2CH2]3N (8c). Compound 8a
(1.62 g, 1.00 mmol) and TFA (20 mL, 240 mmol) in CH2Cl2 (20 mL)
were treated for 3 h at 0 °C, and evaporated. A residue (8b) was treated
with Et3N (4.0 mL) in DMSO (5.0 mL) and N-acetoxysuccinimide53
(0.94 g, 6.0 mmol) in CHCl3 (20 mL) for 3 h at -10 °C and for 50 h
at room temperature and purified by silica gel chromatography with
CHCl3-MeOH (5:1 v/v) to give a compound (8c) in 84% (1.21 g):
1
HPLC Rt 7.7 min; IR (KBr, cm-1) 1650, 1540, 750, 700; H NMR
(DMSO-d6 at 40 °C) δ 1.15 (m, 27H), 1.83 (s, 9H), 2.45 (m, 12H),
3.08 (m, 6H), 3.70-4.00 (m, 6H), 4.32 (m, 6H), 4.85 (s, 3H), 4.95 (m,
6H), 7.40 (m, 15H), 7.71 (t, J ) 6.5, 3H), 7.93 (d, J ) 7.8, 3H), 8.04
(d, J ) 7.8, 3H), 8.08 (d, J ) 7.8, 3H).
[Ac-Ala-Ala-â-(HO)Ala-Ala-NHCH2CH2]3N (1). Compound 8c
(0.40 g, 0.28 mmol) in MeOH (50 mL) was hydrogenated with H2 in
the presence of Pd on carbon (100 mg; 10%) for 24 h and filtered, and
the solvent was evaporated. The product was purified with a Sephadex
LH-20 column to give a white solid (0.20 g, 61%): HPLC Rt 3.8 min;
Optical rotation [R]25 - 83 (c 0.3, H2O); IR (KBr, cm-1) 3300 (N-
D
1
OH), 1650 (CO amide) 1540, 1380, 1120; H NMR (DMSO-d6 at 30
°C) δ 1.17 (d, J ) 7.3, 9H), 1.20 (d, J ) 7.3, 18H), 1.83 (s, 9H), 2.41
(t, J ) 7.0, 6H), 2.50 (br d, J ) 6.5, 6H), 3.10 (m, 6H), 3.71 (m, 6H),
(53) Castro, B.; Dormoy, J. R.; Evin, G.; Selve, C. Tetrahedron Lett.
1975, 1222-1225.
(54) Lindsay, D. G.; Shall, S. Biochem. J. 1971, 121, 737-745.

7256 J. Am. Chem. Soc., Vol. 123, No. 30, 2001
Hara and Akiyama
combining equimolar amounts of 2 and tris(acetylacetonato)Ga(III) in
DMSO-d6 and allowing the solution to equilibrate for 2 days. The degree
of Ga loading was estimated to be 90% in the latter case. Complex
Ga1-2 was soluble in water and DMSO, and separable from insoluble
complex Ga2-2.
Al(III), Ga(III), and In(III) Complexes. In a 10-mm UV-vis cell,
a ligand solution (both 3.9 × 10-7 mol) was mixed with an aqueous
metal(III) ion solution (3.9 × 10-7 mol) prepared from AlCl3 (3.71 ×
10-2 M), Ga(NO3)3 (1.44 × 10-2 M), or In(NO3)3 (8.71 × 10-3 M),
respectively. Each acidic mixture (ca. pH 2) was neutralized to pH 7.0
with KOH (0.1 M) solution and diluted with water to make a solution
of 2.86 mL, half (1.43 mL, 1.95 × 10-7 mol) of which was used each
time. Metal(III) ion complex formation was confirmed by the absence
of insoluble metal hydroxides at pH 7.
For an iron-removal run, the above acidic mixture (pH 2) was diluted
with water to 2.56 mL, half (1.28 mL) of which was used each time.
Kinetics for Iron(III) Loading. In a 10-mm cell were mixed a buffer
solution (1.50 mL) of pH 5.4 (AcOH-AcONa, 0.20 M, maintained at
I ) 0.20 with KCl), and the above aqueous solution (1.43 mL, 1.95 ×
10-7 mol) containing 1 or 2. In the case of a metal(III)-preloaded
solution, each of the above prepared solutions of M-1 or M1-2 (1.43
mL) in a cell was mixed with this buffer solution. To each of these
combined solutions (2.93 mL) was added a solution (0.064 mL, 1.95
× 10-7 mol) of Fe(C2O4)3‚(NH4)3‚3H2O to initiate a 1-equiv iron-
loading reaction. For 2-equiv loading, a solution of 3.90 × 10-7 mol
was added.
