Type II flavin-containing monooxygenases: A new class of biocatalysts that harbors baeyer-villiger monooxygenases with a relaxed coenzyme specificity

Article abstract of DOI:10.1002/cctc.201300550

Within a newly identified set of flavin-containing monooxygenases (FMOs) from Rhodococcus jostii RHA1, we have identified three monooxygenases (FMO-E, FMO-F, and FMO-G) that are effective in catalyzing Baeyer-Villiger oxidations. These type II FMOs display relaxed coenzyme specificity by accepting both NADPH (reduced form of nicotinamide adenine dinucleotide phosphate) and NADH (reduced form of nicotinamide adenine dinucleotide), as a coenzyme, which is a novel and attractive feature among biocatalysts capable of conducting Baeyer-Villiger oxidations. We purified FMO-E and determined that the Michaelis constants for both coenzymes were in the micromolar range, whereas the activity was highest for NADH. By using the stopped-flow technique, formation of a peroxyflavin-enzyme intermediate was observed, which indicated that type II FMOs follow a catalytic mechanism similar to that of other class B flavoprotein monooxygenases. A set of cyclobutanones and cyclohexanones were used to probe the regio- and enantioselectivity of all three recombinant monooxygenases. The biocatalysts readily accepted small cyclic ketones, which enabled the conversion of previously poorly accepted substrates by other monooxygenases (especially norcamphor), and exhibited excellent and unique regio- and enantioselectivities. Sequence analysis revealed that type II FMOs that act as Baeyer-Villiger monooxygenases contain a unique N-terminal domain. Sequence conservation in this protein domain can be used to identify new NADH-dependent Baeyer-Villiger monooxygenases, which would facilitate future biocatalyst discovery efforts. New kid on the block: Members of a newly recognized group of sequence-related flavin-containing monooxygenases can perform Baeyer-Villiger oxidations. Their coenzyme indifference and unique specificity make them attractive biocatalysts.

Full text of DOI:10.1002/cctc.201300550

CHEMCATCHEM
FULL PAPERS
DOI: 10.1002/cctc.201300550
Type II Flavin-Containing Monooxygenases: A New Class of
Biocatalysts that Harbors Baeyer–Villiger Monooxygenases
with a Relaxed Coenzyme Specificity
Anette Riebel,[a] Michael J. Fink,[b] Marko D. Mihovilovic,[b] and Marco W. Fraaije*[a]
Within a newly identified set of flavin-containing monooxyge-
nases (FMOs) from Rhodococcus jostii RHA1, we have identified
three monooxygenases (FMO-E, FMO-F, and FMO-G) that are
effective in catalyzing Baeyer–Villiger oxidations. These type II
FMOs display relaxed coenzyme specificity by accepting both
NADPH (reduced form of nicotinamide adenine dinucleotide
phosphate) and NADH (reduced form of nicotinamide adenine
dinucleotide), as a coenzyme, which is a novel and attractive
feature among biocatalysts capable of conducting Baeyer–Vil-
liger oxidations. We purified FMO-E and determined that the
Michaelis constants for both coenzymes were in the micromo-
lar range, whereas the activity was highest for NADH. By using
the stopped-flow technique, formation of a peroxyflavin–
enzyme intermediate was observed, which indicated that
type II FMOs follow a catalytic mechanism similar to that of
other class B flavoprotein monooxygenases. A set of cyclobuta-
nones and cyclohexanones were used to probe the regio- and
enantioselectivity of all three recombinant monooxygenases.
The biocatalysts readily accepted small cyclic ketones, which
enabled the conversion of previously poorly accepted sub-
strates by other monooxygenases (especially norcamphor), and
exhibited excellent and unique regio- and enantioselectivities.
Sequence analysis revealed that type II FMOs that act as
Baeyer–Villiger monooxygenases contain a unique N-terminal
domain. Sequence conservation in this protein domain can be
used to identify new NADH-dependent Baeyer–Villiger mono-
oxygenases, which would facilitate future biocatalyst discovery
efforts.
Introduction
Flavoprotein monooxygenases (MOs) represent potent catalytic
tools for the production of high-value chemical building blocks
and pharmaceuticals.[1] These oxygenating biocatalysts are able
to perform efficient chemo-, regio-, and/or enantioselective
oxygenation reactions by using only dioxygen and reducing
equivalents such as NADH [reduced form of nicotinamide ade-
nine dinucleotide (NAD+)] and NADPH [reduced form of nicoti-
namide adenine dinucleotide phosphate (NADP+)].[2] Flavopro-
tein MOs can be classified into six structurally and mechanisti-
cally distinct classes (Figure 1). Upon taking into account the
differences in structural and mechanistic properties of the vari-
ous flavoprotein MO classes, class A and B monooxygenases
are especially interesting for biocatalysis, as they are single-
component enzymes with a typically tightly bound flavin ade-
nine dinucleotide (FAD) cofactor.
