Mediated electron transfer with monooxygenases - Insight in interactions between reduced mediators and the co-substrate oxygen

Article abstract of DOI:10.1016/j.molcatb.2014.06.011

One of the most important obstacles to overcome in biocatalysis with monooxygenases is the enzyme's dependency on the costly redox cofactor NAD(P)H. Electrochemical regeneration systems, in which an electrode serves as electron donor, provide an alternative route to enzymatic redox reactions. Mediators are often used to accelerate electron transfer between electrode and enzyme. We investigated the mediated bioelectrochemical conversion of p-xylene to 2,5-dimethylphenol (2,5-DMP) by a P450 BM3 variant and were able to produce 2,5-DMP electrochemically. Due to the fact that mediator reduction is limited by the electrode surface a scale-up was performed. However, increasing the electrode surface area to reactor volume ratio led to a drastic increase in cathodic oxygen reduction, causing a drop in product formation. It was shown that reduced cobalt sepulchrate reacts with the co-substrate oxygen. Furthermore, the reportedly oxygen stable mediator [Cp*Rh(I)(bpy)H] ⁺ was compared to cobalt sepulchrate. While its turnover frequency is of comparable magnitude to cobalt sepulchrate when transferring the electrons between electrode and enzyme, using NADP⁺ as intermediary between the mediator and the enzyme significantly increased the mediator's turnover frequency. The rhodium mediator [Cp*Rh(I)(bpy)H]⁺ does not appear to be significantly more oxygen stable.

Full text of DOI:10.1016/j.molcatb.2014.06.011

Journal of Molecular Catalysis B: Enzymatic 108 (2014) 51–58
Contents lists available at ScienceDirect
Journal of Molecular Catalysis B: Enzymatic
journal homepage: www.elsevier.com/locate/molcatb
Mediated electron transfer with monooxygenases—Insight in
interactions between reduced mediators and the co-substrate oxygen
Andreas Tosstorffa, Alexander Dennigb, Anna Joëlle Ruffb, Ulrich Schwanebergb,
Volker Sieberc, Klaus-Michael Mangoldd, Jens Schradera, Dirk Holtmanna,∗
a DECHEMA Research Institute, Biochemical Engineering Group, Theodor-Heuss-Allee 25, 60486 Frankfurt am Main, Germany
b Institute of Biotechnology, RWTH Aachen University, Worringerweg 3, 52074 Aachen, Germany
c Chemistry of Biogenic Resources, Technische Universität München, Straubing Centre of Science, Schulgasse 16, 94315 Straubing, Germany
d DECHEMA Research Institute, Electrochemistry Group, Theodor-Heuss-Allee 25, 60486 Frankfurt am Main, Germany
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 19 May 2014
Received in revised form 14 June 2014
Accepted 25 June 2014
Available online 3 July 2014
One of the most important obstacles to overcome in biocatalysis with monooxygenases is the enzyme’s
dependency on the costly redox cofactor NAD(P)H. Electrochemical regeneration systems, in which an
electrode serves as electron donor, provide an alternative route to enzymatic redox reactions. Mediators
are often used to accelerate electron transfer between electrode and enzyme. We investigated the medi-
ated bioelectrochemical conversion of p-xylene to 2,5-dimethylphenol (2,5-DMP) by a P450 BM3 variant
and were able to produce 2,5-DMP electrochemically. Due to the fact that mediator reduction is limited
by the electrode surface a scale-up was performed. However, increasing the electrode surface area to
reactor volume ratio led to a drastic increase in cathodic oxygen reduction, causing a drop in product for-
mation. It was shown that reduced cobalt sepulchrate reacts with the co-substrate oxygen. Furthermore,
the reportedly oxygen stable mediator [Cp*Rh(I)(bpy)H]+ was compared to cobalt sepulchrate. While its
turnover frequency is of comparable magnitude to cobalt sepulchrate when transferring the electrons
between electrode and enzyme, using NADP+ as intermediary between the mediator and the enzyme sig-
nificantly increased the mediator’s turnover frequency. The rhodium mediator [Cp*Rh(I)(bpy)H]+ does
not appear to be significantly more oxygen stable.
Keywords:
P450
Cofactor substitution
Redox mediators
Electro-enzymatic process
Oxygen dilemma
© 2014 Elsevier B.V. All rights reserved.
1. Introduction
NAD(P)H cofactor used to activate the molecular oxygen [5]. This
has led to the development of several strategies to reduce the
Cytochrome P450 monooxygenases are heme containing redox
proteins found in human liver, but also in other mammals, plants,
bacteria and fungi. With their ability to hydroxylate saturated car-
bon atoms one of their main purposes in nature is to metabolize
exogeneous molecules. Typically, they catalyze oxidation reactions
such as hydroxylations or epoxidations, but also reductions and
rearrangements of oxygenated species or even isomerization [1,2].
The flavocytochrome P450 BM3 fatty acid hydroxylase from Bacil-
lus megaterium is one of the best understood P450 monooxygenases
[3]. The soluble, 119 kDa enzyme is a fusion protein of a P450 and
a P450 reductase, containing FAD and FMN [4]. Its natural sub-
strates are fatty acids [3]. One of the challenges in biocatalysis
with P450 monooxygenases is the dependence on the expensive
amount of employed cofactor by either replacing or regenerating it.
In most cases, cofactors are regenerated with the help of additional
enzymes, but electrochemical methods are an attractive alternative
to this approach [6]. Due to its favourable characteristics, P450 BM3
has served as a model monooxygenase in several electrochemical
experiments. Mediated electron transfer was employed, just like
direct electron transfer or photo-induced electron transfer with a
fusion protein [7–14]. Different authors described the direct elec-
tron transfer from the electrode to the enzyme by immobilizing
P450 BM3 with a polymer matrix on the electrode, however only
in one case, biocatalytic activity was reported [9,11]. Immobiliz-
ing just the heme domain of P450 BM3 on an electrode apparently
also resulted in a loss of catalytic activity in most cases [15,16].
Only Fantuzzi et al. [17] were able to immobilize the heme domain
of a P450 BM3 under retention of its catalytic activity. As elec-
tron transfer rates from the electrode to the P450 BM3 tend to
be slow given that the electron accepting heme is buried deep in
∗
Corresponding author. Tel.: +49 69 7564 610; fax: +49 69 7564 388.
E-mail address: holtmann@dechema.de (D. Holtmann).
http://dx.doi.org/10.1016/j.molcatb.2014.06.011
1381-1177/© 2014 Elsevier B.V. All rights reserved.

