Reversible biological birch reduction at an extremely low redox potential

Article abstract of DOI:10.1021/ja103448u

The Birch reduction of aromatic rings to cyclohexadiene compounds is widely used in chemical synthesis and requires solvated electrons, the most potent reductants known in organic chemistry. Benzoyl-coenzyme A (CoA) reductases (BCR) are key enzymes in the anaerobic bacterial degradation of aromatic compounds and catalyze an analogous reaction under physiological conditions. Class I BCRs are FeS enzymes and couple the reductive dearomatization of benzoyl-CoA to cyclohexa-1,5-diene-1-carboxyl-CoA (dienoyl-CoA) to a stoichiometric ATP hydrolysis. Here, we report on a tungsten-containing class II BCR from Geobacter metallireducens that catalyzed the fully reversible, ATP-independent dearomatization of benzoyl-CoA to dienoyl-CoA. BCR additionally catalyzed the disproportionation of dienoyl-CoA to benzoyl-CoA/monoenoyl-CoA and the four- and six-electron reduction of benzoyl-CoA in the presence of a reduced low-potential bridged 2,2′-bipyridyl redox dye. Reversible redox titration experiments in the presence of this redox dye revealed a midpoint potential of E⁰′= -622 mV for the benzoyl-CoA/dienoyl-CoA couple, which is far below the values of other known reversible substrate/product redox couples in enzymology. This work demonstrates the efficiency of reversible metalloenzyme catalysis, which in chemical synthesis can only be achieved under essentially irreversible conditions.

Full text of DOI:10.1021/ja103448u

Published on Web 06/28/2010
Reversible Biological Birch Reduction at an Extremely Low
Redox Potential
Johannes W. Kung,† Sven Baumann,‡ Martin von Bergen,§ Michael Mu¨ller,|
Peter-Leon Hagedoorn,⊥ Wilfred R. Hagen,⊥ and Matthias Boll*,†
Institute of Biochemistry, UniVersity of Leipzig, 04103 Leipzig, Germany, Helmholtz Centre
for EnVironmental Research - UFZ, Department of Metabolomics, 04318 Leipzig, Germany,
Helmholtz Centre for EnVironmental Research - UFZ, Department of Proteomics,
04318 Leipzig, Germany, Institute of Pharmaceutical and Medicinal Chemistry,
UniVersity of Freiburg, 79104 Freiburg, Germany, and Department of Biotechnology,
Delft UniVersity of Technology, The Netherlands
Received April 23, 2010; E-mail: boll@uni-leipzig.de
Abstract: The Birch reduction of aromatic rings to cyclohexadiene compounds is widely used in chemical
synthesis and requires solvated electrons, the most potent reductants known in organic chemistry. Benzoyl-
coenzyme A (CoA) reductases (BCR) are key enzymes in the anaerobic bacterial degradation of aromatic
compounds and catalyze an analogous reaction under physiological conditions. Class I BCRs are FeS
enzymes and couple the reductive dearomatization of benzoyl-CoA to cyclohexa-1,5-diene-1-carboxyl-
CoA (dienoyl-CoA) to a stoichiometric ATP hydrolysis. Here, we report on a tungsten-containing class II
BCR from Geobacter metallireducens that catalyzed the fully reversible, ATP-independent dearomatization
of benzoyl-CoA to dienoyl-CoA. BCR additionally catalyzed the disproportionation of dienoyl-CoA to benzoyl-
CoA/monoenoyl-CoA and the four- and six-electron reduction of benzoyl-CoA in the presence of a reduced
low-potential bridged 2,2′-bipyridyl redox dye. Reversible redox titration experiments in the presence of
this redox dye revealed a midpoint potential of E0′) -622 mV for the benzoyl-CoA/dienoyl-CoA couple,
which is far below the values of other known reversible substrate/product redox couples in enzymology.
This work demonstrates the efficiency of reversible metalloenzyme catalysis, which in chemical synthesis
can only be achieved under essentially irreversible conditions.
Introduction
The Birch reduction is a powerful and common synthetic tool
in organic chemistry which is used industrially for dihydro
additions to aromatic rings.1,2 It proceeds via alternate single
electron/proton transfer steps to the aromatic ring yielding
cyclohexadiene compounds, which can be further reduced
(Figure 1a). The rate-determining step is the first electron
transfer yielding a nonaromatic radical anion. The redox
potential of the one-electron reduction of nonactivated aromatic
compounds is below -3 V.3
Considering the totally nonphysiological conditions of the
Birch reduction in organic synthesis, it is surprising that enzymes
exist in nature, which catalyze a similar reaction. For anaerobic
bacteria that degrade aromatic growth substrates, the use of the
dioxygen molecule for an oxidative attack on the aromatic ring
is not an option. Instead, dearomatizing benzoyl-coenzyme A
(benzoyl-CoA 1) reductases (BCRs), key enzymes of the
† University of Leipzig.
Figure 1. Reductive benzene ring dearomatization by the Birch reduction
in organic synthesis and by benzoyl-CoA reductases (BCRs). (a) Reaction
sequence of the Birch reduction of benzoic acid. Solvated electrons serve
as electron donors, and the formation of the kinetically favored cyclohexa-
2,5-diene isomer is shown. (b) Reaction catalyzed by class I BCRs (ATP-
dependent). A reduced ferredoxin serves as an electron donor. (c) Reaction
catalyzed by class II BCRs (ATP independent). The forward reaction of
class II BCRs has not been determined, yet. Aox ) variable artificial one-
electron acceptors.
‡ Helmholtz Centre for Environmental Research - UFZ, Department of
Metabolomics.
