C-H bond activation by a ferric methoxide complex: Modeling the rate-determining step in the mechanism of lipoxygenase

Article abstract of DOI:10.1021/ja016451g

Lipoxygenases are mononuclear non-heme iron enzymes that regio- and stereospecifcally convert 1,4-pentadiene subunit-containing fatty acids into alkyl peroxides. The rate-determining step is generally accepted to be hydrogen atom abstraction from the pentadiene subunit of the substrate by an active ferric hydroxide species to give a ferrous water species and an organic radical. Reported here are the synthesis and characterization of a ferric model complex, [Fe^III(PY5)(OMe)](OTf)₂, that reacts with organic substrates in a manner similar to the proposed enzymatic mechanism. The ligand PY5 (2,6-bis(bis(2-pyridyl)methoxymethane)pyridine) was developed to simulate the histidine-dominated coordination sphere of mammalian lipoxygenases. The overall monoanionic coordination provided by the endogenous ligands of lipoxygenase confers a strong Lewis acidic character to the active ferric site with an accordingly positive reduction potential. Incorporation of ferrous iron into PY5 and subsequent oxidation yields a stable ferric methoxide species that structurally and chemically resembles the proposed enzymatic ferric hydroxide species. Reactivity with a number of hydrocarbons possessing weak C-H bonds, including a derivative of the enzymatic substrate linoleic acid, scales best with the substrates' bond dissociation energies, rather than pK_a's, suggesting a hydrogen atom abstraction mechanism. Thermodynamic analysis of [Fe^III(PY5)(OMe)](OTf)₂ and the ferrous end-product [Fe^II(PY5)(MeOH)](OTf)₂ estimates the strength of the O-H bond in the metal bound methanol in the latter to be 83.5 ± 2.0 kcal mol^-1. The attenuation of this bond relative to free methanol is largely due to the high reduction potential of the ferric site, suggesting that the analogously high reduction potential of the ferric site in LO is what allows the enzyme to perform its unique oxidation chemistry. Comparison of [Fe^III(PY5)(OMe)](OTf)₂ to other coordination complexes capable of hydrogen atom abstraction shows that, although a strong correlation exists between the thermodynamic driving force of reaction and the rate of reaction, other factors appear to further modulate the reactivity.

Full text of DOI:10.1021/ja016451g

Published on Web 12/05/2001
C-H Bond Activation by a Ferric Methoxide Complex:
Modeling the Rate-Determining Step in the Mechanism of
Lipoxygenase
Christian R. Goldsmith, Robert T. Jonas, and T. Daniel P. Stack*
Contribution from the Department of Chemistry, Stanford UniVersity, Stanford, California 94305
Received June 19, 2001
Abstract: Lipoxygenases are mononuclear non-heme iron enzymes that regio- and stereospecifcally convert
1,4-pentadiene subunit-containing fatty acids into alkyl peroxides. The rate-determining step is generally
accepted to be hydrogen atom abstraction from the pentadiene subunit of the substrate by an active ferric
hydroxide species to give a ferrous water species and an organic radical. Reported here are the synthesis
and characterization of a ferric model complex, [FeIII(PY5)(OMe)](OTf)2, that reacts with organic substrates
in a manner similar to the proposed enzymatic mechanism. The ligand PY5 (2,6-bis(bis(2-pyridyl)methoxy-
methane)pyridine) was developed to simulate the histidine-dominated coordination sphere of mammalian
lipoxygenases. The overall monoanionic coordination provided by the endogenous ligands of lipoxygenase
confers a strong Lewis acidic character to the active ferric site with an accordingly positive reduction potential.
Incorporation of ferrous iron into PY5 and subsequent oxidation yields a stable ferric methoxide species
that structurally and chemically resembles the proposed enzymatic ferric hydroxide species. Reactivity
with a number of hydrocarbons possessing weak C-H bonds, including a derivative of the enzymatic
substrate linoleic acid, scales best with the substrates’ bond dissociation energies, rather than pKa’s,
suggesting a hydrogen atom abstraction mechanism. Thermodynamic analysis of [FeIII(PY5)(OMe)](OTf)2
and the ferrous end-product [FeII(PY5)(MeOH)](OTf)2 estimates the strength of the O-H bond in the metal
bound methanol in the latter to be 83.5 ( 2.0 kcal mol-1. The attenuation of this bond relative to free
methanol is largely due to the high reduction potential of the ferric site, suggesting that the analogously
high reduction potential of the ferric site in LO is what allows the enzyme to perform its unique oxidation
chemistry. Comparison of [FeIII(PY5)(OMe)](OTf)2 to other coordination complexes capable of hydrogen
atom abstraction shows that, although a strong correlation exists between the thermodynamic driving force
of reaction and the rate of reaction, other factors appear to further modulate the reactivity.
Scheme 1
Introduction
Lipoxygenases (LOs) are mononuclear non-heme iron en-
zymes, found in both plants and animals, that catalyze the regio-
and stereospecific dioxygenation of cis,cis-1,4-pentadiene-
containing fatty acids to alkyl hydroperoxides. Linoleic acid,
the natural substrate for most plant LOs, is oxygenated with
positional stereospecificity to produce hydroperoxyoctadeca-
dienoic acids, completing the first step in the biosynthesis of
the growth-regulation substance jasmonic acid and of the wound-
healing compounds traumatin and traumatic acid (Scheme 1).1
Mammalian LOs initiate the synthesis of physiological effectors,
such as leukotrienes and lipoxins, from arachidonic acid. The
inhibition of human LO activity is of considerable interest to
pharmaceutical concerns, as various LOs have been implicated
in inflammatory conditions, such as arthritis, bronchial asthma,
and atherosclerosis.1 Recently, LO inhibitors have also been
found to inhibit tumor growth.2,3
The iron-binding ligands of the LOs have been determined
by sequence homology and crystallographic determinations.
X-ray structures of two plant isozymes, soybean lipoxygenase-1
(SLO-1) and soybean lipoxygenase-3 (SLO-3), are known.4-6
Three imidazole nitrogens of histidine residues, an amide oxygen
of an asparagine residue, and the carboxylate oxygen of the
C-terminus residue constitute the endogenous, peptide-based
ligands of the metal center. Protein sequence comparisons of
(2) Castonguay, A.; Rioux, N. Carcinogenesis 1998, 19, 1393-1400.
(3) Adrian, T. E.; Ding, X. Z.; Iverson, P.; Cluck, M. W.; Knezetic, J. A.
Biochem. Biophys. Res. Commun. 1999, 261, 218-223.
* To whom correspondence should be addressed. Tel.: (650) 725-8736.
Fax: (650) 725-0259. E-mail: stack@stanford.edu.
(1) Nelson, M. J.; Seitz, S. P. In ActiVe Oxygen in Biochemistry; Valentine, J.
S., Foote, C. S., Greenberg, A., Liebman, J. F., Eds.; Blackie Academic &
Professional: London, 1995; Vol. 3.
(4) Skrzypczak-Jankun, E.; Amzel, L. M.; Kroa, B. A.; Funk, M. O., Jr.
Proteins: Struct., Funct. Genet. 1997, 29, 15-31.
(5) Minor, W.; Steczko, J.; Stec, B.; Otwinowski, Z.; Bolin, J. T.; Walter, R.;
Axelrod, B. Biochemistry 1996, 35, 10687-10701.
9
10.1021/ja016451g CCC: $22.00 © 2002 American Chemical Society
J. AM. CHEM. SOC. VOL. 124, NO. 1, 2002 83

A R T I C L E S
Goldsmith et al.
the mammalian LOs relative to plant LOs. Selective mutation
of the coordinating asparagine to histidine in SLO-1 supports
this hypothesis. The N694H SLO-1 mutant has a kcat ≈ 10 ( 2
s-1, more similar to the rate of 15-human LO (kcat ≈ 6.2 ( 0.1
s-1) than the wild-type SLO-1 (kcat ≈ 280 ( 8 s-1).16
LO is one of a limited number of mononuclear non-heme
iron enzymes in which catalysis is proposed to occur through
substrate activation rather than O2 activation.17 A kinetic isotope
effect (KIE) of greater than 50 is measured for the LO catalytic
cycle (305 K), indicating that C-H bond cleavage is involved
in the rate-determining step (RDS).18-21 In the enzyme, two
additional processes are found to be partially rate determining
at 298 K in the catalytic cycle: substrate binding and dissocia-
tion, and a yet to be identified solvent-sensitive step.22 As the
solvent temperature is increased, the viscosity of the system
decreases until, at 310 K, the C-H bond cleavage reaction is
fully rate limiting. In the generally accepted mechanism, the
RDS involves hydrogen atom abstraction (HA) from the fatty
acid substrate with concomitant reduction of the ferric-
hydroxide center to a ferrous-water species and formation of
a fatty acid radical (Scheme 2).1 Another mechanism, involving
an organometallic ferric species, has also been proposed.23
Diffusion-limited trapping of the radical intermediate by O2
followed by oxidation of the ferrous-water species to the
ferric-hydroxide species by the peroxyl radical completes the
catalytic cycle.
While stereospecific C-H bond activation of the cis,cis-1,4-
pentadiene subunit has been established during LO activity, the
characterization of the resulting substrate radical is limited.1
Kinetic experiments indicate that C-H bond cleavage occurs
prior to the incorporation of O2 by the activated substrate-
enzyme complex.24,25 Starting with the fatty acid substrate
linoleic acid, the stereospecific product 13(S)-hydroperoxy-
9(Z),11(E)-octadecadienoic acid (13-HPOD) results from SLO-1
activity (Scheme 1); in contrast, autoxidation of linoleic acid
radicals in the absence of enzyme results in multiple regio- and
stereoisomer hydroperoxide products.26,27
Figure 1. Three-dimensional representation of the non-heme iron site of
rabbit 15-lipoxygenase as determined by crystallography, including a scheme
of the ligand coordination set. Fe-N(His) distances average 2.2 Å. The
Fe-O(Ile) distance is 2.1 Å.9
over 20 LOs from both plants and animals reveal the near
conservation of the five structurally characterized iron-binding
ligands. An exception is found for the 15-lipoxygenases from
humans and rabbits, in which the asparagine amide oxygen is
replaced by the imidazole nitrogen of a histidine residue.7,8 The
reported X-ray structure of rabbit 15-lipoxygenase (15-LO)
confirms that the endogenous ligation consists of four imidazole
nitrogens and a carboxylate oxygen (Figure 1).9
Spectroscopic studies have investigated the iron coordination
differences between the resting and active forms of the LO
enzymes. The isolated resting forms of soybean and mammalian
LOs contain a high-spin Fe(II) center.10-12 Spectroscopic data
of SLO-1 indicate that the iron center exists as a mixture of
five- and six-coordinate species in solution with possible
detachment of the asparagine-derived ligand from the iron
center.1 Though not observed in the crystal structures, an
exogenous water molecule likely completes the octahedral
coordination of the Fe(II) center.5 Unlike many of the mono-
nuclear non-heme iron enzymes, the ferrous form, Fe(II), of
LO is inactive, and the iron center must be oxidized to the ferric
state, Fe(III), for catalytic activity.1 Autoactivation of SLO
occurs in the presence of the hydroperoxy product of linoleic
acid.1 Extended X-ray absorption fine structure (EXAFS) and
magnetic circular dichroism studies on the Fe(III) form of SLO-1
indicate a six-coordinate metal center.1 EXAFS studies reveal
the presence of a short iron-ligand bond distance of 1.88 Å
that is attributed to a single hydroxide ligand.13 The histidine-
for-asparagine substitution from plant to mammalian LO results
in a more stabilized six-coordinate Fe(III) species, the increased
stability being attributed to the stronger electron donation of
an imidazole ligand relative to an amide oxygen.14,15 The
stabilized Fe(III) center may explain the reduced reactivity of
While the reaction of free radicals with O2 is well known,
few metal-based systems are known to perform C-H bond
activation through a well-characterized HA mechanism. Such
a mechanism has been identified for the reactions of chromyl
chloride and permanganate metal-oxo species with simple
organic substrates.28-32 More recently, a ferric tris(2,2′-bi-
(15) Pavlosky, M. A.; Zhang, Y.; Westre, T. E.; Gan, Q.-F.; Pavel, E. G.;
Campochiaro, C.; Hedman, B.; Hodgson, K. O.; Solomon, E. I. J. Am.
Chem. Soc. 1995, 117, 4316-4327.
(16) Holman, T. R.; Zhou, J.; Solomon, E. I. J. Am. Chem. Soc. 1998, 120,
12564-12572.
(17) Que, L., Jr.; Ho, R. Y. N. Chem. ReV. 1996, 96, 2607-2624.
(18) Jonsson, T.; Glickman, M. H.; Sun, S. J.; Klinman, J. P. J. Am. Chem.
Soc. 1996, 118, 10319-10320.
(6) Boyington, J. C.; Gaffney, B. J.; Amzel, L. M. Science 1993, 260, 1482-
1486.
(7) Fleming, J.; Thiele, B. J.; Chester, J.; O’Prey, J.; Janetzki, S.; Aitken, A.;
Anton, I. A.; Rapoport, S. M.; Harrison, P. R. Gene 1989, 79, 181-188.
(8) Sigal, E.; Craik, C. S.; Highland, E.; Grunberger, D.; Costello, L. L.; Dixon,
R. A. F.; Nadel, J. A. Biochem. Biophys. Res. Commun. 1988, 157, 457-
464.
(19) Glickman, M. H.; Wiseman, J. S.; Klinman, J. P. J. Am. Chem. Soc. 1994,
116, 793-794.
(20) Hwang, C.-C.; Grissom, C. B. J. Am. Chem. Soc. 1994, 116, 795-796.
(21) Lewis, E. R.; Johansen, E.; Holman, T. R. J. Am. Chem. Soc. 1999, 121,
1395-1396.
(9) Gillmor, S. A.; Villasen˜or, A.; Fletterick, R.; Sigal, E.; Browner, M. F.
Nat. Struct. Biol. 1997, 4, 1003-1009.
(22) Glickman, M. H.; Klinman, J. P. Biochemistry 1995, 34, 14077-14092.
(23) Corey, E. J.; Nagata, R. J. Am. Chem. Soc. 1987, 109, 8107-8108.
(24) Glickman, M. H.; Cliff, S.; Thiemens, M.; Klinman, J. P. J. Am. Chem.
Soc. 1997, 119, 11357-11361.
(10) Slappendel, S.; Malmstro¨m, B. G.; Petersson, L.; Ehrenberg, A.; Veldink,
G. A.; Vliegenthart, J. F. G. Biochem. Biophys. Res. Commun. 1982, 108,
673-677.
(25) Glickman, M. H.; Klinman, J. P. Biochemistry 1996, 35, 12882-
(11) Axelrod, B.; Cheesbrough, T. M.; Laaskso, S. Methods Enzymol. 1981,
71, 441-451.
12892.
(26) Porter, N. A. Acc. Chem. Res. 1986, 19, 262-268.
(27) Porter, N. A.; Wujek, D. G. J. Am. Chem. Soc. 1984, 106, 2626-2629.
(28) Gardner, K. A.; Kuehnert, L. L.; Mayer, J. M. Inorg. Chem. 1997, 36,
2069-2078.
(12) Pistorius, E. K.; Axelrod, B. J. Biol. Chem. 1974, 249, 3183-3186.
(13) Scarrow, R. C.; Trimitsis, M. G.; Buck, C. P.; Grove, G. N.; Cowling, R.
A.; Nelson, M. J. Biochemistry 1994, 33, 15023-15035.
(14) Zhang, Y.; Gan, Q.-F.; Pavel, E. G.; Sigal, E.; Solomon, E. I. J. Am. Chem.
Soc. 1995, 117, 7422-7427.
(29) Cook, G. K.; Mayer, J. M. J. Am. Chem. Soc. 1995, 117, 7139-7156.
(30) Gardner, K. A.; Mayer, J. M. Science 1995, 269, 1849-1851.
9
84 J. AM. CHEM. SOC. VOL. 124, NO. 1, 2002

