Regioselective aliphatic carbon-carbon bond cleavage by a model system of relevance to iron-containing acireductone dioxygenase

Article abstract of DOI:10.1021/ja3038189

Mononuclear Fe(II) complexes ([(6-Ph₂TPA)Fe(PhC(O)C(R)C(O)Ph)]X (3-X: R = OH, X = ClO₄ or OTf; 4: R = H, X = ClO₄)) supported by the 6-Ph₂TPA chelate ligand (6-Ph₂TPA = N,N-bis((6-phenyl-2-pyridyl)methyl)-N-(2-pyridylmethyl)amine) and containing a β-diketonate ligand bound via a six-membered chelate ring have been synthesized. The complexes have all been characterized by ¹H NMR, UV-vis, and infrared spectroscopy and variably by elemental analysis, mass spectrometry, and X-ray crystallography. Treatment of dry CH₃CN solutions of 3-OTf with O₂ leads to oxidative cleavage of the C(1)-C(2) and C(2)-C(3) bonds of the acireductone via a dioxygenase reaction, leading to formation of carbon monoxide and 2 equiv of benzoic acid as well as two other products not derived from dioxygenase reactivity: 2-oxo-2- phenylethylbenzoate and benzil. Treatment of CH₃CN/H₂O solutions of 3-X with O₂ leads to the formation of an additional product, benzoylformic acid, indicative of the operation of a new reaction pathway in which only the C(1)-C(2) bond is cleaved. Mechanistic studies show that the change in regioselectivity is due to the hydration of a vicinal triketone intermediate in the presence of both an iron center and water. This is the first structural and functional model of relevance to iron-containing acireductone dioxygenase (Fe-ARD′), an enzyme in the methionine salvage pathway that catalyzes the regiospecific oxidation of 1,2-dihydroxy-3-oxo-(S)- methylthiopentene to form 2-oxo-4-methylthiobutyrate. Importantly, this model system is found to control the regioselectivity of aliphatic carbon-carbon bond cleavage by changes involving an intermediate in the reaction pathway, rather than by the binding mode of the substrate, as had been proposed in studies of acireductone enzymes.

Full text of DOI:10.1021/ja3038189

Article
pubs.acs.org/JACS
Regioselective Aliphatic Carbon−Carbon Bond Cleavage by a Model
System of Relevance to Iron-Containing Acireductone Dioxygenase
Caleb J. Allpress,† Katarzyna Grubel,† Ewa Szajna-Fuller,† Atta M. Arif,‡ and Lisa M. Berreau*,†
†Department of Chemistry and Biochemistry, Utah State University, Logan, Utah 84322-0300, United States
‡Department of Chemistry, University of Utah, Salt Lake City, Utah 84112-0850, United States
S
* Supporting Information
ABSTRACT: Mononuclear Fe(II) complexes ([(6-Ph2TPA)Fe(PhC(O)-
C(R)C(O)Ph)]X (3-X: R = OH, X = ClO4 or OTf; 4: R = H, X = ClO4))
supported by the 6-Ph2TPA chelate ligand (6-Ph2TPA = N,N-bis((6-phenyl-
2-pyridyl)methyl)-N-(2-pyridylmethyl)amine) and containing a β-diketonate
ligand bound via a six-membered chelate ring have been synthesized. The
1
complexes have all been characterized by H NMR, UV−vis, and infrared
spectroscopy and variably by elemental analysis, mass spectrometry, and X-
ray crystallography. Treatment of dry CH3CN solutions of 3-OTf with O2
leads to oxidative cleavage of the C(1)−C(2) and C(2)−C(3) bonds of the
acireductone via a dioxygenase reaction, leading to formation of carbon monoxide and 2 equiv of benzoic acid as well as two
other products not derived from dioxygenase reactivity: 2-oxo-2-phenylethylbenzoate and benzil. Treatment of CH3CN/H2O
solutions of 3-X with O2 leads to the formation of an additional product, benzoylformic acid, indicative of the operation of a new
reaction pathway in which only the C(1)−C(2) bond is cleaved. Mechanistic studies show that the change in regioselectivity is
due to the hydration of a vicinal triketone intermediate in the presence of both an iron center and water. This is the first
structural and functional model of relevance to iron-containing acireductone dioxygenase (Fe-ARD′), an enzyme in the
methionine salvage pathway that catalyzes the regiospecific oxidation of 1,2-dihydroxy-3-oxo-(S)-methylthiopentene to form
2-oxo-4-methylthiobutyrate. Importantly, this model system is found to control the regioselectivity of aliphatic carbon−carbon
bond cleavage by changes involving an intermediate in the reaction pathway, rather than by the binding mode of the substrate, as
had been proposed in studies of acireductone enzymes.
catalyzed by one of two different dioxygenase enzymes
(Scheme 1).9 In the on-pathway reaction, iron-containing
acireductone dioxygenase (Fe-ARD′) catalyzes the oxidative
cleavage of the C(1)−C(2) bond resulting in the formation of
formic acid and an α-keto acid, the latter of which undergoes
transamination in a subsequent step to regenerate methionine.
In the off-pathway reaction, nickel-containing acireductone di-
oxygenase (Ni-ARD) catalyzes the oxidative cleavage of the
C(1)−C(2) and C(2)−C(3) bonds of the acireductone to form
formic acid, carbon monoxide (CO), and a carboxylic acid.10
The mechanism of Ni-ARD has been the focus of several recent
studies, primarily through model systems, for two main reasons:
(1) Nickel-containing dioxygenases were hitherto unknown;
and (2) the production of CO is of particular current interest
due to its role in cellular signaling.11 In this context, the branch
point in the methionine salvage pathway at which the
acireductone dioxygenases operate represents a combination
of a potential regulatory shunt coupled with the production of a
signaling molecule.
INTRODUCTION
■
One of the most challenging reactions in chemistry and biology
is the selective oxidative cleavage of carbon−carbon bonds. The
enzymes that carry out these reactions are typically dioxy-
genases, which incorporate two oxygen atoms into products via
an oxidative mechanism and often contain a metal cofactor
(typically iron) in a nonheme binding pocket.1 While the mech-
anism of enzymes that cleave aromatic carbon−carbon bonds,
such as the extradiol and intradiol catechol dioxygenases, have
been extensively studied,2,3 enzymes that cleave aliphatic
carbon−carbon bonds have received much less attention until
recently. These enzymes, which include acetylacetone-cleaving
dioxygenase (Dke1),4 2,4′-dihydroxyacetophenone dioxygenase,5
hydroxyethylphosphonate dioxygenase,6 and the acireductone
dioxygenases,7 are the subject of growing interest. The acireduc-
tone dioxygenases, found in the methionine salvage pathway, are
of particular current interest, due to the change in regiospecificity
of the reaction as a function of metal ion bound at the active site.
This differing reactivity within the enzymes as a function of metal
ion identity is unique at present in biology.
Despite catalyzing reactions with differing regiospecificity,
Fe-ARD′ and Ni-ARD contain identical peptide sequences and
bind the divalent metal cofactor with the same four amino acid
The methionine salvage pathway is ubiquitous in biological
systems and is responsible for regenerating methionine from
5′-methylthioadenosine.8 In aerobic systems, this pathway con-
tains a single branch point, wherein an acireductone intermediate
(1,2-dihydroxy-3-oxo-(S)-methylthiopentene) undergoes a reaction
Received: April 20, 2012
© XXXX American Chemical Society
A
dx.doi.org/10.1021/ja3038189 | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Mechanistic studies suggest this reaction proceeds by an initial
net two-electron process to form 1,3-diphenylpropantrione
(a vicinal triketone) and hydroperoxide anion, which may then
combine to form a dioxetane ring and subsequently cleave the
C−C bonds.13 The triketone intermediate was implicated by
the detection of benzil, which is formed in the reaction mix-
ture via a benzoyl migration and decarbonylation (Scheme 2,
bottom).14 We note that the production of benzil was the only
initial evidence for a reaction pathway involving a triketone
intermediate. Benzoyl migration is not possible with the native
ARD substrate, and so such a byproduct would not be pro-
duced in the enzyme, even if the reaction were to proceed via a
triketone intermediate.
Scheme 1. Regiospecificity of Aliphatic Carbon−Carbon
Bond Cleavage by Acireductone Dioxygenases in the
Methionine Salvage Pathway
In the present work, we have investigated the role of the
metal center in the O2 reactivity of acireductones by utilizing an
iron-containing analogue of 1. While spectroscopic studies
provide evidence for a six-membered chelate ring for the coor-
dinated acireductone, we have found that the presence of iron
and water in the reaction mixture opens up a new oxidative
reaction pathway not accessible in our nickel-containing system.