Figure 1. Overview of the different flavoprotein monooxygenase classes; B-V
oxidation=Baeyer–Villiger oxidations, FMN=flavin mononucleotide.
Class A flavoprotein MOs have been shown to be specialized
in hydroxylation and epoxidation reactions.[2] Class B typically
covers Baeyer–Villiger and heteroatom oxidations.[2] Careful
analysis of the sequence and enzymatic properties of the latter
class has revealed several distinct subclasses (Figure 1). An ex-
tensively studied subclass is formed by type I Baeyer–Villiger
monooxygenases (BVMOs): these carbonyl- and heteroatom-
oxidizing enzymes are specific towards NADPH as a cofactor.[3]
Three other subclasses of class B flavoprotein MOs are known
to date. Type I flavin-containing monooxygenases (FMOs), also
NADPH dependent, have mainly been studied for their role in
human metabolism.[4] The human proteome contains six type I
[a] A. Riebel, Prof. Dr. M. W. Fraaije
Molecular Enzymology group
University of Groningen
Nijenborgh 4, 9747 AG Groningen (The Netherlands)
Fax: (+31)503634165
E-mail: m.w.fraaije@rug.nl
[b] Dr. M. J. Fink, Prof. Dr. M. D. Mihovilovic
Institute of Applied Synthetic Chemistry
Vienna University of Technology
Getreidemarkt 9/163-OC, 1060 Vienna (Austria)
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/cctc.201300550.
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2014, 6, 1112 – 1117 1112

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
FMO isoforms, all of which have been reported as important in
the oxidative degradation of nitrogen- and sulfur-containing
compounds. Only recently, some microbial representatives
were discovered and shown to represent interesting biocata-
lysts for oxidizing secondary amines, including the conversion
of indole into indigo blue.[5] Another class B subclass contains
microbial MOs that hydroxylate primary amines, for example,
lysine N6-hydroxylase and ornithine N5-hydroxylase. In refer-
ence to their reactivity, these are named N-hydroxylating MOs
(NMOs). All together, these biocatalyst subclasses share a pro-
nounced preference towards NADPH as a cofactor.
Kinetic characterization of FMO-E
Using purified FMO-E, steady-state kinetic parameters were de-
termined for ketone 1, NADH, and NADPH (Table 1). For both
coenzymes and the ketone, typical Michaelis–Menten behavior
was observed, which yielded the catalytic (kcat) and Michaelis
Table 1. Steady-state kinetic parameters of FMO-E for NADPH, NADH,
and 1.
Substrate/coenzyme
KM [mm]
kcat [sꢀ1
]
Recently, we identified a new group of sequence-related en-
zymes that belong to class B flavoprotein MOs: type II FMOs
(Figure 1). We reported on seven type II FMOs derived from
the genome of Rhodococcus jostii RHA1.[6] Their unique se-
quence profiles as well as their relaxed coenzyme specificity le-
gitimated a classification separate from the other class B flavo-
protein MOs, together with an already previously discovered
MO of Stenotrophomonas maltophilia.[7] Within these seven
newly found type II FMOs, however, we found significant differ-
ences concerning substrate acceptance. FMO-E, FMO-F, and
FMO-G efficiently catalyzed Baeyer–Villiger oxidation of two ke-
tones {i.e., bicyclo[3.2.0]hept-2-en-6-one (1) and phenylacetone
(3)}, a catalytic feature that was so far only known for type I
BVMOs. However, in marked contrast to type I BVMOs, all
type II FMOs were shown to be able to use NADPH or NADH
indifferently. This seems particularly interesting for combining
such enzymes with other redox biocatalysts in multistep bio-
transformation cascades and thus encouraged us to perform
an in-depth study of the newly identified type II FMOs that can
act as bona fide BVMOs.
