Iron(III) Complexes as Superoxide Dismutase Mimics: Synthesis, Characterization, Crystal Structure, and Superoxide Dismutase (SOD) Activity of Iron(III) Complexes Containing Pentaaza Macrocyclic Ligands

Article abstract of DOI:10.1021/ic970861h

Iron(III) complexes with four pentaaza macrocylic ligands, [Fe (L)Cl₂](PF₆), have been synthesized, where the ligand L is 1,4,7,10,13-pentaazacyclopentadecane (L¹), 2R,3R,8S,9S-dicyclohexano-1,4,7,10,13-pentaazacyclopentadecane (L²), 2R,3R,8R,9R -dicyclohexano-1,4,7,10,13-pentaazacyclopentadecane (L³), or 2S,5R,8S,11R,14S-pentamethyl-1,4,7,10,13-pentaazacyclopentadecane (L⁴), respectively. Conductivity measurements in acetonitrile are consistent with two chloro anions coordinated to iron(III). In acetonitrile solution, all four iron complexes exhibit a reversible or quasi-reversible redox couple in the experimental range -1.5 to +1.5 V vs E_Ag/Ag⁺ and the redox potentials for those four complexes are similar (from -0.15 to -0.19 V vs NHE). In aqueous solutions, the electrochemical properties of those complexes are different from those in acetonitrile solution; the redox peaks shift more than 0.5 V more positive. The complexes with L¹, L², and L³ display a reversible redox at 0.35, 0.45, and 0.43 V vs NHE, respectively, while [Fe(L⁴)Cl₂](PF₆) shows a cathodic peak at 0.44 V and two anodic peaks at 0.31 and 0.14 V vs NHE, respectively. The base titration results reveal that two water molecules are coordinated to the iron(III) in these complexes, pK_a1 and pK_a2: 3.46(7) and 7.31(7) for [Fe(L¹)Cl₂](PF₆); 3.7(1) and 7.50(2) for [Fe(L²)Cl₂](PF₆); 4.1(1) and 7.73(2) for [Fe(L³)Cl₂](PF₆); 3.6(7) and 7.4(2) for [Fe(L⁴)Cl₂](PF₆). The Superoxide dismutase (SOD) activity, assessed by stopped-flow experiments, reveals that all four metal complexes catalyze the fast disproportionation of Superoxide in aqueous solution: at pH = 7.8, the catalyic rate constants for the complexes were 0.81 × 10⁷, 1.42 × 10⁷, 1.41 × 10⁷, and 0.29 × 10⁷ M^-1 s^-1 for [Fe(L¹)Cl₂]-(PF₆), [Fe(L²)Cl₂](PF₆), [Fe(L3)Cl₂](PF₆), and [Fe(L⁴)Cl₂](PF₆), respectively. The crystal structure of [Fe(L²)-Cl₂](PF₆) was determined. Crystal data: triclinic, P?1 a = 7.367(2), b = 10.707(2), c = 17.632(2) A?; α = 81.27(2), β = 79.74(2), γ = 83.38(2)°; R = 0.0465 for observed data (I > 2σ (I)). The iron is seven-coordinate with five nitrogen atoms from the pentaaza macrocylic ligand and trans-chloro anions in a pentagonal bipyramidal arrangement. The average bond lengths for Fe-N and Fe-Cl are 2.271 and 2.339 A?, respectively.

Full text of DOI:10.1021/ic970861h

956
Inorg. Chem. 1998, 37, 956-963
Iron(III) Complexes as Superoxide Dismutase Mimics: Synthesis, Characterization, Crystal
Structure, and Superoxide Dismutase (SOD) Activity of Iron(III) Complexes Containing
Pentaaza Macrocyclic Ligands
DeLong Zhang,† Daryle H. Busch,*,† Patrick L. Lennon,‡ Randy H. Weiss,‡
William L. Neumann,‡ and Dennis P. Riley*,‡
Chemistry Department, The University of Kansas, Malott Hall, Lawrence, Kansas 66045, and
Monsanto Corporate Research Division, Monsanto Company, 800 North Lindbergh Boulevard,
St. Louis, Missouri 63167
ReceiVed July 9, 1997
Iron(III) complexes with four pentaaza macrocylic ligands, [Fe(L)Cl2](PF6), have been synthesized, where the
ligand L is 1,4,7,10,13-pentaazacyclopentadecane (L1), 2R,3R,8S,9S-dicyclohexano-1,4,7,10,13-pentaazacyclo-
pentadecane (L2), 2R,3R,8R,9R -dicyclohexano-1,4,7,10,13-pentaazacyclopentadecane (L3), or 2S,5R,8S,11R,14S-
pentamethyl-1,4,7,10,13-pentaazacyclopentadecane (L4), respectively. Conductivity measurements in acetonitrile
are consistent with two chloro anions coordinated to iron(III). In acetonitrile solution, all four iron complexes
+
exhibit a reversible or quasi-reversible redox couple in the experimental range -1.5 to +1.5 V vs EAg/Ag , and
the redox potentials for those four complexes are similar (from -0.15 to -0.19 V vs NHE). In aqueous solutions,
the electrochemical properties of those complexes are different from those in acetonitrile solution; the redox
peaks shift more than 0.5 V more positive. The complexes with L1, L2, and L3 display a reversible redox at 0.35,
0.45, and 0.43 V vs NHE, respectively, while [Fe(L4)Cl2](PF6) shows a cathodic peak at 0.44 V and two anodic
peaks at 0.31 and 0.14 V vs NHE, respectively. The base titration results reveal that two water molecules are
coordinated to the iron(III) in these complexes. pKa1 and pKa2: 3.46(7) and 7.31(7) for [Fe(L1)Cl2](PF6); 3.7(1)
and 7.50(2) for [Fe(L2)Cl2](PF6); 4.1(1) and 7.73(2) for [Fe(L3)Cl2](PF6); 3.6(7) and 7.4(2) for [Fe(L4)Cl2](PF6).
The superoxide dismutase (SOD) activity, assessed by stopped-flow experiments, reveals that all four metal
complexes catalyze the fast disproportionation of superoxide in aqueous solution: at pH ) 7.8, the catalyic rate
constants for the complexes were 0.81 × 107, 1.42 × 107, 1.41 × 107, and 0.29 × 107 M-1 s-1 for [Fe(L1)Cl2]-
(PF6), [Fe(L2)Cl2](PF6), [Fe(L3)Cl2](PF6), and [Fe(L4)Cl2](PF6), respectively. The crystal structure of [Fe(L2)-
Cl2](PF6) was determined. Crystal data: triclinic, P1h; a ) 7.367(2), b ) 10.707(2), c ) 17.632(2) Å; R )
81.27(2), â ) 79.74(2), γ ) 83.38(2)°; R ) 0.0465 for observed data (I > 2σ(I)). The iron is seven-coordinate
with five nitrogen atoms from the pentaaza macrocylic ligand and trans-chloro anions in a pentagonal bipyramidal
arrangement. The average bond lengths for Fe-N and Fe-Cl are 2.271 and 2.339 Å, respectively.
Introduction
targeting, immunogenicity, etc.) synthetic metal complexes offer
considerable promise as SOD catalysts for pharmaceutical
applications.
A considerable amount of experimental data suggests that
inflamed tissue and tissue subjected to ischemia followed by
reperfusion are under severe oxidative stress induced by the
one-electron-reduction product of oxygen, superoxide, produced
by the NADPH oxidase-mediated univalent reduction of mo-
lecular oxygen.1-5 Superoxide dismutase (SOD), which can
destroy the superoxide very rapidly, is Nature’s agent for
protection of the organism from this radical burden. In fact,
native SOD enzymes have been shown in many studies to
exhibit protection in animal models of inflammatory diseases.6
In a variety of scenarios, therapeutic dosage of additional SOD
enzyme has shown promise, but from a number of viewpoints
(stability, tissue permeability, oral bioavailability, specific tissue
Depending on the metal at the active center, there are three
general types of SOD enzymes, namely Cu/Zn-SOD and Mn-
SOD in mammalian systems and Fe-SOD in bacterial systems.