Synthesis and Reactivity of a Lipoxygenase Mimic
A R T I C L E S
Scheme 2
imidazoline) system has been shown to exhibit similar reactiv-
ity.33 In these studies, a thermodynamic bond strength approach,
similar to that used to determine the homolytic X-H bond
dissociation energy in many organic and organometallic34 com-
pounds, provides an understanding of the HA ability of a metal
complex. Three primary factors affect the driving force for
metal-based HA: the redox potential of the metal center, the
strength of the substrate C-H bond broken, and the energy of
the O-H (or N-H) bond formed in the reduced metal complex.
This study focuses on a pentadentate ligand designed and
synthesized to mimic specific attributes of the iron coordination
site in the LOs, namely a five-coordinate, square pyramidal,
endogenous environment that can accommodate a sixth exog-
enous ligand. Using a pentapyridyl ligand, 2,6-bis(bis(2-pyridyl)-
methoxymethane)pyridine (PY5), an Fe(III) complex with an
axial methoxide ligand, [FeIII(PY5)(OMe)](OTf)2 (2) (OTf )
triflate, CF3SO3-), has been synthesized. This model complex
shares many similar structural and spectroscopic properties with
the LOs. Additionally, 2 promotes the cleavage of weak C-H
bonds of substrates with concomitant reduction of the iron center
to the ferrous complex [FeII(PY5)(MeOH)](OTf)2 (1). The
reactivity profile is consistent with a HA mechanism, providing
chemical precedent for the C-H bond activation step in the
proposed mechanism of LO.35 The chemical similarities of this
model complex to the iron site in LOs strongly support the HA
mechanism by a ferric-hydroxide species in the native system.
coordination geometry. [FeII(PY5)(MeOH)](OTf)2 (1) is isolated
directly from an equimolar mixture of FeII(OTf)2 and PY5 in
methanol (MeOH). Addition of either H2O2 or iodosobenzene
to a yellow solution of 1 in MeOH results in the formation of
the orange ferric complex [FeIII(PY5)(OMe)](OTf)2 (2). A UV/
[FeII(PY5)(MeOH)]2+ 0.5H2O2 or C6H5IO8 [FeIII(PY5)(OMe)]2+ (1)
vis spectroscopic titration of the oxidation of 1 indicates that
the reaction is complete after addition of 0.5 equiv of H2O2 or
iodosobenzene. Addition of >1.5 equiv of H2O2 to 1 gives a
thermally unstable blue species that is proposed to be a low-
spin hydroperoxide-bound ferric species, [FeIII(PY5)(OOH)]-
(OTf)2.36,37 The methoxy complex, 2, is isolated as a stable red/
excess H2O2
[FeII(PY5)(MeOH)]2+
8 [FeIII(PY5)(OOH)]2+ (2)
orange powder in high yield (75%). The aerobic and thermal
stability of 2 is unusual, as no other ferric complex with PY5
has been successfully isolated to date.
Solid State Structure. Structural and refinement data for the
crystal structures of the free ligand PY5, 1, and 2 are
summarized in Table 1. The atom labeling scheme of the ligand
in all three structures is congruent in order to facilitate
comparison of the structural parameters. Both iron centers of 1
and 2 are coordinated in an octahedral environment, with the
five nitrogen donors of PY5 in a square pyramidal arrangement
and an oxygen atom of a methanol (1) or a methoxide (2)
completing the coordination sphere. For the purpose of discus-
sion, an equatorial plane (Pleq) is defined by the least-squares
plane of the four nitrogen atoms N2f5 in their respective pyridine
subunits Py2f5, such that the axial positions are occupied by
N1 of Py1 and the heteroatom of the exogenous ligand.
[FeII(PY5)(MeOH)](OTf)2‚(MeOH) was crystallized as a
stable yellow/green prismatic solid with an asymmetric unit
containing the cation [FeII(PY5)(MeOH)]2+, two triflate anions,
and a MeOH solvate molecule. The hydrogen atom of the
coordinated MeOH was located by a difference Fourier map
and is involved in a hydrogen-bonding network with the MeOH
of solvation, which in turn hydrogen bonds to one of the triflate
Results and Analysis
Ferric Complex Synthesis. In an effort to isolate structurally
and functionally relevant analogues of LO enzymes, the
pentadentate ligand, 2,6-bis(bis(2-pyridyl)methoxymethane)-
pyridine (PY5), was designed and synthesized.35,36 The ligand
is composed of five pyridine subunits and accommodates a
single metal ion in a nearly idealized square pyramidal
(31) Cook, G. K.; Mayer, J. M. J. Am. Chem. Soc. 1994, 116, 1855-1868.
(32) Mayer, J. M. Acc. Chem. Res. 1998, 31, 441-450.
(33) Roth, J. P.; Mayer, J. M. Inorg. Chem. 1999, 38, 2760-2761.
(34) Astruc, D.; Trujillo, H. A.; Casado, C. M.; Ruiz, J. J. Am. Chem. Soc.
1999, 121, 5674-5686.
(35) Jonas, R. T.; Stack, T. D. P. J. Am. Chem. Soc. 1997, 119, 8566-8567.
(36) deVries, M. E.; LaCrois, R. M.; Roelfes, G.; Kooijman, H.; Spek, A. L.;
Hage, R.; Feringa, B. L. Chem. Commun. 1997, 1549-1550.
(37) Unpublished results.
9
J. AM. CHEM. SOC. VOL. 124, NO. 1, 2002 85