This new reaction pathway results in the formation of benz-
oylformic acid, the α-keto acid product that would be expected
in a Fe-ARD′ type reaction.
residues (3His, 1Glu).9 Thus, the only constitutive difference
for the two enzymes is the nature of the metal at the active site.
To date, no model complexes have been synthesized to study
the reaction pathway in Fe-ARD′.
EXPERIMENTAL SECTION
■
General Methods. All reagents were obtained from commercial
sources and were used without additional purification unless other-
wise noted. The 1,3-diphenylpropantrione was purchased from TCI
America. Solvents were dried according to published procedures and
were purified by distillation under N2 prior to use.15 Air-sensitive
reactions were performed in an MBraun Unilab glovebox under
a N2 atmosphere or by using standard Schlenk techniques.
Fe(OTf)2·2CH3CN was prepared from Fe powder, and FeCl3 was
prepared from FeCl3·6H2O using known procedures.16,17 The bulky
acireductone 2-hydroxy-1,3-diphenylpropan-1,3-dione was synthesized
by modifying a literature procedure as described below.18 The
6-Ph2TPA (N,N-bis((6-phenyl-2-pyridyl)methyl)-N-(2-pyridylmethyl)-
amine) ligand, [(6-Ph2TPA)Ni(CH3CN)(H2O)](ClO4)2, and 2-oxo-2-
phenylethylbenzoate were synthesized by following previously published
procedures.19−21
The current hypothesis for the difference in regiospecificity is
that a change in the coordination mode of the substrate from a
six-membered chelate in Ni-ARD to a five-membered chelate
in Fe-ARD′ results in the observed differences in reactivity
of bond cleavage (Scheme 1).9 To probe this chelate ring
hypothesis, we have previously prepared a Ni(II)-containing
complex of a bulky acireductone (2-hydroxy-1,3-diphenyl-
propan-1,3-dione), [(6-Ph2TPA)Ni(PhC(O)C(OH)C(O)-
Ph)]ClO4 (1), that is supported by an aryl-appended TPA
ligand (6-Ph2TPA = N,N-bis((6-phenyl-2-pyridyl)methyl)-N-
(2-pyridylmethyl)amine) in order to mimic the hydrophobic
binding pocket found in ARDs.12 X-ray crystallographic and 1H
NMR solution studies of 1 revealed that the acireductone
moiety in this complex is coordinated as a six-membered
chelate ring. Upon treatment of 1 with O2, the C(1)−C(2) and
C(2)−C(3) bonds of the acireductone were cleaved, and CO
was generated, in a Ni-ARD-type reaction (Scheme 2, top).
1
Physical Methods. H NMR spectra of organic compounds were
obtained using a JEOL ECX-300 spectrometer; chemical shifts were
referenced to the residual solvent peak in CD2HCN (1.94 ppm,
1
quintet). H NMR spectra of paramagnetic complexes were obtained
using a Bruker ARX-400 spectrometer and parameters, as previously
described.22 UV−vis data were collected on a HP8453A spectrometer
at ambient temperature. IR spectra were recorded on a Shmidzu FTIR-
8400 spectrometer as KBr pellets. Room temperature magnetic
susceptibilities were determined using the Evans method.23 GC-MS
data were obtained on a Shimadzu GCMS-QP5000 GC-MS with a
GC-17A gas chromatograph, using an Alltech EC5 30 m × 25 mm
Scheme 2. (Top) Reaction of 1 with O2 to Form
Ni-ARD-Type Products via the Formation of an Inter-
mediate Triketone Species and (Bottom) Decarbonylation
of 1,3-Diphenylpropantrione via a Lewis Acid-Mediated
Benzoyl Migration
×25 μm thin film capillary column and temperature program: Tinitial
,
70 °C (5 min); temperature gradient, 23 °C min−1; Tfinal, 250 °C
(10 min). LC-MS data were obtained using negative-ion APCI on a
LCQ Thermo Finnigan MS via a HP1100 with a Betasil C18 10 ×
2.1 mm column; solvent gradient from 5% aqueous methanol to 50%
aqueous methanol. CO was detected using an Agilent 3000A Micro
gas chromatograph. Mass spectral data for metal complexes were
collected by the Mass Spectrometry Facility, University of California,
Riverside. Elemental analysis was performed by Atlantic Microlabs Inc.,
Norcross, GA for all compounds except 3-OTf, which was analyzed by
Canadian Microanalytical Service, Ltd.
Kinetic Studies. Measurements were performed on a HP8453A
spectrometer at 20 °C. All manipulations of 3-X were performed under
a N2 atmosphere. O2-saturated solutions of CH3CN (8.2 mM) were
prepared by bubbling dry O2 through a solution of dry CH3CN.24
Solutions containing lower O2 concentrations were prepared by
B
dx.doi.org/10.1021/ja3038189 | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
diluting the 8.2 mM solution with N2-saturated CH3CN using gastight
syringes.
Anal. calcd for C47H37F3FeN4O6S·CH2Cl2: C, 58.08; H, 4.05; N, 5.77.
Found: C, 58.03; H, 4.29; N, 5.39. μeff = 5.13 μB; UV−vis, nm
(ε, M−1cm−1): 385 (8090); FTIR (KBr, cm−1): 3430 (νOH), 1256
(νOTf), 1227 (νOTf), 1169 (νOTf), 1032 (νOTf).
Caution! Perchlorate salts of metal complexes with organic ligands
are potentially very explosive. Only small amounts of material should
be prepared, and these should be handled with extreme caution.25
2-Hydroxy-1,3-diphenyl-propan-1,3-dione. NaHCO3 (1.68 g,
20.0 mmol) was placed in a flask with 0.10 M RuCl3 (800 μL,
0.080 mmol) and diluted with H2O (7.2 mL), CH3CN (48 mL), and
EtOAc (48 mL). Oxone (24.4 g, 40.0 mmol) was added in one portion
and stirred until a bright-yellow suspension formed and effervescence
ceased. Benzylidene acetophenone (1.66 g, 8.00 mmol) was added in
one portion to initially form a brown solution that became yellow over
time. The progress of reaction was carefully monitored by TLC
(3:1 hexanes:EtOAc), and after 18 min the suspension was diluted
with 50 mL EtOAc, and the solid residue filtered off, washing with a
further 30 mL EtOAc. The filtrate was washed with 40 mL saturated
Na2SO3 and 40 mL H2O. The organic layer was dried over Na2SO4
and filtered, and then the solvent was removed under reduced pressure
to yield the crude product. Recrystallization from hot EtOH afforded
white needle-like crystals that were collected by filtration and washed
with cold EtOH followed by Et2O (0.37 g, 20%). 1H NMR (300 MHz,
CD3CN, 25 °C): δ = 7.99 (d, 3J(H,H) = 7.2 Hz, 4H; Ar−H), 7.66 (t,
3J(H,H) = 7.6 Hz, 2H; Ar−H), 7.52 (t, 3J(H,H) = 7.5 Hz, 4H; Ar−H),
[(6-Ph2TPA)Fe(PhC(O)CHC(O)Ph)]ClO4 (4). Me4NOH·5H2O
(0.0067 g, 0.037 mmol) was dissolved in CH3CN (∼2 mL) and
stirred with dibenzoylmethane (0.0076 g, 0.034 mmol) for ∼1 h under
a N2 atmosphere. This solution was then added to a CH3CN (∼2 mL)
solution of 2-ClO4 (0.034 mmol) and stirred for 18 h to produce a
dark-red solution. The solvent was then removed under reduced
pressure, and the residue was dissolved in CH2Cl2 and filtered through
a glass wool/Celite plug. The filtrate was condensed under reduced
pressure, and precipitation of the product was induced by the addition
of excess hexanes. Recrystallization of the crude product from
CH3CN/Et2O yielded red-brown crystals suitable for X-ray crystallog-
raphy (0.016 g, 57%). Anal. calcd for C45H37ClFeN4O6: C, 65.81; H,
4.54; N, 6.83. Found: C, 65.46; H, 4.57; N, 7.27. HRMS (ESI):
+
m/z calcd for C45H37FeN4O2 : 721.2266 [M−ClO4]+; found:
721.2279. μeff = 5.12 μB; UV−vis, nm (ε, M−1 cm−1): 357 (13400);
FTIR (KBr, cm−1): 1094 (νClO4), 623 (νClO4).