52
A. Tosstorff et al. / Journal of Molecular Catalysis B: Enzymatic 108 (2014) 51–58
the protein, the use of a mediator can increase electron transfer
rates [8]. Mediators can transfer electrons between the electrode
and the enzyme in order to accelerate the electron transfer. Cobalt
sepulchrate, also known as [(1,3,6,8,10,13,16,19-octaazabicyclo-
[6.6.6]eicosane)cobalt(III)]3+, has found broad applications as a
mediator in electrochemical biocatalysis [18,19]. It acts as an
one electron shuttle [7] between electrode and monooxygenase,
thereby cobalt switches between the oxidation states Co(II) and
Co(III). The metallorganic compound pentamethylcyclopentadi-
enyl rhodium 2,2ꢀ-bipyridin ([Cp*Rh(bpy)(H2O)]2+) is one of the
few molecules showing an actual catalytic activity with a TTN
of more than 10. So far the rhodium complex was not used in
combination with a P450 monooxygenase, but it was shown to
react with FAD dependent styrene monooxygenase, making it a
promising candidate for electrochemical biocatalysis with FAD
dependent P450s such as P450 BM3. The rhodium complex has
proven to be stable under biocatalytic conditions and shows cat-
alytic activity towards the reduction of several compounds, such
as NAD(P)+, protons, FAD and FMN [20–22]. The reduction of
NAD(P)+ has furthermore been shown to selectively yield enzy-
matically active 1,4-NAD(P)H [23,24]. Unfortunately, the oxidized
form of the complex tends to deactivate itself and also enzymes
by binding to several amino acids and replacing the coordinated
water molecules. In order to reduce deactivation, the use of
“coordinating buffers”, like TRIS or ammonia buffers, has been
suggested [25]. Oxygen plays an important role as a substrate
in P450 catalysis, but also as a reactant with the mediators dis-
cussed above. It is therefore necessary to understand its behaviour
in an electrochemical cell. The amounts of hydrogen peroxide
that are produced during the electrochemical processes due to
the reduction of oxygen and also by P450 itself require methods
to eliminate it, as it has been shown to deactivate the enzyme.
Deactivation occurs as a result of oxidative damage to the heme
and the protein. To eliminate hydrogen peroxide, there are sev-
eral methods, such as shielding the working electrode with an
additional Pt-mesh or adding catalase [12,26]. Unfortunately, so
far the undesired uncoupling reaction has not been addressed
very much and a general awareness of this oxygen dilemma is
missing [27].
Phenol and its derivatives, e.g. cresols or dimethylphenols are
mainly used for the production of polymers, herbicides and fungi-
cides, but also as disinfectants, plasticizers and solvents [28,29].
Phenol is synthesized through the cumene process that gives ace-
tone as a byproduct [28,30] and alkylphenols are produced by
phenol methylation or chlorination, sulfonation, alkylation and
oxidation of toluene or xylene [28]. The production of phenols
is very energy intensive; furthermore, the annual growth rate of
acetone demand is lower than that of phenol. In consequence
alternative routes for production of phenols are required in order
to avoid unwanted acetone yields and reduce production costs.
Direct aromatic oxidation so far has been ruled out because of
low yields and high production costs [30]. A biocatalytic approach
presents therefore an interesting alternative to the conventional
production of phenols due to its high selectivity and low energy
demands. There have been various reports on direct aromatic ring
hydroxylations with P450 monooxygenases [31]. The mutant P450
BM3 M2 (R47S, Y51W, I401M) employed in our experiments has
been described by Dennig et al. [32] to convert p-xylene effi-
ciently into 2,5-dimethylphenol (2,5-DMP) with a selectivity of 98%
(Fig. 1).
O2
+
NADPH
NADP
H2O
H+
[P450BM3]
OH
Fig. 1. Conversion of p-xylene to 2,5-dimethylphenpol by P450 BM3 monooxyge-
nases.
2. Experimental
2.1. Expression of P450 BM3
Lysogeny broth (LB) medium was prepared with 1.0 g peptone,
0.5 g yeast extract, 1.0 g NaCl and 100 ml double-distilled water
(ddH2O) and sterilized by autoclaving. Terrific broth (TB) medium
consisted of two solutions prepared separately, one containing 12 g
peptone, 24 g yeast extract 4 ml glycerol in 800 ml ddH2O, the other
containing 2.31 g KH2PO4, 12.54 g K2HPO4 in 200 ml ddH2O. The
solutions were combined after autoclaving.
Precultures (Escherichia coli BL21 (DE3) Gold pET28a(+) carrying
the gene, cloned by NcoI/EcoRI for P450 BM3 R47S/Y51W/I401 M
also called P450 BM3 M2) were cultivated for 16 h (37 ◦C,
180 rpm), containing 5 ml LB medium and 50 ␮l kanamycin solu-
tion (35 mg/ml, ddH2O). 100 ml of TB medium containing 1 ml
kanamycin solution were then inoculated with 1 ml of precul-
ture and cultivated (37 ◦C, 180 rpm). After reaching an OD600 of
0.8–1.0, P450 expression was induced by adding 500 ␮M aminole-
vulinic acid, 200 ␮l IPTG (0.5 M, ddH2O). Harvesting occurred 20 h
after induction by centrifugation (3220 rcf, 4 ◦C, 20 min). The pel-
lets were resuspended in potassium phosphate buffer (50 mM, pH
7.5) and centrifuged again (3220 rcf, 4 ◦C, 20 min). The buffer solu-
tion was removed and cell pellets were stored at −20 ◦C. To disrupt
the cells, they were resuspended in potassium phosphate buffer
(50 mM, pH 7.5) and treated with ultrasound (10% amplitude, 1 s
sonication, 2 s pause, 2 min total sonication time). Cell debris was
separated by centrifugation (3220 rcf, 4 ◦C, 30 min) and the super-
natant containing the P450 BM3 was sterilized by filtration (PVDF,
22 ␮M). The concentration of P450 was determined according to
the method described by Omura and Sato [33].
2.2. 2,5-Dimethylphenol detection
Defined volumes of the reaction mixtures were extracted in
glass vials with known amounts of methyl tert-butyl ether (MTBE)
containing 1 mM phenol as internal standard. To achieve phase
separation, the vials were centrifuged (5290 rcf, 20 ◦C, 2 min). The
reaction product was detected via gas chromatography, using an
Agilent J&W DB-WAXetr column (see SI). Phenol had a reten-
tion time of 14.8 min, while 2,5-DMP showed a retention time of
15.8 min. Product concentrations were calculated from a calibra-
tion curve.