bicyclo[3.2.0]hept-2-en-6-one (1):
19.8
2.7
NADPH
NADH
<0.0010
0.0054
2.7
4.3
constants (KM). Even though FMO-E accepts both nicotinamide
coenzymes, the enzyme appears to have a higher affinity (KM)
for NADPH. Yet, the value of KM for NADH is also in the low mi-
cromolar range, and kcat for NADH is higher than that for
NADPH (4.3 vs. 2.7 sꢀ1). This indicates that FMO-E is an efficient
enzyme with NADH as the coenzyme, which is in contrast to
most other class B flavoprotein MOs. Also, all other type II
FMOs have shown coenzyme indifference.[6] The determined
kcat values also confirm that FMO-E is a potent biocatalyst, as
the rate of catalysis is in the same range as that found for
other class B flavoprotein MOs (kcat values for type I BVMOs
phenylacetone MO[8] and cyclohexanone MO[9] are typically in
the range of 1–20 sꢀ1). It is particularly noteworthy that the kcat
values determined for type II FMO from S. maltophilia[7] are two
orders of magnitude lower (0.029 sꢀ1), and this makes the title
enzyme a more likely candidate for biocatalytic processes. In-
terestingly, for both bacterial monooxygenases the rate of cat-
alysis was highest if NADH was used. This may hint to a role of
the coenzyme in the rate-limiting step of the kinetic mecha-
nism of this MO. For type I BVMO cyclohexanone MO, it has
been shown that NADP+ release limits the rate of catalysis.[9]
To establish whether FMO-E follows a catalytic mechanism
similar to that of type I FMOs and type I BVMOs, stopped-flow
experiments were performed. The kinetic mechanism of flavo-
protein MOs can be subdivided into reductive and oxidative
half-reactions. Reduction of the FAD cofactor by the nicotina-
mide coenzyme is considered as the reductive half-reaction.
Anaerobic reduction of FMO-E by NADPH was monitored over
time by using a stopped-flow instrument, which revealed a rela-
tively high rate of reduction (17.5 sꢀ1) at a saturating NADPH
concentration.
Results and Discussion
Within this work, mechanistic, kinetic, and structural aspects of
type II FMO-catalyzed Baeyer–Villiger oxidations were investi-
gated by using FMO-E as a model enzyme. Biocatalytic perfor-
mance was then assessed for all three biocatalysts.
Purification of FMO-E
To allow efficient one-step purification, we produced FMO-E,
FMO-F, and FMO-G with an N-terminal strep-tag (in pBADNS).
Purification trials revealed that only FMO-E could be isolated in
good yields. The other MOs were found to suffer from FAD
loss during the purification procedure. As all three biocatalysts
are sequence related, we decided to use FMO-E as a prototype
for biochemical characterization. From a 1 L culture, 25 mg of
pure FMO-E could be obtained. The purified enzyme displays
In the next step, the reduced flavin of class B flavoprotein
MOs is able to swiftly react with molecular oxygen typically to
yield a relatively stable peroxyflavin intermediate. The ability to
stabilize such a reactive enzyme intermediate in the absence
of a substrate has been the typical hallmark of this biocatalyst
class. For monitoring the above-described oxidative half-reac-
tion, NADPH-reduced FMO-E was rapidly mixed with aerated
buffer. Analysis of the obtained spectra revealed that the data
a
typical flavin absorbance spectrum, with characteristic
maxima at 451 and 381 nm. The extinction coefficient was de-
termined as 11.9 mmꢀ1 cmꢀ1 (e451nm). Furthermore, from the
measured absorbance ratio A280/A451 =10, it was concluded
that the protein was purified as a holoenzyme.
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2014, 6, 1112 – 1117 1113

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
In type I BVMOs, a strictly conserved arginine is crucial for
catalysis; in type I FMOs, an asparagine residue serves such
a role, but in the model of FMO-E no such active residue was
identified in the homologous structural positions. Yet, a histi-
dine and an arginine residue were found to be positioned
close to the isoalloxazine moiety of the FAD cofactor, possibly
assisting in catalysis. Both residues are conserved among
type II FMO sequences. Unfortunately, the protein sequences
of FMO-E and the type II FMO from S. maltophilia[7] are too dis-
similar to locate the mutated residues in the structure of the
latter monooxygenase. To verify whether His222 and Arg563
serve a role in catalysis, the corresponding alanine mutants
were prepared. The first evidence of changed catalytic capaci-
ties of the mutants could directly be seen upon growing cul-
tures under optimal expression conditions. For wild-type FMO-
E, it was observed that the color of the growth medium
changed significantly. This was due to the formation of a blue
pigment that was formed upon oxidative conversion of indige-
nous indole into indigo blue.[6] Expression of the mutants did
not result in any color change in the medium. Furthermore,
purification of the mutant enzymes resulted in only inactive
apoprotein, and the addition of FAD to the apoprotein sam-
ples did not restore any activity. Tests with cleared cell extracts
containing the different mutant enzymes showed that the ac-
tivity was affected dramatically: conversion of thioanisole (2)
dropped form 56[6] to only 6%, whereas phenylacetone (3) and
bicyclo[3.2.0]heptenone (1) were not converted at all under
the standard assay conditions. These results show that the tar-
geted residues are essential for proper FAD binding and, fur-
thermore, provide no conclusive evidence on whether the resi-
dues are important for catalysis.
Figure 2. Deconvoluted flavin absorbance spectra upon mixing FMO-E with
air-saturated buffer. The spectra were reconstructed by using the model A!
