Bioorganometallic chemistry: Biocatalytic oxidation reactions with biomimetic NAD+/NADH co-factors and [Cp*Rh(bpy)H]+ for selective organic synthesis

Article abstract of DOI:10.1016/j.jorganchem.2004.09.044

The biocatalytic, regioselective hydroxylation of 2-hydroxybiphenyl to the corresponding catechol was accomplished utilizing the monooxygenase 2-hydroxybiphenyl 3-monooxygenase (HbpA). The necessary natural 1,4-dihydronicotinamde adenine dinucleotide (NAD

Full text of DOI:10.1016/j.jorganchem.2004.09.044

Journal of Organometallic Chemistry 689 (2004) 4783–4790
www.elsevier.com/locate/jorganchem
Bioorganometallic chemistry: biocatalytic oxidation reactions with
biomimetic NAD+/NADH co-factors and [Cp*Rh(bpy)H]+
for selective organic synthesis
a
Jochen Lutz , Frank Hollmann , The Vinh Ho , Adrian Schnyder ,
a
a
a
b,
a, ,1
*
Richard H. Fish *, Andreas Schmid
a
Institute of Biotechnology, Swiss Federal Institute of Technology, ETH Hoenggerberg, HPT, CH-8093 Zurich, Switzerland
b
Lawrence Berkeley National Laboratory, MS-70-108B, University of California, Berkeley, CA 94720, USA
Received 15 August 2004; accepted 17 September 2004
Abstract
The biocatalytic, regioselective hydroxylation of 2-hydroxybiphenyl to the corresponding catechol was accomplished utilizing the
monooxygenase 2-hydroxybiphenyl 3-monooxygenase (HbpA). The necessary natural 1,4-dihydronicotinamde adenine dinucleotide
(NADH) co-factor for this biocatalytic process was replaced by a biomimetic co-factor, N-benzyl-1,4-dihydronicotinamide, 1b. The
interaction between the flavin (FAD) containing HbpA enzyme and the corresponding biomimetic NADH compound, N-benzyl-
1,4-dihdronicotinamide, 1b, for hydride transfer, was shown to readily occur. The in situ recycling of the reduced NADH biomimic
1b from 1a was accomplished with [Cp*Rh(bpy)H](Cl); however, productive coupling of this regeneration reaction to the enzymatic
hydroxylation reaction was not totally successful, due to a deactivation process concerning the HbpA enzyme peripheral groups; i.e.,
–SH or –NH2 possibly reacting with the precatalyst, [Cp*Rh(bpy)(H2O)](Cl)2, and thus inhibiting the co-factor regeneration proc-
ess. The deactivation mechanism was studied, and a promising strategy of derivatizing these peripheral –SH or –NH2 groups with a
polymer containing epoxide was successful in circumventing the undesired interaction between HbpA and the precatalyst. This latter
strategy allowed tandem co-factor regeneration using 1a or 2a, [Cp*Rh(bpy)(H2O)](Cl)2, and formate ion, in conjunction with the
polymer bound, FAD containing HbpA enzyme to provide the catechol product.
ꢀ 2004 Elsevier B.V. All rights reserved.
Keywords: Biocatalysis; Co-factor regeneration; Chemoenzymatic reactions; Monoxygenase enzymes; NAD+/NADH biomimics; Organorhodium
hydride; [Cp*Rh(bpy)(H)]+; Epoxide polymer
1. Introduction
Despite recent advances in catalytic oxyfunctionaliza-
tion reactions, this class of reactions still represents one
of the major challenges for synthetic organic chemistry
[1]. The design of efficient catalysts has often been in-
spired by examples from nature. However, most bio-
mimics do not meet the performances of their natural
precedents. In particular, the regio-, chemo-, and enan-
tio-specificity of the catalyzed reactions are not as effi-
cient as the natural enzymes, and include low turnover
frequencies (TF), as well as turnover numbers (TN),
*
Corresponding authors. Tel.: +1 510 486 4850; fax: +1 510 486
7303 (R.H. Fish), Tel.: +41 1 633 3691; fax: +41 1 633 1051 (A.
Schmid).
E-mail addresses: rhfish@lbl.gov (R.H. Fish), andreas.schmid@
biotech.biol.ethz.ch, andreas.schmid@bci.uni-dortmund.de
(A. Schmid).
1
Present address: Department of Biochemical and Chemical
Engineering, University of Dortmund, Emil Figge Str. 70, D-44227
Dortmund, Germany. Tel.: +49 231 755 7380; fax: +49 231 755 7382.