2.3. Electrode cleaning and preparation
All platinum electrodes were cleaned by placing them in a mix-
ture of hydrogen peroxide and sulfuric acid (1:1) for 1 min, rinsing
with double-distilled water (ddH2O) after which they were stored
in ethanol. A Luggin capillary containing the Ag/AgCl reference elec-
trode (3 M KCl) was rinsed with ddH2O and stored in 3 M KCl. The
Here we describe the electro-enzymatic conversion of p-xylene
to 2,5-DMP. The reaction product is of high interest for the pro-
duction of temperature stable polymers and as a building block for
“next-generation” plastics [29].

A. Tosstorff et al. / Journal of Molecular Catalysis B: Enzymatic 108 (2014) 51–58
53
glassy carbon electrode was polished before every experiment first
with 1 ␮m, then with 0.1 ␮m alumina/ddH2O slurry. Wires were
attached to the glassy carbon electrode using conductive epoxy glue
and to the platinum electrode by spot wielding. The glassy carbon
foam working electrode was rinsed with ddH2O and ethanol and
stored in ethanol. Before use it was rinsed with the corresponding
buffer solution.
Fig. S1). The reaction solution was prepared by combining 8.5 ml
20 mM stock solution Co(III)Sep in ddH2O, 1% DMSO, 0.1 ␮M P450
M2, 500 ␮l p-xylene and if required 2 ml of catalase stock solution
(50 mM potassium phosphate buffer, pH 7.5, 300 kU/ml) and adding
50 mM potassium phosphate buffer (pH 8) to give a total volume
of 85 ml. Reaction solutions that did not contain any enzyme were
used for several runs. The synthesis reaction was carried out in a
100 ml glass reactor. A potential of −700 mV was applied under
continuous stirring. Aeration with synthetic air was started after
approx. 70 min at 1.6 Nl/h. Samples were taken with a syringe from
the aqueous phase and defined amounts were extracted with MTBE
(1 mM phenol). After 160 min, the remaining 80 ml of solution were
extracted with 15 ml MTBE (1 mM phenol). Oxygen levels could
not be monitored during the reaction, as the sensor is not compat-
ible with p-xylene. All determined reaction rates are initial rates
determined by a least square fit.
2.4. Bioelectrosynthesis in a 2 ml scale
Reactions were conducted at room temperature in a glass
cuvette using an Ag/AgCl (3 M KCl) reference electrode and a plat-
inum wire as counter electrode. Potentials were applied with a
Gamry PCI4 300 potentiostat. P450 BM3 M2 (0.5 ␮M), catalase
(2960 U), 100 ␮l p-xylene, NADP+ (0.5 mM if required) and potas-
sium phosphate buffer (50 mM, pH 8) were incubated for 5 min
before the respective mediator was added and a potential was
applied. A platinum net was used for substrate conversion with
5 mM cobalt sepulchrate as mediator. Reactions with 0.05 mM
[Cp*Rh(bpy)(H2O)]2+ were conducted with a glassy carbon elec-
trode. After the reaction, the mixture was transferred to a vial,
and the concentration of P450 was measured by removing 100 ␮l
from the aqueous phase, following the procedure described above.
The P450 concentration of each sample was determined twice. The
product was extracted from the remaining reaction mixture with
900 ␮l MTBE containing 1 mM phenol as internal standard. When
referring to pentamethylcyclopentadienyl rhodium 2,2ꢀ-bipyridin
or cobalt sepulchrate in this work, they will be abbreviated as
Cp*Rh(bpy) and CoSep as long as their exact oxidation state is not
relevant but only their general function as mediators.
2.7. [Cp*Rh(III)(bpy)(H2O)]2+ reduction
The reduction of [Cp*Rh(III)(bpy)(H2O)]2+ was measured pho-
tometrically in the 100 ml reactor by using a solution of 50 ␮M
mediator in TRIS buffer (100 mM) at pH 7.5 and 5.9 and in potas-
sium phosphate buffer (50 mM, pH 7.5). A potential of −760 mV vs.
Ag/AgCl was applied for 20 min.
As no spectral data for [Cp*Rh(I)(bpy)(H)]+ was measurable,
the complex’s absorption maximum was determined by reduc-
ing [Cp*Rh(III)(bpy)(H2O)]2+ with sodium formate and measuring
the UV/VIS spectrum of the solution after the mixture had turned
blue. Even though [Cp*Rh(III)(bpy)(H2O)]2+ and NADPH show sim-
ilar UV/VIS spectra, NADPH formation by [Cp*Rh(III)(bpy)(H)]+ was
determined photometrically, as no relevant change in the spectrum
was determined when applying a potential of −760 mV to a solution
containing only the rhodium mediator.
2.5. Comparison of electrode material
Reactions were conducted in a cuvette placed in an AvaLight-
DHS-DeepUV + AvaSpec-2048 photometer. To compare electrode
materials in combination with CoSep, platinum and glassy carbon
electrodes were taped with Teflon to give an area of 1.2 cm2. To
compare mediator reduction rates of the materials, the solution was
gassed with nitrogen while oxygen concentrations were monitored
until the minimum oxygen concentration was reached. During the
reaction, the nitrogen inlet was placed above the light path in order
to prevent bubbles from interfering with the light beam. During
the enzymatic reaction, simultaneous gassing with nitrogen and
oxygen monitoring was not possible due to spatial limitations in the
reaction setup. To compare the reduction rate of oxygen between
platinum and glassy carbon electrodes, the oxygen concentration
was monitored while a potential was applied.
3. Results
In a first step the electrochemical conversion of p-xylene was
investigated in the 2.2 ml reactor with a platinum mesh as a
working electrode and cobalt sepulchrate as mediator. The elec-
trochemical potential for the bioelectrosynthesis of 2,5-DMP was
optimized by comparing the product formation after 15 min at dif-
ferent potentials. We were able to show that the electrode-driven
synthesis of 2,5-DMP was possible. An optimized productivity was
found at −700 mV vs. Ag/AgCl and no product was formed in the
absence of cobalt sepulchrate or in the presence of solely the open
circuit potential (OCP). A tenfold increase in productivity can be
observed when increasing the potential from −600 mV to −700 mV
vs. Ag/AgCl (Fig. 2).
2.6. Bioelectrosynthesis in a 100 ml scale
In order to examine the impact of the electrochemical condi-
tions on the enzyme stability, the remaining concentration of the
P450 variant was determined after each reaction. An increased loss
of enzyme was observed with increasing electrochemical poten-
tial. The highest amounts of P450 BM3 M2 were recovered in the
absence of cobalt sepulchrate. Whether p-xylene was present or
not had no significant impact on the amount of recovered enzyme
(Fig. 3). The formation of white precipitate was observed in all
experiments except in the absence of mediator or without an
applied potential. In general, it can be concluded that addition of the
mediator has a negative impact on residual P450 BM3 M2 activity.
In order to compare the electrocatalytic activities of different
electrode materials platinum and glassy carbon electrodes were
used. The Co(III)Sep reduction rates for electrodes of identical sur-
face area were compared under different conditions (Table 1).