2,2-Dihydroxy-1,3-diphenylpropan-1,3-dione. Initially this
triketone hydrate was prepared by exposing 1,3-diphenylpropantrione
to moist air for several weeks, during which time it changed color from
yellow to white. The hydrate may also be synthesized by crystallization
of 1,3-diphenylpropantrione from wet ethanol. Notably the hydra-
tion is reversible, and dehydration will occur over the course of
3
6.34 (d, 3J(H,H) = 7.2 Hz, 1H; CH), 4.68 (d, J(H,H) = 7.0 Hz, 1H;
OH) ppm.
[(6-Ph2TPA)Fe(CH3CN)](ClO4)2 (2-ClO4). Fe(ClO4)2·6H2O
(0.040 g, 0.11 mmol) was dissolved in CH3CN (∼2 mL) and added
to 6-Ph2TPA (0.049 g, 0.11 mmol), and the resulting solution was
stirred for 24 h under an N2 atmosphere. The solution was then
concentrated under reduced pressure, and the metal complex was
precipitated by introducing excess Et2O. The solid was then dried
under reduced pressure (0.059 g, 73%). Et2O diffusion into a CH3CN
solution of 2-ClO4 afforded yellow crystals suitable for X-ray
crystallography. Anal. calcd for C34H32Cl2FeN6O8: C, 52.39; H, 4.14;
N, 10.78. Found: C, 52.12; H, 4.25; N, 11.24. μeff = 5.21 μB; UV−vis,
nm (ε, M−1 cm−1): 285 (14 800); FTIR (KBr, cm−1): 1093 (νClO4),
623 (νClO4).
[(6-Ph2TPA)Fe(CH3CN)](OTf)2·0.5CH2Cl2 (2-OTf). Fe(OTf)2·
2CH3CN (0.11 mmol) was dissolved in CH3CN (∼2 mL) and
added to 6-Ph2TPA (0.11 mmol), and the resulting mixture was stirred
for 24 h under a N2 atmosphere. The solvent was then removed under
reduced pressure, and the metal complex precipitated by addition of
1
several hours when the hydrate is dissolved in dry solvent. H NMR
(300 MHz, CD3CN, 25 °C): δ = 7.98 (d, 3J(H,H) = 7.2 Hz, 4H; Ar−
H), 7.58 (t, 3J(H,H) = 7.4 Hz, 2H; Ar−H), 7.44 (t, 3J(H,H) = 7.5 Hz,
4H; Ar−H), 6.03 (s, 2H; OH) ppm. 13C NMR (100.6 MHz, CD3CN,
25 °C): δ = 196.3, 135.6, 131.3, 130.7, 130.1, 96.8 ppm.
Isomerization of 2-Hydroxy-1,3-diphenyl-propan-1,3-dione
Promoted by 2-ClO4. Complex 2-ClO4 (0.010 mmol) was dissolved
in ∼1 mL CH3CN, and to this solution was added a CH3CN solution
of 2-hydroxy-1,3-diphenyl-propan-1,3-dione (0.010 mmol) and
Me4NOH·5H2O (0.010 mmol) under a N2 atmosphere. The resulting
solution was stirred for 48 h, and the solvent was then removed under
reduced pressure. Analysis of the organic products by GC-MS and 1H
NMR showed the major product was 2-oxo-2-phenylethylbenzoate.
Reaction of 3-X with O2. A 3.0 mL aliquot of 3-X (4.8 mM) in
CH3CN or 95% CH3CN/H2O was placed in a vial. This solution was
then purged with O2 for 15 s, sealed, and stirred for 12 h. The solvent
was then removed under reduced pressure. The organic products were
analyzed by LC-MS as described below.
Control Reaction Testing for Benzoylformic Acid Production
from 2-Hydroxy-1,3-diphenyl-propan-1,3-dione.
Me4NOH·5H2O (0.026 mmol) was dissolved in CH3CN or 85%
CH3CN/H2O (1 mL) and was added to 2-hydroxy-1,3-diphenyl-
propan-1,3-dione (0.026 mmol). This solution was then either
diluted with 2 mL CH3CN or combined with 0.026 mmol of either
[(6-Ph2TPA)Ni(CH3CN)(H2O)](ClO4)2 or 2-ClO4 dissolved in
CH3CN (2 mL) and stirred for 5 min. These solutions were then
purged with O2, sealed, and stirred for 12 h. The solvent was then
removed under reduced pressure. The organic products were
analyzed by LC-MS as described below. Benzoylformic acid was
only detected in the reaction involving 2-ClO4.
excess hexanes to
a CH2Cl2 solution. Anal. calcd for
C34H29F6FeN4O6S2·0.5CH2Cl2: C, 47.07; H, 3.44; N, 7.96. Found:
C, 47.04; H, 3.58; N, 7.89. The presence of 0.5 equiv of CH2Cl2 in the
EA sample was confirmed by integration of the signal of this solvent in
1
the H NMR spectrum of the sample. μeff = 5.03 μB; FTIR (KBr,
cm−1): 1248 (νOTf), 1225 (νOTf), 1167 (νOTf), 1030 (νOTf).
[(6-Ph2TPA)Fe(PhC(O)C(OH)C(O)Ph)]ClO4 (3-ClO4). Me4NOH·
5H2O (0.0049 g, 0.026 mmol) was dissolved in CH3CN (2.0 mL)
and stirred with 2-hydroxy-1,3-diphenyl-propan-1,3-dione (0.0063 g,
0.026 mmol) for 2 min under a N2 atmosphere. This solution was then
added to a CH3CN (1.0 mL) solution of 2-ClO4 (0.026 mmol) and
stirred for 5 min. The solvent was then immediately removed under
reduced pressure. UV−vis, nm (ε, M−1 cm−1): 385 (5080). FTIR
(KBr, cm−1): 3430 (νOH), 1094 (νClO4), 623 (νClO4).
[(6-Ph2TPA)Fe(PhC(O)C(OH)C(O)Ph)]OTf·CH2Cl2(3-OTf).
LiHMDS (0.015 g, 0.091 mmol) was dissolved in Et2O (∼2 mL) and
added to a CH3CN solution of 2-hydroxy-1,3-diphenyl-propan-1,3-dione
(0.022 g, 0.090 mmol) under a N2 atmosphere to form an orange
solution that became cloudy after 1 min. To this solution was added a
CH3CN solution of 2-OTf (0.090 mmol), and the resultant slurry was
stirred for 12 h and then filtered through a glass wool/Celite plug.
The filtrate was then combined with a second slurry of LiHMDS (0.091
mmol) and 2-hydroxy-1,3-diphenyl-propan-1,3-dione (0.091 mmol) in
Et2O/CH3CN and stirred for two days. The solvent was removed under
reduced pressure, and the crude material was redissolved in CH2Cl2 and
filtered through a Celite plug. The compound was then precipitated, first
by vapor diffusion of Et2O into a CH3CN solution, and then by addition
of hexanes to a CH2Cl2 solution to yield a brown solid (0.062 g, 71%).
Control Reaction Testing for Benzoylformic Production
from 1,3-Diphenylpropantrione. Either 1,3-diphenylpropantrione
(0.026 mmol) was dissolved in CH3CN (1 mL) or 2,2-dihydroxy-1,3-
diphenylpropan-1,3-dione was dissolved in 85% CH3CN/H2O (1 mL).
This solution was then either diluted with 2 mL CH3CN or combined
with 0.026 mmol of either [(6-Ph2TPA)Ni(CH3CN)(H2O)](ClO4)2 or
2-ClO4 dissolved in CH3CN (2 mL). To each solution was added 31%
H2O2 (0.026 mmol) and NEt3 (0.026 mmol). The solutions were then
sealed and stirred for 12 h. The solvent was then removed under
reduced pressure. The organic products were analyzed by LC-MS as
described below.
Control Reactions Testing for Benzoylformic Production from
2-oxo-2-Phenylethylbenzoate. Me4NOH·5H2O (0.026 mmol) was
dissolved in CH3CN or 85% CH3CN/H2O (1 mL) and added to
C
dx.doi.org/10.1021/ja3038189 | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
2-oxo-2-phenylethylbenzoate (0.026 mmol). This solution was then
either diluted with 2 mL CH3CN or combined with 2-ClO4 (0.026
mmol) dissolved in CH3CN (2 mL) and stirred for 5 min. These
solutions were then purged with O2, sealed, and stirred for 12 h. The
solvent was then removed under reduced pressure. The organic
products were then analyzed by LC-MS as described below.