0022-328X/$ - see front matter ꢀ 2004 Elsevier B.V. All rights reserved.
doi:10.1016/j.jorganchem.2004.09.044

4784
J. Lutz et al. / Journal of Organometallic Chemistry 689 (2004) 4783–4790
which are also several orders of magnitude lower. This
above-mentioned situation might also be attributed to
the dependency on activated oxygen species, such as per-
oxides, dioxiranes, or high valent halogen compounds,
which are incorporated into the substrate with low selec-
tivity, and frequently at unsatisfactory rates.
Monooxygenase enzymes, on the other hand, utilize
molecular oxygen for the specific incorporation of one
oxygen into unactivated C–H, C(O)–C, and C@C bonds
[2], while minimizing the number of undesired side-
reactions. Furthermore, monooxygenase enzymes have
been reported to exhibit TF of 10–20 sꢀ1 [3], making
them highly important catalysts for synthetic organic
chemistry purposes. The availability of monooxygenase
enzymes, and other aspects associated with preparative
scale reactions, as well as stability under process condi-
tions, are now thought to be considered standard prac-
tices, while the requirements of reducing equivalents
appears to be a continuing challenge [4–6].
are costly and hydrolytically unstable, and therefore,
represent a major barrier for preparative applications
utilizing monooxygenases [4,7]. The ribose, pyrophos-
phate and adenosine groups (Fig. 1) entails the complex
structure of NADH, 2b and NAD(P)H, 3b, thus making
chemical synthesis or isolation from biological sources
tedious, and creates the high costs of these biocatalytic
processes. It has recently been shown by Lo and Fish
[8] that apparently only the nicotinamide ring was re-
quired as the redox active site for co-factor regeneration
during the enzymatic transfer hydrogenation reaction
catalyzed by horse liver alcohol dehydrogenase
(HLADH) enzyme, with the finding that N-benzylnico-
tinamide triflate, 1a (in this paper bromide as the
counter ion instead of triflate), and N-benzyl-1,4-
dihydronicotinamide, 1b, could be biomimics for natu-
ral NAD+/NADH and provide chiral alcohols from
achiral ketones with high enantioselectivity.
To further explore these initial findings on HLADH
enzyme recognition of 1b as an NADH biomimic for
reductions of achiral ketones to chiral alcohols, we
decided to expand the focus of the NADH model, 1b,
In general, monooxygenase enzymes are 1,4-dihydro-
nicotinamde adenine dinucleotide co-factor dependent
enzymes; i.e., NADH or NADPH. These co-factors
O
O
H
H
NH2
NH2
+
-
Br
N
N
1a
1b
O
C
O
C
H
H
NH2
O
P
O
P
NH2
Na+
- O
O
CH2
+
N
-
N
O
O
CH2
O
O
OH OH
OHOH
O
O
NH2
NH2
N
N
N
N
Na+
O
P
O
CH2
N
-
-
Na+
O
P
O
CH2
N
N
N
O
O
O
O
OHOR
OH OR
2a, R = H (NAD+)
3a, R = PO32- (NADP)
2b, R = H (1,4-NADH)
2-
3b, R = PO3 , (NADPH)
Fig. 1. Structures of N-benzylnicotinamide bromide, NAD+, NADP+ (1a, 2a, 3a) and N-benzyl-1,4-dihydronicotinamide, NADH, NADPH (1b, 2b,
3b).

J. Lutz et al. / Journal of Organometallic Chemistry 689 (2004) 4783–4790
4785
300
250
200
150
100
50
O2
H2O
OH
OH
R
OH
R
HbpA
NADH
NAD+
R = Ph, 2'-OH-Ph, 2',2'-(OH)2-Ph, F, Cl, Br, Me, Et, Pr, i-Pr, Bu
4a, R= Ph
4b, R= Ph
Fig. 2. Enzymatic reaction and substrates that are utilized by 2-
hydroxybiphenyl 3-monooxygenase (HbpA).
0
0
4
8
12
16
[1b] [10-3 M]
with a monooxygenase enzyme, 2-hydroxybiphenyl
3-monooxygenase (E.C. 1.14.13.44, HbpA), from Pseu-
domonas azelaica HBP1. The HbpA monooxygenase en-
zyme catalyzes the specific ortho-hydroxylation of a
broad range of a-substituted phenols to corresponding
catechols (4b) (Fig. 2) [9]. The oxidizing species of the
enzyme, designated above, is a 4a-hydroperoxo flavin
[10]. The latter species was formed in a reaction se-
quence consisting of hydride transfer from 2b to the oxi-
dized flavin (FAD), followed by reaction with O2, and
proton abstraction to provide the resulting hydroperox-
ide. Thus, we will report on the substitution of 2b by the
simple model compound, 1b, to sustain the catalytic
cycle of HbpA. Furthermore, we evaluated the in situ
regeneration of 1b from 1a using the precatalyst,
[Cp*Rh(bpy)(H2O)](Cl)2 and sodium formate to catalyt-
ically generate, [Cp*Rh(bpy)(H)](Cl), for the regioselec-
tive hydrogenation of 1a to 1b [14].