Co(III)Sep reduction at −650 mV in the presence of oxygen was
twice as fast for the platinum electrode compared to the glassy
Measurements in the 100 ml glass reactor were conducted
using glassy carbon foam, an Ag/AgCl (3 mM KCl) reference elec-
trode and a platinum counter electrode. The counter electrode was
separated by a glass frit. The electrodes were connected accord-
ingly to the potentiostat and the corresponding potential was
applied to start the reaction. To determine the concentration of
Co(III)Sep, the reaction mixture was pumped through a photo-
metrical flow cell. Co(III)Sep depletion was monitored with an
AvaLight-DHS-DeepUV + AvaSpec-2048 photometer at ꢀ = 475 nm
(ε = 109 M−1 cm−1). A factory calibrated oxygen sensor was placed
in the reaction solution. Nitrogen and synthetic air (20% O2 and
80% N2, Kraiss & Friz, Stuttgart, Germany) flow was controlled with
a flowmeter. In order to separate the pH-meter from the circuit of
the potentiostat, it was connected to an isolation transformer. The
reaction temperature was kept constant at 30 ◦C with a water recir-
culator, while controlling the temperature with a thermostat (see

54
A. Tosstorff et al. / Journal of Molecular Catalysis B: Enzymatic 108 (2014) 51–58
500
400
300
200
100
0
OCP
-500
-600
-700
-800 -700, no CoSep
Potential vs. Ag/AgCl
Fig. 2. Bioelectrochemical 2,5-DMP productivity at different potentials with
Co(III)Sep as mediator and at −700 mV also without mediator. Reaction conditions:
0.5 ␮M P450 BM3 M2, 3000 U catalase, 100 ␮l p-xylene, 5 mM Co(III)Sep, potassium
phosphate buffer (50 mM, pH 8), reaction volume V = 2.2 ml. Pt mesh working elec-
trode, Pt wire counter electrode, Ag/AgCl reference electrode (3 M KCl). Values are
averages of N = 2 (OCP = open circuit potential).
Fig. 4. Change of oxygen and Co(III)Sep concentration during time while a potential
is being applied. Reaction conditions: T = 30 ◦C, 5 mM Co(III)Sep, reaction volume
V = 85 ml, 2 Nl/h aeration with synthetic air, 3D glassy carbon foam working elec-
trode, Pt wire counter electrode, Ag/AgCl reference electrode (3 M KCl), −650 mV
was, applied for 20 min starting at t = 0 min.
110
100
90
80
70
60
50
40
30
20
10
p
significant change is observed, when the potential is increased on
a glassy carbon electrode. Subsequently, the effect of the observed
differences in oxygen and mediator reduction between the elec-
trode materials on the productivity of 2,5-DMP was investigated.
The productivities at 1.2 cm2 platinum and glassy carbon electrode
at −700 mV vs. Ag/AgCl were 37 and 252 ␮M h−1, respectively
(reaction conditions: 0.5 ␮M P450 BM3 M2, 3000 U catalase, 100 ␮l
p-xylene, 5 mM Co(III)Sep, potassium phosphate buffer (50 mM,
pH 8)). By comparing theoretical productivities calculated from the
values in Table 1 and the measured values, it becomes obvious that
actual productivities were comparatively low. By using platinum
or glassy carbon electrode the ratio between these values was 2.1%
or 14.3%, respectively. Further investigation showed a dependency
of the reaction rate not only on the material but also on the surface
area of the electrode. In order to increase the scale of the reaction
system, a commercially available 100 ml reactor with a 3D glassy
carbon foam electrode was used. Effects of aeration and nitrogen
purging on oxygen and Co(III)Sep concentrations in the new
reaction system were examined in detail (Fig. 4) since 2,5-DMP
production by P450 BM3 M2 is oxygen dependent.
0
0
0
e
en
l
Xy
0
P
Se
Co
0
50
70
80
-
-
-
-6
OC
no
p-
,
no
Potential vs. Ag/AgCl
,
-700
700
-
Fig. 3. Detected P450 M2 concentration after the bioelectrosynthesis of 2,5-DMP
after 15 min at different potentials. Furthermore, the influence of the substrate
and the mediator at −700 mV was investigated. Reaction conditions: 0.5 ␮M P450
BM3 M2, 3000 U catalase, 100 ␮l p-xylene, 5 mM Co(III)Sep, potassium phosphate
buffer (50 mM, pH 8), reaction volume V = 2.2 ml. Pt mesh working electrode, Pt wire
counter electrode, Ag/AgCl reference electrode (3 M KCl). Values are averages of N = 2
(OCP = open circuit potential).
Oxygen concentrations decrease rapidly in the 100 ml scale
reactor. After applying −650 mV vs. Ag/AgCl, the oxygen concen-
tration reaches a minimum after 7 min. Afterwards the mediator
reduction can be measured. From these results it can be concluded,
that it is not possible to achieve a detectable mediator reduction
in the presence of oxygen. Once molecular oxygen levels were
reduced to a minimum, quantitative mediator reduction would
take place. Apparently, the reduced mediator interacts with molec-
ular oxygen. After 20 min, the application of a potential was stopped
and Co(III)Sep and O2 concentrations started to increase again due
to the aeration with synthetic air. In the next step both reduction
rates were measured under different conditions (Table 2).
Both, mediator and oxygen reduction rate increase with
decreasing potential. Increasing the aeration rate would further-
more lead to no measurable mediator reduction at all. In all
experiments two phases were observed; in a first time period
the dissolved oxygen was quantitatively reduced and afterwards
reduced mediator was detected. Current efficiencies for the media-
tor reduction depend strongly on the oxygen input. When purging
with nitrogen, an efficiency of up to 93% was measured, while it was
11% when purging with synthetic air at a rate of 0.4 vvm and approx.
0% when the aeration rate was 0.8 vvm. To clarify if the oxygen
reduction is caused by the applied potential or a reduced media-
tor, an experiment was performed with and without the mediator
carbon electrode. While increasing the potential from −650 mV
to −700 mV doubles the oxygen reduction rate on platinum
electrodes, no significant change is observed, when the voltage is
increased on a glassy carbon electrode in the presence of oxygen.
Under anaerobic conditions the mediator reduction at the glassy
carbon electrode was doubled compared to the platinum electrode.
Furthermore, oxygen reduction rates at the electrode materials
were measured (data not shown). A much higher oxygen reduction
rate was observed for the platinum electrode. Oxygen is reduced
up to five times as fast by a platinum electrode compared to glassy
carbon. While increasing the potential from −650 mV to −700 mV
doubles the oxygen reduction rate on platinum electrodes, no
Table 1
Comparison of the initial Co(III)Sep reduction rates for platinum and glassy carbon
electrodes.