A R T I C L E S
Goldsmith et al.
Table 1. Crystallographic Data for PY5, [FeII(PY5)(MeOH)](OTf)2, and [FeIII(PY5)(OMe)](OTf)2
complex
PY5
[Fe(PY5)(MeOH)] (OTf)2‚(MeOH)
[Fe(PY5)(OMe)] (OTf)2‚(MeOH)
formula
C29H25N5O2
475.55
C33H33N5O10F6S2Fe
893.61
C33H32N5O10F6S2Fe
892.60
FW (g mol-1
)
crystal system
space group
a (Å)
b (Å)
c (Å)
monoclinic
Cc (No. 9)
11.017(1)
13.626(1)
16.527(1)
90.00
107.50(3)
90.00
2366.2(5)
4
triclinic
triclinic
P1h (No. 2)
9.532(2)
9.966(1)
19.852(3)
96.17(1)
100.12(2)
92.53(2)
1841.8(5)
2
P1h (No. 2)
12.417(2)
12.672(2)
13.345(1)
80.00(1)
63.509(9)
89.51(1)
1845.1(5)
2
R (deg)
â (deg)
γ (deg)
volume (Å3)
Z
µ
calc (cm-1
)
0.87
6.16
6.20
F
obs (Fcalc) (g cm-3
)
1.320 (1.335)
0.50 × 0.60 × 0.50
10.0° < 2θ < 50.0°
2295
2184 (Rint ) 0.0343)
1379
321
4.30
0.041
0.028
>1.590 (1.608)
0.50 × 0.50 × 0.70
10.0° < 2θ < 50.0°
6790
6472 (Rint ) 0.0579)
5613
514
10.92
0.053
0.074
>1.590 (1.609)
0.20 × 0.50 × 0.70
10.0° < 2θ < 50.0°
7985
6454 (Rint ) 0.0385)
4593
597
7.69
0.059
0.062
crystal size (mm)
2θ range
reflections collected
unique reflections
reflections with (Fo2 > 3.00σ(Fo ))
number of parameters
reflns/params ratio
Ra
2
a
Rw
2
2
2
2
2
a R ) ∑||Fo| - |Fc||/∑|Fo|; Rw ) [∑w(|Fo| - |Fc|)2/∑wFo )1/2, where w ) 4Fo /σ2(Fo ); σ2(Fo ) ) S2(C + R2B) + (pFo )2/(Lp)2, with S ) scan rate,
C ) total integrated peak count, R ) ratio of scan time to background counting time, B ) total background count, Lp ) Lorentz polarization factor, and
p ) p factor (0.002 for PY5, 0.008 for [FeII(PY5)(MeOH)](OTf)2, and 0.006 for [FeIII(PY5)(OMe)](OTf)2).
Table 2. Selected Bond Distances and Bond Angles for
bound PY5 ligand from the idealized C2V symmetry of the free
ligand structure (Figure 2A).
[FeII(PY5)(MeOH)](OTf)2 (1) and [FeIII(PY5)(OMe)](OTf)2 (2)
bond lengths (Å)
bond angles (deg)
In the structure of free ligand, the plane of Py1 and the
methoxy groups are effectively coplanar. The two methoxy-
methyl groups adopt a trans conformation, yielding an overall
pseudo-C2-symmetric conformation. Positioning of the methoxy-
methyl groups in the clefts created between Py2 and Py3 and
between Py4 and Py5 results in predictable structural deviations
1
2
1
2
Fe1-N1 2.097(3) 2.147(4) N1-Fe1-N2
Fe1-N2 2.152(3) 2.154(4) N1-Fe1-N3
Fe1-N3 2.203(3) 2.178(4) N2-Fe1-N3
Fe1-N4 2.217(3) 2.223(4) N4-Fe1-N5
Fe1-N5 2.141(4) 2.107(4) N2-Fe1-N5
Fe1-O3 2.040(3) 1.782(3) N3-Fe1-N4
O3-C30 1.422(6) 1.375(7) N2-Fe1-N4
N3-Fe1-N5
85.0(1)
86.2(1)
83.2(1)
82.1(1)
97.7(1)
95.7(1)
170.5(1) 163.9(1)
171.9(1) 172.1(2)
93.6(1)
90.2(1)
81.2(1)
85.5(1)
82.8(1)
81.4(1)
95.8(2)
97.8(1)
at the quaternary carbon centers (Cquat); the C(Py2)-Cquat
-
C(Py3) and C(Py4)-Cquat-C(Py5) angles are greater than the
anticipated idealized tetrahedral angle. This “noninnocent”
positioning of the methoxy-methyl groups in other related
ligands has been previously noted,44 and these structural
deviations are accentuated upon metal chelation (vide infra).
The most curious aspects of the [FeII(PY5)(MeOH)]2+
structure are the tilting of the Py1 plane relative to the Pleq plane
(19.4° from the perpendicular of Pleq) and the opposing position
of both methoxy-methyl groups to this tilt (Figure 2B). These
structural distortions result from the position of the two
methoxy-methyl groups in the clefts between Py2 and Py1 and
between Py5 and Py1. The C(Py2)-Cquat-C(Py1) and C(Py5)-
Cquat-C(Py1) angles distort from ∼108° in free PY5 to ∼118°
in 1 in order to accommodate the methoxy groups. This tilting
is also associated with consistent metrical distortions in the Fe-
Neq bond distances; the Fe-Neq bonds opposite to the Py1 tilt
(Fe-N2, 2.152(3) Å; Fe-N5, 2.141(4) Å) are significantly
shorter than the other two Fe-Neq distances (Fe-N3, 2.203(3)
Å; Fe-N4, 2.217(3) Å). The noninnocent positioning of the
methoxy-methyl groups provides a mechanism of mechanical
coupling within PY5 such that each half of the ligand works in
concert to ligate a metal.
N2-Fe1-O3
98.8(2)
92.4(2)
N3-Fe1-O3
N1-Fe1-O3
Fe1-O3-C30 134.0(3) 164.9(4)
176.3(1) 177.9(2)
a Estimated standard deviations in the least-squares figure are given in
parentheses.
anions. The anion count and the identity of the exogenous ligand
as MeOH rather than methoxide are consistent with a ferrous
complex. The average value for the five Fe(1)-N(1 f 5) bond
lengths (2.162(3) Å) is also consistent with the assignment of
a high-spin (hs) Fe(II) center in this complex.38,39 Figure 2B
shows an ORTEP representation of [FeII(PY5)(MeOH)]2+ with
selected bond lengths and angles provided in Table 2. The
Fe(1)-O(3) bond length of 2.040(3) Å and the Fe(1)-O(3)-
C(30) bond angle of 134° for the exogenous MeOH ligand lie
at the extremes for characterized divalent transition metals.40-43
The selected view of 1 highlights the deviation of the metal-
(38) Que, L., Jr.; Zang, Y.; Kim, J.; Dong, Y.; Wilkinson, E. C.; Appelman, E.
H. J. Am. Chem. Soc. 1997, 119, 4197-4205.
(39) Que, L., Jr.; Roelfes, G.; Lubben, M.; Chen, K.; Ho, R. Y. N.; Meetsma,
A.; Genseberger, S.; Hermant, R. M.; Hage, R.; Mandal, S. K.; Young, V.
G., Jr.; Zang, Y.; Kooijman, H.; Spek, A. L.; Feringa, B. L. Inorg. Chem.
1999, 38, 1929-1936.
[FeIII(PY5)(OMe)](OTf)2‚(MeOH) was crystallized as a stable
red/orange prismatic solid. The asymmetric unit contains a
dication [FeIII(PY5)(OMe)]2+, two triflate anions, and a MeOH
(40) Bruce, M. I.; Walton, J. K.; Williams, M. L.; Patrick, J. M.; Skelton, B.
W.; White, A. H. J. Chem. Soc., Dalton Trans. 1983, 815-821.
(41) Cotton, F. A.; Duraj, S. A.; Manzer, L. E.; Roth, W. J. J. Am. Chem. Soc.
1985, 107, 3850-3855.
(42) Noro, S.; Kondo, M.; Ishii, T.; Kitagawa, S.; Matsuzaka, H. J. Chem. Soc.,
Dalton Trans. 1999, 1569-1574.
(43) Mandal, S. K.; Que, L., Jr. Inorg. Chem. 1997, 36, 5424-5425.
(44) Jonas, R. T.; Stack, T. D. P. Inorg. Chem. 1998, 37, 6615-6629.
9
86 J. AM. CHEM. SOC. VOL. 124, NO. 1, 2002

Synthesis and Reactivity of a Lipoxygenase Mimic
A R T I C L E S
Figure 2. Representations of the crystal structures of (A) the free ligand PY5, (B) the dication [FeII(PY5)(MeOH)]2+, and (C,D) two 90° views of the
dication [FeIII(PY5)(OMe)]2+. Ellipsoids drawn at the 50% probability level.
Scheme 3
solvate molecule. As with 1, a strong hydrogen-bonding network
is developed among the anions and the MeOH of solvation.
Two orthogonal views of [FeIII(PY5)(OMe)]2+ are shown in
Figure 2C,D, and selected bond lengths and angles are provided
in Table 2. Similar to the structure of [FeII(PY5)(MeOH)]2+
,
the plane of Py1 in [FeIII(PY5)(OMe)]2+ tilts relative to the
perpendicular of Pleq by 20.2°, and the two methoxy-methyl
groups are positioned between Py2 and Py1 and between Py5
and Py1. The C(Py2)-Cquat-C(Py1) and C(Py5)-Cquat-C(Py1)
angles distort from ∼108° in the free ligand to ∼117° in 2, and
the contraction of the bond lengths of Fe-N2 (2.154(4) Å) and
Fe-N5 (2.107(4) Å) relative to Fe-N3 (2.178(4) Å) and Fe-
N4 (2.223(4) Å) correlates with the opposite positioning to the
tilt of Py1. The average value for the five Fe(1)-N(1 f 5) bond
lengths (2.162(4) Å) is consistent with the assignment of a hs
Fe(III) center in this complex.45,46
Comparison of the metrical parameters of 1 and 2 indicates
that the structures are nearly isostructural (Table 2), even though
the metals differ in oxidation state. A RMS error of the structural
overlay of the six-coordinated atoms, N(1 f 5) and O(3), and
the iron centers of 1 and 2 equals 0.21 Å. The average bond
lengths between the iron center in 1 and 2 and N2f5 are nearly
identical (∼2.16 Å). The two largest structural changes that
occur upon oxidation of 1 to 2 are the contraction of the
Fe(1)-O(3) bond length from 2.040(3) Å (1) to 1.782(3) Å (2),
and the opening of the Fe(1)-O(3)-C(30) bond angle from
134° (1) to 165° (2).
Solution Characterization. The solution behavior of the four
iron complexes described in this article is summarized in
Scheme 3. The metrical parameters from the solid-state struc-
tures of 1 and 2 obtained at 203 K suggest a hs ferrous and a
hs ferric compound, respectively. Solid-state magnetic suscep-
tibility measurements support these assignments (1, 298 K,
µeff ) 5.0 µB; 2, 298 K, µeff ) 5.8 µB) but also indicate that 1
and 2 undergo cooperative spin-state transitions at ca. 90 K.
The 1H NMR spectra of 1 and 2 exhibit features of hs complexes
with paramagnetically shifted and broadened peaks in the range
of -7 to 60 ppm. The peaks assigned to the four protons of
each of the pendant pyridyl arms are equivalent, indicating
structural equivalence of Py2f5 on the NMR time scale for both
1 and 2. The solution magnetic susceptibility47 of 1 is equal to
4.7 ( 0.2 ΒΜ in MeOH at ambient temperature (RT), slightly
less than the anticipated spin-only value of 4.9 µB for a hs
ferrous complex. Variable-temperature solution magnetic sus-
(45) Zang, Y.; Kim, J.; Dong, Y.; Wilkinson, E. C.; Appelman, E. H.; Que, L.,
Jr. J. Am. Chem. Soc. 1997, 119, 4197-4205.
(46) Kojima, T.; Leising, R. A.; Yan, S.; Que, L., Jr. J. Am. Chem. Soc. 1993,
115, 11328-11335.
(47) Evans, D. F. J. Chem. Soc. 1959, 2003-2005.
9
J. AM. CHEM. SOC. VOL. 124, NO. 1, 2002 87

A R T I C L E S
Goldsmith et al.
Figure 4. UV/vis spectra of [Fe(PY5)(MeOH)](OTf)2 (1), [Fe(PY5)(OMe)]-
(OTf)2 (2), [Fe(PY5)(MeCN)](OTf)2 (3), and [Fe(PY5)(OMe)](OTf) (4),
all in MeOH, except for 3, which was taken in MeCN.
Figure 3. X-band EPR spectrum of [Fe(PY5)(OMe)](OTf)2 (2) in MeOH
at 77 K.
of the exogenous ligand (Figure 4). The conversion of the axial
MeOH ligand to a methoxide could account for this behavior;
consequently, the novel complex 4 is postulated to be [FeII-
(PY5)(OMe)](OTf). This complex is stable under anaerobic
conditions for hours at RT.
ceptibility measurements of 1 show that µeff decreases to
3.3 ( 0.2 µB at 191 K, behavior associated with spin transition
compounds in solution.48 The solid-state µeff of 5.0 µB for the
solid at RT suggests that 1 exists as an equilibrium mixture of
hs and low-spin (ls) states in solution.49 The µeff of 2 shows
similar temperature-dependent susceptibility in MeOH, decreas-
ing from 5.3 ( 0.2 BM at 291 K to 4.4 ( 0.2 BM at 203 K.49
Fitting the variable-temperature magnetic susceptibility values
of 1 and 2 to a ls-hs equilibrium Boltzmann model gives the
following thermodynamic parameters for the transition from ls
to hs: 1, ∆H° ) 2.4 ( 0.2 kcal mol-1, ∆S° ) 14.0 ( 0.5 eu;
2, ∆H° ) 2.2 ( 0.2 kcal mol-1, ∆S° ) 12.2 ( 0.5 eu. These
values are within the range previously observed for Fe(II)48,50
and Fe(III)51,52 spin transition complexes.
The EPR spectrum of 2 in dry MeOH at 77 K is consistent
with a mixture of hs and ls ferric complexes and corroborates
the observed solution magnetic susceptibility data (vide supra).
The high-field region is tentatively modeled as a rhombic
S ) 1/2 octahedral Fe(III) center with signals at g ) 2.25, 2.16,
and 1.96 (Figure 3). The smaller feature at 1500 G (g ) 4.3)
is consistent with the hs component of 2. The EPR spectrum
in nondried MeOH contains more than a single rhombic
component.
[FeIII(PY5)(OMe)](OTf)2 and [FeII(PY5)(OMe)](OTf) differ
by a single electron, suggesting that the reduction potential of
2 should be equal to the oxidation of potential of 4. This redox
potential is necessary for the thermodynamic analysis of the
reactivity of 2 (vide infra). An irreversible reduction peak at
+0.71 V (vs SHE; Fc/Fc+ (MeOH) ) +0.610 V vs SHE)56 is
found for 2 in MeOH, while 4 shows an irreversible oxidation
peak at +0.76 V (∆E ) 0.100 V). These two potentials differ
by ∼50 mV, a difference potentially attributable to the differ-
ences of the electrochemical solutions. Due to its limited
stability, 4 was generated in situ from 1 by addition of a slight
excess of triethylamine. Attempts to perform electrochemistry
on 2 with a similar amount of triethylamine immediately
decomposed the ferric complex. Given the small difference
between these experimental values, the average, +0.73 V, is
taken as the reduction potential of 2. [FeII(PY5)(MeOH)](OTf)2
shows a highly irreversible oxidation potential approximately
200 mV greater than that for 4 (at 0.93 V).
Thermodynamic Justification for Potential Hydrogen
Atom Abstraction. Structurally and electronically, complexes
1 and 2 differ only by the net addition of a single hydrogen
atom. Based on enthalpic considerations, a bimolecular reaction
of 2 with a substrate containing a weak C-H bond may proceed
via hydrogen atom abstraction (HA) if the bond strength of the
O-H bond created in 1 is greater than or approximately equal
to the bond strength of the C-H bond broken. However, an
enthalpically unfavored reaction can also proceed to completion
if a second fast irreversible process follows the initial HA, such
as a radical coupling reaction or a fast secondary oxidation.
The O-H bond strength of the ligated MeOH in the product,
1, can be calculated from a thermodynamic cycle (Scheme 4)
that includes the reduction potential of 2 (Ered ) 0.73 V) and
the proton affinity of 1 (pKa ) 9.1 ( 0.2) in MeOH.57 The last
two equations in Scheme 4, which describe the free energy of
formation of a hydrogen atom and its solvation in MeOH,
complete the thermodynamic cycle. As the solvation of a proton
in MeOH and that of a proton in H2O differ, the standard
reduction potential of H2 in MeOH can be calculated using the
Solutions of 2 in MeOH, MeCN, and acetone are light orange
and are aerobically stable at RT. The UV/vis absorption spec-
trum in MeOH shows a high-energy feature at 337 nm (ꢀ )
3600 M-1 cm-1) that likely results from a ligand-to-metal charge
transfer (LMCT) transition between the pyridine ligands of PY5
and the Fe(III) center (Figure 4). Solutions of 1 in MeOH are
green-yellow and are aerobically stable for days at RT. The UV/
vis spectrum of 1 shows a high-energy feature at 370 nm (ꢀ )
1650 M-1 cm-1) that is similar in terms of both energy and
intensity to other complexes with hs Fe(II)-pyridine bonds
(Figure 4).53,54 This band is tentatively assigned as a pyridine
MLCT band.55 Titration of 1 with a strong base shows clean
isobestic conversion to a novel complex 4. The shift to a lower
energy band from 1 to 4 is consistent with improved σ-bonding
(48) Gutlich, P.; Hauser, A.; Spiering, H. Angew. Chem., Int. Ed. Engl. 1994,
33, 2024-2054.
(49) See Supporting Information.
(50) Chang, H.-R.; McCusker, J. K.; Toftlund, H.; Wilson, S. R.; Trautwein,
A. X.; Winkler, H.; Hendrickson, D. N. J. Am. Chem. Soc. 1990, 112,
6814-6827.
(51) Neya, S.; Tsubaki, M.; Hori, H.; Yonetani, T.; Funasaki, N. Inorg. Chem.
2001, 40, 1220-1225.
(52) Beattie, J. K.; Binstead, R. A.; Dewey, T. G.; Turner, D. H. J. Am. Chem.
Soc. 1980, 102, 6442-6451.
(55) Borovik, A. S.; Papaefthymiou, V.; Taylor, L. F.; Anderson, O. P.; Que,
L., Jr. J. Am. Chem. Soc. 1989, 111, 6183-6195.
(53) Hagen, K. S.; Diebold, A. Inorg. Chem. 1998, 37, 215-223.
(54) Que, L., Jr. In Iron Carriers and Iron Proteins; Loehr, T. M., Ed.; VCH
Publishing: New York, 1989; pp 467-524.
(56) Zuman, P.; Meites, L. Electrochemical Data; John Wiley & Sons: New
York, 1974; Vol. A.
(57) The calculated pKa differs slightly from that previously reported in ref 35.
9
88 J. AM. CHEM. SOC. VOL. 124, NO. 1, 2002