Largely due to toxicity issues, we have focused our efforts on
the design of Mn(II)-based complexes as SOD mimics and have
achieved remarkable success in generating highly stable Mn-
(II) complexes with outstanding catalytic SOD activity.7,8 The
ligands utilized in these synthetic Mn(II)-based mimics are of
the same class as we report here; namely, the 1,4,7,10,13-
pentaazacyclopentadecane (L1) type ligand and its C-substituted
derivatives. Since iron is a most relevant metal for the design
of synthetic SOD catalysts, we also embarked on an effort to
discover macrocyclic Fe complexes capable of dismuting
superoxide at bimolecular rates in excess of 1 × 10+6 M-1 s-1
(for comparison the Mn(II) complexes of the [15]aneN5 ligands
exhibit rates generally in excess of 10+7 M-1 s-1).7,8 While a
* To whom correspondence should be addressed.
† The University of Kansas.
‡ Monsanto Co.
(1) McCord, J. M. New Horizons 1993, 1, 70.
(2) Kilgore, K. S.; Lucchesi, B. R. Clin. Biochem. 1993, 26, 359.
(3) Halliwell, B. Lancet 1994, 344, 721.
(4) Grisham, M. B. Lancet 1994, 344, 859.
(5) Chen, L. Y.; Nichols, W. W.; Hendricks, J.; Mehta, J. L. Am. Heart
J. 1995, 129, 211.
(7) Riley, D. P.; Weiss, R. H. J. Am. Chem. Soc. 1994, 116, 387.
(8) Riley, D. P.; Henke, S. L.; Lennon, P. L.; Weiss, R. H.; Neumann,
W. L.; Rivers, W. J.; Aston, K. W.; Sample, K. R.; Rahman, H.; Ling,
C.-S.; Shieh, J.-J.; Busch, D. H.; Szulbinski, W. Inorg. Chem. 1996,
35, 5213.
(6) McCord, J. M. Free Radicals Biol. Med. 1986, 2, 307.
S0020-1669(97)00861-6 CCC: $15.00 © 1998 American Chemical Society
Published on Web 02/04/1998

Fe(III) Complexes as Superoxide Dismutase Mimics
Inorganic Chemistry, Vol. 37, No. 5, 1998 957
(e) [Fe(L4)Cl2](PF6). The crude product of this complex was
prepared by using a procedure similar to that for [Fe(L3)Cl2](PF6).
Crystals were obtained by recrystallization of the crude product from
a mixture of acetonitrile and ethanol. Anal. Calc for C15H35N5Cl2F6-
FeP: H, 6.34; C, 32.37; N, 12.59. Found: H, 6.30; C, 32.44; N, 12.40.
Instrumentation. Electrochemical experiments were performed on
a Princeton Applied Research model 175 programmer and model 173
potentiostat. The output was recorded using a Houston Instruments
recorder. A glassy carbon electrode, a silver wire, and a platinum wire
were used as working, reference, and secondary electrodes, respectively.
Under nitrogen, acetonitrile solutions of the complexes (1-2 mM) with
tetrabutylammonium tetrafluoroborate(0.1 M) as a supporting electrolyte
were used in the experiments. The potentials vs NHE were determined
by using ferrocene as internal reference in organic solvents, typically
acetonitrile, and adding 0.41 V to the observed voltages vs EFc/Fc+. In
aqueous solution, a saturated calomel electrode was used as the
reference electrode, and the potential vs NHE was calibrated by the
SCE potential (0.24 vs NHE).
few FeII and FeIII complexes have been reported to behave as
SOD mimics, their SOD activity was not directly measured.9-12
Herein we report the synthesis, characterization, and SOD
activity of four new FeIII complexes with [15]aneN5 type ligands.
The SOD activities of these complexes have been measured by
the stopped-flow technique13 and are also reported here.
Experimental Section
Synthesis. L1 was synthesized as reported in ref 7, L2 was
synthesized as reported in ref 8, and ligands L3 and L4 were synthesized
according to the methods of ref 14 .
(a) [Fe(L1)Cl](PF6). In a glovebox, ligand L1 (0.50 mmol) was
dissolved in 0.5 mL of methanol, and 2 mL of an FeCl2 (0.50 mmol)-
methanol solution was slowly added under stirring. The mixture was
stirred for 2 h before it was filtered to remove any insolubles. To the
filtrate was added a clear solution of ammonium hexafluorophosphate
(1 mmol). The resulting solution was slowly cooled to room temper-
ature and allowed to stand overnight. The off-white solid was collected
by filtration and washed with 2 mL of methanol. Yield: 0.185 g (0.41
mmol) or 82%. Anal. Calc for C10H20N5ClFeF6P(CH3OH): H, 6.04;
C, 27.32; N, 14.48. Found: H, 6.00; C, 27.18; N, 14.70.
A YSI model 35 conductance meter was used to measure the molar
conductivity. Usually 5 mL of an acetonitrile solution of the complex
was prepared in a 25 mL test tube, and the typical concentration for
the solutions was around 1 × 10-3 M.
The proton dissociation constants of the water molecules coordinated
to metal centers were determined by a potentiometric method. Typi-
cally, about 3 mg of sample was dissolved in 2 mL of boiled deionized
water under nitrogen, and a NaOH solution (1.010 M) in a 10 µL
syringe was used to titrate the complex solution, while the pH was
measured by an Accumet pH meter, model 805 MP, from Fisher
Scientific. The titration was carried out under a nitrogen atmosphere.
The pH of the solution was recorded after every addition of 0.2 µL of
NaOH solution, and the data were processed by the PKAS program.15
The rate constant for the catalyzed superoxide dismutation reaction
of superoxide was measured by the stopped-flow spectrophotometric
methods reported previously13 and by utilizing an OLIS rapid-scan
spectrophotometer/stopped-flow system. In general, the FeIII(L) com-
plexes were dissolved in 80 mM HEPES buffer solutions, and the
saturated KO2 solution (∼0.2 M) in DMSO was freshly prepared. On
the instrument, an aqueous solution of the complex and DMSO solution
of KO2 (mixing ratio: 19/1 water to DMSO) were injected into the
mixing cell from two syringes, and the decay of absorbance of
superoxide (λmax ) 245 nm) was monitored. The background decay
of superoxide itself is second-order and at the pH range utilized in
these studies (7.4-8.1) is much slower than the catalytic rates measured
with these catalysts when they are utilized at the 10-6 M concentration
range.13 Additionally, the control reactions of free ligand and FeCl3
in buffer at pH ) 7.4 show no catalytic activity (the ferric chloride
solutions age rapidly generating olated insoluble products). The data
were exported as text files which were processed in the graphic program
Cricket Graph. The variation of the absorbance with time provided
(b) [Fe(L1)Cl2](PF6). In a glovebox, FeCl3 (0.5 mmol) was dissolved
in 2 mL of pyridine. This solution was added to 0.5 mmol of ligand
in methanol under stirring, and the resulting solution was heated for 2
h. After the solution was cooled to room temperature and filtered, a
clear solution of ammonium hexafluorophosphate was added to obtain
a yellow precipitate. Yield: 0.170 g (0.338 mmol) or 68%. Anal.
Calc for C10H20N5FeF6P(CH3OH)0.5: H, 5.41; C, 25.07; N, 13.92.