Fig. 3. Dependence of HbpA activity and stability on the concentra-
tion of reduced NADH model 1b. General conditions: 50 mM KPi-
buffer (pH 7.5), T = 30 ꢁC; [HbpA] = 0.66 · 10ꢀ6 M, [2-hydroxybiphe-
nyl] = 2 · 10ꢀ3 M, 4a, reaction time = 15 min. TNHbpA = turnover
number of HbpA (as determined by the amount of product, 4b,
formed).
As shown in Fig. 3, a Michaelis–Menten like activity
profile was observed with 1b concentrations up to
4 mM. Thus, a Km value of 3.77 mM (0.25 mM 6 [1b] 6
4 mM) was estimated. This corresponds to an approxi-
mately 130-fold decrease in affinity of HbpA towards
the NADH model, 1b, as compared to 27 lM for natu-
ral NADH, and readily explains the lower activity ob-
served for HbpA in presence of 1b. Therefore, at very
high concentrations of 1b, maximum catalytic activity
might be expected. Surprisingly, the opposite effect
was observed, as further increases of 1b concentration
decreased the initial HbpA activity. Furthermore, we
observed that the 1b oxidation rate was far higher than
catechol formation. Similar effects, though to a far lesser
extent, had also been observed with 2b as the reductant,
which was explained by the uncoupling of 2b oxidation
from substrate hydroxylation, leading to the formation
of hydrogen peroxide [10,11].
2. Results and discussion
Preliminary experiments on the specific ortho-
hydroxylation of 4a (2 mM) using HbpA as the enzyme
catalyst (substrate/catalyst ratio (S/C): 8300), with 2 mM
1b as the stoichiometric source of reducing equivalents,
resulted in approximately a 6.3% yield of 4b after 3 h.
Thus, the HbpA monooxygenase enzyme performed
Fig. 4 shows a typical reaction profile of the HbpA-
catalyzed hydroxylation of 4a with 1b as reductant. Less
1.5
1
12
10
8
more than 500 cycles at an average TF of 2.9 minꢀ1
.
Interestingly, and in contrast to the results of Fish and
co-workers [8] with horse liver alcohol dehydrogenase,
only a fraction of the potential enzymatic activity of
HbpA, (896 minꢀ1), was obtained [11].
6
To clarify this phenomenon, we further investigated
the kinetic properties of HbpA with the reduced nicotin-
amide model, 1b. According to the Michaelis–Menten
theory, the enzymatic rate is dependent on the binding
affinity of HbpA for the substrate 1b, as well as on the
maximum rate for substrate conversion (v = vmax Æ S/
(Km + S)). Both parameters (Km, vmax) were expected
to differ between the natural and the biomimetic co-fac-
tor due to the apparent structural differences. Thus, we
examined the influence of varying the concentration of
1b on the conversion rate of 4a to 4b (Fig. 3).
0.5
0
4
2
0
0
1
2
3
4
5
t [h]
Fig. 4. Reaction profile of a typical enzymatic hydroxylation of 2-
hydroxybiphenyl leading to the corresponding catechol (r), with
HbpA promoted by stoichiometric amounts of 1b, h. General
conditions: 50 mM Tris-buffer (pH 7.5), T = 30 ꢁC, [HbpA] = 1 · 10ꢀ6
M, [2-hydroxybiphenyl] = 2 · 10ꢀ3M.

4786
J. Lutz et al. / Journal of Organometallic Chemistry 689 (2004) 4783–4790
than 20% of the reducing equivalents provided to HbpA
by 1b were used for productive O2-activation, i.e.,
hydroxylation of 4a. In the case of NADH, 2b, this va-
lue was greater than 85%.
Therefore, we are suggesting that an increased inter-
action of 1b with FAD, resulted in hydrogen peroxide
formation, rather than in catechol product formation,
4b. Upon reaction with FAD alone, 1b was consumed
approximately 40 times faster than native NADH, 2b.
Therefore, we concluded that hydride transfer from
the reduced pyridine ring to the oxidized FAD alloxa-
zine moiety occurred more efficiently with 1b than with
2b.