Benzoylformic acid was not detected for either of these reactions.
Reaction of 1,3-Diphenylpropantrione with Ferric Salts.
Under a nitrogen atmosphere, 1,3-diphenylpropantrione or 2,2-dihydroxy-
1,3-diphenylpropan-1,3-dione (0.026 mmol) was dissolved in CH3CN
or 85% CH3CN/H2O (1 mL), respectively, and to these solutions was
added FeCl3 or Fe(ClO4)3·6H2O (0.105 mmol) in CH3CN (2 mL).
The solutions were then sealed and stirred for 12 h. The solvent was
then removed under reduced pressure. The organic products were
then analyzed by LC-MS as described below.
Figure 1. Thermal ellipsoid representation of the cationic portions of
2-ClO4 (left) and 4 (right). Ellipsoids are drawn at 50% probability.
Hydrogen atoms are omitted for clarity.
Reaction of 2-Hydroxy-1,3-diphenylpropan-1,3-dione with
Fe(ClO4)3·6H2O. Under a nitrogen atmosphere, Me4NOH·5H2O
(0.026 mmol) was dissolved in 85% CH3CN/H2O (1 mL) and added
to 2-hydroxy-1,3-diphenyl-propan-1,3-dione (0.026 mmol). To this
solution was added Fe(ClO4)3·6H2O (0.105 mmol) in CH3CN
(2 mL). The solution was sealed and stirred for 12 h. The solvent was
then removed under reduced pressure. The organic products were
then analyzed by LC-MS as described below.
the generation and isolation of analytically pure [(6-Ph2TPA)-
Fe(PhC(O)C(OH)C(O)Ph)]OTf (3-OTf). Thus far X-ray
quality crystals of 3-X (X = ClO4 or OTf) have not been
obtained. Therefore, we have synthesized [(6-Ph2TPA)Fe-
(PhC(O)CHC(O)Ph)]ClO4 (4) by combining 2-ClO4 with
1 equiv of the anion of dibenzoylmethane, as outlined in
1
Scheme 3, to use as a structural and H NMR spectroscopic
General Procedure for Organic Product Recovery and
Analysis. To each crude product mixture, 1 mL of 10 mM HCl
and 3 mL Et2O were added and stirred for 3 h. The organic layer was
then decanted, and the aqueous layer extracted with a further 3 mL
Et2O. The organic fractions were combined, and solvent evaporated
under reduced pressure. Recovery of the organic material, determined
as percent mass of the acireductone or triketone starting material, was
typically ∼80%. For further analysis, the organic products were redissolved
in either CH3CN (GC-MS) or 1:1 MeOH:H2O (LC-MS). Products were
identified by comparison to the retention times and fragmentation
patterns of authentic compounds. The ratio of benzoic acid to
benzoylformic acid was determined from a calibration curve based on
peak area in the LC-MS spectrum.
a
Scheme 3. Synthesis of 2−4
18O Labeling Studies. For H218O labeling, a 3.0 mL aliquot of
4.8 mM 3-ClO4 in CH3CN was combined with a 10 μL aliquot of
H218O under an nitrogen atmosphere. This solution was then exposed
to 16O2, sealed, and stirred for 12 h. The solvent was then removed
under reduced pressure.
For 18O2 labeling, a 3.0 mL aliquot of 4.8 mM 3-ClO4 was placed in
a solvent transfer flask from which the atmosphere was removed by
three freeze−pump−thaw cycles. 18O2 was then introduced into the
flask, after which it was resealed and allowed to stir for 12 h. The
solvent was then removed under reduced pressure. 18O incorporation
levels in benzoic acid and benzoylformic acid were determined from
the relative intensities of the [M − 1]−, [M + 1]−, and [M + 3]−
molecular ions in the LC-MS spectrum.
a
All reactions were performed under anaerobic conditions. CH3CN
was used as the solvent unless otherwise noted.
RESULTS
■
Complex Synthesis and Characterization. In our
initial attempts to generate an iron-containing analogue of 1,
admixture of equimolar amounts of Fe(ClO4)2·6H2O and
6-Ph2TPA in CH3CN enabled the facile generation of
[(6-Ph2TPA)Fe(CH3CN)](ClO4)2 (2-ClO4). This compound
has been isolated and comprehensively characterized (X-ray
model to evaluate the coordination mode of the acireductone
ligand in 3-X. X-ray quality crystals of 4 were grown by
diffusion of Et2O into a CH3CN solution of the complex (see
Table 1).
X-ray Crystallography. As shown in Figure 1, X-ray
crystallographic studies of 2-ClO4 revealed a cationic portion
containing a single molecule of acetonitrile bound to the iron
center, resulting in a trigonal bipyramidal geometry (τ =
0.97).26 By contrast, 4·0.5CH3CN has a cationic portion
containing a distorted octahedral Fe(II) center, with the
dibenzoylmethane anion coordinated in a bidentate fashion via
a six-membered chelate ring. As is expected for the fully
delocalized diketonate anion, the C−O bond lengths (O(1)−
C(31) 1.278(5) and O(2)−C(39) 1.270(5) Å) are similar, as
are the C−C bond lengths within the chelate ring (C(31)−
C(38) 1.408(6) and C(38)−C(39) 1.403(6) Å) (see Table 2).
1
crystallography (Figure 1), elemental analysis, H NMR, IR,
and a magnetic susceptibility measurement). When 2-ClO4 is
combined with the monoanion of the bulky acireductone in dry
acetonitrile, a new complex, [(6-Ph2TPA)Fe(PhC(O)C(OH)-
C(O)Ph)]ClO4 (3-ClO4) is formed, however it has not been
obtained in analytically pure form. Therefore, in an alterna-
tive synthetic approach, we combined the anhydrous salt
Fe(OTf)2·2CH3CN with 6-Ph2TPA in dry CH3CN, which led
to the facile generation of [(6-Ph2TPA)Fe(CH3CN)](OTf)2
(2-OTf). Reaction of this complex with an excess of LiHMDS
and acireductone under strictly anhydrous conditions allowed
D
dx.doi.org/10.1021/ja3038189 | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Table 1. Summary of X-ray Data Collection and
Refinement
Table 2. Selected Bond Distances (Å) and Angles (°) for 2-
ClO4 and 4·0.5CH3CN
a
2-ClO4
4·0.5CH3CN
2-ClO4
Fe(1)−N(1)
Fe(1)−N(2)
Fe(1)−N(3)
Fe(1)−N(4)
2.090(2)
2.190(2)
2.149(2)
2.123(2)
2.096(2)
108.19(9)
97.13(9)
76.11(9)
N(1)−Fe(1)−N(2)
N(1)−Fe(1)−N(3)
N(5)−Fe(1)−N(2)
N(4)−Fe(1)−N(3)
N(5)−Fe(1)−N(3)
N(5)−Fe(1)−N(4)
N(4)−Fe(1)−N(2)
78.81(9)
formula
Mr
C36H35Cl2FeN7O8
820.46
C90H74Cl2Fe2N8O12·CH3CN
123.11(9)
172.11(9)
113.93(9)
100.93(9)
111.60(9)
76.21(8)
1683.23
tetragonal
I-4
crystal system
space group
a , Å
monoclinic
P21/c
Fe(1)−N(5)
13.7317(3)
19.1640(3)
14.2994(4)
90
34.4839(4)
34.4839(4)
13.5578(2)
90
N(1)−Fe(1)−N(4)
N(1)−Fe(1)−N(5)
N(3)−Fe(1)−N(2)
4·0.5CH3CN
Fe(1)−N(1)
Fe(1)−N(2)
Fe(1)−N(3)
Fe(1)−N(4)
Fe(1)−O(1)
Fe(1)−O(2)
O(1)−C(31)
O(2)−C(39)
O(2)−Fe(1)−O(1)
O(2)−Fe(1)−N(1)
O(1)−Fe(1)−N(1)
b, Å
c, Å
α, °
β, °
γ, °
V, Å3
a
94.5399(10)
90
90
2.156(3)
2.194(3)
2.371(4)
2.293(4)
2.012(3)
1.979(3)
1.278(5)
1.270(5)
88.82(12)
98.08(12)
172.73(13)
O(2)−Fe(1)−N(2)
O(1)−Fe(1)−N(2)
N(1)−Fe(1)−N(2)
O(2)−Fe(1)−N(4)
O(1)−Fe(1)−N(4)
N(1)−Fe(1)−N(4)
N(2)−Fe(1)−N(4)
O(2)−Fe(1)−N(3)
O(1)−Fe(1)−N(3)
N(1)−Fe(1)−N(3)
N(2)−Fe(1)−N(3)
169.06(13)
94.92(12)
78.70(13)
111.32(13)
95.81(13)
79.61(13)
78.59(14)
96.72(12)
86.26(12)
95.13(12)
73.32(13)
90
3751.14(15)
4
16122.1(4)
8
Z
Dc, Mg m−3
1.453
1.387
T, K
150(1)
150(1)
red-brown
plate
color
yellow
crystal shape
crystal size, mm
μ, mm−1
F(000)
θ range, °
prism
0.28 × 0.23 × 0.15
0.606
0.35 × 0.35 × 0.05
0.497
1696
6992
3.75−27.48
2.00−26.00
99.3
a
completeness to θ, % 99.2
Data for one of two cations present in asymmetric unit.