B!C!D with rates of 7.2, 1.2, and 0.5 sꢀ1, respectively. A) Reduced state,
B) first peroxyflavin intermediate, C) second peroxyflavin intermediate,
D) oxidized state.
could be best fitted with an irreversible three-step model (A!
B!C!D, Figure 2). Directly after mixing with oxygen, spe-
cies B is rapidly formed with a rate of 7.2 sꢀ1, as evidenced by
an absorbance band at approximately 365 nm. This perfectly
matches the spectral features of a peroxyflavin intermediate
and confirms that FMO-E operates through a catalytic mecha-
nism similar to that of other class B flavoprotein MOs. This in-
termediate (spectrum B) undergoes slow and modest spectral
change (1.2 sꢀ1) to form a less-intense absorbance band (spec-
trum C). This transition may represent a change in the protona-
tion state of the peroxyflavin or a conformational change and
has also been observed for type I BVMOs.[10,11] In a subsequent
slow process (0.5 sꢀ1), the flavin is oxidized (spectrum D),
which is accompanied by the formation of hydrogen peroxide.
These kinetic and spectral properties closely resemble the ki-
netic behavior of other class B flavoprotein MOs.
Intriguingly, on the protein sequence level, the type II FMOs
described in this paper all share an extra N-terminal extension
of approximately 160 residues in comparison with the other
four rhodococcal type II FMOs and the type II FMO from S. mal-
tophilia. This aberrant sequence feature coincides with the po-
tency to efficiently catalyze Baeyer–Villiger oxidations. The mo-
lecular basis for this correlation is unclear. However, it may be
used to identify new type II FMOs that can be exploited as
Structural analysis of FMO-E
To obtain a better view on how FMO-E is able to catalyze
Baeyer–Villiger oxidations, we compared a structural model of
FMO-E with the crystal structures of a type I BVMO (i.e.,
PDB 1W4X)[12] and a type I FMO (i.e., PDB 2VQ7).[13] Comparison
of the active sites clearly showed that all three enzymes
employ different strategies to catalyze such oxygenations
(Figure 3).
NADH-driven biocatalysts for Baeyer–Villiger oxidations.
A
BLASTP search by using the N-terminal sequence of FMO-E
(170 residues) resulted in the identification of 545 type II FMOs
containing the typical N-terminal extension. All these hits dis-
played >25% sequence identity in the N terminus, and they
also contain the typical GxGxxG Rossmann fold motifs in the
C-terminal region, which confirmed that they are FAD-contain-
ing NAD(P)H-dependent monooxygenases. These putative MOs
are mainly found in fungal genomes and none of them has
been explored as a biocatalyst so far.
Biocatalytic exploration
As type II FMOs are fundamentally different from type I
BVMOs, we set out to explore the substrate acceptance range
and stereoselective properties of all three FMOs studied as
whole-cell biocatalysts. For this, 14 ketones [6 prochiral cyclo-
butanones 4–9, 4 prochiral cyclohexanones 10–13, 3 racemic
Figure 3. Comparison of the active site architectures of a) PDB:1W4X as
a type I BVMO, b) 2VQ7 as a type I FMO, and c) FMO-E as the type II FMO.
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2014, 6, 1112 – 1117 1114

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
fused cyclobutanones 1, 14, 15, and racemic norcamphor (16)]
were tested as substrates (see Tables 2 and 3 for their structur-
al formulas).
preference in all cases, which enabled the synthesis of both
optical antipodes in enantioenriched form.
Motivated by the good results from prochiral cyclobutanone
transformations, we decided to test the performance of these
new Baeyer–Villiger catalysts in another reaction mode: regio-
divergent conversion of racemic fused cycloketones 1 and 14–
16. Gratifyingly, the MOs were capable of converting all four
tested ketones (Table 3), which resulted in preferential forma-
tion of normal lactones (in this case, the rearrangement was
governed by higher nucleophilicity, that is, a higher degree of
substitution of the migrating
Desymmetrization of prochiral cyclobutanones and cyclohex-
anones revealed a clear preference (Table 2). All three monoox-
ygenases efficiently converted most cyclobutanones, whereas
the cyclohexanones were poorly accepted as substrates or not
at all. The corresponding lactones were formed with moderate
to good enantioselectivity (from 5 to 79%ee, median 45%ee).