Found: H, 5.60; C, 25.18; N, 13.89. The product was recrystallized
from acetonitrile solution. Anal. Calc for C10H20N5Cl2FeF6P: H, 4.18;
C, 24.92; N, 14.53. Found: H, 4.99; C, 25.00; N, 14.60. In the mass
spectrum, two peaks at m/e ) 306 and 341 were observed and assigned
to ([Fe(L1)Cl]2+ + e)+ and [Fe(L1)Cl2]+, respectively.
(c) [Fe(L2)Cl2](PF6). In a glovebox, 0.3 mmol of L2 was dissolved
in 1 mL of methanol, and 0.3 mmol of FeCl3 in pyridine (1.5 mL) was
added to the ligand under stirring. After 3 h of heating, the solution
was filtered. To the filtrate was added a clear solution containing 0.1
g (0.6 mmol) of NH4PF6. The mixture was evaporated to dryness.
Methanol (2 mL) was added to the residue, and the mixture was stirred
for 2 h. The greenish yellow solid was collected via filtration and
washed with methanol. Yield: 0.10 g or 56%. Anal. Calc for C18-
H37N5Cl2FeF6P: H, 6.28; C, 36.35; N, 11.78. Found: H, 6.22; C,
36.16; N, 11.59. The product was recrystallized from a mixture of
acetonitrile and ethanol to obtain crystals of X-ray quality.
(d) [Fe(L3)Cl2](PF6). In a glovebox, 0.3 mmol of FeCl3 was added
to 1 mL of pyridine, and 0.3 mmol of L3 in methanol (1 mL) was
added. The mixture was heated and stirred for 3 h to obtain a brownish
solution, which was filtered after cooling to room temperature. To
the filtrate was added a clear solution containing 0.12 g (0.7 mmol) of
NH4PF6. The mixture was evaporated to dryness. Acetonitrile (2 mL)
was added to the residue, and the mixture was stirred for 2 h before it
was filtered. The filtrate was evaporated to obtain a solid, which was
then dissolved in methanol. Upon slow evaporation, a greenish yellow
solid formed and was collected by filtration. The product was dissolved
in ethanol, and another greenish solid was obtained upon evaporation
of the solvent. The solid was filtered off and dried. Yield: 75 mg or
42%. Anal. Calc for C18H37N5Cl2FeF6P: H, 6.28; C, 36.35; N, 11.78.
Found: H, 6.10; C, 36.22; N, 11.58.
the observed rate constant kobs
.
X-ray Crystal Structure. A single crystal of [Fe(L2)Cl2](PF6) of
0.6 × 0.3 × 0.1 mm dimensions was mounted on a glass fiber in
random orientation. Data collection was performed on a Siemens P4RA
automated single-crystal X-ray diffractometer using graphite-mono-
chromated Mo KR radiation (λ ) 0.710 73 Å) at 25 °C. Autoindexing
of 15 centered reflections from the rotation photograph indicated a
triclinic lattice. Equivalent reflections were checked to confirm the
Laue symmetry, and fractional index search was conducted to confirm
the cell lengths (XSCANS, Siemens Analytical Instruments, Madison,
WI, 1994). Final cell constants and orientation matrix for data
collection were calculated by least-squares refinement of the setting
angles for 36 reflections (9° < θ < 25°). Intensity data were collected
using ωj-2θ scans with variable scan speed. Three representative
reflections measured every 97 reflections showed 17.6% variation
during data collection. Crystal data and intensity data collection
parameters are listed in Table 3.
(9) Bull, C.; McClune, G. J.; Fee, J. A. J. Am. Chem. Soc. 1983, 105,
5290.
(10) Nagano, T.; Hirano, T.; Hirobe, M. J. Biol. Chem. 1989, 264, 9243.
(11) Nagano, T.; Hirano, T.; Hirobe, M. Free Radical Res. Commun. 1991,
12, 221.
(12) Iuliano, L.; Pedersen, J. Z.; Ghiselli, A.; Pratico, D.; Rotilio, G.; Violi,
F. Arch. Biochem. Biophys. 1992, 293, 153.
(13) Riley, D. P.; Rivers, W. J.; Weiss, R. H. Anal. Biochem. 1991, 196,
344.
Data reduction was carried out using XSCANS, and structure solution
and refinement were carried out using the SHELXTL-Plus (5.03)
(14) Lennon, P. J.; Rahman, H.; Aston, K. W.; Henke, S. L.; Riley, D. P.
Tetrahedron Lett. 1994, 35, 853.
(15) Martell, A. E.; Motekaitis, R. J. Determination and Use of Stability
Constants; VCH Publishers, Inc.: New York, 1988; p 159.

958 Inorganic Chemistry, Vol. 37, No. 5, 1998
Zhang et al.
Table 4. Selected Bond Lengths (Å) for [Fe(L2)Cl2](PF6)
Table 1. Proton Dissociation Constants for the Aqua Ligands of
the FeIII Complexes
Fe-N(2)
2.250(3)
2.275(2)
2.290(3)
2.3501(8)
1.480(4)
1.468(4)
1.482(4)
1.480(4)
1.478(4)
1.498(5)
1.526(4)
1.534(6)
1.522(5)
1.513(4)
1.521(6)
1.521(4)
Fe-N(4)
Fe-N(3)
2.259(3)
2.280(3)
2.3287(8)
1.472(4)
1.465(4)
1.475(4)
1.474(4)
1.475(4)
1.513(4)
1.513(4)
1.520(6)
1.512(6)
1.506(4)
1.533(4)
1.520(5)
Fe-N(5)
complex
pKa1
pKa2
complex
pKa1
pKa2
Fe-N(1)
Fe-Cl(2)
[FeL1Cl2](PF6) 3.46(7) 7.31(7) [FeL3Cl2](PF6) 4.1(1) 7.73(2)
[FeL2Cl2](PF6) 3.7(1) 7.50(5) [FeL4Cl2](PF6) 3.6(3) 7.4(2)
Fe-Cl(1)
N(1)-C(1)
N(2)-C(3)
N(3)-C(4)
N(4)-C(11)
N(5)-C(12)
C(1)-C(2)
C(5)-C(10)
C(6)-C(7)
C(8)-C(9)
C(11)-C(12)
C(13)-C(14)
C(16)-C(17)
N(1)-C(18)
N(2)-C(2)
N(3)-C(5)
N(4)-C(10)
N(5)-C(13)
C(3)-C(4)
C(5)-C(6)
C(7)-C(8)
C(9)-C(10)
C(13)-C(18)
C(15)-C(16)
C(17)-C(18)
Table 2. Redox Potentials for the FeIII Complexes in Acetonitrile
and Aqueous Solution with the Peak Separation (mV) in
Parentheses
E1/2 (V vs NHE)
complex
CH3CN
H2O
[FeL1Cl2](PF6)
[FeL2Cl2](PF6)
[FeL3Cl2](PF6)
[FeL4Cl2](PF6)
-0.17 (75)
-0.15 (80)
-0.18 (90)
-0.19 (75)
0.35 (75)
0.45 (80)
0.43 (80)
0.44 (Eap),a 0.31 (Ecp), 0.14 (Ecp)
a Eap and Ecp are the potentials for anodic and cathodic peaks,
respectively.