§ Helmholtz Centre for Environmental Research - UFZ, Department of
Proteomics.
| University of Freiburg.
⊥ Delft University of Technology.
(1) Birch, A. J. Q. ReV. 1950, 4, 69–93.
(2) Birch, A. J.; Rao, S. AdV. Org. Chem. 1972, 1–65.
(3) Mortensen, J.; Heinze, J. Angew. Chem., Int. Ed. 1984, 23, 84–85.
9
9850 J. AM. CHEM. SOC. 2010, 132, 9850–9856
10.1021/ja103448u  2010 American Chemical Society

Reversible Biological Birch Reduction
A R T I C L E S
anaerobic aromatic metabolism, reduce the aromatic ring of the
substrate with two electrons and specifically form cyclohexa-
1,5-diene-1-carboxyl-CoA (dienoyl-CoA 2, Figure 1b).4,5 In
most cases, no further reduction to the corresponding cyclic
monoenoyl-CoA or cyclohexanecarboxyl-CoA is observed
indicating a tight coordination of substrate binding, transfer of
two electrons, and product release. The one-electron reduction
potential of the benzoyl-CoA analogue S-ethyl thiobenzoic acid
in an aprotic solvent to the corresponding radical anion is -1.9
V.6 This value is more positive than that of the nonactivated
benzoic acid due to the extended stabilization possibilities of
the radical anion intermediate at the carbonyl functionality of
the thioester.7,8 The redox potential of the reaction benzoyl-
CoA + 2e- + 2H+ f dienoyl-CoA in an aqueous solution is
unknown.
[4Fe-4S] cluster; BamC is an electron transferring subunit
containing three further FeS clusters.17 Class II BCRs are
proposed not to require ATP hydrolysis, and it was hypothesized
that the BamDEFGHI components are involved in an unchar-
acterized electron activation process. Using Ti(III)-citrate or
dithionite as electron donors, BamBC did not catalyze the
reductive dearomatization of benzoyl-CoA.17
In this work, we demonstrate that BamBC was capable of
catalyzing both the aromatization and dearomatization reactions
in the absence/presence of external electron acceptors/donors,
respectively, without coupling to an activation reaction. The
redox potential of the benzoyl-CoA/dienoyl-CoA conversion was
determined to be the most negative described for a fully
reversible substrate/product couple in enzymatic catalysis. The
results obtained provide initial insights into the energetics of
the key reaction in anaerobic aromatic catabolism.
There are two different classes of BCRs, which both form
the identical dienoyl-CoA product (Figure 1b): class I BCRs
couple electron transfer from a reduced ferredoxin to the
aromatic ring to a stoichiometric ATP hydrolysis (2ATP/2e-).5,9,10
For this reason, the reactions catalyzed by class I BCRs are
considered essentially irreversible. Class I BCRs have an Rꢀγδ-
subunit composition and contain three [4Fe-4S] clusters. The
Rδ-subunits bind one ATP each and are responsible for ATP-
driven electron transfer to the active site containing ꢀγ-subunits.
ATP hydrolysis-dependent electron transfer to a chemically inert
substrate is a common feature of benzoyl-CoA reductases and
nitrogenases.11 ATP-dependent low potential electron transfers
have also been described for the activating enzymes of 2-hy-
droxyacyl-CoA dehydratases12,13 or B12-dependent dehaloge-
nases.14 BCR from the facultatively anaerobic Thauera aro-
matica is the best studied enzyme of class I BCRs. For this
enzyme, initial evidence has been obtained that the stereospecific
trans-dihydro addition of hydrogen atoms to benzoyl-CoA
proceeds via radical intermediates according to the classical
Birch reduction.15,16
Materials and Methods
Previously Described Methodologies. Previous papers provide
details for the cultivation of G. metallireducens19 and the purifica-
tion of BamBC under strictly anaerobic conditions.17 The synthesis
and purification of benzoyl-CoA 1, dienoyl-CoA 2, the three
cyclomonoenoyl-CoA isomers, and [ring-13C]-benzoyl-CoA/di-
enoyl-CoA was as described earlier.20,21 Synthesis and H NMR/
1
13C NMR purity control of 2,12-dimethyl-7,8-dihydro-6H-dipyri-
do[1,2-a:2′,1′-c][1,4]diazepinediium dibromide 4 was as described.22
BamBC Reduction and Disproportionation Assay. BamBC
(1-5 µM) was reduced in 500 µL of reaction buffer containing
150 mM Mops/KOH, 15 mM MgCl2, 150 mM NaCl, 5 mg/mL
BSA, pH 6.8, with 0.15 mM dienoyl-CoA in the absence of an
external electron acceptor. The decrease in absorption at 409 nm
was followed by UV/vis spectroscopy. Samples (25 µL) were taken
at different time points and analyzed by HPLC coupled to diode
array detection as described.16 The amounts of benzoyl-CoA and
monoenoyl-CoA were determined by their extinction coefficients
(ε[benzoyl-CoA]260 ) 21 300 M-1 cm-1 17, ε[dienoyl-CoA]260
20 900 M-1 cm-1 17 ε[cyclohex-1-ene-1-carboxyl-CoA]260 ) 20 700
M-1 cm-1, this work).