Interestingly, type II FMOs displayed complementary enantio-
carbon atom). This kind of regio-
selectivity is known from type I
Table 2. Desymmetrization of cyclobutanones and cyclohexanones by FMO-E, FMO-F, and FMO-G by using
BVMOs of the cyclopentanone
monooxygenase cluster.[15] Most
interestingly, the title FMOs are
the only Baeyer–Villiger biocata-
lysts known to be able to per-
form kinetic resolution of nor-
camphor (16) to the normal lac-
tone with both excellent stereo-
selectivity and yield. The best re-
whole cells.[a]
Compound Substrate
structure
R
FMO-E
Rel. conv. ee
FMO-F
Rel. conv. ee
FMO-G
Rel. conv. ee
[%]
[%]
[%]
[%]
[%]
[%]
4
5
6
7
8
9
Bn
Ph
pClC6H4
pMeOC6H4CH2
3,4-OCH2O-C6H3CH2
3,4,5-(MeO)3C6H2CH2
>99
>99
>99
98
73
10
16 (S)
69 (R)
79 (R)
5 (S)
30 (R)
13 (R)
90
96
76
75
83
n.c.
40 (R) 95
78 (S) 93
64 (S) 67
18 (S) 99
50 (S) 40
50 (R)
67 (S)
47 (S)
42 (R)
40 (S)
n.a.
n.a.
n.c.
ported
enzyme
for
this
transformation, type I BVMO cy-
clododecanone MO from Rhodo-
coccus ruber SC1,[16] catalyzes the
formation of both regioisomers
of norcamphor with excellent
enantioselectivity (>98%), but
in almost equimolar ratio (57:43
10
11
12
Me
CO2Et
Ph
n.c.
12
n.c.
n.a.
n.c.
n.a.
n.a.
n.a.
n.c.
n.c.
n.c.
n.a.
n.a.
n.a.
99 (+)[b] n.c.
n.a.
n.c.
13
n.c.
n.a.
n.c.
n.a.
n.c.
n.a.
[a] The relative conversion (Rel. conv.) and ee values of the lactone products were determined by chiral-phase
GC-FID after a biotransformation time of 24 h. The absolute configuration or sign of the optical rotation of the
lactone products was assigned on the basis of published reference biotransformations.[14] n.a.=Not applicable,
n.c.=no conversion. [b] Absolute configuration of the product is not known; the sign of the optical rotation is
given instead.
normal/abnormal)
and
at
a rather low rate. Specifically,
FMO-F and FMO-G display excel-
lent enantioselectivities (E>200)
and regioselectivities (>99:1 for
Table 3. Regiodivergent oxidation of racemic fused cyclobutanones rac-1, 14, 15 and norcamphor (rac-16) by FMO-E, FMO-F, and FMO-G by using whole
cells.[a]
Compound Substrate structure
FMO-E
FMO-F
FMO-G
Rel. conv. [%] eeN [%]
EN
Rel. conv. [%] eeN [%]
EN
Rel. conv. [%] eeN [%]
EN
N/ABN
eeABN [%]
EABN N/ABN
eeABN [%]
EABN
N/ABN
eeABN [%]
EABN
97
14:86
46 (3aR,6aS) n.d.
3 (3aR,6aS) n.d.
92
12:88
19 (3aS,6aR)
6 (3aR,6aS)
n.d.
n.d.
81
20:80
21 (3aS,6aR)
rac
n.d.
n.d.
1
81
95:5
39 (3aS,7aS) n.d.
98 (3aR,7aS) n.d.
77
92:8
44 (3aS,7aS)
71 (3aR,7aS)
137
59
91
92:8
18 (3aS,7aS)
75 (3aR,7aS)
n.d.
n.d.
14
>99
>99:1
rac
n.a.
n.d.
n.a. >99:1
77
91[b]
n.d.
99
99
rac
n.a.
n.d.
n.a.
15
16
n.d. >99:1
14
>99:1
94 (1S,5R)
n.a.
36
22
>99 (1S,5R)
n.a.
>200
n.a.
47
99:1
99 (1S,5R)
n.d.
>200
n.a.
n.a. >99:1
[a] The relative conversion (Rel. conv.) and ee values of the lactone products were determined by chiral-phase GC-FID after a biotransformation time of
24 h. Selectivity values E in kinetic resolution-type reactions were calculated according to Sih’s equation: E=[ln(1ꢀeeS)ꢀln(1+eeS/eeP)]/[ln(1+eeS)ꢀln(1+eeS/
eeP)], in which S=substrate and P=product. N=normal lactones, ABN=abnormal lactones are given in italics. Absolute configuration or sign of the opti-
cal rotation of the lactone products was assigned on the basis of published reference biotransformations.[15] n.a.=Not applicable, n.d.=not determined,
rac=racemic. [b] Product regioisomers were not separable by preparative low-pressure chromatography; for assignment of the regio- and stereoisomers,
see Ref. [15].
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2014, 6, 1112 – 1117 1115

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
normal/abnormal lactone) for the oxygenation of this natural
compound.
catalytic residues that are employed in type I BVMOs and
type I FMOs. By site-directed mutagenesis, we determined that
His222 and Arg563 are crucial for the functioning of FMO-E.