Synthesis and Reactivity of a Lipoxygenase Mimic
A R T I C L E S
Scheme 4
a ∆G1solv assumed to be equal to ∆G2
.
b This constant contains the entropic free energy associated with the solvation of a hydrogen atom. Subtraction
solv
of this term from the sum of the free energies above gives the resulting enthalpic bond dissociation energy. c Species followed by the notation (me) are in
MeOH solution.
known energy of the H2(g)/H+
couple relative to the
with a second equivalent of 2. Quantitation of the EPR signal
indicates that a substoichiometric amount of TTBP radical is
formed, but this reaction clearly indicates that 2 can react
through a formal hydrogen atom transfer reaction.
(MeOH)
hydrogen electrode in water.58 The H2 reduction potential, the
H2 bond dissociation energy (BDE), and the entropic contribu-
tion at 298 K are conveniently combined into a single constant
C,59 and equal 54.4 kcal mol-1. The energy calculation for the
overall reaction is an enthalpic BDE.
Quantifying the rate of reduction of 2 by optical spectroscopy
is faciliated in MeCN, as the product 3 has much more intense
MLCT bands than 1. At 298 K, the anaerobic decay of 2 in
MeCN follows clean first-order kinetics under conditions of
excess TTBP (>40 equiv), with the rate constant linearly
dependent on the concentration TTBP. A calculated second-
order rate constant is k2 ) 0.60 ( 0.05 M-1 s-1. The slower
reactivity with TTBP deuterated in the phenolic position (k2 )
0.30 ( 0.05 M-1 s-1) gives a kinetic isotope effect of 2.0,
suggesting that O-H(D) bond cleavage is involved in the rate-
determining step.
An independent check of the value of C in MeOH can be
determined from any chemical with a known pKa in MeOH, a
well-defined oxidation potential for its conjugate base in MeOH,
and a known BDE. Phenol is a convenient reference; BDE )
88.6 kcal mol-1 60,61 pKa(MeOH) ) 14.3.62 Using an experi-
,
mentally determined oxidation potential of the phenolate anion
in MeOH (Ered ) 0.640 V vs SHE), the value of C calculated
from the phenol reference is nearly identical (54.4 kcal mol-1
)
to the first value. By both methods, the O-H bond strength of
the MeOH ligand in 1 is calculated as 83.5 ( 2.0 kcal mol-1
.
Oxidation of Cyclohexadiene. The reaction of 2 with
substrates containing weak C-H bonds under N2 results in the
reduction of the Fe(III) center and oxidation of the substrates.
The nature of the mechanism of substrate activation is of obvious
interest, given the structural similarities of 2 and activated LO.
Natural substrates of LOs, such as linoleic acid and arachidonic
acid, contain a cis,cis-1,4-pentadiene subunit where stereo-
specific C-H bond cleavage is proposed to occur. A minimal
substrate model for the cis,cis-1,4-pentadiene subunit is 1,4-
cyclohexadiene (CHD), which has a C-H BDE of 75 ( 4 kcal
This O-H BDE is dramatically reduced relative to that of
unligated MeOH (102 kcal mol-1).63
Oxidation of 2,4,6-Tri-tert-butylphenol. The ability of 2
to formally remove a hydrogen atom was initially probed
by using 2,4,6-tri-tert-butylphenol (TTBP) as a substrate. The
TTBP radical is relatively stable in MeOH (t1/2 ) 1 h) and is
readily identifiable due to its strong UV/vis absorption bands.64
Additionally, TTBP has a relatively weak O-H bond of 81 kcal
mol-1 61
. The weak O-H BDE combined with the faster reaction
of phenols relative to C-H substrates of similar BDE65 makes
a formal hydrogen atom transfer rapid compared to the normal
radical decay, allowing a measurable amount of the radical to
form in solution. The reaction of 10 equiv of TTBP with 2 in
MeOH under N2 at RT generates the characteristic blue-green
color of the TTBP radical. EPR analysis at 77 K of the reaction
aliquots shows rapid decay of the ferric species followed by
the growth of the TTBP radical, g ) 2 signal. However, the
radical intensity develops only after most of 2 is consumed,
implicating a fast secondary reaction of the phenoxyl radical
mol-1 66,67
CHD slowly reduces 2 to 1 in MeOH, as monitored
.
by UV/vis spectroscopy. If the reaction occurs in MeCN, the
reduced product, 1, immediately undergoes solvent exchange
to form ls [FeII(PY5)(MeCN)](OTf)2 (3), as indicated by its
diamagnetic 1H NMR spectrum and the increased intensity and
shifting of the MLCT band (Figure 4). [FeII(PY5)(MeCN)]-
(ClO4)2 has been previously characterized.36 In MeCN solution,
3 is indefinitely stable, and the vastly different UV/vis spectra
of 2 and 3 facilitate the monitoring of these reactions. The
progress of the reaction can also be monitored by 1H NMR. 1H
NMR analysis of the organic products shows 0.5 equiv of
benzene produced for each equivalent of the ferric complex
reduced (>95%). As measured optically in both MeOH and
MeCN at 298 K, the anaerobic decay of 2 follows first-order
kinetics under conditions of excess substrate with the rate
constant linearly dependent on the concentration of the substrate
(58) Johnsson, M.; Persson, I. Inorg. Chim. Acta 1987, 127, 15-24.
(59) Bordwell, F. G.; Cheng, J. P.; Harrelson, J. A., Jr. J. Am. Chem. Soc. 1988,
110, 1229-1231.
(60) DosSantos, R. M. B.; Simoes, J. A. M. J. Phys. Chem. Ref. Data 1998, 27,
707-739.
(61) Pedulli, G. F.; Lucarini, M.; Pedrielli, P.; Cabiddu, S.; Fattuoni, C. J. Org.
Chem. 1996, 61, 9259-9263.
(62) Lange’s Handbook of Chemistry, 14th ed.; Dean, J. A., Ed.; McGraw-
Hill: New York, 1992.
(63) Kerr, J. A. Chem. ReV. 1966, 66, 465-500.
(64) Cook, C. D.; Kuhn, D. A.; Fianu, P. J. Am. Chem. Soc. 1955, 78, 2002-
2005.
(65) Lockwood, M. A.; Blubaugh, T. J.; Collier, A. M.; Lovell, S.; Mayer, J.
M. Angew. Chem., Int. Ed. 1999, 38, 225-227.
(66) Mulder, P.; Laarhoven, L. J. J.; Wayner, D. D. M. Acc. Chem. Res. 1999,
32, 342-349.
(67) Burkey, T. J.; Majewski, M.; Griller, D. J. Am. Chem. Soc. 1986, 108,
2218-2221.
9
J. AM. CHEM. SOC. VOL. 124, NO. 1, 2002 89