)
,
The class II BCRs were recently discovered in the Fe(III)-
respiring, obligately anaerobic Geobacter metallireducens.17 In
this organism, BCR is proposed to constitute a complex of
eight benzoate inducible gene products (BamBCDEFGHI). The
BamBC components have been isolated and characterized using
an assay that followed the reverse reaction, the electron acceptor
dependent aromatization of the product dienoyl-CoA.17,18 The
BamB is proposed to contain an active site tungstopterin and a
Isotope Exchange Assay. In the isotope exchange experiments,
0.25 mM dienoyl-CoA and 0.5 mM [ring-13C]-benzoyl-CoA were
added to 0.45 µM BamC in 500 µL of reaction buffer. Samples
(25 µL) were taken at different time points and mixed with 100
µL of methanol. The supernatants were centrifuged twice at
16 000g. The samples were desalted by application to a preparative
C18 reverse-phase HPLC column (Knauer) using an HPLC separa-
tion module (Waters 2695) as described.16 Final elution was carried
out with a mixture of 30% acetonitrile and 70% water. The samples
were freeze-dried overnight.
Benzoyl-CoA Reduction Assay with Ti(III)-Reduced (4). The
assay contained 5 mM Ti(III)-citrate, 5 mM 4, and 0.4 mM benzoyl-
CoA in 400 µL of reaction buffer. The reaction was started by
addition of 1-3 µM BamBC (30 °C). Samples of 50 µL were taken
at different time points and stopped by addition of 100 µL of
methanol. After centrifugation, the samples were applied onto an
HPLC (Waters 2695) in a gradient from 12% to 20% acetonitrile
in 50 mM potassium phosphate, pH 6.8, within 13 min.
(4) Boll, M. J. Mol. Microbiol. Biotechnol. 2005, 10, 132–42.
(5) Boll, M. Biochim. Biophys. Acta 2005, 1707, 34–50.
(6) Boll, M.; Laempe, D.; Eisenreich, W.; Bacher, A.; Mittelberger, T.;
Heinze, J.; Fuchs, G. J. Biol. Chem. 2000, 275, 21889–95.
(7) Buckel, W.; Golding, B. T. Annu. ReV. Microbiol. 2006, 60, 27–49.
(8) Buckel, W.; Keese, R. Angew. Chem., Int. Ed. 1995, 34, 1502–1506.
(9) Boll, M.; Fuchs, G. Eur. J. Biochem. 1995, 234, 921–33.
(10) Unciuleac, M.; Boll, M. Proc. Natl. Acad. Sci. 2001, 98, 13619–24.
(11) Seefeldt, L. C.; Hoffman, B. M.; Dean, D. R. Annu. ReV. Biochem.
2009, 78, 701–22.
(12) Buckel, W. Angew. Chem., Int. Ed. Engl. 2009, 48, 6779–87.
(13) Buckel, W.; Hetzel, M.; Kim, J. Curr Opin Chem Biol 2004, 8,
462–7.
HPLC-MS/MS Analysis of CoA Esters. All experiments were
carried out on an Agilent 1100 series binary HPLC system (Agilent
Technologies) coupled with a 4000 QTRAP linear ion trap mass
(14) Ferguson, T.; Soares, J. A.; Lienard, T.; Gottschalk, G.; Krzycki, J. A.
J. Biol. Chem. 2009, 284, 2285–95.
(15) Mo¨bitz, H.; Boll, M. Biochemistry 2002, 41, 1752–8.
(16) Thiele, B.; Rieder, O.; Golding, B. T.; Mu¨ller, M.; Boll, M. J. Am.
Chem. Soc. 2008, 130, 14050–1.
(19) Lovley, D. R.; Giovannoni, S. J.; White, D. C.; Champine, J. E.;
Phillips, E. J.; Gorby, Y. A.; Goodwin, S. Arch. Microbiol. 1993, 159,
336–44.
(17) Kung, J. W.; Lo¨ffler, C.; Dorner, K.; Heintz, D.; Gallien, S.; Van
Dorsselaer, A.; Friedrich, T.; Boll, M. Proc. Natl. Acad. Sci. 2009,
106, 17687–92.
(20) Laempe, D.; Eisenreich, W.; Bacher, A.; Fuchs, G. Eur. J. Biochem.
1998, 255, 618–27.
(18) Wischgoll, S.; Heintz, D.; Peters, F.; Erxleben, A.; Sarnighausen, E.;
Reski, R.; Van Dorsselaer, A.; Boll, M. Mol. Microbiol. 2005, 58,
1238–52.
(21) Thiele, B.; Rieder, O.; Jehmlich, N.; von Bergen, M.; Mu¨ller, M.;
Boll, M. J. Biol. Chem. 2008, 283, 20713–21.
(22) Homer, R. F.; Tomlinson, T. E. J. Chem. Soc. 1960, 2498–2503.
9
J. AM. CHEM. SOC. VOL. 132, NO. 28, 2010 9851

A R T I C L E S
Kung et al.
spectrometer (AB Sciex) equipped with a TurboIon spray. The MS
was operated in negative ion and selected reaction monitoring
(SRM) mode. Separation of CoA esters was achieved on a
Chromolith SpeedROD RP-18e 50 mm × 4.6 mm from Merck at
40 °C with eluent A (95% 20 mM aqueous ammonium acetate,
5% acetonitrile) and eluent B (50% 20 mM aqueous ammonium
acetate, 50% acetonitrile) at a flow rate of 400 µL/min using
gradient elution. The characteristic mass transitions for 12C benzoyl-,
dienoyl-, monoenoyl-, and cyclohexanoyl-CoA were monitored at
870.2/408.1 m/z, 872.2/408.1 m/z, 874.2/408.1 m/z, and 876.2/408.1
m/z, respectively; those for the [ring-13C]-equivalents were six mass
units higher.