reflections collected
15 757
13 636
13 631
independent
reflections
8519
Rint
0.0365
0.0525
data/restraints/
parameters
GoF; F2
8519/0/595
13631/0/1057
1.022
1.012
R1, wR2; I > 2σ(I)
R1, wR2; all data
0.0542, 0.1379
0.0811, 0.1569
0.9146/0.8487
0.0515, 0.0914
0.0817, 0.1028
0.9756/0.8454
max/min
transmission
Δρmax/min, eÅ−3
0.696/−0.594
0.701/−0.395
a
Radiation used: Mo Kα (λ = 0.71073 Å); diffractometer: Nonius
KappaCCD.
The overall structural features of 4·0.5CH3CN, as well as the
bond distances involving the coordinated diketonate anion, are
highly similar to those found for the Ni(II)-containing
acireductone complex 1.12
Figure 2. Comparison of selected paramagnetically shifted features
of 2−4.
1H NMR Spectroscopy. A previous study on the
UV−vis and Infrared Spectroscopy. The absorption
spectra of 3-ClO4 and 3-OTf exhibit features with λmax = 385
nm, albeit with differing extinction coefficients depending on
the counterion (5080 and 8090 M−1 cm−1, respectively). We
attribute this absorption band primarily to a π−π* transition of
the acireductone diketonate that is coordinated to the iron
center, consistent with our structural proposal of the acireduc-
tone being bound via a six-membered chelate ring. Similar
features are found for both 4 (λmax = 357 nm; hypsochromically
shifted, as expected in the absence of the hydroxyl group) and 1
(λmax = 399 nm) when dissolved in acetonitrile (Figure S1).12
We note that a shoulder feature on the longer wavelength side
of the π−π* transition of 3-X and 4 may be due to an MLCT
transition, as has been identified in spectroscopic studies of
Dke1 and relevant model compounds.28,29
1
paramagnetically shifted features in the H NMR spectra for
a variety of nickel complexes with the same chelate ligand
(6-Ph2TPA) has demonstrated the sensitivity of the peak
distribution pattern to the coordination of different anions.22 As
shown in Figure 2, complex 3-ClO4 1exhibits signals in the
paramagnetically shifted region of the H NMR spectrum that
are very similar to those exhibited by the dibenzoylmethane
complex 4, albeit the signals for 3-ClO4 are shifted slightly
upfield.27 By contrast, the solvate complex 2-ClO4 has a
distinctly different pattern of peaks in the paramagnetic region.
It is also worth noting that 2-OTf and 3-OTf have the same
peak distribution patterns as 2-ClO4 and 3-ClO4, respectively.
On the basis of these spectroscopic comparisons, we formulate
the solution structure of 3-X as [(6-Ph2TPA)Fe(PhC(O)-
C(OH)C(O)Ph)]X, wherein similar to the Ni(II)-analog 1, the
bulky acireductone is bound via a six-membered chelate ring.
We note that chemical shifts of 1H NMR resonances of 2-OTf,
2-ClO4, 3-OTf, and 4 are available in Table S1.
The proposed six-membered chelate is additionally sup-
ported by the IR spectral features of 3-X, wherein a O−H
stretch is present at ∼3430 cm−1, and no free CO stretch is
found at ∼1700 cm−1. If the acireductone were coordinated as a
E
dx.doi.org/10.1021/ja3038189 | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
five-membered ring, the hydroxyl group would instead likely be
deprotonated and a free CO moiety would be present.
Thus, in contrast to the proposed binding mode of the
acireductone in the enzyme−substrate complex, wherein
changing the metal center from Ni to Fe changes the bind-
ing mode of the acireductone from a six-membered to a five-
membered chelate ring, in our synthetic systems, our acireduc-
tone binds via a six-membered chelate when either metal ion is
present. This is unsurprising, as the difference in Lewis acidity
between Ni(II) and Fe(II) is modest and is not proposed as the
factor that differentiates the binding mode in the enzyme. It is,
rather, changes in the tertiary structure of the enzyme, resulting
in a more open binding pocket in Fe-ARD′ that are proposed
to direct the acireductone to bind via a five-membered chelate.30
Our model systems have the same secondary structural features,
as defined by having the same 6-Ph2TPA chelate ligand, and thus
exhibit the same binding mode for the acireductone. Therefore,
our model system is an ideal case to test the chelate hypothesis
for the differing reactivity of Fe-ARD′ and Ni-ARD; if the
proposal is correct, our iron-containing complexes 3-X should
exhibit the same dioxygenase reactivity as our previously reported
nickel-containing system 1.
Anaerobic Reactivity. Our inability to generate analytically
pure 3-ClO4 has been due to an anaerobic water-promoted
isomerization reaction of the acireductone ligand. The exclusion
of water in the synthesis of 3-OTf was likely the factor that
allowed its generation in analytically pure form. Evidence for the
formation of 3-ClO4, followed by its subsequent decay, was
found by monitoring (UV−vis) of a reaction mixture, wherein an
acetonitrile solution of 2-ClO4 was combined with the bulky
acireductone anion under anaerobic conditions. Slow decay of
the 385 nm absorption band was observed upon prolonged
stirring, and analysis of the organic products by GC-MS showed
that the ester 2-oxo-2-phenylethylbenzoate, an isomer of the
bulky acireductone, had been produced. This same isomeriza-
tion reactivity has been previously reported for the Co(II) analog
[(6-Ph2TPA)Co(PhC(O)C(OH)C(O)Ph)]ClO4 (5), wherein a
water and Lewis acid-mediated isomerization reaction of the
acireductone ligand resulted in the formation of the same ester
(2-oxo-2-phenylethylbenzoate, Scheme 4).21 By contrast, the
Figure 3. UV−vis spectra of 3-ClO4 before (solid line) and after
(dashed line) the addition of O2.
oxidation state from Fe(II) to Fe(III). Our attempts to isolate
and characterize the iron-containing products of these oxygen-
ation reactions have thus far been unsuccessful, presumably due
to the poor affinity of Fe(III), a hard Lewis acid, for the aryl-
appended TPA ligand. However, acidification of the crude
reaction mixture followed by extraction with Et2O has allowed
us to analyze the organic compounds produced in the decom-
position of the acireductone. In the reaction of 3-OTf with O2,
the major products observed were benzoic acid and benzil,
along with small amounts of the ester. These products were the
same as were observed in the reaction of the Ni analogue 1 with
20
O2 and thus are generally consistent with the chelate hypo-
thesis, which predicts that 3-OTf and 1, having the same six-
membered chelate ring for acireductone binding, should
produce the same products in a dioxygenase reaction.
It was therefore very surprising to discover that benzoyl-
formic acid is produced in the reaction of 3-ClO4 with O2, in
addition to the other products produced in the reaction of
3-OTf (Scheme 5; Figure S2). Benzoylformic acid is an α-keto
Scheme 5. Organic Products Detected in the Reaction of 3-X
with O2 in CH3CN
Scheme 4. Proposed Mechanism for the Isomerization of
2-Hydroxy-1,3-diphenyl-propan-1,3-dione To Form 2-Oxo-
2-phenylethylbenzoate
a
a
Benzoylformic acid (red) was only detected when X = ClO4.
nickel-containing complex 1 does not efficiently promote this
isomerization chemistry under wet conditions. We also note that
in the O2 reactivity studies of 3-ClO4 described below, the
reaction to generate ester is always operative.