A R T I C L E S
Goldsmith et al.
Table 3. Second-Order Rate Constants for the Reactions of
Hydrocarbon Substrates with [FeIII(PY5)(OMe)](OTf)2 (2) in MeOH
and MeCN
substrate
k2corr(MeOH)a
k2corr(CH CN)a
KIEb
3
toluene
ethylbenzene
2,3-dimethyl-2-butene
(1.5 ( 0.4) × 10-4 (1.9 ( 0.4) × 10-4 6.5
(2.3 ( 0.3) × 10-4 (2.3 ( 0.2) × 10-4
(2.8 ( 0.2) × 10-4 (3.0 ( 0.3) × 10-4
-
-
-
4-methyl-1,3-pentadiene (2.3 ( 0.6) × 10-3 (2.1 ( 0.5) × 10-3
9,10-dihydroanthracene (4.9 ( 0.3) × 10-3 (5.0 ( 0.1) × 10-3 5.5
1,4-cyclohexadiene
(2.7 ( 0.4) × 10-2 (3.0 ( 0.4) × 10-2 2.7
a Second-order rate constants per available C-H bond were measured
at 323 K in units of M-1 s-1 and were adjusted for reaction stoichiometry
Figure 5. First-order rate constants for the reduction of [FeIII(PY5)(OMe)]-
(OTf)2 (2) to 1 in the presence of 1,4-cyclohexadiene in MeOH (0) and
MeCN (O). All data taken with greater than 40-fold excess of cyclohexa-
diene under anaerobic conditions at 298 K. k2corr(MeCN) ) 2.8 × 10-4
to yield k2corr.
b Kinetic isotope effect determined in MeOH at 298 K.
fold slower than that measured under O2-free conditions, with
a slightly greater effect in MeOH as compared to that in MeCN.
M-1 s-1, k2corr(MeOH) ) 2.4 × 10-4 M-1 s-1
.
If the hydrogen-transfer reaction proceeds via a charged
species, then the observed rate constants would be expected to
be orders of magnitude different in protic versus aprotic
solvents.69 The similarity of k2corr in MeOH and MeCN for CHD
supports a radical-based process. The consistent kinetic behavior
of the reduction of 2 in both MeOH and MeCN and the
perturbation of reaction rate in the presence of O2 support a
rate-limiting HA step in the oxidation process.
Scheme 5
Other Hydrocarbon Substrate Reactivity. The linear cor-
relation of the log of the C-H bond cleavage reaction rate
constants with C-H BDE has been established for a number
of metal-oxo complexes, including chromyl chloride and
permanganate species.28-31,70 This empirical correlation between
the activation energy (Ea) and the enthalpic driving force (∆H)
is consistent for primary, secondary, and tertiary C-H bonds
and appears to be independent of substrate spatial demands. The
Evans-Polanyi relationship suggests that in HA reactions, the
entropies of C-H bond activation are essentially constant,
resulting in a linear relationship between log(k2corr) and ∆H.71,72
Thus, at parity of substrate, the primary determinant of HA
reactivity for a metal-oxo complex is the strength of the O-H
bond created.
(Figure 5). The observed rate constant does not change as the
concentration of 2 is systematically varied between 0.16 mM
and 0.23 mM with 0.066 M fluorene, suggestive of a first-
order dependence on 2 and an overall second-order rate law
(eq 3). The second-order rate constants need to be statistically
d[FeII(PY5)(MeOH)2+
dt
]
) k2[FeIII(PY5)(OMe)2+][CHD] (3)
corrected for the number of weak C-H bonds (e.g., CHD
has four equivalent weak C-H bonds) and for secondary
reactions: k2corr(MeOH) ) (2.4 ( 0.2) × 10-4 M-1 s-1
,
Further support for a HA mechanism in the reduction of 2
by CHD results from the reactivity profiles of other substrates
containing weak C-H bonds with 2. Because the reaction rates
are slow at 298 K, the reactions were performed at 323 K in
both MeOH and MeCN. Under pseudo-first-order conditions
of substrate, the reduction of 2 was examined using the
following substrates: toluene, ethylbenzene, 2,3-dimethyl-2-
butene, 4-methyl-1,3-pentadiene, 9,10-dihydroanthracene (DHAn),
1,4-cyclohexadiene, xanthene, fluorene, tetralin, diphenyl-
methane, triphenylmethane, and cumene. The first six substrates
were analyzed in both MeOH and MeCN, while the last five
were analyzed only in MeCN. As with CHD at 298 K, all
substrates examined in MeOH and MeCN exhibit similar rates
in both protic and aprotic solutions (Table 3). In addition, the
deuterated analogues toluene-d8 and 9,10-dihydroanthracene-
d4 were examined in MeOH to assess the nature of the rate-
limiting step. Oxidations of both deuterated analogues were
significantly slower than for the nondeuterated substrates,
k2corr(MeCN) ) (2.8 ( 0.4) × 10-4 M-1 s-1, 298 K. The per-
deuterated analogue of CHD (CHD-d8) reacts at a significantly
reduced rate relative to CHD: k2corr(MeOH) ) (9.0 ( 0.3) ×
10-5 M-1 s-1, k2corr(MeCN) ) (8.0 ( 0.3) × 10-5 M-1 s-1
,
298 K. The ratios of the rate constants (kCHD/kCHD-d ) are
8
2.7 and 3.4 in MeOH and MeCN, respectively, suggestive of
a primary KIE for the observed reactivity.
The combination of a second-order rate law and a primary
KIE using deuterated substrates indicates that the reduction of
2 is most easily understood as a bimolecular process in which
C-H bond cleavage is the rate-determining step. The collision
of 2 with a weak C-H group on CHD results in 1 equiv of 1
and a CHD substrate radical (Scheme 5). A fast secondary
reaction is proposed to follow, as the C-H BDE of the radical
is thought to be significantly reduced from that of CHD, and
the radical is quenched with a further equivalent of 2. As a test
for the presence of radicals, the reactions were run in the
presence of O2. Because O2 is known to react with radicals near
the diffusion limit,68 it should compete with 2 for CHD radicals.
Indeed, the decay rate of 2 for the reactions under O2 is ∼2-
(69) Avila, D. V.; Brown, C. E.; Ingold, K. U.; Lusztyk, J. J. Am. Chem. Soc.
1993, 115, 466-470.
(70) Mayer, J. M.; Wang, K. J. Am. Chem. Soc. 1997, 119, 1470-1471.
(71) Tedder, J. M. Angew. Chem., Int. Ed. Engl. 1982, 21, 401-410.
(72) Ingold, K. U.; Russel, G. A. In Free Radicals; Kochi, J. K., Ed.; Wiley:
New York, 1973; pp 283-293.
(68) Maillard, B.; Ingold, K. U.; Scaiano, J. C. J. Am. Chem. Soc. 1983, 105,
5095-5099.
9
90 J. AM. CHEM. SOC. VOL. 124, NO. 1, 2002

Synthesis and Reactivity of a Lipoxygenase Mimic
A R T I C L E S
Table 4. Corrected Second-Order Rate Constants for the
Reactions of Hydrocarbon Substrates with [FeIII(PY5)(OMe)](OTf)2
(2) in MeCN
to better simulate imidazole ligation. However, several essential
features of the endogenous ligation sphere of LO are effectively
reproduced. The pentadentate PY5 ligand enforces a square
pyramidal endogenous ligation of the enzyme that leaves an
exogenous ligand site similar to the enzyme. Additionally, the
endogenous ligands of both the synthetic complex and the
enzyme are relatively uncharged, enabling very positive reduc-
tion potentials for the ferric complexes (vide supra). Both
the structure and the reactivity of [FeII(PY5)(MeOH)](OTf)2 (1)
and [FeIII(PY5)(OMe)](OTf)2 (2) suggest that PY5 creates a
reasonable model for LO by simple mimicry of the coordination
site of the enzyme.
c
substrate
k2corr(CH CN)a
BDEb
ref
∆H° d
3
toluene
ethylbenzene
tetralin
(1.9 ( 0.4) × 10-4 88 ( 1 108
+4.5
(2.3 ( 0.2) × 10-4 85 ( 1 93, 109 +1.5
(2.9 ( 0.2) × 10-4 85 ( 1 109
(9.5 ( 0.4) × 10-3 83 ( 1 109
+1.5
-0.5
cumene
2,3-dimethyl-2-butene
diphenylmethane
triphenylmethane
(3.0 ( 0.3) × 10-4 82 ( 2 110, 111 -1.5
(8.9 ( 0.4) × 10-4 82 ( 1 107, 113 -1.5
(1.8 ( 0.4) × 10-3 81 ( 1 107
-2.5
4-methyl-1,3-pentadiene (2.1 ( 0.5) × 10-3 81 ( 2 93, 114 -2.5
fluorene
(3.3 ( 0.2) × 10-3 80 ( 2 107, 112 -3.5
9,10-dihydroanthracene (5.0 ( 0.1) × 10-3 78 ( 2 107
-5.5
-8.5
-8.5
1,4-cyclohexadiene
xanthene
(3.0 ( 0.4) × 10-2 75 ( 4 66, 67
The most dramatic structural similarity between the model
complexes and LO is the short Fe(1)-O(3) distance observed
in the crystal structure of 2; this distance is reminiscent of the
short Fe-O distance (1.88 Å) observed in the EXAFS infor-
mation for the proposed ferric-hydroxide species in activated
LO.13 Both the Fe(1)-O(3) bond length (1.782(3) Å) and the
Fe(1)-O(3)-C(30) bond angle (165°) in 2 are near the extremes
for terminally bound methoxide anions in structurally character-
ized iron complexes.39,73,74 These metrical parameters are
generally associated with more electron-deficient metals such
as titanium75-77 and zirconium.78 The neutral PY5 ligation to
the Fe(III) center generates an extremely Lewis-acidic metal
site that reduces its electron deficiency by π-bonding to the lone
pair electrons of the exogenous methoxide ligand. This multiple-
bond character for the Fe-OMe bond is similar to that proposed
to occur for the oxide bridge in µ-oxo-ferric dimer com-
plexes.39,79 The π and σ-bonding effects and negative charge
of the methoxide ligand in 2 are thought to be important in
the stabilization of this higher oxidation state. As a result of
the multiple-bond character, the Fe(1)-O(3) bond contracts
significantly. The proposed hydroxide ligand in the ferric active
form of LO shows a similar Fe(III)-O contraction from the
ferrous-H2O form and may act in an analogous manner to
stabilize the ferric active species.14,80,81
(2.8 ( 0.2) × 10-2 75 ( 1 112
a Second-order rate constants per available C-H bond were measured
at 323 K in units of M-1 s-1 and were adjusted for reaction stoichiometry.
b For the weak C-H bond in units of kcal mol-1 c References are for
.
literature BDE of substrates. d Relative to the bond dissociation energy of
the O-H bond in the bound MeOH in 1 (83.5 kcal mol-1).
Figure 6. Dependence of the k2corr of the reduction of [FeIII(PY5)(OMe)]-
(OTf)2 (2) on ∆H. All data taken with greater than 40-fold excess of
substrate under anaerobic conditions in MeCN at 323 K. ∆H is relative to
83.5 kcal mol-1, the bond dissociation energy of the O-H bond in the
bound MeOH in 1.
with kC
/kC
) 5.5 ( 0.5 for DHAn and kC H /kC D )
14H12
14H8D4
7
8
7 8
Electronically, 2 also resembles LOs. The crystal structure
and both solid-state and solution magnetic susceptibility mea-
surements show that 2 is predominantly high-spin at 298 K, as
are the ferric sites of SLO-1 and 5-human lipoxygenase (5-
HLO).82,83 The LOs differ from 2 in that the native enzymes
do not exhibit spin-crossover behavior. The electrochemical
behavior of 2 is also similar to that of LO. The redox poten-
tial of 2 at ∼0.73 V is in the same range as that measured
for SLO-1 (0.6 ( 0.1 V vs SHE).1 These potentials are
much higher than that of the simplest ferric model complex of
6.5 ( 1 for toluene in MeOH at 298 K.
A plot of log(k2corr) versus the pKa of each substrate49 does
not show a linear relationship, as is found between log(k2corr
)
and the C-H bond BDE (Table 4) of each substrate. This latter
linear relationship supports a HA mechanism and is graphically
presented as a plot of log(k2corr) versus ∆H° of the reaction
(Figure 6), where ∆H° is defined as the strength of the O-H
bond in 1 subtracted from the C-H BDE of each substrate.
The correlation appears unaffected by steric crowding within
each substrate, as evidenced by the normal reactivity of
extremely bulky substrates such as triphenylmethane.
(73) Lecomte, C.; Chadwick, D. L.; Coppens, P.; Stevens, E. D. Inorg. Chem.
1983, 22, 2982-2992.
(74) Hoard, J. L.; Hamor, M. J.; Hamor, T. A.; Caughey, W. S. J. Am. Chem.
Soc. 1965, 87, 2312-2319.
Discussion
(75) Aslan, H.; Sielisch, T.; Fischer, R. D. J. Organomet. Chem. 1986, 315,
C69-C72.
Modeling of metalloenzymes provides the opportunity to
create not only spectroscopic and structural models of an active
site but also functional analogues. The ability to systematically
vary the coordination sphere with a synthetic model is a
powerful probe for elucidating the role of the ligation sphere
in conferring specific reactivity to a metal site. The ligand PY5
most closely resembles the coordination sphere of mammalian
lipoxygenases (LO) that contain four imidazoles from histidines
and one carboxylate; certainly, a better model would include a
carboxylate ligand and less π-back-bonding nitrogenous ligands
(76) Boreham, C. J.; Buisson, G.; Duee, E.; Jordanov, J.; Latour, J.-M.; Marchon,
J.-C. Inorg. Chim. Acta 1983, 70, 77-82.
(77) Wright, D. A.; Williams, D. A. Acta Crystallogr., Sect. B 1968, 24, 1107-
1114.
(78) Karia, R.; Willey, G. R.; Drew, M. G. B. J. Chem. Soc., Dalton Trans.
1986, 2493-2495.
(79) Holm, R. H. Chem. ReV. 1987, 87, 1401-1449.
(80) Zhang, Y.; Gebhard, M. S.; Solomon, E. I. J. Am. Chem. Soc. 1991, 113,
5162-5175.
(81) Deaton, J. C.; Gebhard, M. S.; Solomon, E. I. Inorg. Chem. 1989, 28, 877-
889.
(82) Cheesbrough, T. M.; Axelrod, B. Biochemistry 1983, 22, 3837-3840.
(83) Chasteen, N. D.; Grady, J. K.; Skorey, K. I.; Neden, K. J.; Riendeau, D.;
Percival, M. D. Biochemistry 1993, 32, 9763-9771.
9
J. AM. CHEM. SOC. VOL. 124, NO. 1, 2002 91