Determination of the Extinction Coefficients and Redox
Potential of (4). For the determination of the extinction coefficients
of 4, UV-vis spectra were recorded with a UV-1650PC Shimadzu
spectrophotometer in gastight quartz cuvettes under anaerobic
conditions. For reduction of 4, a solution in reaction buffer without
BSA was complemented with 0.15 mM dienoyl-CoA in the
presence of BamBC (0.04 µM). The extinction coefficients were
determined by measuring the change of absorbance upon addition
of distinct amounts of a 0.64 µM 2,6-dichlorophenolindophenol
stock solution.
The determination of the one-electron redox potential of 4 was
performed by cyclic voltammetry on a 25 µL droplet containing
0.5 mM 4 in 50 mM Mops, pH 7.5, using an argon-purged
electrochemical cell, with a nitric acid activated glassy carbon
working electrode (Le Carbon Lorraine), a Ag/AgCl reference
electrode (Radiometer), and a platinum counter electrode as pre-
viously described.23 The electrodes were connected to a PSTAT10
potentiostat (Ecochemie) controlled by GPES software (version 4.7).
The scan rate was 10 mV/s. All reported potentials have been
recalculated to the normal hydrogen electrode (NHE) by the addition
of +0.197 V.
Redox Titration of the Benzoyl-CoA/Dienoyl-CoA Couple.
The equilibrium concentrations of dienoyl-CoA and benzoyl-CoA
were determined in the presence of BamBC and 4 in reaction buffer.
Initial concentrations for dienoyl-CoA varied from 0.05 to 3.0 mM
and for oxidized 4 from 0.5 to 20 mM. At different time points,
25-50 µL samples were taken and mixed with 100 µL of methanol.
The concentrations of benzoyl-CoA and dienoyl-CoA were deter-
mined by HPLC as described. The respective amount of reduced
electron acceptor was determined spectrophotometrically using ε512
) 3700 M-1 cm-1 at concentrations below 4 mM and ε555 ) 2370
M-1 cm-1 at higher concentrations. The equilibrium redox potential
was determined according to the Nernst equation for two reversible
redox pairs using E0′ ) -673.5 mV for the one-electron reduction
potential of 4. The ratio benzoyl-CoA/dienoyl-CoA was plotted
against the redox potential, and the curve obtained was fitted to
the Nernst equation using the GraphPad Prism4 software package.
Inhibition of BamBC-Catalyzed Dienoyl-CoA Aromatiza-
tion by Benzoyl-CoA and Monoenoyl-CoA. The determination
of the Ki values were performed according to a Dixon plot analysis.
The assay was carried out with 0.1, 0.2, and 0.4 mM dienoyl-CoA.
In each case, benzoyl-CoA and monoenoyl-CoA were added in
concentrations varying from 0 to 2 mM. The reciprocals of the
initial reaction rates were plotted against the concentration of
the inhibitor (Dixon plot). The Ki values were determined using
the GraphPad Prism4 software package.
Figure 2. FeS cluster reduction and disproportionation reaction of BamBC.
Dotted line: reduction of the iron sulfur clusters as followed by absorbance
change at 409 nm (left y-axis). Straight lines: increase/decrease of CoA
esters as determined by HPLC analysis (right y-axis): O dienoyl-CoA 2;
[, benzoyl-CoA 1; ], monoenoyl-CoA 3; b, FeS cluster reduction. The
data were fitted to exponential curves using the GraphPad software package.
Disproportionation Reaction. A previous study revealed that
only dienoyl-CoA was capable of reducing the four FeS clusters
and the tungstopterin cofactor of BamBC. This finding was
explained by an extremely low redox potential of both the redox
centers of BamBC and the benzoyl-CoA/dienoyl-CoA couple.17
To investigate further the electron transfer from dienoyl-CoA
to oxidized BamBC, the stoichiometry and kinetics of dienoyl-
CoA oxidation/BamBC cofactor reduction were studied in the
absence of an external electron acceptor by HPLC (CoA esters)
and UV/vis spectroscopy (FeS cluster reduction). The redox
centers of purified BamBC were reduced by dienoyl-CoA in a
slow reaction with an initial rate of 11 nmol/mg/min FeS clusters
reduced as monitored by the time-dependent decrease of
absorption at 409 nm (Figure 2). Surprisingly, an approximately
10-fold excess of dienoyl-CoA was required to reduce the FeS
clusters of BamBC completely. HPLC analysis of the CoA esters
revealed that during BamBC reduction dienoyl-CoA consump-
tion was accompanied by the formation of benzoyl-CoA and
cyclohex-1-ene-1-carboxyl-CoA (monoenoyl-CoA 3, see also
Figure 6, Discussion section).
Obviously, the dienoyl-CoA-reduced tungsten cofactor of
BamBC transferred electrons to both the FeS clusters and, after
benzoyl-CoA/dienoyl-CoA exchange, to a second dienoyl-CoA,
thereby reducing it to monoenoyl-CoA. This disproportionation
reaction followed the same time course as the FeS cluster
reduction but at a 5-fold higher rate; the initial rate was 0.1%
of the rate of dienoyl-CoA oxidation with methyl viologen as
an electron acceptor (57.2 µmol/min/mg). In a typical experi-
ment, the stoichiometry of the disproportionation reaction was
0.58 mol of benzoyl-CoA and 0.42 mol of monoenoyl-CoA
formed per mole of dienoyl-CoA oxidized; it remained constant
in the course of the reaction. The imbalance between benzoyl-
CoA and monoenoyl-CoA formation can be rationalized by the
electrons used for FeS cluster reduction. This finding explains
why an excess of dienoyl-CoA was required to reduce BamBC
completely. The total recovery of electrons derived from the
disproportionation reaction/BamBC reduction was maximally
91%.