Elucidation of the structure of FMO-E would shed light on the
structural and catalytic properties of this NAD(P)H-dependent
monooxygenase. At this point, the striking discovery of the N-
terminal extension in Baeyer–Villiger active FMOs can serve as
a search template in biocatalytically uncharted space: many
more type II FMOs await exploration and future research will
reveal how they operate at the molecular level.
Clearly, the studied type II FMOs display biocatalytic features
that set them apart from type I BVMOs. Although the enantio-
selectivity is far from pharmaceutical standards in many cases
and higher substrate titers were not investigated within this
project, these enzymes nevertheless represent a new and
promising group of oxidative biocatalysts.
Conclusions
Within the newly described class of type II FMOs, we identified
three FMOs that could efficiently catalyze Baeyer–Villiger oxida-
tions. The exemplary kinetic analysis of FMO-E revealed that it
is catalytically competent by displaying high affinities towards
both nicotinamide coenzymes NADH and NADPH with signifi-
cantly higher catalytic rates than that of a recently reported
type II FMO from S. maltophilia. The pre-steady-state kinetic
analysis of FMO-E revealed a catalytic mechanism that was sim-
ilar to that of other class B MOs:
Experimental Section
Reagents and enzymes
Oligonucleotide primers were purchased from Sigma, dNTPs and
the In-Fusion 2.0 CF Dry-Down PCR Cloning Kit were purchased
from Clontech, and Phusion polymerase was purchased from Finn-
zymes. All other chemicals were obtained from Acros Organics,
ABCR, Sigma–Aldrich, TCI Europe, and Roche Diagnostics GmbH.
Bacterial strains and Plasmids
1) Oxidized FMO-E is rapidly reduced by a reduced coenzyme
without the need for substrate binding.
Escherichia coli TOP10 from Invitrogen was used as a host for DNA
manipulations and protein expression. The used expression vector
was a modified pBAD vector (pBADN) in which the NdeI site was
replaced by the original NcoI site.[19] For this study, we introduced
a strep-tag binding site (TGG AGC CAC CCG CAG TTT GAA AAA)
followed by a TEV protease cleavage site (GAG AAT TTA TAT TTT
CAA GGT) between the promoter and the start codon of the gene
to be cloned to yield the pBADNS vector.
2) The reduced FMO-E rapidly reacts with molecular oxygen.
3) FMO-E is able to form a stable peroxyflavin intermediate.
For all other studied class B monooxygenases, it was shown
that the oxidized nicotinamide coenzyme remained bound to
the active site until the peroxyflavin had oxygenated the sub-
strate. Only then was it released, often in a relatively slow ki-
netic event. Such a scenario fits with the steady-state and pre-
steady-state kinetic data of FMO-E in which NADP+/NAD+ was
released in a relatively slow process. More detailed kinetic and
structural studies will settle this issue.
Sequence analysis, cloning, expression, and protein
purification
The NCBI server (blast.ncbi.nlm.nih.gov/Blast.cgi) was used for DNA
sequence retrieval and BLAST searches. The EBI server (www.ebi.a-
c.uk/Tools/msa/clustalw2/) was used for multiple sequence align-
ment by CLUSTALW.[19] The FMO-E sequence was used as input to
prepare a structural model by using the CPH modeling server
(www.cbs.dtu.dk/services/CPHmodels/).[20] The model was made on
the basis of the structure of phenylacetone MO (PDB 1W4X).
A major disadvantage of the well-studied type I BVMOs is
their high specificity for NADPH, a relatively expensive cofactor
in comparison to NADH. Moreover, attempts to engineer
a NADH-accepting type I BVMO have so far not been success-
ful.[17,18] Given that the studied FMO-E, FMO-F, and FMO-G en-
zymes can act as BVMOs with NADH as the coenzyme, they
represent highly sought-after biocatalysts. This way, a signifi-
cantly better balanced exploitation of the available reduction
equivalents within a metabolically active cell may be achieved,
which would alleviate the metabolic burden of such systems.