Platinum (␮M min−1
)
Glassy carbon (␮M min−1
)
O2, −650 mV
N2, −650 mV
O2, −700 mV
14.0
20
29.3
7.0
40
6.5
Reaction conditions: oxygen or nitrogen saturated solution, 1.2 cm2 electrode sur-
face area, Pt wire counter electrode, Ag/AgCl reference electrode (3 M KCl).

A. Tosstorff et al. / Journal of Molecular Catalysis B: Enzymatic 108 (2014) 51–58
55
Table 2
Table 3
Comparison of the reduction rates of oxygen and Co(III)Sep at glassy carbon foam
electrode in the linear range in the 100 ml reactor.
Bioelectrochemical formation of 2,5-DMP by P450 M2 with Cp*Rh(bpy) as a medi-
ator and Cp*Rh(bpy) combined with NADP+.
Co(III)Sep reduction
O2 reduction rate in
2,5-DMP productivity
Mediator turnover
frequency in h−1
rate in ␮M min−1
␮M min−1
in ␮M h−1
−600 mV, 0.4 vvm
synthetic air
1
3
4
22
36
2
2
50 ␮M Cp*Rh(bpy)
50 ␮M Cp*Rh(bpy),
500 ␮M NADP+
200 ␮M
2.7 1.6
28.9 7.8
0.05
0.52
−650 mV, 0.4 vvm
34
synthetic air
36.8 0.6
0.18
−700 mV, 0.4 vvm
synthetic air
80 12
46 0.1
Cp*Rh(bpy),
500 ␮M NADP+
−650 mV, 0.8 vvm
4
2
3
29
1
Reaction conditions: TRIS buffer (100 mM, pH 7.5), −760 mV vs. Ag/AgCl for 15 min,
Glassy carbon working electrode, 1.2 cm2 surface, Pt wire counter electrode, Ag/AgCl
reference electrode (3 M KCl). Values are averages of N = 2.
synthetic air
−650 mV, 0.4 vvm
nitrogen
68
n.a.
Reaction conditions: T = 30 ◦C, 5 mM Co(III)Sep, reaction volume V = 85 ml, 3D glassy
carbon foam working electrode, Pt wire counter electrode, Ag/AgCl reference elec-
trode (3 M KCl), vvm = volume of gas per volume reaction volume and minute.
formate to its characteristically blue form absorbing at 620 nm,
by using electrochemical reduction no increase at 620 nm or any
other wavelength was observed, even under anaerobic conditions.
Only a slight decrease in absorption at 360 nm was seen. Never-
theless, a cyclic voltammogram shows a reduction at −760 mV on
a glassy carbon electrode (data not shown). As [Cp*Rh(bpy)H]+ is
known to reduce NADP+ to NADPH, the mediator reduction was
monitored indirectly by measuring the concentration of NADPH at
340 nm. Adding NADP+ to the reaction mixture, led to the reduc-
tion of NADP+ to NADPH, clearly showing that mediator reduction
takes place, as cathodic NADP+ reduction occurs at much more neg-
ative potentials. In general, the electrode-driven reduction rate of
NADP+ in the presence of oxygen is slow. The reduction rate of
NADP+ was observed to be independent of the aeration rate, how-
ever there was a five-fold increase when purging with nitrogen
from 0.15 to 0.72 mM/h (see SI). While the rhodium mediator has
not yet been used in combination with a P450 monooxygenase, it
has been used in combination with other heme containing proteins.
We were able to synthesize 2,5-DMP by using Cp*Rh(bpy) solely as a
mediator and in another experiment by combining it with NADP+.
Using NADP+ showed a more than tenfold increase in productiv-
ity (Table 3). Nevertheless, in cobalt sepulchrate driven reactions,
the productivities were of more than 250 ␮M h−1. In this aspect,
it is important to also take the mediator efficiency into consider-
ation, which can be done by determining the turnover frequency
(n(product)/n(mediator)t). While cobalt sepulchrate has a slightly
higher turnover frequency (0.09 h−1) than the rhodium mediator,
adding NADP+ to the rhodium mediator, drastically increased its
efficiency. Nevertheless, also in this case the turnover frequency it
can be regarded as inefficient.
(Fig. 5). In both cases a decrease of the oxygen concentration in
the buffer system can be measured. Nevertheless, in the absence
of the mediator the oxygen depletion was much lower and during
the whole experiment oxygen was present in high concentrations.
Therefore, it becomes clear that Co(III)Sep functions as a catalyst,
strongly accelerating oxygen reduction.
To investigate the influence of the detected effects on the prod-
uct formation of 2,5-DMP a cobalt sepulchrate driven synthesis
was performed in the large scale setup (5 mM Co(III)Sep, 0.1 ␮M
P450 M2, 1% DMSO, catalase (850,000 U), 85 ml reaction volume,
−700 mV vs. Ag/AgCl for 20 min, 3D glassy carbon foam working
electrode, Pt wire counter electrode, Ag/AgCl reference electrode).
The measured product formation rate in this reaction system was
12.2 ␮M h−1. Surprisingly, this value is much lower compared to the
productivity in the small-scale system (∼450 ␮M h−1) mentioned
above. While in the small scale reactor oxygen remains in the reac-
tion system, the increased electrode surface in the scaled-up system
leads to an oxygen-limited enzymatic process.
Due to the apparent sensitivity of reduced cobalt sepulchrate
towards oxygen, the bioelectrochemical synthesis was also con-
ducted with Cp*Rh(bpy), which was reported to be more oxygen
stable [5]. A quantitative electrochemical reduction of the rhodium
mediator could not be achieved under nitrogen purge, neither
at pH 7.5 or 5.5 in TRIS solution nor in potassium phosphate
buffer at pH 7.5. While the mediator is readily reduced by sodium
4. Discussion
4.1. Electrode materials
Even though the influence of electrode materials on redox
systems is well known [34], this issue is often ignored in mediated
bioelectrosynthesis. Comparing a platinum and glassy carbon
electrode of the same macroscopic area showed a clear difference
in the mediator and oxygen reduction between the two materials.