Results
With the recent discovery of BamBC as the first member of the
ATP-independent class II BCRs, a dearomatizing enzyme system
became available that opened a door to elucidate the thermody-
namic basis of the Biological Birch reduction. The reaction
catalyzed by BCRs was considered to operate at the negative redox
potential limit in enzyme catalysis.
(23) Hagen, W. R. Eur. J. Biochem. 1989, 182, 523–30.
9
9852 J. AM. CHEM. SOC. VOL. 132, NO. 28, 2010

Reversible Biological Birch Reduction
A R T I C L E S
dienoyl-CoA-reduced BamBC. After completion of the reaction,
the initial 2:1 ratio of 13C/12C added was equally distributed
between benzoyl-CoA and dienoyl-CoA. The initial rate of the
isotope exchange was 19.4 µmol/min/mg, which corresponds
to 34% of the dienoyl-CoA aromatization rate with methyl
viologen as the electron acceptor (57.2 µmol/min/mg). After
prolonged incubation, the formation of [ring-13C]-monoenoyl-
CoA was observed, which can be explained by the slow,
BamBC-catalyzed disproportionation of the [ring-13C]-dienoyl-
CoA formed.
Midpoint Potential of the Benzoyl-CoA/Dienoyl-CoA Re-
dox Couple. The results obtained indicate that in the absence
of an external electron acceptor the conversion of dienoyl-CoA
to benzoyl-CoA by BamBC was fully reversible. To determine
the two-electron reduction potential of benzoyl-CoA in aqueous
solution, redox titration experiments were carried out by the
determination of the equilibrium concentrations of benzoyl-CoA/
dienoyl-CoA and the oxidized/reduced forms of an appropriate
redox dye in the presence of BamBC. Methyl viologen (E0′ )
-456 mV, 0.5 mM)24 was completely reduced by 0.2 mM
dienoyl-CoA/BamBC without reaching equilibrium. When
methyl viologen was replaced by 1,1′,2,2′-tetramethyl viologen
(0.2 mM, E0′ ) -536 mV as determined by cyclic voltammetry,
kind gift of R.K. Thauer, Marburg), still 90-95% of the redox
dye was reduced by 0.2 mM dienoyl-CoA in the presence of a
surplus of benzoyl-CoA (1 mM) indicating that the equilibrium
was just reached. For this reason, a number of bridged 2,2′-
bipyridyl salts with redox potentials much more negative than
those of nonbridged 4,4′-bipyridyl salts such as methyl viologens
were synthesized as potential redox dyes for the redox titration
experiment. The 4,4′-dimethyl derivative of the propylene-
bridged 2,2′-biypridyl, 2,12-dimethyl-7,8-dihydro-6H-dipyrido-
[1,2-a:2′,1′-c][1,4]diazepinediium dibromide 4, was chosen as
a promising candidate for the redox titration experiments as it
was readily reduced by dienoyl-CoA/BamBC to the colored free
radical (Figure 4a,b). The one-electron reduction potential of 4
in the buffer used was E0′ ) -673.5 mV as determined by cyclic
voltammetry (Supporting Information Figure S2). One-electron
reduced 4 decayed to an unknown species in a slow, linear
reaction as evidenced by a time-dependent decrease of the UV/
vis spectrum. The rate of this linear background reaction reached
maximally 10% of the initial rate of the dienoyl-CoA-dependent
reduction of 4, which allowed a reliable correction for this decay.
The redox titration typically started with the fully oxidized
form of 4, and the time-, dienoyl-CoA-, and protein-dependent
reduction of 4 was followed until the equilibrium was reached.
At saturating dienoyl-CoA concentrations, the rate of the
reduction of 4 by dienoyl-CoA was nearly identical to the one
obtained with methyl viologen as acceptor (57.6 µmol/min/mg).
The equilibrium concentrations of dienoyl-CoA and benzoyl-
CoA were determined by HPLC analysis and those of the
oxidized/reduced form of 4 by UV/vis spectroscopy. By varying
the initial concentrations of dienoyl-CoA, benzoyl-CoA, and
4, the equilibrium redox potential was poised to values between
-660 and -540 mV. Reduction of 4 by dienoyl-CoA was barely
achieved at potentials below -650 mV. Addition of benzoyl-
CoA to an equilibrated sample shifted the equilibrium toward
the oxidized state of 4. This finding indicated the presence of a
true redox equilibrium which could be reached from the
oxidative and reductive direction. The data obtained were fitted
Figure 3. Time course of 12C/13C isotope exchange reaction of BamBC.
The reaction was started with a 2:1 ratio of [ring-13C]-benzoyl-CoA (0.5
mM) and 12C-dienoyl-CoA (0.25 mM) in the presence of BamBC. Samples
taken were analyzed by HPLC-MS/MS; for SRM chromatograms, see
Supporting Information Figure S1. The y-axis shows the difference in the
concentrations of the individual CoA esters. Reduction of BamBC by
dienoyl-CoA (O) yielded benzoyl-CoA (]) (straight lines). Reduced BamBC
converted [ring-13C]-benzoyl-CoA ([) to [ring-13C]-dienoyl-CoA (b)
(dotted lines), demonstrating the reversibility of the reaction. The data were
fitted to exponential curves using the GraphPad software package.