The dramatically increased rate of 1b oxidation in
comparison to 2b may be explained by steric, electronic,
and conformational effects that facilitate a hydride
transfer between 1b and FAD. Further investigations
to clarify this mechanistic point will be accomplished
in the future. Based on these results, the interesting
dependency of HbpA activity and the concentration of
1b may be explained by two opposing effects. Firstly,
the formation of reduced FAD at the active site was rate
limiting in the catalytic mechanism of HbpA. Thus,
according to the Michaelis–Menten model, increasing
the concentration of 1b should have increased the rate
of the hydroxylation reaction. Moreover, the increased
rate of formation of FADH2 also supports the nonpro-
ductive generation of hydrogen peroxide (H2O2). We
suggest that catalytically relevant residues of HbpA,
e.g., cysteine SH groups, at the active site of HbpA, were
directly affected by in situ generated H2O2.
reactivity of FDH compared to HbpA might be ex-
plained by the different enzymatic mechanisms of the
FDH reaction for the regeneration of 1b by formate.
For example, Labrou and co-workers described
Lys360 to be involved in molecular recognition of 2a
via the ribose ring [13]. FDH brings 2a and formate in
close proximity via specific non-covalent binding of both
substrates, which stabilizes the transition state. Thus,
the reaction of formate and 2a was found to be acceler-
ated by several orders of magnitude by FDH. In the case
of HbpA, as shown above, the chemical regeneration of
FADH2, which is the reactive species for the monooxyg-
enase enzyme, proceeds very rapidly even in the absence
of enzyme. Therefore, in the case of 1b, HbpA was not
necessary for the reductive half reaction, i.e., the transfer
of hydride to FAD.
The active site of HbpA controls the regiospecific
hydroxylation reaction of the substrate via its proximity
to the FADH hydroperoxide. Recently, Lo and Fish [8]
proposed a possible role for the zinc metal ion center at
the active site of HLADH, which was thought to facili-
tate the stereospecific hydride transfer from 1b to the
achiral ketone substrate for chiral alcohol synthesis.
Moreover, in contrast to HLADH, FDH possesses no
metal centers to which 1a might bind, and therefore,
be in position in proximity to the formate ion to form
1b. More importantly, the formate driven regeneration
of reduced nicotinamide models, 1a–1b or 2a–2b, has
been reported to be catalyzed by the organorhodium hy-
dride, [Cp*Rh(bpy)(H)]+ [14], with catalytic activities in
the range of 11 hꢀ1 being observed (Fig. 5).
Apparently, the use of 1b as a reductant does not
yield the high correlation between consumption of
reducing equivalents and substrate hydroxylation that
was observed with natural NADH, 2a [11]. One plausi-
ble approach to rationalising these results assumes that
the NADH backbone, presumably via multiple H-bond
and other non-covalent interactions, induces major
structural changes of the active site geometry, which
positions the phenolic substrate in close proximity to
the activated 4a-hydroperoxoflavin, thereby minimizing
undesired side reactions [12]. Since 1b lacks the back-
bone of NADH, none of the aforementioned effects pre-
sumably occurs, and the 4a-hydroperoxoflavin becomes
susceptible to hydrolysis to H2O2 in competition with
ortho hydroxylation of substrate. Further experiments
are currently underway to establish this hypothesis.
We next envisioned in situ regeneration of the biomi-
metic co-factor, 1b, using [Cp*Rh(bpy)(H2O)](Cl)2 as
the precatalyst, with formate as the hydrogen source,
to form 1b from 1a [14a]. The usual reduction procedure
used by enzymologists entails the role of formate dehy-
drogenase (FDH) to regenerate native NADH, 2b.
However, even after several hours in the presence of
1 mM 1a, no formation of 1b was detectable, with
FDH as the reducing catalyst. This apparent lack of
Thus, we decided to couple the 1b regeneration as
shown in Fig. 5 with the HbpA catalyzed hydroxylation
of 2-hydroxybiphenyl. We found only small amounts of
product (in the lM range) when combining biomimetic
co-factor regeneration with the enzymatic hydroxylation
reaction. Further analysis revealed that this lack of cate-
chol product could be explained by the loss of activity of
both the organorhodium precatalyst, [Cp*Rh(bpy)-
(H2O)](Cl)2, and the HbpA monooxygenase enzyme un-
der the reaction conditions. Since the general reaction
conditions show that each catalytic species was compar-
atively stable [15,16], we propose an inhibitory interac-
tion between both reactants. In fact, we observed a
deactivation of HbpA in the presence of the precatalyst
[Cp*Rh(bpy)(H2O)](Cl)2, accompanied by the forma-
tion of a yellow precipitate (Table 1).