Co(III)Sep reduction was twice as fast on glassy carbon compared
to platinum. On the other hand, oxygen levels decreased much
faster when using a platinum electrode. As a consequence, when
using glassy carbon, an increased productivity can be expected, and
was also observed experimentally; glassy carbon electrode shows
the same productivity as a platinum electrode with twice the sur-
face area. Platinum and glassy carbon also differ in their reactivity
when using Cp*Rh(bpy). While it was not possible to quantitatively
reduce this mediator with neither of the electrodes, it was not
even possible to detect a reduction signal when conducting a cyclic
voltammetry experiment with a platinum electrode (data not
Fig. 5. Comparison of oxygen depletion in the presence and absence of cobalt sepul-
chrate. Reaction conditions: T = 30 ◦C, 2 mM Co(III)Sep, potassium phosphate buffer
(50 mM, pH 8), reaction volume V = 85 ml, 0.4 vvm aeration with synthetic air, 3D
glassy carbon foam working electrode, Pt wire counter electrode, Ag/AgCl reference
electrode (3 M KCl).

56
A. Tosstorff et al. / Journal of Molecular Catalysis B: Enzymatic 108 (2014) 51–58
shown). This was only possible when using a glassy carbon elec-
trode. While a cyclic voltammogram of the rhodium complex with
a platinum electrode was reported, the experiment was conducted
under very different reaction conditions than in our case, using a
mixture of acetonitrile and tetrabutylammonium hexafluorophos-
phate as solvent [35]. One aspect which is especially important
for enzymatic reactions is the increased selectivity of glassy
carbon electrodes towards the formation of hydrogen peroxide
when reducing oxygen. Nevertheless, the question remained unan-
swered whether the total amount of hydrogen peroxide produced is
actually higher than on a platinum electrode, given the fact that the
total amount of oxygen reduced is higher on a platinum electrode.
When furthermore taking into consideration that glassy carbon is a
much cheaper material than platinum, one can conclude that glassy
carbon is a superior material when it comes to bioelectrosyntheses
using cobalt sepulchrate or Cp*Rh(bpy) as mediators.
making the reaction with oxygen spin forbidden [36,37]. In contrary
to our observations, Kölle et al. [35] described the electrochemical
reduction of [Cp*Rh(bpy)H2O]2+ resulting in a brownish solution
at pH 6–8. Several differences in their reaction setup may have
caused this observation (much smaller reaction volumes, purged
with argon and used a mercury cathode). When comparing the
mechanism of the electrochemical reduction of [Cp*Rh(bpy)H2O]2+
to the formate driven reduction, there is one striking difference. The
electrochemical reduction occurs via two sequential one electron
transfers, whereas formate reduces the substance by transferring
a hydride ion [20]. As was shown for the rhodium complex, but
also in this work for the cobalt sepulchrate mediator, the enzyme
reaction depends on the macroscopic electrode surface area. While
this observation served as a motivation to increase the surface
to volume ratio, resulting in the unfortunate rapid oxygen loss
described above, it is also helpful to understand how the mediated
reaction works in the presence of oxygen. Due to the surface area
dependence of the enzymatic reaction and the observation that no
significant amount of reduced mediator is present in the aqueous
bulk, one can assume that the electron transfer from the mediator
to the enzyme takes place at the interface of the electrode surface
and the aqueous solution. Heterogeneous reactions usually show
concentration gradients at the interface of two phases. While only
the mediator reduction and the cathodic oxygen reduction are actu-
ally heterogeneous reactions, the oxidized enzyme is expected to
show such an interfacial concentration gradient as well, because
the amount of reduced mediator increases when approaching the
electrode’s surface.
There are several parameters influencing the exact nature of
these gradients, such as diffusion coefficients, absorption and
desorption rates, electron transfer rate constants and reaction rate
constants. We were able to show, that in the 2.2 ml reactor oxygen
is present in concentrations of at least 100 ␮M, whereas no reduced
cobalt sepulchrate was detected. The assumption can therefore be
made that the mediator catalyzed enzyme reduction takes place at
the electrode-bulk interface. The exact course of the concentration
gradients is therefore crucial for the development of the over-
all reaction. Following this argumentation, oxygen and oxidized
P450 compete at the interface for reduced mediator. Assuming that
mediator oxidation is of second order, a rate law can be suggested
(Eq. (1)), where kox is the reaction rate constant for the reaction
between mediator and oxygen, ke is the reaction rate constant for
the reaction between mediator and enzyme and rel is the reaction
rate for the heterogeneous electron transfer between electrode and
mediator.
4.2. The oxygen dilemma
A main challenge is the oxidative uncoupling of the mediator
oxidation from the enzymatic reaction as molecular oxygen can
re-oxidize the mediator. This causes a lower energy efficiency, as
reduced mediator is lost and furthermore leads to the formation
of harmful reactive oxygen species [5]. When scaling up the reac-
tion and increasing the ratio of electrode surface to reactor volume,
the fast decrease in oxygen concentrations is most striking. Besides
resulting in a very inefficient process, as large amounts of elec-
trons are passed directly to oxygen, this will also slow down the
oxygen dependent hydroxylation reaction. Initially, the idea was
to find reaction conditions under which an optimal concentration
ratio of reduced cobalt sepulchrate and oxygen would establish.
However, one can clearly observe how reduced cobalt sepulchrate
concentrations in the bulk start to increase just when all of the
oxygen has been consumed. Furthermore all of the reduced medi-
ator will be re-oxidized before oxygen levels start to rise. Oxygen
absorption is apparently much slower than the oxidation of cobalt
sepulchrate. The uncoupling of the mediator reduction from the
enzymatic reaction leads to a further decrease in energy effi-
ciency. The dependency of monooxygenases on molecular oxygen
inevitably causes this oxygen dilemma. While increasing the oxy-
gen concentration by simple aeration is possible in the bulk, but will
of course increase the rate of mediator oxidation and thus uncou-
pling. Fig. 6 shows a postulated reaction network for mediator
driven reaction with P450 monooxygenases.
The mediator reduction rate can be increased by applying
more negative potentials which will lead to an increased oxygen
reduction rate. However, an increase of the oxygen reduction rate
was only observed when changing the potential from −600 mV
to −650 mV. Further changing the potential from −650 mV to
−700 mV showed no effect, pointing at a mass transport or adsorp-
tion/desorption limitation of the oxygen reduction at this point.
One could therefore argue that it might be possible to achieve
high concentrations of reduced cobalt sepulchrate by increasing
the mediator reduction to a point where it is faster than the media-
tor oxidation by molecular oxygen. Combining this approach with
intensive aeration might actually lead to the desired conditions
with a high monooxygenase activity. However, extensive aeration
could denature the protein and could lead to foam formation. Fur-
thermore, the overall process would not be energy efficient at all,
as uncoupling and cathodic oxygen reduction would be occurring
at a maximum rate. Apparently, cobalt sepulchrate is not the most
favourable mediator for electrochemical reactions with monooxy-
genases. In order to overcome this oxygen dilemma, another
mediator was investigated. [Cp*Rh(bpy)(H2O)]2+ can be reduced to
[Cp*Rh(bpy)H]+, which was reported to be relatively oxygen sta-
ble, as the additional electrons are stabilized in a hydrogen bond,
∂[Medox
]
= kox[Medred][O2] + kE[Medred][Eox] − rel
(1)
∂t
Obviously, not only the rate constants are of importance
here, but especially the ratio of oxygen concentration to the
concentration of oxidized P450 [O2]/[Eox] at the interface will be
decisive for the outcome of the competition for reduced mediator
[Medred]. We observed a tenfold increase in productivity when
changing the applied potential from −600 to −700 mV. While
this might be explained with an increased mediator reduction
rate, the increased cathodic oxygen reduction at platinum elec-
trodes we observed when changing the potential from −650 to
−700 mV and as a consequence a reduced value for [O2]/[Eox
]
might be a superposed effect further accelerating the reaction.