The monoenoyl-CoA formed during the disproportionation
reaction was not further reduced to cyclohexanecarboxyl-CoA
5 by reduced BamBC. It was not oxidized to dienoyl-CoA or
benzoyl-CoA using oxidized BamBC and high-potential artificial
electron acceptors such as 2,6-dichlorophenol indophenol. These
findings suggest that the redox potential of the dienoyl-CoA/
monoenoyl-CoA couple was too positive for electron transfer
to the active site of BamBC. Both monoenoyl-CoA and benzoyl-
CoA served as competitive inhibitors of dienoyl-CoA aroma-
tization with apparent Ki values of 0.2 ( 0.02 mM and 0.34 (
0.02 mM, respectively. These findings indicate that benzoyl-
CoA and monoenoyl-CoA have a considerable affinity for
oxidized BamBC.
Aromatization/Dearomatization Reaction without External
Acceptors/Donors. To test whether dienoyl-CoA reduced
BamBC was competent to catalyze benzoyl-CoA dearomatiza-
tion, the enzyme was incubated with a mixture of 12C-dienoyl-
CoA (0.25 mM) and [ring-13C]-benzoyl-CoA (0.5 mM) in the
absence of an external electron acceptor. The higher concentra-
tion of benzoyl-CoA was used to minimize the disproportion-
ation side reaction. HPLC analysis in conjunction with mass
spectrometric analysis of samples taken at different time points
revealed the BamBC- and time-dependent decrease of 12C-
dienoyl-CoA and [ring-13C]-benzoyl-CoA and the concomitant
increase of both 12C-benzoyl-CoA and [ring-13C]-dienoyl-CoA
(Figure 3, Supporting Information Figure S1).
Formation of 12C-benzoyl-CoA from 12C-dienoyl-CoA was
via the aromatization reaction described above, whereas the
formation of [ring-13C]-dienoyl-CoA can only be explained by
the reductive dearomatization of [ring-13C]-benzoyl-CoA by
(24) Stombaugh, N. A.; Sundquist, J. E.; Burris, R. H.; Orme-Johnson,
W. H. Biochemistry 1976, 15, 2633–41.
9
J. AM. CHEM. SOC. VOL. 132, NO. 28, 2010 9853

A R T I C L E S
Kung et al.
Figure 4. Redox titration of benzoyl-CoA/dienoyl-CoA redox couple in the presence of 4 as redox dye. (a) Structure of 4 used for the redox titration. (b)
UV/vis spectrum of BamBC reduced 4 to the free radical with maxima at 385 and 512 nm. (c) Redox titration of the benzoyl-CoA/dienoyl-CoA couple in
the presence of 4. Benzoyl-CoA and dienoyl-CoA concentrations were determined by HPLC, and the concentrations of the oxidized and reduced forms of
4 were determined by UV/vis spectroscopy; the percentage of benzoyl-CoA concentration at the equilibrium is shown (oxidized form of the redox couple).
E0′ of the one-electron reduction potential of 4 was 673.5 mV (Supporting Information Figure S2). The values obtained were fitted to a Nernst equation with
E0′ ) -622 mV and n ) 2 with r2 ) 0.95.
to a Nernst equation with E0′ (benzoyl-CoA/dienoyl-CoA) )
-622 mV, n ) 2, and r2 ) 0.95 (Figure 4c).
citrate at pH values above 725 had no stimulatory effect on the
benzoyl-CoA reduction rate.
Dearomatization Reaction with an External Electron Donor.
The full reversibility of the BamBC reaction during the redox
titration experiment suggested that dienoyl-CoA-reduced 4 was
competent to serve as an electron donor for benzoyl-CoA
reduction by BamBC. To test this assumption further, we
partially reduced 5 mM 4 with 5 mM Ti(III)-citrate (E0′ ∼ -720
mV at pH 7)25 and used it as the ultimate electron donor for
benzoyl-CoA reduction. Notably, Ti(III)-citrate in the absence
of 4 did not serve as an electron donor for BamBC-catalyzed
benzoyl-CoA dearomatization which is in agreement with earlier
observations.17 HPLC analysis of samples taken at different time
points revealed the time-, protein-, and electron donor dependent
decrease of benzoyl-CoA and the formation of reduced cyclic
products (Figure 5). The expected dienoyl-CoA product was
only transiently formed in traces at the beginning of the reaction.
In contrast, the four-electron reduced 1- and 2-isomers of
monoenoyl-CoA continuously increased and reached a maximal
concentration. While the latter remained constant, prolonged
incubation resulted in a decrease of the 1-monoenoyl-CoA
isomer formed with a concomitant increase of an apolar product,
which was identified by mass spectrometry as the six-electron-
reduced cyclohexanecarboxyl-CoA (Figures 5 and 6). When
benzoyl-CoA was replaced by dienoyl-CoA, a similar product
pattern was observed including the transient formation of the
1-monoenoyl-CoA isomer and its conversion to cyclohexan-
ecarboxyl-CoA. This finding confirms that dienoyl-CoA was
an intermediate during benzoyl-CoA reduction to monoenoyl-
CoA/cyclohexanecarboxyl-CoA.
Discussion
The reactions catalyzed by the class II BamBC from G.
metallireducens and the corresponding rates determined in this
work are summarized in Figure 6. The reversible benzoyl-CoA
dearomatization/dienoyl-CoA aromatization in the presence of
4ox/red as electron acceptor/donor (redox titration experiment)
(Figure 6a), the reversible aromatization/dearomatization reac-
tions in the absence of an electron acceptor (12/13C exchange
between dienoyl-CoA and benzoyl-CoA) (Figure 6b), the
dearomatization of benzoyl-CoA with Ti(III)-citrate reduced 4
as electron donor (Figure 6c), and the disproportionation of
dienoyl-CoA to benzoyl-CoA and monoenoyl-CoA in the
absence of an external electron acceptor (Figure 6d) are shown.
In this work, the forward reaction for a tungsten-containing class
II BCR has been demonstrated for the first time.