Table 5. Selected Bond Angles (deg) for [Fe(L2)Cl2](PF6)
N(2)-Fe-N(4)
N(4)-Fe-N(5)
N(4)-Fe-N(3)
N(2)-Fe-N(1)
N(5)-Fe-N(1)
N(2)-Fe-Cl(2)
N(5)-Fe-Cl(2)
N(1)-Fe-Cl(2)
N(4)-Fe-Cl(1)
N(3)-Fe-Cl(1)
Cl(2)-Fe-Cl(1)
C(1)-N(1)-Fe
C(3)-N(2)-C(2)
C(2)-N(2)-Fe
C(4)-N(3)-Fe
144.97(9) N(2)-Fe-N(5)
69.11(9) N(2)-Fe-N(3)
72.67(9) N(5)-Fe-N(3)
73.37(9) N(4)-Fe-N(1)
72.39(9) N(3)-Fe-N(1)
84.88(7) N(4)-Fe-Cl(2)
89.13(7) N(3)-Fe-Cl(2)
93.28(7) N(2)-Fe-Cl(1)
94.96(7) N(5)-Fe-Cl(1)
86.15(7) N(1)-Fe-Cl(1)
176.77(4) C(1)-N(1)-C(18)
113.2(2) C(18)-N(1)-Fe
114.4(2) C(3)-N(2)-Fe
113.7(2) C(4)-N(3)-C(5)
112.5(2) C(5)-N(3)-Fe
144.79(9)
73.72(10)
141.44(10)
141.44(9)
145.23(9)
88.26(7)
94.69(7)
92.38(7)
92.15(7)
84.29(7)
115.6(2)
113.1(2)
112.1(2)
115.2(2)
111.0(2)
111.7(2)
Table 3. Crystal Data and Structure Refinement Parameters for
[Fe(L2)Cl2](PF6)
empirical formula: C18.50H39Cl2F6FeN5O0.5
formula weight: 611.27
temperature: 293(2) K
P
wavelength: 0.710 73 Å
crystal system: triclinic
space group: P1h
unit cell dimensions: a ) 7.3671(5) Å, b ) 10.7063(7) Å,
c ) 17.631(2) Å
R ) 81.269(8)° â ) 79.736(8)°, γ ) 83.388(7)°
volume, Z: 1347.0(2) Å3, 2
density (calcd):1.507 g/cm3
absorption coefficient: 0.878 mm-1
C(11)-N(4)-C(10) 112.3(2) C(11)-N(4)-Fe
F(000): 636
C(10)-N(4)-Fe
C(12)-N(5)-Fe
N(1)-C(1)-C(2)
N(2)-C(3)-C(4)
N(3)-C(5)-C(10)
C(10)-C(5)-C(6)
C(6)-C(7)-C(8)
C(8)-C(9)-C(10)
N(4)-C(10)-C(9)
N(4)-C(11)-C(12) 110.6(3) N(5)-C(12)-C(11) 111.5(2)
N(5)-C(13)-C(18) 108.5(2) N(5)-C(13)-C(14) 113.6(3)
C(18)-C(13)-C(14) 111.4(3) C(15)-C(14)-C(13) 109.9(3)
C(16)-C(15)-C(14) 111.3(3) C(17)-C(16)-C(15) 111.1(3)
C(16)-C(17)-C(18) 111.2(3) N(1)-C(18)-C(13) 106.1(2)
N(1)-C(18)-C(17) 115.0(3) C(13)-C(18)-C(17) 111.0(3)
116.8(2) C(12)-N(5)-C(13) 112.8(2)
crystal size: 0.6 × 0.3 × 0.1 mm
112.6(2) C(13)-N(5)-Fe
106.8(3) N(2)-C(2)-C(1)
106.7(3) N(3)-C(4)-C(3)
106.7(2) N(3)-C(5)-C(6)
115.9(2)
107.6(2)
107.4(3)
114.9(3)
109.9(3)
110.6(3)
107.9(2)
111.2(3)
θ range for data collection: 1.93-28.00°
limiting indices: -1 e h e 10, -14 e k e 14, -24 e l e 24
no. of reflections collected: 7989
no. of independent reflections: 6498 (Rint ) 0.0239)
absorption correction: semiempirical from ψ-scans
max and min transmission: 0.8802 and 0.7396
refinement method: full-matrix least-squares on F2
data/restraints/parameters: 6426/1/339
goodness-of-fit on F2: 1.029
final R indices [I > 2σ(I)]: R1 ) 0.0465, wR2 ) 0.1161
R indices (all data): R1 ) 0.0782, wR2 ) 0.1366
largest diff peak and hole: 0.544 and -0.341 e Å-3
10.6(3)
C(7)-C(6)-C(5)
111.2(3) C(9)-C(8)-C(7)
110.9(4) N(4)-C(10)-C(5)
113.9(3) C(5)-C(10)-C(9)
software package (G. M. Sheldrick, Siemens Analytical Instruments,
Madison, WI, 1995). Linear decay correction was applied to the data,
and absorption correction was carried out using Ψ-scan and equivalent
reflections. The structure was solved by direct methods and refined
successfully in the triclinic space group P1h. Full-matrix least-squares
the iron complexes are expected to exhibit very low FeII/III redox
potentials, resulting in FeII complexes which are readily oxidized
and should be very unstable in air. Accordingly, we observe
that the off-white FeII complex with L1 quickly turns brown
after exposure to air. Consequently, for reasons of stability and
ease of handling, we prepared the FeIII complexes of these
pentaazacyclopentadecane ligands for the studies on their SOD
activity.
2
refinement was carried out by minimizing ∑w(Fo - Fc2)2. The non-
hydrogen atoms were refined anisotropically to convergence. The
hydrogen atoms bonded to the N atoms were refined freely with
isotropic thermal parameters. The rest of the hydrogens were refined
using appropriate riding models. The final residual values were R(F)
) 4.65% for 4424 observed reflections [I > 2σ(I)] and Rw(F2) ) 13.7%;
σ ) 1.03 for all the data. Structure refinement parameters are listed
in Table 3. The atomic coordinates (×104) and equivalent isotropic
displacement parameters (Å2 × 103) are listed in Table 6. The
compound crystallizes with half a molecule of methanol per molecule
of complex. The solvent molecule is disordered. The occupancy factor
could not be refined as the solvent molecule is located near the center
of inversion and occupies one of the two symmetry-related sites 50%
of the time.
Both Fe(II) and Fe(III) complexes were prepared in a
glovebox under a nitrogen atmosphere because of oxygen
sensitivity issue and because of the fact that the starting materials
including ligands and FeCl3 are very hygroscopic. The Fe(II)
complex was prepared from the reaction between the ligand
and Fe(II) in methanol solution, and the complex was precipi-
tated with PF6- as the counteranion after addition of ammonium
hexafluorophosphate. The complexation reaction between FeCl3
and ligand was carried out in pyridine: typically, an FeCl3
pyridine solution (0.5 mL) is added to a vial containing the
Results and Discussion
-
ligand in methanol. All the complexes were prepared with PF6
Synthesis and Characterization. Since the pentaazacy-
cloalkane ligand has five saturated secondary amine N donors,
as the counterion.