As shown in Table 1, the extent of decreasing HbpA
activity depended on both the HbpA and the
[Cp*Rh(bpy)(H2O)](Cl)2 concentrations, and suggested
a stoichiometric interaction of both components causing
this inhibition reaction. In a similar situation to the inhi-
bition of HbpA, the [Cp*Rh(bpy)(H2O)](Cl)2 activity
for the reduction of N-benzylnicotinamide bromide,
1a, in the presence of formate ion, was also decreased.
Therefore, we suggest that the potential nucleophilic

J. Lutz et al. / Journal of Organometallic Chemistry 689 (2004) 4783–4790
4787
H2O
O
OH
OH
NH2
HCO2H
+
N
2+
4b
N
N
1a
HbpA
Rh
OH2
O
H
H
NH2
OH
CO2
N
O2
4a
1b
Fig. 5. Tandem co-factor regeneration, using [Cp*Rh(bpy)(H2O)](Cl)2 and formate with 1a (bromide counter ion) to provide 1b, with a flavin
containing and NAD+ dependent, monooxygenase enzyme, 2-hydroxybiphenyl 3-monooxygenase (HbpA), for selective hydroxylation of 2-
hydroxybiphenyl, 4a, to its catechol derivative, 2,3-dhydroxybiphenyl, 4b.
Table 1
Deactivation of HbpA and the precatalyst, [Cp*Rh(bpy)(H2O)](Cl)2,
at different concentrations
0.4
0.3
c([Cp*Rh(bpy)(H2O)](Cl)2)
(mM)
c(HbpA)
(U mlꢀ1
Residual HbpA-activity
(%)a
)
0
0.11
0.11
0.11
100
28
8
0.2
0.1
0
0.02
0.04
0.02
0.22
0.44
1.52
16
42
68
General conditions: KPi buffer (50 mM, pH 7.5), T = 30 ꢁC.
Determined after a 60 min incubation followed by a spectroscopic
assay described previously [16].
0
50
100
150
200
a
Incubation time / min
Fig. 6. Influence of various NHþ4 concentrations on the residual HbpA
activity in the presence of [Cp*Rh(bpy)(H2O)](Cl)2. General condi-
tions: 50 mM phosphate buffer (pH 7.5, T = 30 ꢁC); [HbpA] = 0.39
lM, [[Cp*Rh(bpy)(H2O)](Cl)2] = 0.04 mM, [FAD] = 20 lM,
[NH4+] = 0 (m), 10 (h) 20 (n), 50 (}), and 100 (r) mM. At intervals,
samples were withdrawn, supplemented with NADH (final 0.2 mM)
and 2-hydroxybiphenyl (final 1 mM) and analyzed at k = 340 nm.
residues on HbpA, such as lysines (–NH2) or cysteines
(–SH), may possibly coordinate to the Cp*Rh metal
ion center. A strongly binding amino acid probably will
not be displaced by formate ion, therefore, inhibiting the
rate of the organorhodium hydride regeneration reac-
tion, and may also explain the deactivation of HbpA.
This was demonstrated via the influence of ammonium
ions on the HbpA activity in the presence of
[Cp*Rh(bpy)(H2O)](Cl)2, and is shown in Fig. 6; a clear
decrease in activity was noted with increasing ammo-
nium ion concentrations. Therefore, the peripheral
groups on HbpA could bind to the [Cp*Rh(bpy)-
(H2O)](Cl)2 dication. via aqua ligand displacement,
and possibly affect the ternary structure of HbpA.
Alternatively, we experimented to ascertain if the
nucleophilic functional groups on HbpA could be deriva-
tized, and thus prevent the putative [Cp*Rh(bpy)(H2O)]-
(Cl)2 deactivation reaction by covalently immobilizing
HbpA with Eupergit C, a polymer with epoxide groups
that would react with the –NH2 or –SH groups on the
HbpA enzyme. Using natural NAD+, 2a, as the co-factor,
with in situ regeneration catalyzed by [Cp*Rh(bpy)H]+,
resulted in the expected reactivity of HbpA and hydroxy-
lation of 4a to 4b with a TF of 4 hꢀ1 (Fig. 7); the low TF
being a consequence of presumable mass transfer effects
associated with the polymer supported enzyme. Further-
more, in similar experiments with 1a, only trace amounts
of 4b were observed, and we are now conducting experi-
ments to understand the reasons why there was such
low product formation.