The full reversibility of benzoyl-CoA dearomatization in
aqueous solution (Figure 6a,b) is particularly remarkable as the
reversal of the analogous Birch reduction in organic synthesis
would couple cyclohexadiene aromatization to the reduction of
alkali cations to the corresponding metals. The redox potential
of the benzoyl-CoA/dienoyl-CoA couple appears to be at the
negative limit for a fully reversible substrate/product couple
redox in biochemistry. Due to the lack of electron donors with
potentials more negative than -500 mV, electron transfer
reactions in living systems below this potential are usually driven
by exergonic reactions such as photoactivation [e.g., photosys-
tem I26] or ATP hydrolysis [e.g., nitrogenase].11,13 They are
therefore all considered as essentially irreversible under physi-
ological conditions. However, very low potential reversible
electron transfer has been reported between redox intermediates
in the catalytic cycle of radical/SAM superfamily enzymes. In
the case of lysine 2,3-aminomutase, reversible electron transfer
for the reductive cleavage of S-adenosylmethionine occurs at a
The initial rate of two-electron transfer to the substrates
benzoyl-CoA and dienoyl-CoA was estimated to be 25% of the
aromatization/dearomatization with 4 as electron acceptor/donor
(14.7 µmol/min/mg). The rate of benzoyl-CoA reduction was
highest at pH 7 and could not be increased at higher pH values.
Thus, the expected lowering of the redox potential of Ti(III)-
(25) Hans, M.; Buckel, W.; Bill, E. J. Biol. Inorg. Chem. 2008, 13, 563–
(26) Webber, A. N.; Lubitz, W. Biochim. Biophys. Acta 2001, 1507, 61–
74.
79.
9
9854 J. AM. CHEM. SOC. VOL. 132, NO. 28, 2010

Reversible Biological Birch Reduction
A R T I C L E S
Figure 6. Reactions catalyzed by BamBC. (a) Aromatization/dearomati-
zation of dienoyl-CoA/benzoyl-CoA in the presence of an external electron
acceptor/donor (A). With A ) methyl viologen, the reaction was essentially
irreversible in the direction of benzoyl-CoA formation, whereas with A )
4, it was fully reversible. The specific activities were similar with both
viologens (∼57 µmol/min/mg, referred to as 100%). (b) Reversible
aromatization/dearomatization reaction in the absence of an external electron
as evidenced by the 12C/13C isotope exchange reaction. (c) Dearomatization
of benzoyl-CoA in the presence of Ti(III)-reduced 4 (A) to four- and six-
electron reduced products. (d) Disproportionation of dienoyl-CoA in the
absence of an external electron acceptor with oxidized BamBC. During
this reaction, a portion of electrons are transferred to oxidized BamBC;
therefore, the ratio of benzoyl-CoA/monoenoyl-CoA formed depended on
the enzyme concentration used. The numbers presented below the individual
reactions refer to relative specific activities.
benzoyl-CoA have to be considered as artificial reactions in the
presence of excess artificial reductant as the subsequent
enzymatic steps of the benzoyl-CoA degradation pathway rely
on the release of the dienoyl-CoA product.28,29
BamB belongs to the family of aldehyde:ferredoxin oxi-
doreductases (AOR), which usually oxidize aldehydes to the
corresponding carboxylic acids with a ferredoxin as the electron
acceptor.30 As the E0′ values of typical carboxylic acid/aldehyde
couples are between -530 and -560 mV, such reactions were
not considered to be reversible under physiological conditions.31
However, previous work provided evidence that this reaction
can be at least partially reversed in vitro using low potential
artificial electron donors and high concentrations of the car-
boxylic acid.32,33 With BamBC, a member of the AOR family
catalyzes a fully reversible redox reaction at the even more
negative redox potential of E0′ ) -622 mV. Tungsten cofactor
containing enzymes of the AOR family appear to be the
biocatalysts of choice for electron transfer reactions at the
negative redox potential limit in enzymatic catalysis.
Figure 5. Benzoyl-CoA dearomatization by BamBC with Ti(III)-citrate-
reduced 4 as electron donor. HPLC diagrams of CoA ester analysis are
shown after (a) 0.5 min, (b) 4 min, (c) 10 min, (d) 21 min, (e) 40 min, and
(f) 64 min of incubation. (I) ) benzoyl-CoA, (II) ) dienoyl-CoA, (III) )
1-monoenoyl-CoA, (IV) ) 2-monoenoyl-CoA, (V) ) cyclohexanecarboxyl-
CoA. The 1- and 2-isomers of monoenoyl-CoA and cyclohexanecarboxyl-
CoA were identified by coelution with standards and by mass spectrometric
analysis.
[4Fe-4S] cluster at -600 mV,27 a value comparable to that of
the benzoyl-CoA/dienoyl-CoA redox couple. Although benzoyl-
CoA dearomatization must be coupled to a currently unknown
activation reaction in vivo, the results from both the isotope
exchange activity of BamBC and the reversible redox titration
in the presence of 4 clearly demonstrate the full reversibility of
the reaction in vitro.
Depending on the type and concentration of the electron donor
used, BamBC catalyzed either a specific reversible two-electron
reduction yielding dienoyl-CoA or the irreversible four- or six-
electron reduction to different products. The latter process can
be rationalized by an uncoupling of electron transfer to the active
site of BamBC and dienoyl-CoA release. Obviously over-
reduction of the bound dienoyl-CoA is favored in reduced versus
oxidized BamBC. The four- and six-electron reductions of
(28) Kuntze, K.; Shinoda, Y.; Moutakki, H.; McInerney, M. J.; Vogt, C.;
Richnow, H. H.; Boll, M. EnViron. Microbiol. 2008, 10, 1547–56.