A R T I C L E S
Goldsmith et al.
0.60 M-1 s-1) precludes the former mechanism of fast electron
transfer as an initial step. The high oxidation potential of TTBP
(1.60 V vs SHE in MeCN)90 along with the positive reduction
potential of 2 predicts an intermediate with ∆G° ) 19.8 kcal
mol-1, a free energy higher than that of the empirical transition
state (∆Gq ) 17.5 kcal mol-1). Similar analysis of the reactivity
of 2 with DHAn (vide supra) precludes fast initial electron
transfer for hydrocarbons. Using a lower limit estimate E° of
∼2.0 V vs SHE for the oxidation potential of DHAn,91 an initial
outer-sphere oxidation of DHAn by 2 would be ca. 1.3 V uphill
(∆G° ) 30 kcal mol-1), much higher than the observed ∆Gq
of 20.8 kcal mol-1 at 298 K. Analysis of the reactivity of 2
with a series of hydrocarbons provides evidence against a slow
initial proton transfer as a mechanistic step. In all cases, the
rate is first-order with respect to both the concentration of 2
and the concentration of substrate. The empirical second-order
rate constants, corrected for the number of weak C-H bonds
in the substrate and the reaction stoichiometry, yield an adjusted
second-order rate constant, k2corr. The nonlinear relation between
Figure 7. Thermodynamic cycle for the [FeIII(H2O)5(OH)]2+/[FeII(H2O)6]2+
system in H2O. aFrom ref 85. bCalculated from the above literature values.
cCalculated from the equation BDE ) 1.37 pKHA + 23.06Eo(A-) + C,
with a C value of 56.115
LO, namely the monodeprotonated ferric hexaaquo complex,
[FeIII(H2O)5(OH)]2+ (0.32 V vs SHE, Figure 7).
The pKa of 1 in MeOH is 9.1 ( 0.2, while the estimated pKa
for the ferrous-H2O form of human LO is >10.84 For
comparison, [FeII(H2O)6]2+ has a pKa of 9.7.85 The very positive
reduction potential of 2 combined with the basicity of the
methoxide ligand of 4 allows the methoxide ligand of 2 to be
a suitable model for the proposed hydroxide ligand in the active
native species. A thermodynamic analysis shows that the
conversion of 2 to 1 in MeOH results in the formation of an
O-H bond with a bond dissociation energy (BDE) of 83.5 (
2.0 kcal mol-1. A similar analysis of the O-H BDE in the
ferrous-H2O form of LO and [FeII(H2O)6]2+ gives 85 and 77
kcal mol-1, respectively.86
In the proposed mechanism of LO, the ferric-hydroxide
species cleaves a substrate C-H bond via HA with concomitant
reduction of the active site to a ferrous-water species (Scheme
2); 2 appears to react with substrates in an analogous manner.
2,4,6-Tri-tert-butylphenol (TTBP) reduces 2 to generate a tri-
tert-butylphenoxyl radical, demonstrating the ability of 2 to
formally accept a hydrogen atom. However, this reactivity does
not distinguish between a concerted hydrogen atom transfer
mechanism and other mechanisms consisting of discrete proton
and outer-sphere electron-transfer steps. An inner-sphere mech-
anism involving a ferric 2,4,6-tri-tert-butylphenoxide complex
is unlikely, given that there are no structurally characterized
octahedral metal complexes that include a 2,6-disubstituted tert-
butylphenoxide ligand.87,88 The observed kinetic isotope effect
(KIE) of 2.0 ( 0.3 for the reduction of 2 by TTBP is large for
a secondary isotope effect, particularly given that the KIE for
TTBP proton self-exchange is 1.24,89 and suggests that C-H
bond cleavage may be involved in the RDS. In addition, the
observation of large primary KIEs for cyclohexadiene, dihydro-
anthracene (DHAn), and toluene implicates C-H cleavage in
the RDS for the reduction of 2 by hydrocarbons.
49
log(k2corr) and the pKa precludes initial proton abstraction as
a rate-limiting step.
The elimination of the likely alternative mechanisms leaves
HA as the probable mechanistic pathway. Evidence for HA as
a mechanistic pathway includes the similar rates of reaction in
MeOH and MeCN, suggesting that the reactions progress
without change in the charge distribution. The most convincing
argument for HA over other potential mechanisms is the strongly
linear relationship between log(k2corr) and the substrate’s C-H
BDE.
Given this linear relationship, an unknown C-H BDE of a
substrate that reduces 2 can be measured using an empirically
determined reaction rate. The LO substrate, linoleic acid,
contains a cis,cis-1,4-pentadiene subunit where stereospecific
C-H bond cleavage takes place. The bond dissociation energy
(BDE) for the weak C-H bond of 1,4-pentadiene is found to
be 77 kcal mol-1 when the molecule adopts a planar conforma-
tion.92,93 The geometric constraints of a cis,cis-diene fatty acid
effectively preclude the formation of a planar 1,4-pentadiene
subunit and will certainly increase the C-H BDE of linoleic
acid above 77 kcal mol-1 94
The anaerobic reaction of excess
.
ethyl linoleate, the ester-protected linoleic acid, and 2 produces
1 and several isomeric trienes in a ∼50% yield, as identified
by GC-MS. The second-order rate constant for C-H bond
activation yields k2corr ) (9.0 ( 0.5) × 10-4 M-1 s-1. By
comparison with the rates of substrates with calculated and/or
experimentally determined BDE values, the C-H bond strength
of the C11 hydrogen (Scheme 1) is estimated as 83 ( 1 kcal
mol-1. This BDE is 6 kcal mol-1 greater in energy than the
estimated value from 1,4-pentadiene, suggesting that only one
of the olefins of linoleic acid significantly conjugates to the
radical in the transition state when reacting with 2. Linoleic
acid appears to react with the enzyme in a similar manner. EPR
studies show an intermediate organic radical species that is more
consistent with delocalization over three, not five, carbon atoms.1
Two likely alternative mechanisms consistent with the rate-
determining O-H or C-H bond cleavage are either fast electron
transfer followed by a rate-limiting proton abstraction, or a rate-
limiting proton abstraction followed by fast electron transfer.
The fast rate of the reduction of 2 by TTBP in MeCN (k2 )
(84) Holman, T. R. Personal communication about pKa of ferrous human LO.
(85) Martell, A. E.; Smith, R. M. Critical Stability Constants; Plenum: New
York, 1976; Vol. IV, p 1.
(90) Bordwell, F. G.; Cheng, J. P. J. Am. Chem. Soc. 1991, 113, 1736-1743.
(91) Schlesener, C. J.; Amatore, C.; Kochi, J. K. J. Am. Chem. Soc. 1984, 106,
3567-3577.
(86) Given a C value of 56 kcal mol-1 (ref 115).
(87) Darensbourg, D. J.; Zimmer, M. S.; Rainey, P.; Larkins, D. L. Inorg. Chem.
1998, 37, 2852-2853.
(92) Clark, K. B.; Culshaw, P. N.; Griller, D.; Lossing, F. P.; Simo˜es, J. A. M.;
Walton, J. C. J. Org. Chem. 1991, 56, 5535-5539.
(93) McMillen, D. F.; Golden, D. M. Annu. ReV. Phys. Chem. 1982, 33, 493-
532.
(88) Geerts, R. L.; Huffman, J. C.; Caulton, K. G. Inorg. Chem. 1986, 25, 1803-
1805.
(89) Arick, M. R.; Weissman, S. I. J. Am. Chem. Soc. 1968, 90, 1654.
(94) Fort, R. C., Jr.; Hrovat, D. A.; Borden, W. T. J. Org. Chem. 1993, 58,
211-216.
9
92 J. AM. CHEM. SOC. VOL. 124, NO. 1, 2002

Synthesis and Reactivity of a Lipoxygenase Mimic
A R T I C L E S
Table 5. Reaction Parameters for Coordination Complexes’
Reactivity with 9,10-Dihydroanthracene via Hydrogen Atom
Abstractiona
coordination complexes as found for simple radicals. One
complicating factor may be the differing steric hindrance around
the hydrogen abstracting moieties. The methyl moiety of the
methoxide ligand and the bulk of the PY5 ligand, for instance,
may make the reactive methoxide oxygen in 2 less accessible
than the four unshielded oxygens of MnO4-. A second and
perhaps more important factor introduced by Mayer and Wang
involves the differing structural reorganizational energies for
each system affecting the rate of HA.70
complex
ref
∆Hq b ∆Sq c
BDEd
∆H° e
[Fe(Hbim)(H2bim)2]2+
33
32
70
11.6 -36 76 ( 2
13.8 -15 80 ( 3
14.5 -23 79 ( 2
16.0 -21 75 ( 2
+2
-2
-1
+3
-
MnO4
[L2Mn(O)2MnL2]3+
[L2Mn(O)(OH)MnL2]3+ 70
2
this work 13.4 -25 83.5 ( 2 -5.5
a Measured in MeCN under anaerobic conditions over the temperature
range 298-343 K. b Measured in kcal mol-1 c Measured in e.u. d Energy
.
Summary
of the X-H bond formed upon reduction of the metal complex in kcal
mol-1 e Relative to the BDE of dihydroanthracene (78 kcal mol-1).
.
A new ligand, 2,6-bis(bis(2-pyridyl)methoxymethane)pyridine
(PY5), provides the necessary ligation to mimic the five-
coordinate endogenous coordination sphere in the iron active
site in LOs, more closely resembling the active site of
mammalian LOs than plant LOs. An Fe(III) complex of PY5
with an axial methoxide ligand, [FeIII(PY5)(OMe)](OTf)2 (2),
was synthesized from the oxidation of [FeII(PY5)(MeOH)]-
(OTf)2 (1) with H2O2 or iodosobenzene. The ferric complex
shares many of the structural and spectroscopic attributes of
native LO. In the most widely accepted mechanism for the
function of LOs, the rate-limiting step involves abstraction of
a hydrogen atom from the cis,cis-1,4-pentadiene subunit of the
fatty acid substrate with concomitant reduction of the ferric-
hydroxide center to a ferrous-water species and formation of
a fatty acid radical. Substrates with a weak C-H bond reduce
2 to generate a radical species and 1. The reactivity profile with
a variety of substrates is consistent with a HA mechanism and
provides chemical precedent for the C-H bond activation step
in the proposed mechanism of LO. The ability to predict
substrate C-H bond dissociation energies, such as that of
linoleic acid, with the bond strength approach is a useful tool
in the understanding of metal-mediated HA in metalloenzymes.
This model reactivity supports the currently proposed mech-
anism of LO and provides a thermodynamic justification for
HA by simple ferric-hydroxide species. Comparisons to other
hydrogen abstracting species suggest that sterics and/or structural
reorganization complicate the relation between ∆H° and the rate
of HA as compared to organic radicals.
Furthermore, SLO-1 can oxygenate monounsaturated fatty acids,
albeit at rates much slower than that for the natural substrate,
raising the possibility that the role of the second olefinic unit is
one of recognition.95
Attempts to reproduce the enzymatic catalytic activity with
2 under O2 to produce alkyl peroxide products are complicated
by the lack of a substrate binding site in 2. In the enzymatic
mechanism, the peroxyl radical that results from the reaction
of O2 with the initial alkyl radical abstracts a hydrogen atom
from the water attached to the reduced metal center to re-form
the oxidized metal site (Scheme 2). In the case of 2, the
intermediate radicals should be free to diffuse from the reduced
metal complex 1 before subsequent reaction with O2 to create
the alkyl peroxyl radical. An alkyl peroxyl radical should be a
more reactive oxidant than 2, not only because the O-H BDE
of an alkyl peroxide is significantly greater than that of 1 (ca.
90 kcal mol-1, ROOH; 83.5 kcal mol-1, 1) but also because
the reactive oxygen in the alkyl peroxyl radical is less sterically
hindered than the methoxide oxygen in 2. In the presence of
excess substrate, the formed alkyl peroxyl radical would likely
start a radical chain reaction. Since the products of the metal-
based and organic radical-based reactions are identical, the
resultant catalytic activity cannot be attributed to turnover at
the metal complex as it can for the enzyme.
The HA ability of other simple metal coordination complexes
has been determined.32,33,70 The reaction of 9,10-dihydroan-
thracene (DHAn) is often used as a standard to gauge relative
reactivity since identification of the organic product anthracene
is straightforward. Generally, log(k2corr) scales well with ∆H°,
Experimental Section
particularly in the cases of oxygen-based organic radicals and
Syntheses. All starting materials were purchased from Aldrich and
used without further purification unless noted otherwise. FeII(OTf)2 was
synthesized according to a literature method.96 Sodium hydride (60%
oil dispersion) was washed with hexanes and dried in a vacuum. All
solvents and gases were of analytical grade and were purified by
literature methods.97 CH2Cl2 and MeCN were distilled from CaH2 under
N2 and stored over 4 Å molecular sieves. MeOH was distilled from
Mg(OMe)2 under N2 and stored in darkness over 4 Å molecular sieves.
Anhydrous diethyl ether (ether) was stored over 4 Å molecular sieves.
THF was distilled from Na under N2. 1,4-Cyclohexadiene and ethyl
linoleate were dried over CaH and vacuum transferred under N2.
9,10-Dihydroanthracene was sublimed and recrystallized from EtOH.
Fluorene and xanthene were recrystallized twice from EtOH. 2,3-
Dimethyl-2-butene, 4-methyl-1,3-pentadiene, and tetrahydronaphthalene
were dried over CaSO4 and distilled under N2. Toluene, toluene-d8,
and ethylbenzene were washed with cold concentrated H2SO4, water,
- 33
metal-oxo complexes such as MnO4
.
A closer inspection
of the reactivity, however, reveals that other metal complexes
deviate slightly from this correlation. The ferric biimidazole
system of Roth and Mayer abstracts hydrogen atoms at about
the same rate as [FeIII(PY5)(OMe)](OTf)2, although the N-H
bond formed (76 ( 2 kcal mol-1) is roughly 7 kcal mol-1
lower.33
An Eyring plot analysis of the reaction of 2 with DHAn over
the range of 298-343 K49 indicates a ∆Hq of 13.4 ( 0.5 kcal
mol-1 and a ∆Sq of -25 ( 3 eu. An analysis of the two kinetic
parameters and ∆H° for the two iron systems as well as three
manganese systems shows little correlation between ∆Hq and
∆H° (Table 5). ∆Sq varies independently of ∆H° as well.
Consequently, there does not appear to be a simple linear
correlation between ∆H° and the rate of HA among the studied
(96) Haynes, J. S.; Sams, J. R.; Thompson, R. C. Can. J. Chem. 1981, 59, 669-
678.
(97) Perrin, D. D.; Armarego, W. L. F. Purification of Laboratory Chemicals,
1st ed.; Pergamon Press: New York, 1988.
(95) Novak, M. J.; Clapp, C. H.; Senchak, S. E.; Stover, T. J.; Potter, T. C.;
Findeis, P. M. J. Am. Chem. Soc. 2001, 123, 747-748.
9
J. AM. CHEM. SOC. VOL. 124, NO. 1, 2002 93