Additionally, higher compatibility and flexibility may be ach-
The target genes were amplified by PCR by using genomic DNA of
R. jostii RHA1 as a template, and they were subsequently cloned
into pBADNS by using the In-Fusion PCR Cloning kit by following
the recommendations of the manufacturer. The mutations were in-
troduced through the QuikChange Site-Directed Mutagenesis Kit
from Stratagene by following the recommendations of the
manufacturer.
ieved upon designing redox cascades in
environment.
a
cell-free
This study showed that the tested type II FMOs could
indeed be used as efficient biocatalysts. By testing a set of
small cycloketones we unveiled new and excellent stereoselec-
tivities; for example, the conversion of norcamphor (16) by
FMO-F proceeds with >99% regioselectivity and >99% enan-
tioselectivity. Such selectivity has not been observed before
with previously identified BVMOs. This may be explained by
the fact that type II FMOs are only distantly related to type I
BVMOs. As a consequence, they will exhibit totally different
active-site architectures. In fact, FMOs seem to lack the typical
For expression, cells were grown in baffled flasks with lysogeny
broth (LB) medium (800 mL) supplemented with ampicillin and ara-
binose (50 mgmLꢀ1) concentrations of 0.002% (FMO-E and FMO-G)
and 0.2% (FMO-F) by shaking at 200 rpm at 248C for 32 h, based
on the optimal expression conditions previously determined.[6] Cell
extracts were prepared by sonication with subsequent centrifuga-
tion to obtain the cleared cell extracts. Purification of the enzymes
from the cleared cell extracts was done with Strep-Tactinꢁ Sephar-
ose from IBA GmbH by following the recommendations of the
manufacturer. As a minor modification, FAD (10 mm), dithiothreitol
(DTT; 1.0 mm), and 10% glycerol were added to the buffer.
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2014, 6, 1112 – 1117 1116

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
reference biotransformations for substrates 4–9,[14] for substrate
11,[16] and for substrates 1 and 14–16.[15]
Determination of kinetic parameters of FMO-E
Activities of purified FMO-E were determined spectrophotometri-
cally by monitoring the increase or decrease in NADPH or NADH in
time at 340 nm. The reaction mixture (1.0 mL) typically contained
50 mm Tris/Cl, pH 7.5 (10% glycerol, 1.0 mm DTT, 1.0 mm EDTA,
10 mm FAD), 150 mm coenzyme NADPH or NADH, 50 mm ketone
substrate, 5.0% dioxane, and 0.10 mm enzyme at 258C. Kinetic pa-
rameters were obtained by fitting the data as previously
described.[21]
Acknowledgements
This research was financially supported by the Integrated Biosyn-
thesis Organic Synthesis (IBOS) program of the Netherlands Or-
ganisation for Scientific Research (NWO) and the European
Union’s Seventh Framework Programme (FP7/2007–2013) under
grant agreement n8212281 (project OXYGREEN).
Stopped-flow experiments
The reductive and oxidative half-reactions of FMO-E were both an-
alyzed by using an Applied Photophysics SX17 MV stopped-flow
instrument equipped with a photodiode array detector. All experi-
ments described were performed at 258C in 50 mm Tris/Cl (10%
glycerol, 1.0 mm DTT, 1.0 mm EDTA) at pH 7.5. All concentrations
stated are those in the reaction chamber after mixing. Anaerobic
conditions were achieved by flushing the system and solutions
with N2 and removing trace amounts of oxygen upon addition of
1.0 mm putrescine and a catalytic amount of putrescine oxidase.
The rate of reduction of FMO-E by NADPH was assessed under
anaerobic conditions by adding equimolar amounts of NADPH.
The reoxidation of the reduced flavin was achieved by mixing
NADPH-reduced FMO-E with aerated buffer and following the
spectral changes of the flavin cofactor in time. The obtained spec-
tra were analyzed by means of numerical integration methods by
using Pro-K (Applied Photophysics Ltd.), which yielded the ob-
served rate constants.
Keywords: Baeyer–Villiger monooxygenases · biocatalysis ·
cofactors
·
enantioselectivity
·
flavin-containing
monooxygenases
[1] D. E. Torres PazmiÇo, M. Winkler, A. Glieder, M. W. Fraaije, J. Biotechnol.
2010, 146, 9–24.
[2] W. J. H. van Berkel, N. M. Kamerbeek, M. W. Fraaije, J. Biotechnol. 2006,
124, 670–689.
[3] G. de Gonzalo, M. D. Mihovilovic, M. W. Fraaije, ChemBioChem 2010, 11,
2208–2231.
[4] D. M. Ziegler, Trends Pharmacol. Sci. 1990, 11, 321–324.
[5] A. Rioz-Martꢂnez, M. Kopacz, G. de Gonzalo, D. E. Torres PazmiÇo, V.
Gotor, M. W. Fraaije, Org. Biomol. Chem. 2011, 9, 1337.
[6] A. Riebel, G. de Gonzalo, M. W. Fraaije, J. Mol. Catal. B 2013, 88, 20–25.
[7] C. N. Jensen, J. Cartwright, J. Ward, S. Hart, J. P. Turkenburg, S. T. Ali,
M. J. Allen, G. Grogan, ChemBioChem 2012, 13, 872–878.
[8] M. W. Fraaije, J. Wu, D. P. H. M. Heuts, E. W. van Hellemond, J. H. L. Spel-
berg, D. B. Janssen, Appl. Microbiol. Biotechnol. 2005, 66, 393–400.