Fe(III) Complexes as Superoxide Dismutase Mimics
Inorganic Chemistry, Vol. 37, No. 5, 1998 959
Table 6. Atomic Coordinates (×104) and Equivalent Isotropic
Displacement Parameters (Å2 × 103) for [Fe(L2)Cl2](PF6)a
x
y
z
U(eq)
Fe
1087(1)
-1791(1)
3992(1)
2271(3)
1455(4)
-416(4)
310(4)
1810(3)
3029(5)
1692(5)
62(5)
3887(1)
4860(1)
2917(1)
5172(2)
2740(2)
2112(2)
3839(2)
5678(2)
4476(3)
3508(3)
1828(3)
1146(3)
1648(3)
528(3)
2976(1)
3500(1)
2524(1)
3656(1)
4126(2)
3063(2)
1798(2)
2154(2)
4334(2)
4718(2)
4390(2)
3714(2)
2298(2)
2281(2)
1464(3)
874(3)
35(1)
49(1)
48(1)
39(1)
41(1)
44(1)
41(1)
39(1)
48(1)
48(1)
51(1)
54(1)
46(1)
63(1)
76(1)
86(1)
68(1)
48(1)
54(1)
52(1)
42(1)
57(1)
70(1)
65(1)
54(1)
41(1)
51(1)
94(1)
88(1)
73(1)
58(1)
140(1)
171(2)
162(2)
261(12)
323(21)
Cl(1)
Cl(2)
N(1)
N(2)
N(3)
N(4)
N(5)
C(1)
C(2)
C(3)
C(4)
C(5)
C(6)
C(7)
C(8)
C(9)
C(10)
C(11)
C(12)
C(13)
C(14)
C(15)
C(16)
C(17)
C(18)
P(1)
7(5)
-174(4)
-1265(5)
-943(6)
-1474(7)
-418(6)
-718(4)
-566(5)
391(5)
2402(4)
3528(5)
4096(6)
5163(6)
4046(5)
3467(4)
5000
4624(4)
3464(3)
6564(3)
5000
4699(5)
5995(7)
3077(5)
4188(23)
4063(26)
167(4)
1296(4)
2414(4)
2767(3)
5074(3)
6145(3)
6695(3)
7640(3)
8645(3)
8040(4)
7093(3)
6090(3)
0
-1266(2)
-267(3)
-777(2)
5000
5040(4)
3662(4)
4504(5)
689(20)
955(20)
902(2)
1723(2)
1489(2)
1659(2)
2509(2)
1923(2)
2342(3)
2989(2)
3562(2)
3150(2)
5000
5577(2)
4535(2)
4463(1)
10000
Figure 1. Structures of the pentaaza macrocyclic ligands utilized for
the synthesis of Fe(III) complexes.
All four of the new FeIII complexes described here were found
to be high-spin d5 complexes on the basis of the solution
magnetic susceptibilities determined by NMR and Evans’
method,16 which were observed to be approximately 5.6 µB for
the each of the four new complexes. The electronic spectrum
of each new FeIII complex was also recorded in HEPES buffer
in water at a concentration of 1.0 × 10-4 M over a pH range of
6.0-8.1. The spectra of all the complexes were virtually
identical with a λmax at 294 nm (pH ) 8.1) with an extinction
coefficient of ∼9400 cm-1 M-1. When the pH is lowered, the
band max shifts to slightly lower energy with a lower molar
extinction (λmax at 306 nm ꢀ ∼3100). The high-pH absorption
is due to the presence of the [FeIII(L)(OH)2]+ complex, and the
low-pH absorption is due to the presence of the [FeIII(L)(OH)-
(OH2)]2+ complex (vide infra).
Acidity of the Coordinated Water Molecules. Molar
conductivity measurements in acetonitrile solution reveal that
the FeIII complexes [Fe(L)Cl2](PF6) behave as 1:1 electrolytes
(120-160 Ω-1 cm2 mol-1). In contrast, in aqueous solutions
the chloride anions are subject to dissociation to form the
corresponding aquo/hydroxo complexes. The resulting aquo
complexes can deprotonate to form hydroxo complexes, and
those derivatives should have very different physical properties,
e.g., redox potentials. Consequently, it is very important to
know the proton dissociation constants of the coordinated water
molecules. The deprotonation constants determine the distribu-
tion of different species at each pH and, consequently, establish
the coordination environment around the metal center, i.e.,
ligands present and the coordination number. In Figure 2 the
titration curve for a 2.31 mM aqueous [Fe(L1)Cl2](PF6) solution
with 1.010 M NaOH is presented. In Table 1 the pKa’s for the
FeIII complexes in aqueous solutions obtained from the titration
studies are presented. Each of the complexes exhibits two pKa
values, which indicates that two water molecules coordinate to
the metal center in the Fe(III) complexes. Therefore, the
complexes are seven-coordinate in aqueous solutions, assuming
that all five nitrogens on the pentaaza macrocylic ligand
coordinate with iron. This implies that the chloride anions are
replaced by water molecules. The pKa values are quite similar
for all of the complexes, falling in the ranges 3.5-4.1 and 7.3-
F(1)
F(2)
F(3)
P(2)
F(4)
F(5)
F(6)
O(1)
C(19)
9142(2)
9945(3)
10287(2)
9961(10)
10774(8)
a U(eq) is defined as one-third of the trace of the orthogonalized Uij
tensor.
The greenish yellow FeIII complex, [Fe(L1)Cl2](PF6), is
soluble in water and acetonitrile. The corresponding Fe(III)
complexes of the two stereoisomers L2 and L3 have markedly
different solubilities. The complex [Fe(L2)Cl2](PF6) was much
more soluble in polar organic solvents than [Fe(L3)Cl2](PF6),
although they have the same composition and are of the same
electrolyte type, 1:1. For example, [Fe(L2)Cl2](PF6) is only
slightly soluble in methanol while [Fe(L3)Cl2](PF6) is very
soluble. This difference in the solubility has also been observed
for the tosylated ligands. The solubility of [Fe(L4)Cl2](PF6) was
similar to that of [Fe(L3)Cl2](PF6). Stability is a major design
consideration for candidates for human pharmaceutical applica-
tions. Thus, the determination of the stability of these Fe(III)
complexes in aqueous media is of considerable importance. The
complex [Fe(L1)Cl2](PF6) is stable in aqueous solution for
several hours at physiological pH, but some precipitate does
form after several days, probably due to hydrolysis. In contrast,
the FeIII complexes with the other three ligands are stable in
aqueous solution under physiological conditions with the
aqueous solutions of the complexes with L2, L3, and L4
remaining clear for several months. [Fe(L2)Cl2](PF6) is also
stable in mildly acidic solution and only decomposes slowly in
4.6 N H2SO4 solution. The decomposition of the complexes is
evidenced by the decrease in the absorbance for the complexes
(vide infra). It was found that only about 8% of [Fe-
(L2)Cl2](PF6) decomposes in 4.6 N H2SO4 over a period of 10
h. This indicates that the iron(III) ion forms very stable
complexes with pentaazacycloalkanes.
(16) Evans, D. F. J. Chem. Soc. 1959, 2003.

960 Inorganic Chemistry, Vol. 37, No. 5, 1998
Zhang et al.
Figure 3. Projection view of the molecule [Fe(L2)Cl2](PF6) with non-
hydrogen atoms represented by 50% probability ellipsoids.
Figure 2. Experimental and calculated titration curves for 2.00 mL of
2.31 mM [Fe(L1)Cl2](PF6).
the iron complex catalyzes the decomposition of superoxide.
In acetonitrile solution, the redox potentials for the other three
complexes are quite similar to that for [Fe(L1)Cl2](PF6) (Table
2).
7.7 for the first and second proton dissociations, respectively
(Table 1). These results suggest that the complexes are stronger
acids than acetic acid. The pKa’s for the complexes with L2
and L3 are slightly higher than those for [Fe(L1)Cl2](PF6) and
[Fe(L4)Cl2](PF6), and this is reflected in the pH values found
for aqueous solutions of the complexes at comparable concen-
trations. Consequently, under the reaction conditions of interest
(i.e., at or near physiological pH, ∼7.4), these FeIII(L) complexes
exist as an equilibrium mixture of the [Fe(L)(OH)2]+ and the
[Fe(L)(OH2)(OH)]2+ complexes.