As odd as it may seem, increasing the cathodic oxygen reduction
rate might actually lead to an increased productivity, as long as
there is enough oxygen available in the bulk solution. However
this will also lead to a reduced energy efficiency of the overall
process. When comparing cobalt sepulchrate and Cp*Rh(bpy) as
mediators, the volumetric productivities (c(product)/t) obviously
favour cobalt sepulchrate. Even though no quantitative amounts

A. Tosstorff et al. / Journal of Molecular Catalysis B: Enzymatic 108 (2014) 51–58
57
Product
H2O
O2
H2O
H2O2 or H2O
O2
P450ox
H2O2 or H2O
H2O
O2
Substrate
Product
Su
bs
tra
te
Fig. 6. Postulated reaction network of electrochemically driven P450 biocatalysis.
of [Cp*Rh(bpy)H]+ are formed during the course of the reaction,
[Cp*Rh(bpy)H]+ formation is not rate limiting per se, as can be
seen when adding NADP+ to the reaction mixture. The drastic
increase in productivity observed here might be caused by a faster
diffusion of NADP+ to the electrode, assuming that the reaction
was previously limited by enzyme diffusion. Given the large dif-
lies in crucial electron pathways, specifically, the cathodic oxygen
and mediator reduction and the oxygen and enzyme reduction by
the reduced mediator. Influencing the reaction rates of each of these
electron transfer steps is required in order to channel the electrons
to the enzyme and eventually accelerate product formation. Two
goals can be defined in order to drive the overall reaction in the
desired direction:
ferences in diffusion coefficients between enzymes (10−7 cm2 s−1
)
and organic molecules (10−5 cm2 s−1), it is no surprise that using
NADP+ as an intermediate between the mediator and the enzyme
will enhance the reaction [38]. However, as was described before,
[Cp*Rh(bpy)H]+ is also not completely oxygen stable. By adding
NADP+ to the reaction mixture, there is an additional competitor
for reduced mediator available in much larger amounts than either
the oxidized P450 or molecular oxygen. As can be seen in Eq. (2), an
additional term for the reaction between NADP+ and the mediator
appears, where kN is the corresponding reaction rate constant.
Therefore much more electrons are channelled into the desired
electron pathway resulting in higher mediator efficiency.
•
•
Increasing the selectivity and rate of the mediator reduction;
Increasing the selectivity and rate of the enzyme reduction.
Even though it was not possible to achieve these goals in this
work, there is still a great amount of tools available to eventu-
ally do so. In order to achieve a faster and more efficient mediator
reduction, several approaches are possible. As the mediator reduc-
tion is an electron transfer reaction between the electrode and
the oxidized mediator, both the mediator and the electrode can
be modified in order to increase the rate of this electron trans-
fer. At the same time, it is desirable to reduce the cathodic oxygen
reduction in order to increase energy efficiency. Another option is
to rule out the possibility of any cathodic oxygen reduction by con-
ducting the electrochemical mediator regeneration in an anaerobic
environment. This can be achieved by proper process design. Inde-
pendently on how efficient and fast a mediator is regenerated, in
order to achieve a high productivity, the electron transfer from the
mediator to the enzyme is just as important to achieve the overall
goal of high product formation. Modifications of the electron donor
and acceptor are means to achieve a high selectivity and rate of the
enzyme reduction as reported for heterogeneous electron trans-
fer step. In order to completely understand a bioelectrochemical
process, it is not sufficient to characterize a mediator by its inter-
action with an enzyme or coenzyme. The charge transfer between
the electrode and the mediator is equally important and can have
consequences for the overall reaction.
∂[Medox]∂t = kox[Medred][O2] + kE[Medred][Eox
+ kN[Medred][NADP+] − rel
]
(2)
Another explanation could be a higher affinity of the rhodium
complex for NADP+ than for the enzyme. While the reduction of
NADP+ takes place at the electrodes surface, enzyme reduction
itself is now largely homogeneous and no longer rate limiting. The
fact that increasing the Rh-mediator concentration by a factor of 4
did not result in a significant increase in productivity shows that
mediator mass transport to the electrode surface might not be rate
limiting, but rather another reaction step. On the other hand, given
that using cobalt sepulchrate in much larger concentrations than
[Cp*Rh(bpy)(H2O)]2+ resulted in higher volumetric productivities
suggests that the opposite is true. In summary, further insight,
especially in the reaction mechanisms of both mediators is there-
fore required to understand the mechanism and to improve the
mediated electron transfer with oxygen-depended enzymes.
Appendix A. Supplementary data
Supplementary data associated with this article can be found,
in the online version, at http://dx.doi.org/10.1016/j.molcatb.
2014.06.011.
5. Conclusion and outlook
It is well known that the combination of electrochemical and
enzymatic reactions to bioelectrosynthetic system enables efficient
processes, e.g. with peroxidases [39–41], alcohol dehydrogenases
[42,43] or enoate reductases [44,45]. As discussed above, when
performing bioelectrosynthesis with an enzyme depending on
molecular oxygen, one is dealing with a very complex reaction net-
work (Fig. 6). However, as shown in this work, the overall challenge
References
[1] F.P. Guengerich, Chem. Res. Toxicol. 14 (6) (2001) 611–650.
[2] F.P. Guengerich, A.W. Munro, J. Biol. Chem. 288 (24) (2013) 17065–17073.
[3] A.W. Munro, D.G. Leys, K.J. McLean, K.R. Marshall, T.W.B. Ost, S. Daff, C.S. Miles,
S.K. Chapman, D.A. Lysek, C.C. Moser, C.C. Page, P.L. Dutton, Trends Biochem.
Sci. 27 (5) (2002) 250–257.

58
A. Tosstorff et al. / Journal of Molecular Catalysis B: Enzymatic 108 (2014) 51–58
[4] F. Hannemann, A. Bichet, K.M. Ewen, R. Bernhardt, Biochim. Biophys. Acta
[24] E. Steckhan, S. Herrmann, R. Ruppert, E. Dietz, M. Frede, E. Spika,
Organometallics 10 (5) (1991) 1568–1577.