Aerobic Reactivity. Exposure of acetonitrile solutions of
3-X to O2 at ambient temperature results in the rapid bleaching
of the 385 nm absorption feature (Figure 3), indicating de-
composition of the acireductone anion. Analysis of the
headspace gas of the reaction vessel shows that CO has been
produced qualitatively. After prolonged exposure to O2, total
loss of the well-defined paramagnetically shifted features in the
1H NMR spectrum is observed, consistent with a change in
acid and is the expected product in the oxidation of the bulky
acireductone if it were undergoing Fe-ARD′-type reactivity.
Given that all our spectroscopy detailed above strongly suggests
that 3-OTf and 3-ClO4 have the same solution structure,
precluding a five-membered chelate, this change in reactivity
was very interesting. Our initial hypothesis was that the ester
initially formed by an isomerization reaction could undergo a
separate oxygenation reaction to generate benzoyl-2-oxo-2-
phenylethanoate. This anhydride would subsequently undergo
hydrolysis to generate the observed benzoylformic acid (and an
equivalent of benzoic acid, Scheme S1). However, a control
F
dx.doi.org/10.1021/ja3038189 | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
reaction in which the ester was exposed to O2 in the presence
of 2-ClO4 did not lead to the generation of benzoylformic acid.
Rather only trace amounts of the hydrolytic products benzoic
acid and 2-hydroxyacetophenone were detected in addition to
unreacted ester.
Table 3. Summary of Production of Benzoylformic Acid
Kinetics. In order to gain more insight into why 3-ClO4
exhibits different regioselectivity in C−C bond cleavage than 1
and 3-OTf, we undertook mechanistic studies. For both 3-ClO4
and 3-OTf, the reaction is found to be first order in each of 3-X
and O2, with an overall rate constant, k2 = 0.40 M−1 s−1
(Figures S3 and S4). The similarity of the rate in each case
implies they have the same rate-determining step, and the
differentiation in terms of reactivity likely occurs after the
acireductone itself has been consumed. It is also worth noting
that these reactions are slightly slower than was observed in the
analogous reaction of 1 (k = 1.7 M−1 s−1).13
Investigations of Fe/O2 Reactivity. A common paradigm
in dioxygenase reactions involving a ferrous center is the direct
metal-centered activation of O2, forming a ferric-superoxo
species. Speculating that such a species could be partially
responsible for the differentiation in reactivity between 1 and
3-ClO4, we investigated its feasibility. The formation of such a
species would require the dissociation of one of the phenyl-
appended pyridyl arms of the chelate ligand to open up a
a
Relative percentage of the total amount of benzoic acid and
benzoylformic acid determined by LCMS peak area. Benzil and ester
b
−
are additional products not presented in this table. HO2 generated
in situ by combining 1.1 equivalents of 30% H2O2 (aq) and NEt3.
1
coordination position. Low-temperature H NMR studies of 4
in CD3CN (−40 °C) and CD3OD (−70 °C) have shown no
evidence of loss of the Cs symmetry of the complex, suggesting
that dissociation of a chelate ligand arm is unlikely to be
occurring. Additionally, in the reaction of 3-X with O2 in
acetonitrile at −40 °C, we have observed no intermediates
consistent with the formation of a ferric-superoxo species by
UV−vis. Attempts to intercept a superoxo intermediate by
hydrogen atom abstraction from common probes, such as
dihydroanthracene, 2,4-ditbutylphenol and 2,4,6-tritbutylphenol
have all yielded a negative result.31 Based on these results we
conclude that it is unlikely that the reaction proceeds via a
ferric-superoxo intermediate.
Isotopic Labeling. Labeling studies of the reaction of
3-ClO4 with 18O2 show a modest incorporation of a single label
into both benzoic acid (36%) and benzoylformic acid (53%).
We hypothesized that water in the reaction mixture (due to the
use of the pentahydrate base Me4NOH·5H2O to generate
3-ClO4) was at least partially responsible for the loss of label, as
addition of H218O to the 16O2 reaction results in modest label
incorporation (Scheme S3). To probe whether water could
have a role in the reaction, we repeated the reaction of 3-OTf
with O2 in the presence of 5% added water. Gratifyingly, we ob-
served the production of benzoylformic acid (Table 3), therefore
indicating that the presence of water had been the reason for the
differing observed reactivities of 3-OTf and 3-ClO4.
Scope of α-Keto Acid Formation. Having established the
importance of water in the generation of the α-keto acid, we
conducted a series of control reactions specifically searching for
benzoylformic acid formation. Most importantly, we have
reinvestigated the products generated in the reaction of the
nickel complex 1 with O2 in both pure CH3CN and 95%
CH3CN/H2O using the workup procedure and analysis
methods (GC-MS and LC-MS) employed for the O2 reaction
of 3-X. Notably, no benzoylformic acid formation was detected
starting from 1 (Scheme S2 and Figure S2 (LC-MS)). Addi-
tionally, reaction of the triketone (1,3-diphenylpropantrione) in
the presence of the hydroperoxide anion and [(6-Ph2TPA)Ni-
(CH3CN)(H2O)](ClO4)2 resulted in no detectable benzoylformic
acid formation (Scheme S2). Treatment of the tetramethylammo-
nium salt of the bulky acireductone anion with O2 also failed to
yield any α-keto acid formation. Taken together, these results
imply that both water and an iron center are required for α-keto
acid formation.
Raising the amount of water present in the reaction of 3-OTf
with O2 results in a marked increase in the amount of benzoyl-
formic acid produced, as summarized in Table 3. However, we
have been unable to generate equimolar amounts of benzoic
acid and benzoylformic acid regardless of the amount of water
present in the system. Interestingly, reaction of the triketone
(the two-electron oxidized form of the acireductone) with
−
HO2 in the presence of a ferrous complex also results in the
generation of benzoylformic acid. When the same reaction is
repeated, starting from the hydrated triketone (2,2-dihydroxy-
1,3-diphenylpropan-1,3-dione) an increase in the amount of
benzoylformic acid produced is observed (Table 3). These
results suggest that the water sensitivity of the oxidation
chemistry is due to the hydration of a triketone intermediate.
We have also investigated the role of ferric species in the
reaction chemistry. Attempts to isolate ferric complexes by the
combination of Fe(ClO4)3·6H2O, 6-Ph2TPA, Me4NOH·5H2O,
and the bulky acireductone have been unsuccessful. Monitoring
1
these reactions by H NMR in the paramagnetic region shows
the growth of peaks consistent with the reduction of FeIII to
form the ferrous species 2-ClO4 (Figure S5). Subsequent
analysis of the organic products of this reaction shows the
production of the ester as a major species as well as triketone
and trace amounts of benzoic acid (Scheme 6). Given that
ferric/ferrous couple is a one-electron process, we next com-
bined 4 equiv of a ferric salt with the acireductone anion or
2 equiv with the triketone. As shown in Table 3, these reactions
also lead to carbon−carbon bond cleavage, with the
regioselectivity influenced by the water content of the reaction
mixture. We note that the ferric-mediated oxidative cleavage of
vicinal triketones in water has previously been reported,32 and
the oxidation of acireductone analogues, such as ascorbic acid
by ferric ions, is well documented.33,34 This oxidation of an
G
dx.doi.org/10.1021/ja3038189 | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
oxidative pathway is accessible, leading to a change in the
regiospecificity of carbon−carbon bond cleavage that does not
require distinct acireductone coordination motifs, in contrast to
the chelate hypothesis.
Scheme 6. Attempted Synthesis of [(6-Ph2TPA)Fe-
(PhC(O)C(OH)C(O)Ph)](ClO4)2, and the Resulting
Products Including Oxidized Organic Species
In the reaction of 3-X with O2 in the presence of water we
never observe the formation of benzoic acid and benzoylformic
acid in an equimolar ratio (Table 3). Thus, there are always at
least two oxidative pathways operative. Regardless of
conditions, it appears that an O2 reaction akin to that found
for 1 is always operative in the Fe(II)-containing system. This
reaction leads to the formation of 2 equiv of benzoic acid and
carbon monoxide. When water is added, a new reaction
pathway is enabled, wherein 1 equiv each of benzoic acid and
benzoylformic acid are generated. Because complex 1 exhibits
only the first type of reactivity, comparative studies of the
reactivity of 1 and 3-X enable us to probe for the chemical
factors that enable α-keto acid formation. From these studies
we find that regardless of metal ion or water content of the
system, the initial step is the two-electron oxidation of the
acireductone by dioxygen to form a triketone and the hydro-
peroxide anion as intermediates. This proposal is supported by
our previous mechanistic studies of the O2 reaction of 1 as well
as kinetic studies that show a similar rate-determining step for
the oxidation of 1 and 3-X in the presence or absence of water.