Electrochemistry. The enzymatic reactions between SOD
and the superoxide anion are redox reactions. The oxidized
form of the enzyme SOD’s, such as an FeIII-SOD, oxidize the
superoxide anion to molecular oxygen and the reduced form of
SOD reduces superoxide (or its conjugate acid, HO2• 17) to
peroxide. This implies that a requirement for SOD activity be
that the redox potential for the metal complex should fall
between the redox potentials for the reduction and oxidation of
In unbuffered aqueous solutions (pH ∼ 3), three of the
complexes, [Fe(L1)Cl2](PF6), [Fe(L2)Cl2](PF6), and [Fe(L3)Cl2]-
(PF6), display well-defined redox couples in the scan range from
-0.75 to +0.50 V vs SCE. The redox potential for [Fe(L1)-
Cl2](PF6) is 0.11 V vs SCE or 0.35 V vs NHE with a peak
separation of 75 mV. The potential shifts 0.52 V toward more
positive potentials, compared to the values observed in aceto-
nitrile solution. Although a potential shift has been observed
for other FeII complexes with pentadentate ligands derived from
triazacycloalkanes,19 such a large shift is unexpected. This large
potential shift is attributed to the replacement of the two anions
from the coordination sphere by two neutral water ligands. The
FeII/FeIII redox potentials for [Fe(L2)Cl2](PF6) and [Fe(L3)Cl2]-
(PF6) are 100 mV higher than that for [Fe(L1)Cl2](PF6), again
contrasting with the similarity of the redox potentials for the
complexes in acetonitrile solutions. At pH ) 2.6, [Fe(L1)Cl2]-
(PF6) only shows one reduction peak; the disappearance of the
anodic peak may be due to the dissociation of the FeII complex
at low pH. In contrast, [Fe(L2)Cl2](PF6) exhibits reversible
redox behavior even at pH ) 2.09, but the anodic peak almost
disappears at pH ) 1.60. This implies that the more highly
preorganized and rigid macrocycle containing the two trans-
cyclohexano groups on the ligand stablizes the resultant FeII
complex. In the scan range from -0.50 to +0.50 V vs SCE,
[Fe(L4)Cl2](PF6) has one oxidation peak at 0.44 V vs NHE and
two reduction peaks at 0.31 and 0.14 V vs NHE, respectively.
Crystal Structure of [Fe(L2)Cl2](PF6). The hexafluoro-
phosphate salt of the FeIII complex of L2, [Fe(L2)Cl2](PF6),
crystallizes in the triclinic space group P1h (see Table 3 for
pertinent crystal data). A projection view of the molecule with
non-hydrogen atoms represented by 50% probability ellipsoids
is presented in Figure 3, and selected bond lengths (Table 4),
bond angles (Table 5), and atom coordinates (Table 6) are listed.
The FeIII is seven-coordinate with five nitrogen atoms from the
macrocyclic ligand and two trans-chloro anions. The five N
atoms are almost planar; thus, the coordination sphere can be
best described as a pentagonal bipyramid. There are five chelate
rings formed by Fe and N-C-C-N units. In these chelate
rings, the N-C-C-N torsion prefers the gauche conformation,
as observed in metal complexes with ethylenediamine and with
the same family of ligands with Mn(II) complexes.7,8 Four
N-C-C-N torsion angles in this complex have a gauche
conformation, while the fifth torsion N(4)-C(11)-C(12)-N(5)
has a almost eclipsed conformation. For five planar N atoms,
it is difficult for all five N-C-C-N torsion angles to have
ideal gauche conformation (around 60°) at the same time. Two
II
III
•-
-/HO2
superoxide anion, i.e., EO -/O < EFe /Fe < EO
(-0.3 to
2
2
2
+0.9 V17,18). From the data presented in this study (vide infra),
the iron couple involved in this catalytic process clearly falls
within this range. In acetonitrile solution, each of the four FeIII
complexes exhibits only one reversible redox wave in the scan
+
range from -1.5 to +1.5 V vs EAg/Ag , which can be assigned
to the FeII/FeIII couple. The cyclic voltammogram for [Fe(L1)-
Cl2](PF6) in an acetonitrile solution purged with nitrogen exhibits
+
a reversible couple at -0.19 V vs EAg/Ag , or -0.17 V vs NHE,
obtained by calibration with ferrocene. The cyclic voltammo-
gram for the complex [Fe(L1)Cl2](PF6) in an acetonitrile solution
that has been saturated with air is strikingly different. When
the scan starts at 0 V and proceeds toward negative potentials,
the complex is reduced to an FeII species and exhibits a cathodic
peak at -0.23 V. As the scan continues, molecular oxygen is
reduced to superoxide at -1.15 V. When the scan is then
returned to 0 from -1.4 V, no corresponding anodic peaks are
found, in contrast to the reversible redox behaviors for both
the FeIII complex and dioxygen in acetonitrile solutions. The
disappearance of the oxidation peak for the FeII species can be
rationalized. The concentration of the FeIII complex (0.8 mM)
is less than that of oxygen; thus, the FeII species is present at
lower concentration than superoxide ion, as was observed from
the reduction currents in the cyclic voltammogram. Even
without considering the possible oxidation of the FeII species
by dissolved oxygen, it is impossible for FeII to consume all of
the superoxide through a 1:1 chemical reaction. Hence, the
disappearance of the oxidation peak for superoxide suggests that
(17) Sawyer, D. T.; Valentine, J. S. Acc. Chem. Res. 1981, 14, 393.
(18) Sawyer, D. T.; Tsang, P. K. S. Free Radical Res. Commun. 1991,
12-13, 75.
(19) Zhang. D. Ph.D. Dissertation, University of Kansas, 1994.

Fe(III) Complexes as Superoxide Dismutase Mimics
Inorganic Chemistry, Vol. 37, No. 5, 1998 961
rigid cyclohexyl groups further force the N-C-C-N torsion
connecting them to adopt an eclipsed conformation. This forces
the molecule to adopt a pseudomirror symmetry, which passes
through Fe, N(2), two chloro atoms, and the centroid of the
bond C(11)-C(12). The eclipsed conformation increases strain
in the N(4)-C(11)-C(12)-N(5) unit and results in the relatively
larger N-C-C angle (111.1°), compared to the average angle
of 107.1° in the other two diazaethane units.
The five Fe-N bond lengths are 2.290(3), 2.250(3), 2.280-
(3), 2.259(3), and 2.275(2) Å, with an average of 2.271 Å. The
N atom not connected to the cyclohexyl groups possesses a
shorter coordination bond, possibly due to its relative freedom.
The five chelate angles N-Fe-N are 73.37(9), 73.72(10), 72.67-
(9), 69.11(9), and 72.39(0)°, with an average of 72.25°. The
sum of the chelate angles is 361.2°, very close to 360° for an
ideal planar structure. It is interesting to compare this structure
with that of the corresponding [MnII(L2)Cl2] complex containing
the high-spin d5 MnII ion in the identical ligand environment:
the MnII-N bond distances are slightly longer, with the average
distance at 2.31 Å, and the Mn-Cl distances are 2.56 and 2.63
Å,20 whereas the FeIII analogue with L2 has Fe-Cl bond lengths
of 2.33 and 2.35 Å. The shorter bond lengths undoubtedly are
due to the smaller size/greater charge of the +3 high-spin d5
FeIII center.
Figure 4. Dependence of pH on the kcat for the new FeIII complexes.
react with the superoxide, suggesting that catalytic SOD activity
may be possible. The SOD activity assay/experiment is
designed so that the initial concentration of superoxide . [FeIII-
(L)]. Under these conditions, the two general reactions govern-
ing the overall catalytic redox reaction are reduction of FeIII
(eq 1) and oxidation of a reduced FeII species with the strong
k1
9
O2- + [FeIII(L)]
8 O2 + [FeII(L)]
(1)
(2)
Kinetics of Superoxide Decay and Mechanism Studies.
Spontaneous superoxide decomposition in aqueous solution is
a second-order reaction.13 In the presence of superoxide
dismutase, the decomposition of superoxide is much faster (at
pH > 7) and the catalyzed reaction is first-order in superoxide
concentration. For example, the reported rate for bacterial Fe-
k2
HO2 + [FeII(L)]
9
8 HO2- + [FeIII(L)]
•
• 20
oxidant HO2 . Since the initial concentration of O2- is much
-
larger than the catalyst concentration, the amount of O2
SOD is on the order of 3 × 10+8 M-1 s-1 21
.