4788
J. Lutz et al. / Journal of Organometallic Chemistry 689 (2004) 4783–4790
4. Experimental
2
4.1. Chemicals and enzymes
1.5
1
All chemicals, benzyl bromide, nicotinamide, sodium
dithionite, NADH, FAD, sodium formate, 2-hydroxybi-
phenyl and 2,3-dihydroxybiphenyl, dioxane, methanol,
acetone and buffers, were obtained from Fluka, as well
as the enzymes Candida boidinii formate dehydrogenase
(EC 1.2.1.2) and bovine liver catalase (EC 1.11.1.6) in
the highest available quality.
0.5
0
The HbpA (EC 1.14.13.44) was partially purified
from recombinant E. coli JM101 (pHBP461) [16,17],
via anionic expanded bed adsorption chromatography
and subsequent desalting, saturation with FAD and dia-
lyzed against 50 mM sodium phosphate buffer of pH 7.5
using a VariPerm L hollow fiber module from Stagroma
(Reinach, Switzerland), and stored at ꢀ20ꢁC. The spe-
cific activity was calculated to be 0.78 U/mg correspond-
ing to approximately 22% purity. UV activity assays
were accomplished as described in the following
reference [11].
0
1
2
3
4
5
6
7
8
t [h]
Fig. 7. Chemoenzymatic hydroxylation of 2-hydroxybiphenyl with
immobilized HbpA and [Cp*Rh(bpy)(H2O)](Cl)2 using formate as the
source of reducing equivalents, and natural NAD+, 2a, as the co-
factor. General conditions: potassium phosphate buffer (50 mM, pH
7.5, T = 25 ꢁC), c(NaHCO2) = 150 mM, c(HbpA) = 0.175 lM (immo-
bilized on 150 mg Eupergit C), c(NAD+) = 0.2 mM, c(catalase) = 50
U mlꢀ1, c([Cp*Rh(bpy)(H2O)](Cl)2) = 0.04 mM (h, 2-hydroxybiphe-
nyl; r, 2,3-dihydroxybiphenyl).
3. Conclusions
4.2. Synthesis of 1a, 1b, and [Cp*Rh(bpy)(H2O)](Cl)2
In the present study, we have extended the use of the
biomimetic NAD+/NADH models, 1a and 1b, from pre-
viously described horse liver alcohol dehydrogenase to
the FAD containing monooxygenase, HbpA [15].
Although the activity of HbpA with the biomimetic
NADH co-factor, 1b, was in the lower percentage range
compared to natural NADH, the ease of chemical syn-
thesis of this biomimic, as well as the increased rates
of hydride transfer from the reduced biomimetic co-fac-
tor, 1b, directly to FAD provides a simple process for in
vitro applications of this and other FAD containing
monooxygenases (e.g., styrene monooxygenase, cyclo-
hexanone monooxygenase).
While formate dehydrogenase, FDH, from Candida
boidinii did not reduce the oxidized biomimic, 1a, the
[Cp*Rh(bpy)(H2O)](Cl)2/formate system appears to
deactivate the HbpA monooxygenase enzyme via bind-
ing to functional groups, such as –NH2 and –SH. This
deactivation process was somewhat circumvented via
derivatization of the enzyme with an epoxide containing
polymer to provide, with co-factor 2a, the hydroxylation
product catechol, 4b, while with biomimic co-factor, 1a,
minor product formation was observed.
The N-benzylnicotinamide bromide (1a) was synthe-
sized by dissolving 10 g nicotinamide in 200 ml 1,4-diox-
ane/50 ml methanol, followed by the addition of 11.6 ml
benzyl bromide, and heating for 5 h at 80 ꢁC under re-
flux. The precipitate was filtered and washed three times
with dioxane, recrystallized from methanol, and filtered
and dried under vaccum and stored at ꢀ20 ꢁC. The bro-
mide anion was not exchanged by precipitaiton with sil-
ver triflate (Yield 64% based on nicotinamide), as was
described by Fish and coworkers [14a]. The reduced
N-benzyl-1,4-dihydronicotinamide (1b) was obtained
by dithionite reduction [18]. Therefore, 2.5 g of 1a, in
presence of 4.6 g sodium carbonate, was dissolved in
60 ml water and reduced by slow addition of sodium
dithionite in a 4-fold molar excess for 2 h. The precipi-
tate was filtered, thoroughly washed with water, dried
(Yield 82%), stored at ꢀ20 ꢁC, and checked by 1H
1
NMR. The H NMR spectrum of 1b corresponds to
the published spectral data [14a]. The 100 mM stock
solutions of 1a and 1b in methanol were prepared fresh.
The [Cp*Rh(bpy)(H2O)](Cl)2 complex was synthesized
as published elsewhere [5c,14c].