[9] D. Sheng, D. P. Ballou, V. Massey, Biochemistry 2001, 40, 11156–11167.
[10] I. A. Mirza, B. J. Yachnin, S. Wang, S. Grosse, H. Bergeron, A. Imura, H.
Iwaki, Y. Hasegawa, P. C. K. Lau, A. M. Berghuis, J. Am. Chem. Soc. 2009,
131, 8848–8854.
[11] R. Orru, H. M. Dudek, C. Martinoli, D. E. Torres PazmiÇo, A. Royant, M.
Weik, M. W. Fraaije, A. Mattevi, J. Biol. Chem. 2011, 286, 29284–29291.
[12] E. Malito, A. Alfieri, M. W. Fraaije, A. Mattevi, Proc. Natl. Acad. Sci. USA
2004, 101, 13157–13162.
[13] A. Alfieri, E. Malito, R. Orru, M. W. Fraaije, A. Mattevi, Proc. Natl. Acad.
Sci. USA 2008, 105, 6572–6577.
Bioconversions with cleared cell extracts
Conversions were performed as previously described.[6] For deter-
mining the exact concentration of each enzyme in the respective
extract, a recently developed method was used that relies on the
decrease in the absorbance at 450 nm upon NADPH-mediated re-
duction of the flavin cofactor.[22]
[14] F. Rudroff, J. Rydz, F. H. Ogink, M. Fink, M. D. Mihovilovic, Adv. Synth.
Catal. 2007, 349, 1436–1444.
Whole-cell conversions
[15] M. D. Mihovilovic, P. Kapitꢃn, P. Kapitꢃnovꢃ, ChemSusChem 2008, 1,
143–148.
Bioconversion of a set of 14 target compounds was performed by
using cells expressing FMO-E, FMO-F, and FMO-G by employing the
above-mentioned expression conditions. Cells were grown with
optical density OD590 =0.2–0.6 at 378C. The cultures were then
cooled to 248C and l-arabinose was added to the appropriate final
concentration. Cyclodextrin (4.0 mm) was supplemented as a cell-
membrane transfer agent. This mixture was thoroughly shaken and
then divided into aliquots (1.0 mL) in a 24-well plates. Substrates
were added as solutions in 1,4-dioxane (5.0 mL, 0.8m, 5.0 mm final
concentration), and the plates were sealed with adhesive film
(EasySEAL, Greiner Bio One) and incubated in an orbital shaker
(Infors HT Multitron 2) at 248C for 24 h. Analytical samples were
prepared by extraction of the biotransformation culture (0.5 mL)
with EtOAc (1.0 mL, supplemented with 1.0 mm methyl benzoate
as internal standard) after centrifugal separation of the cell mass
(ꢁ15 kRCF, 1 min, RT). All transformations were performed as tech-
nical triplicates; conversion and selectivity are reported as arith-
metic mean values. Absolute configuration or optical rotation
values of lactone products were assigned on the basis of published
[16] M. J. Fink, D. V. Rial, P. Kapitꢃnovꢃ, A. Lengar, J. Rehdorf, Q. Cheng, F.
Rudroff, M. D. Mihovilovic, Adv. Synth. Catal. 2012, 354, 3491–3500.
[17] H. M. Dudek, D. E. Torres PazmiÇo, C. Rodrꢂguez, G. Gonzalo, V. Gotor,
M. W. Fraaije, Appl. Microbiol. Biotechnol. 2010, 88, 1135–1143.
[18] N. M. Kamerbeek, M. W. Fraaije, D. B. Janssen, Eur. J. Biochem. 2004, 271,
2107–2116.
[19] J. D. Thompson, D. G. Higgins, T. J. Gibson, Nucleic Acids Res. 1994, 22,
4673–4680.
[20] M. Nielsen, C. Lundegaard, O. Lund, T. N. Petersen, Nucleic Acids Res.
2010, 38, W576–81.
[21] D. E. Torres PazmiÇo, R. Snajdrova, B.-J. Baas, M. Ghobrial, M. D. Mihovi-
lovic, M. W. Fraaije, Angew. Chem. 2008, 120, 2307–2310; Angew. Chem.
Int. Ed. 2008, 47, 2275–2278.
[22] A. Riebel, H. M. Dudek, G. Gonzalo, P. Stepniak, L. Rychlewski, M. W.
Fraaije, Appl. Microbiol. Biotechnol. 2012, 95, 1479–1489.
Received: July 10, 2013
Revised: August 19, 2013
Published online on October 7, 2013
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2014, 6, 1112 – 1117 1117