Examples of
consumed by the FeIII species must equal that reacting with FeII,
assuming no reaction between the iron complexes and other
compounds, such as hydrogen peroxide. From eq 3, it can been
synthetic low molecular weight superoxide dismutase catalysts
employing iron as the redox-active metal are relatively sparse.
Only a few synthetic complexes have been reported, and these
have been shown to possess modest catalytic SOD activity. For
example, FeIII(EDTA) has been reported to catalyze the dis-
mutation at a rate approximately 103 times slower than the native
enzymes.22 Synthetic porphyrins containing iron also have been
reported but suffer from catalytic deactivation owing to dimer
formation.23 We have previously reported that the Fe(III)
complex of tris[N-(2-pyridylmethyl)-2-aminoethyl]amine (Fe-
TPAA) was the most active Fe-based mimic which we had
d[O2 ]/dt ) {k1[FeIII(L)] + k2([FeII(L)][H+]/Ka])}[O2 ] (3)
-
-
Ka ) [O2 ][H+]/[HO2]
-
kobs ) k1[FeIII(L)] + k2′[FeII(L)] )
(k1RFeIII(L) + k2′RFeII(L))[Fe]tot (4)
k2′ ) k2[H+]/Ka
studied with a kcat of 2.16 × 10+6 M-1 s-1 23,24
.
We have found that the Mn(II) complexes with pentaazacy-
cloalkane ligands display very high SOD activity and catalyze
the first-order decomposition of superoxide with rate constants
as high as 1.2 × 108 M-1 s-1 at pH ) 7.4.7,8 For this reason
we synthesized the corresponding Fe(III) complexes with some
of our representative pentaazacycloalkane ligands in order to
determine whether active synthetic Fe-based mimics could be
synthesized.
kcat ) kobs/[Fe]tot ) (k1RFeIII(L) + k2′RFeII(L)
)
(5)
(6)
kcat ) 2k1RFeIII(L) ) 2k2RFeII(L)
seen that the decomposition of superoxide catalyzed by metal
complexes is first-order in superoxide. The first-order rate
constant contains the constant concentration of the catalyst.
The kobs values measured for these four new FeIII complexes
under conditions of varying pH all show pure first-order
superoxide decay behavior. The kcat for a particular FeIII
complex is obtained from the plot of the kobs value versus the
initial concentration of FeIII complex. The slopes of these plots
are equal to the kcat value at a given pH.13 In Figure 4 are shown
the values for kcat for the four new complexes over a range of
pH. For three of the four complexes, the kcat increases linearly
as the pH decreases. In contrast, the complex Fe(L4) is
insensitive to the pH. This pH-dependent phenomenon has also
been observed for the manganese(II) complexes with the same
three pentaazacycloalkane ligands, and pH independence was
also observed with the analogous MnII complex of ligand L4.
For the MnII(L) complexes, this behavior has been rationalized
The new Fe(III) complexes reported here are the first iron
complexes to be synthesized with pentaazacycloalkanes and to
be investigated as potential SOD mimics. Their redox potentials
lie in a range such that both oxidized and reduced species can
(20) Riley, D. P.; Lennon, P. J.; Neumann, W. L.; Weiss, R. H. J. Am.
Chem. Soc. 1997, 119, 6522.
(21) (a) Fee, J. A.; McClune, G. J.; O’Neill, P.; Fielden, E. M. Biochem.
Biophys. Res. Commun. 1981, 100, 377. (b) Lavelle, F.; McAdam,
M. E.; Fielden, E. M.; Puget, K.; Michelson, A. M. Biochem. J. 1977,
161, 3.
(22) McClune, G. J.; Fee, J. A.; McClusky, G. A.; Groves, J. T. J. Am.
Chem. Soc. 1977, 99, 5220.
(23) Weiss, R. H.; Flickinger, A. G.; Rivers, W. J.; Hardy, M. M.; Aston,
K. W.; Ryan, U. S. J. Biol. Chem. 1993, 268, 23049.
(24) Nagano, T.; Hirano, T.; Hirobe, M. J. Biol. Chem. 1989, 264, 9243.

962 Inorganic Chemistry, Vol. 37, No. 5, 1998
Zhang et al.
in terms of a mechanism that is a combination of two separate
rate-determining steps: a pH-dependent and a pH-independent
process. The pH-dependent process is not operative with the
pentamethyl ligand L48, but with the remaining complexes, it
can play a dominant role in the rate at physiological pH. This
pH-dependent process involves reduction of protonated super-
an inner-sphere binding to FeIII. This requires the loss of a
bound axially coordinated water molecule. The rates of the
water exchange on the FeIII complex may then be rate-limiting,
and the generation of the vacant coordination site governs the
catalytic rate. In general, water exchange rates on high-spin
FeIII in acid are measured to be faster than 10+3 s-1 and several
orders of magnitude faster than this when the pH is higher and
a hydroxo ligand is bound to FeIII, labilizing the trans-
aquoligand to dissociation.28 The rate of catalysis observed with
these complexes requires that koff for the bound water be on
the order of 10+7 s-1, a value entirely consistent with a water
bound to the high-spin FeIII center with a trans-hydroxosas
occurs with this system.
To probe this further, the catalytic rate of superoxide
disproportionation was studied in D2O at pH ) 7.55.29 In D2O,
the rate of water exchange with the bound FeIII may show a
slight secondary isotope effect, since the isotope effect observed
in H2O (D2O) for water exchange on +3 metal ions is expected
to be small, as is observed with +3 lanthanide ions.30 In
contrast, a large inverse isotope effect should be expected
(perhaps as large as 3). This unusual effect would be predicted
because the pKa of a coordinated heavy water will be higher
than that of a bound light water; thus, the equilibrium
concentration of the [FeIII(L)(OH)(OH2)]2+ complex will be
shifted, so that at any pH in the range 7.4-8.1 the D2O system
will contain a higher concentration of active catalytic species,
[FeIII(L)(OH)(OH2)]2+. Thus, a faster apparent rate should be
observed in D2O because there will be a higher concentration
of the active form of the catalyst. At pH ) 7.60 the rate of
dismutation of superoxide was measured with the complex [FeIII-
(L2)Cl2]+, and in D2O the kcat was found to be 4.41 × 10+7
M-1 s-1scorresponding to a rate enhancement of 2.13 times
that observed in light water. This result provides additional
support for the conclusion that generation of a vacant coordina-
tion site on the FeIII followed by rapid binding to FeIII by
superoxide with subsequent reduction of FeIII (eqs 7-9) is
•
oxide HO2 via proton-coupled electron transfer from Mn(II),
• 20
i.e., H atom transfer from a bound water to HO2 . Proton-
coupled electron transfer is gaining experimental support as an
important pathway of electron transfer in biological systems such
as photosystem II,25 as this path does not require charge
separation to build up in the transition state. Recent thermo-
dynamic data show that the bond dissociation energy (BDE) of
an O-H bond of a water bound to a MnIII(Salen) derivative is
•
86 kcal/mol.26 Thus, H atom transfer to HO2 is energetically
allowed as the BDE of the O-H bond of hydrogen peroxide is
reported to be 87.2 kcal/mol.27
In contrast, the pH-independent step involves prior binding
of superoxide anion to a vacant coordination site on Mn(II)
followed by fast oxidation of the MnII complexsa process
dependent on the rate of generation of vacant coordination sites
on the Mn(II) center. However, for the FeIII complexes, the
deprotonation of coordinated water molecules must be taken
into account, since the aquo Fe(III) complexes described are
much stronger acids than the corresponding aquo Mn(II)
complexes, for which the first pKa of the axial bound water is
∼11.2.8 Thus, the axial-site ligand speciation is much more
critical for these Fe(III) complexes. At the pH range in which
the kcat determinations are made (pH ∼ 7.6-8.1) most of the
FeIII(L) complex is in the dihydroxo form, [Fe(L)(OH)2]+.