In future studies, various strategies will be tested,
such as site directed mutagenesis of nucleophilic amino
acids, protection of exposed amino acid functional
groups by other modifications (acylation, etc.), or direc-
ted evolution approaches, to reduce enzyme deactiva-
tion by [Cp*Rh(bpy)(H2O)](Cl)2, while increasing the
activity towards the reduced biomimic, 1b, should also
allow satisfactory product formation with other chemo-
enzymatic cell-free hydroxylation systems.
4.3. Biotransformation, co-factor regeneration by [CpꢁRh-
ðbpyÞðH2OÞ](Cl)2
In typical biotransformations, either 50 mM potas-
sium phosphate or 50 mM Tris buffer of pH 7.5, was
used. The 2 mM 2-hydroxybiphenyl (4a) (100 mM stock
solution in methanol) and typically 2–4 mM 1b (without
in situ co-factor regeneration) or 2–4 mM 1a in presence

J. Lutz et al. / Journal of Organometallic Chemistry 689 (2004) 4783–4790
4789
of 10–50 lM [Cp*Rh(bpy)(H2O)](Cl)2 and 150 mM so-
dium formate (with in situ co-factor regeneration) were
used at 30 ꢁC and 200–250 rpm. Enzyme concentrations
varied from 1/1000 to 1/10 dilutions of an enzyme stock
solution of 6.1 U/ml and 7.7 mg/ml protein, but mostly
HbpA was diluted 1/15.
The UV/Vis spectra were recorded at 15 min intervals.
No peak formation at 350 nm (characteristic for 1b)
could be detected during 2 h of incubation, while di-
rectly after additon of either 0.5 mM 2a or 10 lM
[Cp*Rh(bpy)(H2O)]2+ to the above assay, the 350 nm
peak was observed.
4.4. Eupergit immobilisation
Acknowledgements
The 2 ml HbpA stock solution and 1ml of potassium
phosphate buffer (pH 7.5) were incubated at 4 ꢁC for 2
days with 0.4 g Eupergitꢂ C. The immoblisation effi-
cency was calculated from UV enzyme assays [16] and
Bradford protein assays [19] of the supernatant, before
and after the immobilisation. Immobilisation efficiencies
were in the range of 20%. After washing the immobili-
sates, they were used in biotransformations on a 3 ml
scale (see above).
We gratefully acknowledge Martin Neuenschwander
for fruitful discussions and for the initial synthesis of
N-benzyl-1,4-dihydronicotinamide, and Philipp Angerer
1
for support with the H NMR experimental measure-
ments. RHF gratefully acknowledges Department of
Energy funding to LBNL from the Advanced Energy
Projects and Technology Research Division, Office of
Computational and Technology Research, under Con-
tract No. DE AC03-76SF00098.
4.5. Analytical techniques
To determine product formation in aqueous solu-
tions, the enzymatic reaction was stopped by the addi-
tion of 0.5% (v/v) perchloric acid and the resulting
precipitate was removed by centrifugation (20000g, 5
min). The samples were analyzed on a LiChroCARTꢂ
column (125-4 RP-18) using a Merck LaChrom D-
7000 high pressure liquid chromatography system
(VWR international AG, Switzerland) running under
isocratic conditions at 40% acetonitrile and 60% water
(0.1% perchloric acid) and 0.75 ml/min flow rate.
References
[1] (a) E.N. Jacobsen, Acc. Chem. Res 33 (2000) 421;
(b) K.B. Sharpless, Chem. Br. 22 (1986) 38.
[2] (a) V. Ullrich, Angew. Chem., Int. Ed. Engl. 11 (1972) 701;
(b) S. Murai, Activation of Unreactive Bonds and Organic
Synthesis, Springer, Verlag, Berlin Heidelberg, 1999;
(c) H.L. Holland, Curr. Opin. Chem. Biol. 3 (1999) 22;
(d) H.L. Holland, H.K. Weber, Curr. Opin. Biotechnol. 11 (2000)
547.
[3] W.A. Duetz, J.B. van Beilen, B. Witholt, Curr. Opin. Biotechnol.
12 (2001) 419.
[4] H.K. Chenault, G.M. Whitesides, App. Biochem. Biotech. 14
(1987) 147.
4.6. Spectroscopy
[5] (a) H.M. Zhao, W.A. van der Donk, Curr. Opin. Biotech. 14
(2003) 583;
The 1b concentrations were measured at 350 nm, and
2b concentrations at 340 nm using a Cary 1E (Varian,
Switzerland) spectrophotometer. The extinciton coeffi-
(b) M.D. Leonida, Curr. Med. Chem. 8 (2001) 345;
(c) F. Hollmann, P.C. Lin, B. Witholt, A. Schmid, J. Am. Chem.