(BBA): Gen. Subj. 1770 (3) (2007) 330–344.
[5] F. Hollmann, K. Hofstetter, A. Schmid, Trends Biotechnol. 24 (4) (2006)
163–171.
[25] M. Poizat, I.W.C.E. Arends, F. Hollmann, J. Mol. Catal. B: Enzym. 63 (3–4) (2010)
149–156.
[6] S.J. Sadeghi, A. Fantuzzi, G. Gilardi, Biochim. Biophys. Acta (BBA): Proteins Pro-
teomics 1814 (1) (2011) 237–248.
[26] V. Reipa, M.P. Mayhew, V.L. Vilker, Proc. Natl. Acad. Sci. U. S. A. (1997)
13554–13558.
[7] V.V. Shumyantseva, T.V. Bulko, A.I. Archakov, J. Inorg. Biochem. 99 (5) (2005)
1051–1063.
[27] D. Holtmann, M.W. Fraaije, I.W.C.E. Arends, D.J. Opperman, F. Hollmann, Chem.
Commun. (2014).
[8] A.K. Udit, F.H. Arnold, H.B. Gray, J. Inorg. Biochem. 98 (9) (2004) 1547–1550.
[9] A.K. Udit, M.G. Hill, H.B. Gray, Langmuir 22 (25) (2006) 10854–10857.
[10] I.F. Sevrioukova, C.E. Immoos, T.L. Poulos, P. Farmer, Isr. J. Chem. 40 (1) (2000)
47–53.
[11] D. Holtmann, K.-M. Mangold, J. Schrader, Biotechnol. Lett. 31 (5) (2009)
765–770.
[28] L. Pilato, Phenolic Resins: A Century of Progress, Springer, Heidelberg, 2010.
[29] Y. Shibasaki, K. Hoshi, E. Suzuki, Y. Shiraishi, Y. Norisue, Y. Oishi, Polym. J. 41
(12) (2009) 1136–1143.
[30] R.J. Schmidt, Appl. Catal. A: Gen. 280 (1) (2005) 89–103.
[31] F. Hollmann, I.W.C.E. Arends, K. Buehler, A. Schallmey, B. Buhler, Green Chem.
13 (2) (2011) 226–265.
[12] S. Zengin C¸ ekic¸ , D. Holtmann, G. Güven, K.-M. Mangold, U. Schwaneberg, J.
Schrader, Electrochem. Commun. 12 (11) (2010) 1547–1550.
[13] F.W. Ströhle, S.Z. Cekic, A.O. Magnusson, U. Schwaneberg, D. Roc-
catano, J. Schrader, D. Holtmann, J. Mol. Catal. B: Enzym. 88 (0) (2013)
47–51.
[14] C. Ley, H. Schewe, F.W. Ströhle, A.J. Ruff, U. Schwaneberg, J. Schrader, D. Holt-
mann, J. Mol. Catal. B: Enzym. 92 (0) (2013) 71–78.
[15] B.D. Fleming, Y. Tian, S.G. Bell, L.-L. Wong, V. Urlacher, H.A.O. Hill, Eur. J.
Biochem. 270 (20) (2003) 4082–4088.
[16] A.K. Udit, M.G. Hill, V.G. Bittner, F.H. Arnold, H.B. Gray, J. Am. Chem. Soc. 126
(33) (2004) 10218–10219.
[17] A. Fantuzzi, Y.T. Meharenna, P.B. Briscoe, C. Sassone, B. Borgia, G. Gilardi, Chem.
Commun. 12 (2006) 1289–1291.
[18] K.M. Faulkner, M.S. Shet, C.W. Fisher, R.W. Estabrook, Proc. Natl. Acad. Sci. U. S.
A. 92 (1995) 7705–7709.
[19] U. Schwaneberg, D. Appel, J. Schmitt, R.D. Schmid, J. Biotechnol. 84 (2000)
249–257.
[20] F. Hollmann, B. Witholt, A. Schmid, J. Mol. Catal. B: Enzym. 19–20 (2002)
167–176.
[21] F. Hollmann, A. Schmid, E. Steckhan, Angew. Chem. Int. Ed. 40 (1) (2001)
169–171.
[32] A. Dennig, J. Marienhagen, A.J. Ruff, L. Guddat, U. Schwaneberg, ChemCatChem
4 (6) (2012) 771–773.
[33] T. Omura, R. Sato, J. Biol. Chem. 239 (1964) 2370–2378.
[34] P. Chen, R.L. McCreery, Anal. Chem. 68 (22) (1996) 3958–3965.
[35] U. Kölle, B.-S. Kang, P. Infelta, P. Comte, M. Grätzel, Chem. Ber. 122 (10) (1989)
1869–1880.
[36] W. Adam, Chem. unserer Zeit 15 (6) (1981) 190–196.
[37] P. Lechtken, Chem. unserer Zeit 8 (1) (1974) 11–16.
[38] M.E. Young, P.A. Carroad, R.L. Bell, Biotechnol. Bioeng. 22 (5) (1980)
947–955.
[39] L. Getrey, T. Krieg, F. Hollmann, J. Schrader, D. Holtmann, Green Chem. 16 (3)
(2014) 1104–1108.
[40] T. Krieg, S. Huttmann, K.-M. Mangold, J. Schrader, D. Holtmann, Green Chem.
13 (10) (2011) 2686–2689.
[41] S. Lütz, K. Vuorilehto, A. Liese, Biotechnol. Bioeng. 98 (3) (2007) 525–534.
[42] S. Kochius, M. Paetzold, A. Scholz, H. Merkens, A. Vogel, M. Ansorge-
Schumacher, F. Hollmann, J. Schrader, D. Holtmann, J. Mol. Catal. B: Enzym.
103 (2014) 61–66.
[43] S. Kochius, J.B. Park, C. Ley, P. Könst, F. Hollmann, J. Schrader, D. Holtmann, J.
Mol. Catal. B: Enzym. 103 (2014) 94–99.
[44] H. Simon, H. Günther, J. Bader, W. Tischer, Angew. Chem. Int. Ed. 20 (10) (1981)
861–863.
[22] F. Hollmann, A. Schmid, Towards, J. Inorg. Biochem. 103 (3) (2009) 313–315.
[23] H.C. Lo, O. Buriez, J.B. Kerr, R.H. Fish, Angew. Chem. Int. Ed. 38 (10) (1999)
1429–1432.
[45] K. Fisher, S. Mohr, D. Mansell, N.J. Goddard, P.R. Fielden, N.S. Scrutton, Catal.
Sci. Technol. 3 (6) (2013) 1505–1511.