To probe the mechanism of this system further, we utilized
the complex with L2 for further spectrophotometric studies. The
intense (ꢀ ∼9400) and isolated absorption band (λmax ) 294
nm) of this complex made it possible to probe the nature of the
rate-determining step in this catalytic system. Using [Fe(L2)-
Cl2]+ in water with HEPES buffer at pH ) 7.55, we studied
the decay of superoxide under catalytic conditions ([O2•-]/[Fe]
> 8) in which the spectral region from 230 to 420 nm is
monitored using an OLIS rapid-scan spectrophotometer stopped-
[FeIII(L)(OH)(OH2)]2+ f [FeIII(L)(OH)]2+ + H2O (7)
[FeIII(L)(OH)]2+ + O2•- f [FeIII(L)(OH)(O2)]+ (8)
flow system. In this way, both the superoxide decay (λmax
)
245 nm, ꢀ ∼2200) and any change in the spectrum of the FeIII
complex can be monitored simultaneously at 294 nm. In this
series of experiments at pH ) 7.55, the FeIII complex is present
as an approximately 1:1 mixture of the [Fe(L)(OH)2]+ and [Fe-
(L)(OH)OH2)]2+ complexes. During the entire course of the
first-order decay of the superoxide (complete in 40 ms), no
change is observed in the spectrum of the FeIII complex. This
result clearly indicates that reduction of high-spin d5 FeIII to
FeII is rate-determining during this catalytic process and that
throughout the reaction most of the Fe is present as FeIII. This
is the inverse of the situation for the MnII system, in which
oxidation of the corresponding high-spin d5 metal ion is rate-
determining.
If reduction of an FeIII complex to an FeII complex is rate-
determining, it is logical, on the basis of electrochemical data,
that since the [FeIII(L)(OH)(OH2)]2+ complex is far easier to
reduce than the dihydroxo FeIII(L) complex, it is indeed the
catalytically active species in this system. This also means that
the reduction of an aquo hydroxo FeIII(L) complex occurs by
[FeIII(L)(OH)(O2)]+ f [FeII(L)(OH)]+ + O2
(9)
operating in this catalytic system. Thus, if [Fe(L)]tot ∼ [FeIII-
(L)], then kobs is approximated by the expression of eq 10. Since
kobs ) k1[FeIII(L)(OH)(OH2)]2+
(10)
the concentration of the aquo hydroxo species is related to the
total [FeIII ] by the Ka′ of the second bound water, the kobs
becomes as defined in eq 11. [Fe]total([H+]/([H+] + Ka′) is equal
kobs ) k1[FeIII(L)]tot([H+]/([H+] + Ka′)
(11)
to the concentration of aquo hydroxo species. From these
equations, kcat is pH dependent, and the plot of [H+]/([H+] +
Ka) versus [H+] would generate a curve over a wide enough
pH range. However, if [H+] , Ka, kcat is equal to 2k1[H+]/
(Ka), which has a linear relationship with [H+]. On the other
(25) Haganson, C. W.; Lydakis-Simantris, N.; Tang, X.; Tommos, C.;
Warncke, K.; Babcock, G. T.; Diner, B. A.; McCracken, J.; Styring,
S. Photosynth. Res. 1995, 46, 177.
(26) Caudle, M. T.; Pecoraro, V. L. J. Am. Chem. Soc. 1997, 119, 3415.
(27) Howard, C. J. J. Am. Chem. Soc. 1980, 102, 6937.
(28) Basolo, F.; Pearson, R. G. Mechanisms of Inorganic Reactions; John
Wiley & Sons: New York, 1967; Chapter 3, pp 145-158.
(29) Covington, A. K.; Paabo, M.; Robinson, R. A.; Bates, R. G. Anal.
Chem. 1968, 40, 501.
(30) Reidler, J.; Silber, H. B. J. Phys. Chem. 1973, 77, 1275.

Fe(III) Complexes as Superoxide Dismutase Mimics
Inorganic Chemistry, Vol. 37, No. 5, 1998 963
activity with a pH behavior unlike the other complexes, as its
redox potential and bound water pKa values are virtually
identical. Possible explanations are (1) that this complex readily
undergoes ligand modification under conditions of catalysis
(presence of an excess of strong oxidant + strong reductant)
producing an altered catalyst (e.g., imine formation) with
different bound water pKa values and a slower inherent rate for
superoxide dismutation and (2) that the complex may more
readily form oxo-bridged dimers with lower inherent kcat values
than a monomeric structure and pKa values not in the range of
the pH of interest for these studies. The latter explanation is
not consistent with the observed first-order rate dependence of
kcat on the concentration of [Fe(L4)Cl2]+, but the presence of
other species arising from ligand oxidation processes is con-
sistent with the observation of two additional reduction processes
observed in the cyclic voltammetry for this complex. None of
the other complexes exhibit the new waves. Thus, it is likely
that the behavior observed with the complex derived from L4
may arise from redox chemistry occurring on the ligand itself.
From our studies, we find that the pH of the solution
influences kcat by changing the distribution of the different
solvated FeIII complexes. If other factors are similar, a higher
pKa for the aquo ligand will raise the kcatsas is observed here.
This mechanistic insight should aid in the design of ligands
which will promote enhanced SOD activity by decreasing the
acidity of the second bound water on the FeIII center. In
conclusion, we are continuing to pursue the goal of enhanced
catalytic activity for these FeIII-based SOD catalysts by modify-
ing the pentaaza macrocyclic ligand structure with substituent
variations which allow us to alter the bound water pKa while
retaining ligands which are not subject to oxidative change
themselves.
Figure 5. Relationship between the concentration of the [Fe(L)(OH)-
(OH2)]2+ species and the pH of the solution.
hand, when [H+] . Ka, kcat is equal to 2k1 and is pH
independent. In this system near physiological pH (7.4), [H+]
is approximately the value of Ka′ for these four FeIII complexes.
As discussed earlier, the FeIII complexes are seven-coordinate
with five nitrogen atoms from the macrocyclic ligand and two
oxygens from water molecules or hydroxide ions. The pKa’s
for the water molecules indicate that [FeIIIL(H2O)(OH)]2+ and
[FeIIIL(OH)2]+ are dominant species in solution under the
conditions of interest near physiological pH. Assuming that
reduction (eq 1) is the rate-determining step and [FeIII(L)(H2O)-
(OH)]2+ is the active species, the plots of the fraction of the
aquo hydroxo species ([H+]/([H+] + Ka′) against [H+] are
presented in Figure 5 for the four FeIII complexes using their
experimentally derived Ka′ values (Table 1). It can be seen from
the data over this pH range that the [H+] dependence on the
k
cat rate for these complexes is indeed expected to be linear on
the basis of the dependence of the rate on the concentration of
the active catalytic species, [FeIII(L)(H2O)(OH)]2+
Acknowledgment. The authors thank Mr. Karl Aston, Mr.
Hayat Rahman, Mr. Willie Rivers, and Mr. Kirby Sample for
their outstanding technical assistance.
.
Three of these complexes (with L1, L2, L3) are in excellent
agreement with this rationale for pH dependence. The complex
with the pentamethyl ligand ([Fe(L4)Cl2]+) clearly does not
appear to fit with this explanation of pH dependence, since this
complex has virtually no pH dependence on its catalytic rate.
It is not clear why this complex possesses a lower catalytic
Supporting Information Available: Tables listing the positional
and isotopic displacement coefficients for hydrogen atoms and the
anisotropic displacement coefficients for the non-hydrogen atoms (2
pages). See any current masthead page for ordering information.
IC970861H