(29) Peters, F.; Shinoda, Y.; McInerney, M. J.; Boll, M. J. Bacteriol. 2007,
189, 1055–60.
(30) Roy, R.; Menon, A. L.; Adams, M. W. Methods Enzymol. 2001, 331,
132–44.
(31) Bol, E.; Bevers, L. E.; Hagedoorn, P. L.; Hagen, W. R. J. Biol. Inorg.
Chem. 2006, 11, 999–1006.
(32) Strobl, G.; Feicht, R.; White, H.; Lottspeich, F.; Simon, H. Biol. Chem.
Hoppe-Seyler 1992, 373, 123–32.
(33) White, H.; Strobl, G.; Feicht, R.; Simon, H. Eur. J. Biochem. 1989,
184, 89–96.
(27) Wang, S. C.; Frey, P. A. Biochemistry 2007, 46, 12889–95.
9
J. AM. CHEM. SOC. VOL. 132, NO. 28, 2010 9855

A R T I C L E S
Kung et al.
The determination of the redox potential of the benzoyl-CoA/
dienoyl-CoA couple provides novel insights into the energetics
of a key reaction in the anaerobic aromatic metabolism. If H2
(E0′ ) -414 mV for the 2H+/H2 redox couple), or a reduced
ferredoxin at a similar redox potential, served as the electron
donor for benzoyl-CoA reduction (E0′ ) -622 mV), the
resulting ∆G0 for the two-electron reduction would yield a value
of around +40 kJ mol-1. Thus, hydrolysis of a single MgATP
to MgADP + Pi (∆G0 ∼ -50 kJ mol-1 under cellular
concentrations)34 would be thermodynamically sufficient to drive
benzoyl-CoA dearomatization. However, class I BCRs couple
electron transfer from reduced ferredoxin (E0′ ) -435 mV)35
to benzoyl-CoA to the hydrolysis of even two MgATP, which
can be rationalized based on the mechanism (one ATP hydro-
lyzed for each electron transferred). Accordingly, benzoyl-CoA
dearomatization in ATP-dependent BCRs is essentially irrevers-
ible. Class I BCRs are distributed in facultative anaerobes such
as denitrifyers or bacteria with an anoxic photosynthesis and
so they can afford an ATP hydrolyzing BCR.
the transport of two Na+/H+ across the membrane to the
cytoplasm would be sufficient to drive benzoyl-CoA reduction
with reduced ferredoxin as the electron donor (assuming ∆G0
) -20 kJ/mol for the H+/Na+ transport). In the case of an
electron bifurcation process, an endergonic electron transfer is
tightly coupled to an exergonic one.37,38 Accordingly, reduction
of benzoyl-CoA by reduced ferredoxin has to be coupled to that
of a second electron acceptor with E0′ g -220 mV. If a low-
potential ferredoxin with E0′ ∼ -500 mV, e.g., generated by
2-oxoglutarate:ferredoxin oxidoreductase, served as an electron
donor, even NAD+ could serve as an appropriate second electron
acceptor for a bifurcation driven benzoyl-CoA reduction process.
In this context, it is interesting to note that the class II BCR complex
from G. metallireducens is predicted to contain three modules
(BamGHI) with similarities to the soluble NAD+/NADH-binding
modules of NADH:quinone oxidoreductases.17,18
Acknowledgment. We acknowledge Nasser Gad’on, Georg
Fuchs (Freiburg), and Simon Wischgoll (Leipzig) for help with the
cultivation of cells. We thank Lydia Walter (Freiburg) for help with
the chemical synthesis of 4. We are grateful to Gary Sawers (Halle)
for careful proofreading of the manuscript. This work was funded
by the German research council (DFG), project BO 1565/10-1.
Genes coding for ATP-independent class II BCRs have so
far only been found in genomes of obligately anaerobic bacteria
where energy conservation is at a premium.17,18,36 In these
enzymes, two possible ATP-independent processes for an elec-
tron transfer to the aromatic ring were recently discussed and
are driven either by a membrane potential or by electron
bifurcation.17 In the case of the former scenario, a coupling to
Supporting Information Available: The HPLC/MS analysis
of the 12/13C exchange reaction (Figure S1) and the cyclic
voltammetric determination of the midpoint potential of 4
(Figure S2). This material is available free of charge via the
Internet at http://pubs.acs.org.
(34) Thauer, R. K.; Jungermann, K.; Decker, K. Bacteriol. ReV. 1977, 41,
100–80.
JA103448U
(35) Boll, M.; Fuchs, G.; Tilley, G.; Armstrong, F. A.; Lowe, D. J.
Biochemistry 2000, 39, 4929–38.
(36) McInerney, M. J.; Rohlin, L.; Mouttaki, H.; Kim, U.; Krupp, R. S.;
Rios-Hernandez, L.; Sieber, J.; Struchtemeyer, C. G.; Bhattacharyya,
A.; Campbell, J. W.; Gunsalus, R. P. Proc. Natl. Acad. Sci. U.S.A.
2007, 104, 7600–5.
(37) Li, F.; Hinderberger, J.; Seedorf, H.; Zhang, J.; Buckel, W.; Thauer,
R. K. J. Bacteriol. 2008, 190, 843–50.
(38) Thauer, R. K.; Kaster, A. K.; Seedorf, H.; Buckel, W.; Hedderich, R.
Nat. ReV. Microbiol. 2008, 6, 579–91.
9
9856 J. AM. CHEM. SOC. VOL. 132, NO. 28, 2010