Ammonium ferric oxalate remained intact at least hours under the
experimental conditions, although much less stable than ferric hydrox-
amates. Formation of iron(III) complexes with respect to iron(III) and
ligand (or metal-loaded ligand) was followed by monitoring an increase
in absorbance at 425 nm with time (over 60 s), and the rate was
determined by plotting the data according to the second-order kinetic
equation, [Atꢀ /A∞(A∞ - At)] ) k2appt. An excellent linear plot was
obtained for more than three half-lives. The rate constant was an average
of at least two determinations and contains errors of (10%.
As a sample of an iron-removal run, a buffered FeM-2 solution
(2.844 mL) in a cell was obtained, by mixing the above M1-2 solution
(1.28 mL), the buffer solution (1.50 mL), and the iron solution (0.064
mL). These solutions were used in 1 and 24 h after mixing, respectively.
A solution (2.844 mL) of Fe-1, Fe2-2, or Fe1-2 was similarly obtained.
Electrospray Ionization Mass Spectrometry. Ferric hydroxide was
obtained by neutralizing an aqueous ferric chloride solution with NaOH
to pH 7.0. The precipitates were separated and washed with water.
Ligand 2 and excess Fe(OH)3 (precipitates) were stirred together in
water for 50 h and filtered. A sample was obtained by evaporation of
the filtrate.
the precipitate was stirred in water to give an M1-2 solution, and then
1 equiv of ferric ammonium oxalate was added to this solution. These
FeM-2 solutions were used directly for ES-MS determinations. Samples
of FeIn-2 were measured within a few hours after preparation.
ESI cone voltage, + 30 V. Fe2-2 m/z: [Fe2-2 + H2O + 3H]3+, 661.6;
[Fe2-2 + Na + H]2+, 995.1; [Fe2-2 + 2Na]2+, 1006.1; [Fe2-2 + Na]+,
1989.9. FeAl-2 m/z: [FeAl-2 + Na + H]2+, 980.7; [FeAl-2 + 2Na]2+
,
992.3; [FeAl-2 + Na]+, 1960.0. FeGa-2 m/z: [FeGa-2 + Na + H]2+,-
1001.8; [FeGa-2 + 2Na]2+, 1021.3; [FeGa-2 + 2Η2O + Na + H]2+
,
1019.7. FeIn-2 m/z: [FeIn-2 + Η2O + Na]+, 2065.9: ESI cone voltage,
- 15 V. FeIn-2 m/z: [FeIn-2 + 2ÃΗ]2-, 1028.9; [FeIn-2 + 2Η2O +
2OH]2-, 1046.5.
Rates of Iron(III) Removal by EDTA. To initiate a reaction, EDTA
(0.156 mL, 3.9 × 10-3 mol to become 1.3 × 10-3 M) was added to
each of the buffered FeM-2, Fe1-2, and Fe2-2 solutions (2.844 mL,
1.95 × 10-7 mol) in a cell at 25.0 °C. The reaction was monitored at
425 nm to give absorbance changes versus time curves. The collected
data were analyzed by the curve-fitting procedure using eqs 7, 8, or 9,
up
respectively (see text), to generate first-order rate constants, k1 and
k1low. The reaction was duplicated for reproducibility. The limits of
estimated deviations were determined through the same curve fits
up
((20% for k1 and (5% for k1low).
For the case of Fe-1 (1.3 × 10-4 M), a solution of EDTA (2.6 ×
10-3 M) was used. A pseudo-first-order rate constant for Fe-1 was
obtained from a semilogarithmic plot of absorbance versus time, the
reaction being followed more than three half-lives. The observed rate
(kobs) was obtained as an average value with error limits of (5%.
Determination of pK’s. Potentiometric titration was performed by
a previously described procedure.28 The pK’s of 1 and 2 were calculated
using programs PKAS and HYPERQUAD,55,56 with σ < 0.01 for 1
and σ < 0.05 for 2. The pK value was an average of two determinations
with error limits of (0.05 pK unit.
Equilibrium Competition Reactions With EDTA. The reactions
were carried out essentially by the same procedure previously reported.28
Acknowledgment. We thank Dr. Kazuhisa Hiratani of the
National Institute for Advanced Interdiciprinary Research for
his courtesy in taking measurements of mass spectrometry.
Supporting Information Available: Table S1: 1H NMR
data of ligands 1 and 2 (PDF). This material is available free
of charge via the Internet at http://pubs.acs.org.
JA003251G
(55) Martell, A. E.; Motekaitis, R. J. Determination and Use of Stability
Constants, 2nd ed.; VCH Publishers: New York, 1992.
(56) Gans, P.; Sabatini, A.; Vacca, A. Talanta 1996. 43, 1739-1753.
For an FeM-2 sample, 1 equiv of M(III) hydroxide (M ) Al, Ga,
and In) was similarly prepared as precipitates, and a mixture of 2 and