A R T I C L E S
Goldsmith et al.
10% NaHCO3, and again water until the washings were neutral. After
drying with MgSO4, these substrates were distilled from Na. All iron
complexes for crystallographic analysis were synthesized and handled
under a N2 inert atmosphere using a MBraun Labmaster 130 glovebox
or standard Schlenk-line techniques. Flash column chromatography was
performed using Silica Gel 60, 230-400 mesh from EM Science
(Gibbstown, NJ) using standard techniques.98
were grown from an acetone/ether solution. 1H NMR (400 MHz,
CDCl3): δ (ppm) 3.07 (6 H, s, C-OMe), 7.10 (4 H, t of d, J1 ) 3.0
Hz, J2 ) 1.2 Hz, 5-Hpy-a), 7.34 (4 H, d, J ) 5.0 Hz, 3-Hpy-a), 7.42
(2 H, d, J ) 5.6 Hz, 3-Hpy-b), 7.52 (4 H, t, J ) 5.0 Hz, 4-Hpy-a),
7.64 (1 H, t, J ) 5.6 Hz, 4-Hpy-b), 8.33 (4 H, d, J ) 3.0 Hz, 6-Hpy-
a). 13C NMR (400 MHz, CDCl3): δ (ppm) 53.2, 121.4, 121.7, 121.8,
125.0, 135.5, 136.9, 148.2. Mass spectroscopy (FAB+, MH+): m/e
476.1 (EM ) 475.2).
Instrumentation. 1H NMR spectra were recorded on a Varian
Gemini-400 (400 MHz) NMR spectrometer at RT, and chemical shifts
are reported in ppm downfield from an internal TMS reference.
Electronic spectra at RT were measured on a Polytec X-dap fiber-optics
UV/vis diode array spectrophotometer. Electronic paramagnetic reso-
nance (EPR) spectra were recorded on a Bruker ER 220D-SRC
instrument as frozen solutions at 77 K at X-band frequency in quartz
tubes. EPR spectra were fit using Bruker Simfonia 1.25. Electrochemical
measurements were recorded at 100 mV/s under N2 at RT using a
Bioanalytical Systems, Inc. CV-50W voltammetric analyzer, a platinum
working electrode, a platinum wire auxiliary electrode, 0.1 M (n-Bu4N)-
(ClO4) supporting electrolyte, and a silver/silver chloride wire reference
electrode, with all potentials referenced to the ferrocenium/ferrocene
couple (in MeOH ) +0.610 V vs SHE, ∆E ) 0.080 V).56 Solution
magnetic moments were determined in MeOH-d4, acetone-d6, and
CDCl3 at RT by the Evans method.47 Solid-state magnetic susceptibility
measurements were performed with a SQUID magnetometer on a
Magnetic Property Measurement System model 1822 at 298 K. Gas
chromatography-mass spectroscopy (GC-MS) data were collected on
a Hewlett-Packard 5890 system. Mass spectroscopy data (positive FAB
and LSIMS) were collected by the Mass Spectrometry Facility,
Department of Pharmaceutical Chemistry, University of California, San
Francisco. Elemental analyses were performed by Midwest Microlabs
(Indianapolis, IN).
Metal Complex Syntheses. [Fe(PY5)(MeOH)](OTf)2 (1). Equi-
molar amounts of PY5 (0.51 g) and Fe(OTf)2 (0.38 g) were dissolved
in MeOH (25 mL) under N2. Addition of ether resulted in the precipi-
tation of a yellow compound in nearly quantitative yield (0.85 g, 92%).
Yellow-green crystals suitable for X-ray analysis were obtained from
a MeOH/ether solution of the complex, [Fe(PY5)(MeOH)](OTf)2‚
(MeOH). Absorption spectrum (MeOH): λmax (nm), ꢀ (M-1 cm-1) 370,
1650; 785, 14; 865, 11. Cyclic voltammetry (MeOH): +0.930 V vs
SHE (∆E ) 0.140 V). Solution magnetic moment (acetone-d6, 292
1
K): µeff ) 4.7 µB. H NMR (400 MHz, acetone-d6): δ (ppm) -11.3,
13.7, 15.8, 42.4, 53.5, 57.0. Elemental analysis: calcd for C32H33N5O11-
F6S2Fe C, 42.83; H, 3.70; N, 7.80; found C, 42.46; H, 3.38; N, 7.76.
[Fe(PY5)(OMe)](OTf)2 (2). Addition of 0.5 equiv of iodoso-
benzene (0.044 g) to a solution of [FeII(PY5)(MeOH)](OTf)2 (0.32 g)
in MeOH (20 mL) gave a red-orange ferric species at RT. The solution
was removed and the orange crystalline material washed with ether.
Recrystallization from a MeOH/ether solution of the complex resulted
in red-orange crystals suitable for X-ray analysis (0.24 g, 75%), [FeIII-
(PY5)(OMe)](OTf)2‚MeOH. Absorption spectrum (MeOH): λmax (nm),
ꢀ (M-1 cm-1) 337, 3600. Electron paramagnetic resonance (MeOH):
(T ) 77 K) g ) 2.25, 2.16, and 1.96. Cyclic voltammetry (MeOH):
irreversible reduction potential ) +0.710 V vs SHE (Fc+/Fc in MeOH
) +0.610 V vs SHE). Solution magnetic moment (CD3OD, 292 K):
µeff ) 5.3 µB. 1H NMR (400 MHz, CDCl3): δ (ppm) -7.1, 12.5, 15.7,
45.3, 55.6, 60.2. Elemental analysis: calcd for C32H30N5O10F6S2Fe C,
43.76; H, 3.44; N, 7.97; found C, 43.57; H, 3.28; N, 7.76.
Ligand Syntheses. 2,6-Bis(bis(2-pyridyl)carbinol)pyridine (PY5-
OH). A dioxane solution containing 2,6-pyridinedicarboxylic acid (5.16
g) and excess thionyl chloride was heated at reflux for 4 h, and the
solvent was removed to give a quantitative yield of 2,6-pyridinedicar-
bonyl chloride.99 A THF solution (500 mL) of 2-bromopyridine (19.75
g) was cooled to -78 °C, and n-BuLi (50 mL, 2.5 M in hexane) was
added dropwise to maintain the temperature below -60 °C. Slow
addition of a THF solution (50 mL) of the acid chloride (6.26 g) at
-78 °C to this lithopyridine solution was followed by quenching with
MeOH (50 mL) and warming the solution to RT. After the addition of
water (50 mL) and 10% HCl (100 mL), the organic components were
removed, and the aqueous solution was washed with CH2Cl2. NaOH
was added to the solution until basic, and the product was extracted
with CH2Cl2. Evaporation of the CH2Cl2 solution gave a crude product
that did not readily crystallize. The product was purified by flash column
chromatography (5% MeOH/CH2Cl2) followed by crystallization from
[Fe(PY5)(CH3CN)](OTf)2 (3). Equimolar amounts of PY5 (0.11
g) and Fe(OTf)2 (0.09 g) were dissolved in CH3CN (5 mL) under
N2 to give a deep orange-brown solution. A brown crystalline solid
was isolated after addition of ether (0.15 g, 75%). If dried under vacuum
or kept in an unsealed container, loss of the coordinated CH3CN
results in a pale yellow compound. Orange-brown crystals grown in
an CH3CN/ether solution are stable in ether solution. Absorption
spectrum (CH3CN): λmax (nm), ꢀ (M-1 cm-1), 362, 6900; 384, 6900;
424, 5880; 554, 175. Cyclic voltammetry (CH3CN): +1.150 V vs
SHE (∆E ) 0.090 V). Solution magnetic moment (CD3CN, 292 K):
1
µeff ) 0 µB. H NMR (400 MHz, CD3CN): δ (ppm) 4.08 (6 H, s,
C-OMe), 7.74 (4 H, d of d, J ) 4.1 Hz, 3-Hpy-a), 8.01 (4 H, t, J )
7.6 Hz, 4-Hpy-a), 8.20 (1 H, t, J ) 9.2 Hz, 4-Hpy-b), 8.30 (4 H, d,
J ) 7.3 Hz, 5-Hpy-a), 8.37 (2 H, d, J ) 8.6 Hz, 3-Hpy-b), 9.97 (4 H,
d, J ) 4.6 Hz, 6-Hpy-a). The perchlorate analogue of this compound
has been previously reported.36
1
acetone/ether (7.05 g, 50%). H NMR (400 MHz, CDCl3): δ (ppm)
7.15 (4 H, t of d, J1 ) 3.0 Hz, J2 ) 1.0 Hz, 5-Hpy-a), 7.18 (2 H, d, J
) 1.6 Hz, 3-Hpy-b), 7.42 (4 H, d, J ) 5.2 Hz, 3-Hpy-a), 7.50 (1 H, t,
4-Hpy-b), 7.52 (4 H, t of d, J1 ) 5.2 Hz, J2 ) 1.0 Hz, 4-Hpy-a), 7.74
(2 H, s, C-OH), 8.49 (4 H, d, J ) 3.0 Hz, 6-Hpy-a). 13C NMR (400
MHz, CDCl3): δ (ppm) 120.6, 122.3, 123.3, 136.2, 137.5, 147.5, 162.6.
Mass spectroscopy (FAB+, MH+): m/e 448.1 (EM ) 447.2).
2,6-Bis(bis(2-pyridyl)methoxymethane)pyridine (PY5). Dimethyl-
ation of PY5-OH (6.18 g) was performed in DMF (100 mL) using 2
equiv of NaH (0.70 g) and 2 equiv of iodomethane (4.12 g) at RT for
1 h. The solution was acidified with 10% HCl (30 mL) to dissolve the
product in the aqueous layer, and the DMF was removed by extraction
with CH2Cl2. NaOH was added to the aqueous solution until basic,
and the product was extracted with CH2Cl2. The crude product collected
from evaporation of the CH2Cl2 solution was purified by crystallization
from cold acetone/ether (5.65 g, 85%). Colorless X-ray quality crystals
[Fe(PY5)(OMe)](OTf) (4). Addition of triethylamine (0.3 mL, 18
equiv) to a solution of [Fe(PY5)(MeOH)](OTf)2 (0.11 g) in MeOH
(10 mL) under N2 resulted in a cloudy red-orange solution at RT. After
filtration of the solution, addition of ether resulted in the precipitation
of an orange compound (0.024 g, 25%). Absorption spectrum
(MeOH): λmax (nm), ꢀ (M-1 cm-1), 340, 1300; 440, 1700; 715, 30.
Cyclic voltammetry (MeOH): +0.760 V vs SHE (∆E ) 0.100 V).
Solution magnetic moment (acetone-d6, 292 K): µeff ) 4.6 µB. 1H NMR
(400 MHz, acetone-d6): δ (ppm) -10.8, 14.1, 15.8, 39.0, 54.3, 60.1.
Substrate Syntheses. 2,4,6-Tri-tert-butylphenol-d (TTBP-d). 2,4,6-
Tri-tert-butylphenol (0.13 g) was dissolved in DMSO-d6 (3.0 mL) along
with 0.010 g of NaH under N2. The solution was stirred overnight and
then quenched with 5.0 mL of D2O. White precipitate was collected,
washed with D2O, and dried under vacuum (0.12 g, 90%). GC-MS
indicated deuteration of >99%. 1H NMR (400 MHz, CDCl3): δ (ppm)
7.20 (2H, s), 1.45 (18 H, s), 1.30 (9 H, s).
(98) Still, W. C.; Kahn, M.; Mitra, A. J. Org. Chem. 1978, 43, 2923-2925.
(99) Bessard, Y.; Crettaz, R. Heterocycles 1999, 51, 2589-2602.
9
94 J. AM. CHEM. SOC. VOL. 124, NO. 1, 2002