Soc. 125 (2003) 8209;
cient of 1b was determined to be 5.76 l/(mmol cm)ꢀ1
.
R. Mertens, L. Greiner, E.C.D. van den Ban, H.B.C.M. Haaker,
A. Liese, J. Mol. Catal. B: Enzyme 24-25 (2003) 39;
(d) F. Hollmann, A. Schmid, Biocatal. Biotransform. 22 (2004)
63.
Spectra were recorded in quartz cuvettes from 200–600
nm in 15 min intervals.
[6] (a) K. Hofstetter, J. Lutz, I. Lang, B. Witholt, A. Schmid,
Angew. Chem., Int. Ed. 43 (2004) 2163;
4.7. Interaction of 1b or 2b with FAD
(b) F. Zambianchi, P. Pasta, G. Carrea, S. Colonna, N. Gaggero,
J.M. Woodley, Biotechnol. Bioeng. 78 (2002) 488;
(c) U. Schwaneberg, A. Sprauer, C. Schmidt-Dannert, R.D.
Schmid, J. Chromatogr. A 848 (1999) 149.
The 0.2 mM FAD, in the presence of either 0.3 mM
1b or 2b, were incubated at 30 ꢁC in Tris buffer and
the spectra over time were recorded at 15 min intervals.
As a control experiment, FAD, 1b, and 2b were incu-
bated separately. The time course at 350 nm for 1b
and 340 nm for 2b, demonstrated the decreases in con-
centrations of 1b and 2b in the presence of FAD.
[7] L.J. Rover, J.C.B. Fernandes, G. de Oliveira Neto, L.T.
Kubota, E. Katekawa, S.H.P. Serrano, Anal. Biochem. 260
(1998) 50.
[8] H.C. Lo, R.H. Fish, Angew. Chem., Int. Ed. 41 (2002) 478.
[9] A. Schmid, H.-P.E. Kohler, K.-H. Engesser, J. Mol. Catal. B:
Enzyme 5 (1998) 311;
M. Held, A. Schmid, H.-P.E. Kohler, W. Suske, B. Witholt,
M.G. Wubbolts, Biotech. Bioeng. 62 (1999) 641.
[10] W.A. Suske, W.J.H. van Berkel, H.-P.E. Kohler, J. Biol. Chem.
274 (1999) 33355.
4.8. Enzymatic regeneration of 1b by formate dehydrogenase
The 2.5 U formate dehydrogenase was incubated in
the presence of 1 mM 1a and 160 mM sodium formate
in 50 mM Tris buffer, pH 7.5 at 30 ꢁC, in a 1 ml cuvette.
[11] W.A. Suske, M. Held, A. Schmid, T. Fleischmann, M.G.
Wubbolts, H.-P.E. Kohler, J. Biol. Chem. 272 (1997) 24257.

4790
J. Lutz et al. / Journal of Organometallic Chemistry 689 (2004) 4783–4790
[12] (a) A. Meyer, A. Schmid, M. Held, A.H. Westphal, M. Roeth-
lisberger, H.P. Kohler, W.J. van Berkel, B. Witholt, J. Biol.
Chem. 277 (2002) 5575;
(c) E. Steckhan, S. Herrmann, R. Ruppert, E. Dietz, M. Frede,
E. Spika, Organometallics 10 (1991) 1568.
[15] F. Hollmann, B. Witholt, A. Schmid, J. Mol. Catal. B: Enzyme
19–20 (2002) 167.
(b) A. Meyer, Thesis 2002, Swiss Federal Institute of Technology
Zurich, Zurich, 2002.
[16] A. Schmid, I. Vereyken, M. Held, B. Witholt, J. Mol. Catal. B:
Enzyme 11 (2001) 455.
[13] N.E. Labrou, D.J. Rigden, Y.D. Clonis, Eur. J. Biochem. 267
(2000) 6657.
[17] J. Lutz, V.V. Mozhaev, Y.L. Khmelnitsky, B. Witholt, A. Schmid,
J. Mol. Catal. B: Enzyme 19–20 (2002) 177.
[14] (a) H.C. Lo, O. Buriez, J.B. Kerr, M.M. Olmstead, R.H. Fish,
Inorg. Chem. 40 (2001) 6705;
[18] E. Pop, M.E. Brewster, M.J. Huang, N. Bodor, Theochem.-J.
Mol. Struc. 102 (1993) 27.
[19] M.M. Bradford, Anal. Biochem. 72 (1976) 248.
(b) D. Westerhausen, S. Herrmann, W. Hummel, E. Steckhan,
Angew. Chem., Int. Ed. Engl. 31 (1992) 1529;