The involvement of triketone and hydroperoxide intermediates
is also supported by studies involving authentic triketone
and hydrogen peroxide in the presence of the corresponding
[(6-Ph2TPA)M(CH3CN)x]2+ (M = Ni (x = 2) or Fe (x = 1))
complex. In these reactions we observed similar regioselectivity,
as in the reaction of 1 or 3-X, indicating that it is at the
triketone level that a differentiation in the chemistry occurs.
In the absence of water, the triketone intermediate formed in
the O2 reactions of 1 and 3-X will undergo reaction with the
hydroperoxide anion to selectively cleave the C(1)−C(2) and
C(2)−C(3) bonds and release CO. This chemistry is consistent
with that previously reported by Pochapsky, wherein 2,3,4-
pentanetrione was shown to undergo reaction with H2O2 to
give 2 equiv of acetic acid and carbon monoxide.36 For 3-X in
the presence of water, we propose that an additional reaction
pathway is operative due to hydration of the triketone inter-
mediate. Triketone hydration is a well-known process, and the
central carbonyl of 1,3-diphenyltriketone is the most electrophilic
and therefore will be the site of initial hydration.37 Once formed,
we propose that the hydrated triketone may interact with the
iron center to form a five-membered chelate ring (Scheme 7),
acireductone in the absence of molecular oxygen is particularly
interesting due to the recent discovery of the operation of a
methionine salvage pathway under anaerobic conditions.35
DISCUSSION
■
The chelate ring hypothesis was formulated on the basis of a
number of observations in studies of Ni-ARD and Fe-ARD′.9
First, oxygen uptake studies, as well as spectroscopic studies
performed in the presence of oxygen, provided no evidence for
oxygen binding at the active site in either metal-containing form
of the enzyme. Additionally, Fe-ARD′-type reactivity was found
to occur when the enzyme is reconstituted with magnesium,
implying that iron-centered redox activity may not be
important in directing the regiospecificity of the reaction.
Second, NMR studies provided evidence for a metal-dependent
entropy switch that converts the active site tertiary structure
between a closed (Ni-ARD) and open (Fe-ARD′) form. Docking
studies show that the acireductone would likely coordinate via a
five-membered chelate in Fe-ARD′, but in the more congested
Ni-ARD active site, it would bind via a less sterically demanding
six-membered chelate.30 These binding modes would activate the
C(1)/C(2) or C(1)/C(3) positions, respectively, for reaction
with oxygen (Scheme 1), leading in turn to the proposed
products for the Fe-ARD′ and Ni-ARD catalyzed reactions.
Notably, the structural data currently available for enzyme−
substrate (ES) adducts of Fe-ARD′ or Ni-ARD, respectively,
are limited to XAS data, which does not provide definitive proof
of the substrate coordination motif.9 Additionally, the UV−vis
absorption features of the ES adducts are similar, which may
actually indicate that there is no difference in binding mode.
Our previous studies of a nickel-containing complex (1)
supported the chelate hypothesis in terms of the coordination
mode of the bulky acireductone ligand, which is akin to that
proposed for the ES complex of Ni-ARD. The O2 reactivity of
1, while resulting in the formation of Ni-ARD type products,
proceeded via a different pathway than that being proposed for
the enzyme. Specifically, the initial reaction in the model system
leads to the formation of triketone and hydroperoxide
intermediates from which carbon−carbon bond cleavage was
found to occur. In the enzyme, the coordinated acireductone is
proposed to react directly with O2 to give a coordinated cyclic
peroxide species from which aliphatic C−C bond cleavage
occurs. The results described herein show that a simple change
in the metal ion from Ni(II) in 1 to Fe(II) in 3-X, while
maintaining the same supporting ligand coordination environ-
ment, has no effect on the coordination mode of the bulky
acireductone. However, despite the congruence of the structure
of 1 and 3-X, α-keto acid formation was found to occur upon
reaction of 3-X with O2 in the presence of water. Thus, an
Scheme 7. Proposed Reaction Pathway Involving Triketone
Hydration As a Means To Generate Fe-ARD′-Type Products
Containing Oxygen Atoms Derived From O2
H
dx.doi.org/10.1021/ja3038189 | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
which becomes susceptible to attack by hydroperoxide at an
activated terminal carbonyl moiety. Formation of a four-
membered dioxetane ring via loss of water would generate the
species from which C(1)−C(2) bond cleavage and α-ketoacid
formation could occur.
Scheme 8. Proposed Sequence for Hydrated Triketone
Aliphatic Carbon−Carbon Bond Cleavage Promoted by
Ferric Iron Species
The reaction pathway outlined in Scheme 7 shows that it is
the hydration and subsequent coordination of an intermediate,
not differences in chelation involving the acireductone substrate
itself (chelate hypothesis), that determine the regioselectivity of
the reaction. This is in contrast to the reaction pathways of
other metal-containing dioxygenases and their model systems,
wherein changes in the primary and/or secondary environ-
ment of the metal center determine the regioselectivity of
carbon−carbon bond cleavage.38 In terms of ARD enzymes, it
is important to note that the active site in Fe-ARD′ is much
more open and thus solvent-exposed when compared to the
Ni-ARD active site, wherein a tryptophan loop maintains a
hydrophobic microenvironment.30 The reaction pathway
shown in Scheme 7 provides a means for 18O incorporation
from 18O2 in the α-ketoacid product. The substoichiometric
18O incorporation that is found in the α-keto acid generated in
the O2 reaction of 3-X (∼50%) and in the dioxygenase reac-
tivity of Fe-ARD′ (∼78%)39 suggests that water exchange may
be important for both the model system and the enzyme. The
involvement of a triketone-type intermediate in the enzymatic
system cannot currently be ruled out on the basis of either
experimental or computational studies. Therefore, our future
work will involve approaches toward examining the feasibility of
a triketone-type pathway in reactions involving C(1)−
H-containing ARD substrates. We note that recent advances
in the synthesis of C(1)−H triketones will be instrumental in
carrying out this work.40
Of course the question then arises as to why the C(1)−C(2)
cleavage pathway leading to α-keto acid formation is not
operative for the nickel-containing complex 1. Our working
hypothesis is that the Ni(II) center does not effectively mediate
the hydration of the triketone intermediate, preventing this
species from forming in significant amounts during the reaction
progression. This hypothesis is supported by the differing
anaerobic, water-dependent bulky acireductone isomerization
chemistry of the nickel and iron complexes 1 and 3-X. While 1
does not undergo isomerization of the acireductone, 3-X under-
goes rapid isomerization in the presence of water (Scheme 4).
Similar water-dependent isomerization of the acireductone to
the ester, as well as subsequent hydrolysis, was also observed in
solutions containing the cobalt complex 5. Interestingly, the
solid-state structures of the solvate complexes [(6-Ph2TPA)M-
(CH3CN)x]2+ are pentacoordinate, with a single solvent
molecule when M = Fe or Co but hexacoordinate with two
solvent molecules when M = Ni.19 The differing coordination
preferences for the metal ions in these systems may be
responsible for the differences in Lewis acid activation for the
triketone and/or water that would influence the formation of
hydrated triketone and acireductone isomerization.
wherein 18O from 18O2 is incorporated into the α-keto acid
product.41 In this regard, our studies do not exclude a pathway,
wherein the HO2− produced in the reaction of 3-X with O2 first
oxidizes Fe(II) to Fe(III), which then acts as Lewis acid to
mediate triketone hydration. The hydrated triketone could then
−
react with an additional equivalent of HO2 to give Fe-ARD′
type products, wherein the 18O label from 18O2 would be
incorporated.
CONCLUSION
■
The chelate hypothesis had remained unchallenged in the
literature to date as an explanation for the differing
regiospecificity of Ni-ARD and Fe-ARD′ without the need to
invoke metal-centered redox chemistry, which has not been
observed in the native enzymes. In this first study of an iron-
containing model system, designed to directly probe this
hypothesis, we have found that the chelate hypothesis is not
necessary to explain the differentiation in reactivity. Rather, in
our model system, a difference between nickel and iron in the
hydration of a triketone intermediate allows a change in regio-
selectivity of the reaction. This is an alternative proposition for
the Fe-ARD′ reaction, as it would be accounted for by the
differences in solvent accessibility in the active sites of Ni-ARD
and Fe-ARD′ without a need to dismiss the similarity in
absorption spectra for the enzyme−substrate adducts. The
notion of hydration of an intermediate as the key fac-
tor differentiating regioselectivity also provides a potential
framework for understanding how oxidation of an acireduc-
tone may occur in anaerobic systems using oxidants other
than dioxygen.