Synthesis and Reactivity of a Lipoxygenase Mimic
A R T I C L E S
9,10-Dihydroanthracene-d4 (DHAn-d4). 9,10-Dihydroanthracene
(3.02 g) was dissolved in DMSO-d6 (8.0 mL) along with 0.19 g of
NaH under N2. After the deep red solution was stirred at RT for 2 h,
the reaction was quenched with D2O (5 mL), and the crude product
was filtered and washed with copious amounts of H2O. 1H NMR
(CDCl3) indicated deuteration of ∼90%. The above reaction was
repeated, and the deuterated product was recrystallized twice from EtOH
(2.78 g, 90%). GC-MS indicated deuteration of >99%. 1H NMR (400
MHz, CDCl3): δ (ppm) 7.18 (4H, m, 1-H), 7.33 (4H, m, 2-H), no
resonance peak at 3.95, as in DHAn.
lographic software package from Molecular Structure Corporation.
Specific details for each of the crystal structures are found in the
Supporting Information.
pKa Measurements. The pKa of [FeII(PY5)(MeOH)]2+ was deter-
mined by spectrophotometric titrations with two bases: 2,6-lutidine
(pKa(MeOH) of cation ) 6.8)105 and N-methylmorpholine (pKa(MeOH)
of cation ) 8.4 as measured against bromocresol green, pKa(MeOH)
) 9.8106). Each scan was fit to the best linear combination of pure 1
and 4 through a least-squares fit criterion in an Excel spreadsheet. From
the relative concentrations of 1 and 4 and the known stoichiometry of
the base added, a pKa can be obtained.
1,4-Cyclohexadiene-d8 (CHD-d8). A modified Birch reduction of
benzene was used.100 To anhydrous liquid NH3 (75 mL), cooled to 233
K, was added benzene-d6 (3.25 g), ethanol-d6 (5.46 g), and diethyl ether
(8.45 g). Sodium metal (2.61 g) was added slowly over 2 h, so as to
maintain the temperature of the reaction below the boiling point of
NH3 (-33 °C). The reaction was quenched with Decalin (15 mL) and
ice water (40 mL). The organic layer was washed with copious amounts
of water followed by 1 M HCl. After drying with MgSO4, the product
was distilled from Decalin under N2, discarding the first fraction of
ethanol. A second distillation under N2 gave CHD-d8 in good purity
Kinetic Measurements. Kinetic experiments were performed in
rigorously dried solvents. MeOH (Aldrich, 99.8+%) was refluxed over
Mg turnings and iodine until the iodine color disappeared. The dried
solvent was distilled from Mg(OMe)2 under N2 and stored in darkness
in the drybox over molecular sieves, 4 Å. CH3CN (99.8%) was refluxed
over CaH and then distilled under N2. The dried solvent was stored in
darkness in the drybox over molecular sieves, 4 Å.
The kinetic data were measured with a Polytec X-dap fiber-optics
UV/vis diode array spectrophotometer. The reaction cell was maintained
at a constant temperature using an FTS Systems model MC480A1 multi-
temperature bath. MeOH and MeCN solutions of 2 were prepared in
the drybox or on a N2 line using standard Schlenk techniques directly
before use. Substrate was added directly to the solvent, and the solution
was transferred to the reaction vessel. A fiber-optic dip-probe was
inserted into the solution (path length ) 0.1 or 1.0 cm), and the vessel
was carefully sealed. For kinetic experiments in the presence of O2,
the initial N2 was removed by continuous purging of the reaction vessel
with O2. Data collection was delayed for 5 min to allow for complete
diffusion of O2 into the reaction mixture.
All kinetic runs were performed with an excess of substrate to
achieve pseudo-first-order conditions for the reduction of the ferric
complex. The reduction of [FeIII(PY5)(OMe)]2+ was measured over
the wavelength range from 300 to 450 nm, which includes the band
maxima for all three relevant ferric and ferrous complexes. A constant
concentration of 2 (2.25 × 10-3 M) was used in all reactions, while
the substrate concentration ranged from 0.20 to 2.00 M and was adjusted
to achieve a reasonable rate of ferric complex reduction. For the
reactions run in MeOH, the reduction of 2 to 1 was monitored, and in
MeCN, the reduction of 2 to 3 was monitored. The reactions were run
to near completion (greater than 5 half-lives of 2) in all kinetic
measurements except for that of toluene. These reactions were
exceptionally slow and were only run to greater than 2 half-lives of 2.
All reactions remained homogeneous throughout the data collection.
Spectral deconvolution and kinetic analyses were determined using
the Specfit software package. The data were corrected to remove
substrate or product peaks in the iron region of the spectra when these
features interfered with the kinetic analyses. The rate constants were
determined from absorption data collected in the wavelength range
300-450 nm. For each substrate, at least four measurements were made
of kobs with differing substrate concentrations in the range of 0.20-2.0
1
(2.44 g, yield ) 75%). GC-MS indicated deuteration of >99%. H
NMR (400 MHz, CDCl3): δ (ppm) no resonance peaks.
X-ray Crystallography. ORTEP representations with a detailed
numbering scheme and complete tables of positional parameters, bond
lengths, bond angles, and anisotropic thermal factors for the crystal
structures described below are provided in the Supporting Information.
For each of the X-ray crystal structures presented, the data set was
collected at a temperature of 203 K using the ω scan technique. A
suitably sized crystal was mounted in paratone oil on a glass fiber and
placed in a cold stream of N2 on an Enraf-Nonius CAD-4 diffractometer
with graphite-monochromated Mo KR radiation (λ ) 0.71073 Å). Unit
cell parameters and an orientation matrix for data collection were
typically obtained from a least-squares refinement using the setting
angles of 25 carefully centered reflections in the range 19.00° < 2θ <
36.00°. Structural and refinement data for PY5, 1, and 2 are summarized
in Table 2. Moving-crystal moving-counter background measurements
were made by scanning an additional 25% above and below the scan
range. The counter aperture consisted of a variable horizontal slit with
a width ranging from 2.0 to 2.5 mm and a vertical slit set to 2.0 mm.
Three intensity-check reflections were monitored hourly to determine
if there was any significant crystal decay due to X-ray exposure. Unless
noted below, a correction factor was not applied. An empirical
absorption correction based on azimuthal scans of several appropriate
reflections was applied to all of the data sets. The data were corrected
for Lorentz and polarization effects. The structures were solved by direct
methods and expanded using Fourier techniques. All non-hydrogen
atoms were refined anisotropically, unless noted below. Hydrogen atoms
were located by difference Fourier maps but included at an idealized
position 0.95 Å from their parent atoms for the final refinement.
Isotropic thermal parameters 1.2 times the parent atoms were assumed.
Unless otherwise noted, the remaining significant peaks on the final
difference Fourier maps of 1 and 2 were located near the triflate anion-
(105) Covington, A. K.; Dickinson, T. Physical Chemistry of Organic SolVent
Systems; Plenum Press: London, 1973; p 823.
(106) Dean, J. A. Lange’s Handbook of Chemistry, 15th ed.; McGraw-Hill: New
York, 1999; Vol. 1.
(s). Neutral atom scattering factors were taken from Cromer and
102
Waber.101 Anomalous dispersion effects were included in Fcalc
,
the
values for ∆f ′ and ∆f ′′ were those of Creagh and McAuley.103 The
values for the mass attenuation coefficients are those of Creagh and
Hubbell.104 All calculations were performed using the teXsan crystal-
(107) Bordwell, F. G.; Cheng, J.-P.; Ji, G.-Z.; Satish, A. V.; Zhang, X. J. Am.
Chem. Soc. 1991, 113, 9790-9795.
(108) Berkowitz, J.; Ellison, G. B.; Gutman, D. J. Phys. Chem. 1994, 98,
2744-2765.
(109) Arends, I. W. C. E.; Mulder, P.; Clark, K. B.; Wayner, D. D. M. J. Phys.
Chem. 1995, 99, 8182-8189.
(100) Brandsma, L.; van Sodingen, J.; Andringa, H. Synth. Commun. 1990, 20,
2165-2168.
(101) Cromer, D. T.; Waber, J. T. International Tables for X-ray Crystal-
lography; Kynoch: Birmingham, 1974; Vol. IV.
(110) Luo, Y.-R.; Holmes, J. L. J. Phys. Chem. 1994, 98, 303-312.
(111) Rodgers, A. S.; Wu, M. C. R. J. Am. Chem. Soc. 1973, 95, 6913-
6917.
(102) Ibers, J. A.; Hamilton, W. C. Acta Crystallogr. 1964, 17, 781-782.
(103) Creagh, D. C.; McAuley, W. J. In International Tables for Crystallography;
Wilson, A. J. C., Ed.; Kluwer Academic Publishers: Boston, 1992; Vol.
C, pp 219-222, Table 4.2.6.8.
(104) Creagh, D. C.; Hubbell, J. H. In International Tables for Crystallography;
Wilson, A. J. C., Ed.; Kluwer Academic Publishers: Boston, 1992; Vol.
C, pp 200-206, Table 4.2.4.3.
(112) Stein, S. E.; Brown, R. L. J. Am. Chem. Soc. 1991, 113, 787-793.
(113) Rossi, M. J.; McMillen, D. F.; Golden, D. M. J. Phys. Chem. 1984, 88,
5031-5039.
(114) Pedley, J. B.; Naylor, R. D.; Kirby, S. P. Thermochemical Data of
Organic Compounds; Chapman and Hall: London, 1986.
(115) Lind, J.; Shen, X.; Eriksen, T. E.; Mere´nyi, G. J. Am. Chem. Soc. 1990,
112, 479-482.
9
J. AM. CHEM. SOC. VOL. 124, NO. 1, 2002 95

A R T I C L E S
Goldsmith et al.
M. The absolute second-order rate constants k2 were calculated by a
least-squares fitting plot of kobs versus substrate concentration.
Kinetic experiments were analyzed for products following each run.
The reaction mixture was concentrated to near dryness, dissolved in
ether, and washed with 0.1 M NaOH followed by 1 M HCl solutions
to remove the iron and ligand. Following this workup, the following
products were identified by GC-MS (starting substrate in parenthe-
ses): anthracene (DHAn), benzene (CHD), 1,2-dihydronaphthalene
(tetralin), and R-methylstyrene (cumene). The other organic products
either were generated in quantities below the level of detection or were
removed in the base/acid workup. When a proposed radical intermediate
could lose an additional hydrogen to form an olefin, the resultant olefin
was proposed to be the product, based on the aforementioned products
for similar reactions. Because of the large excess of substrate needed
to run the reactions on a convenient time scale, accurate yields could
not be measured for most kinetic experiments.
Acknowledgment. We thank Dr. Uma Sundaram for useful
discussions. We are grateful to the National Institutes of Health
(GM50730) and the Stanford Graduate Fellowship Fund (C.R.G.)
for financial support of this work.
Supporting Information Available: Experimental data for
variable-temperature solution magnetic susceptibility measure-
ments for 1 and 2; kinetic data for Table 3; plot of k2corr versus
substrate pKa; Eyring plot of the DHAn reduction of 2 over
298-343 K; structure reports for PY5, 1, and 2 (PDF). This
material is available free of charge via the Internet at
http://pubs.acs.org.
JA016451G
9
96 J. AM. CHEM. SOC. VOL. 124, NO. 1, 2002