ASSOCIATED CONTENT
* Supporting Information
■
S
Crystallographic files in CIF format; table of chemical shifts of
1H NMR resonances for Fe(II) compounds; LC-MS data for
organic products of reactions of 1 and 3 with O2; schemes
describing control reactions and 18O-labeling studies. This
material is available free of charge via the Internet at http://
pubs.acs.org.
We note that an alternative route for α-keto acid formation
could involve the direct oxidation a hydrated triketone by ferric
ions generated in solution (Scheme 8). In this regard, it would
be expected that ferric ions could rapidly promote hydration of
the triketone, and previous studies have demonstrated ferric-
mediated oxidative cleavage of vicinal triketones in aqueous
solutions to give α-keto acid and carboxylic acid products.32
While this type of reactivity is certainly viable in our systems, it
would not explain the observed isotope incorporation data,
AUTHOR INFORMATION
Corresponding Author
lisa.berreau@usu.edu
■
Notes
The authors declare no competing financial interest.
I
dx.doi.org/10.1021/ja3038189 | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
(32) Mecinovic,
Ed. Engl. 2009, 48, 2796−2800.
́
J.; Hamed, R. B.; Schofield, C. J. Angew. Chem., Int.
ACKNOWLEDGMENTS
■
The authors thank Dr. Dale R. Gardner (USDA-ARS,
Poisonous Plant Research Lab, Logan, UT) for assistance
with LC-MS experiments and the National Science Foundation
(Grant CHE-0848858 to LMB) for support of this research.
(33) Taqui Khan, M. M.; Martell, A. E. J. Am. Chem. Soc. 1967, 89,
4176−4185.
(34) Kim, Y. J.; Feng, X.; Lippard, S. J. Inorg. Chem. 2007, 46, 6099−
6107.
(35) (a) Singh, J.; Tabita, F. R. J. Bacteriol. 2010, 192, 1324−1331.
́
(b) Sekowska, A.; Denervaud, V.; Ashida, H.; Michoud, K.; Haas, D.;
REFERENCES
■
Yokota, A.; Danchin, A. BMC Microbiology 2004, 4, 9. (c) Imker, H. J.;
Fedorov, A. A.; Fedorov, E. V.; Almo, S. C.; Gerlt, J. A. Biochemistry
2007, 46, 4077−4089.
(1) Lange, S. J.; Que, L., Jr. Curr. Opin. Chem. Biol. 1998, 2, 159−172.
(2) Bugg, T. D. H.; Ramaswamy, S. Curr. Opin. Chem. Biol. 2008, 12,
134−140.
(3) Broderick, J. B. Essays Biochem. 1999, 34, 173−189.
(4) Straganz, G. D.; Glieder, A.; Brecker, L.; Ribbons, D. W.; Steiner,
W. Biochem. J. 2003, 369, 573−581.
(5) Hopper, D. J.; Kaderbhai, M. A. Biochem. J. 1999, 344, 397−402.
(6) Cicchillo, R. M.; Zhang, H.; Blodgett, J. A.; Whitteck, J. T.; Li, G.;
Nair, S. K.; van der Donk, W. A.; Metcalf, W. W. Nature 2009, 459,
871−874.
(7) Wray, J. W.; Abeles, R. H. J. Biol. Chem. 1993, 268, 21466−
21469.
(36) Dai, Y.; Pochapsky, T. C.; Abeles, R. H. Biochemistry 2001, 40,
6379−6387.
(37) Rubin, M. B.; Gleiter, R. Chem. Rev. 2000, 100, 1121−1164.
(38) Costas, M.; Mehn, M. P.; Jensen, M. P.; Que, L., Jr. Chem. Rev.
2004, 104, 939−986.
(39) Wray, J. W.; Abeles, R. H. J. Biol. Chem. 1995, 270, 3147−3153.
(40) Goswami, S.; Maity, A. C.; Fun, H.-K.; Chantrapromma, S. Eur.
J. Org. Chem. 2009, 1417−1426.
(41) No isotope incorporation from dioxygen into the ferric-oxidized
products was observed in the previous study.32
(8) Albers, E. IUBMB Life 2009, 61, 1132−1142.
(9) Pochapsky, T. C.; Ju, T.; Dang, M.; Beaulieu, R.; Pagani, G. M.;
OuYang, B. In Metal Ions in Life Sciences; Sigel, A., Sigel, H., Sigel, R. K.
O., Eds.; Wiley-VCH: Weinheim, Germany, 2007; Vol. 2, pp 473−500.
(10) Dai, Y.; Wensink, P. C.; Abeles, R. H. J. Biol. Chem. 1999, 274,
1193−1195.
(11) Motterlini, R.; Otterbein, L. E. Nat. Rev. Drug Discov. 2010, 9,
728−743.
(12) Szajna, E.; Arif, A. M.; Berreau, L. M. J. Am. Chem. Soc. 2005,
127, 17186−17187.
(13) Berreau, L. M.; Borowski, T.; Grubel, K.; Allpress, C. J.;
Wikstrom, J. P.; Germain, M. E.; Rybak-Akimova, E. L.; Tierney, D. L.
Inorg. Chem. 2011, 50, 1047−1057.
(14) Roberts, J. D.; Smith, D. R.; Lee, C. C. J. Am. Chem. Soc. 1951,
73, 618−625.
(15) Armarego, W. L. F.; Perrin, D. D. Purification of Laboratory
Chemicals, 4th ed.; Butterworth−Heinemann, Boston, MA, 1996.
(16) Hagen, K. S. Inorg. Chem. 2000, 39, 5867−5869.
(17) So, J.-H.; Boudjouk, P. Inorg. Chem. 1990, 29, 1592−1593.
(18) Plietker, B. J. Org. Chem. 2003, 68, 7123−7125.
(19) Makowska-Grzyska, M. M.; Szajna, E.; Shipley, C.; Arif, A. M.;
Mitchell, M. H.; Halfen, J. A.; Berreau, L. M. Inorg. Chem. 2003, 43,
7472−7488.
(20) Szajna-Fuller, E.; Rudzka, K.; Arif, A. M.; Berreau, L. M. Inorg.
Chem. 2007, 46, 5499−5507.
(21) Grubel, K.; Ingle, G. K.; Fuller, A. L.; Arif, A. M.; Berreau, L. M.
Dalton Trans. 2011, 40, 10609−10620.
(22) Szajna, E.; Dobrowolski, P.; Fuller, A. L.; Arif, A. M.; Berreau, L.
M. Inorg. Chem. 2004, 43, 3988−3997.
(23) Evans, D. F. J. Chem. Soc. 1959, 2003−2005.
(24) Achord, J. M.; Hussey, C. L. Anal. Chem. 1980, 52, 601−602.
(25) Wolsey, W. C. J. Chem. Educ. 1973, 50, A335−A337.
(26) Addison, A. W.; Rao, T. N.; Reedijk, J.; van Rijn, J.; Vershcoor,
G. C. J. Chem. Soc., Dalton Trans. 1984, 2177−2184.
(27) X-ray crystallographically-characterized Ni(II) bulky acireduc-
tone complex 1 and its dibenzoylmethane analog, [(6-Ph2TPA)
1
Ni(PhC(O)CHC(O)Ph)]ClO4, exhibited a similar H NMR spectral
congruence.20
(28) (a) Park, H.; Baus, J. S.; Lindeman, S. V.; Fiedler, A. T. Inorg.
Chem. 2011, 50, 11978−11989. (b) Park, H.; Bittner, M. M.; Baus, J.
S.; Lindeman, S. V.; Fiedler, A. J. Inorg. Chem. 2012, 51, 10279−10289.
(29) Diebold, A. R.; Neidig, M. L.; Moran, G. R.; Straganz, G. D.;
Solomon, E. I. Biochemistry 2010, 49, 6945−6952.
(30) Ju, T.; Goldsmith, R. B.; Chai, S. C.; Maroney, M. J.; Pochapsky,
S. S.; Pochapsky, T. C. J. Mol. Biol. 2006, 363, 823−834.
(31) Mukherjee, A.; Cranswick, M. A.; Chakrabarti, M.; Paine, T. K.;
Fujisawa, K.; Munck, E.; Que, L., Jr. Inorg. Chem. 2010, 49, 3618−
̈
3628.
J
dx.doi.org/10.1021/ja3038189 | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX