Expression of recombinant human flavin monooxygenase and moclobemide-N-oxide synthesis on multi-mg scale

Article abstract of DOI:10.1039/c2cc17878h

A panel of human flavin monooxygenases were heterologously expressed in E. coli to obtain ready-to-use biocatalysts for the in vitro preparation of human drug metabolites. Moclobemide-N-oxide (65 mg) was the first high-priced metabolite prepared with recombinant hFMO3 on the multi-milligram scale.

Full text of DOI:10.1039/c2cc17878h

View Article Online / Journal Homepage / Table of Contents for this issue
Dynamic Article Links
ChemComm
Cite this: Chem. Commun., 2012, 48, 6001–6003
www.rsc.org/chemcomm
COMMUNICATION
Expression of recombinant human flavin monooxygenase and
moclobemide-N-oxide synthesis on multi-mg scalew
Steven P. Hanlon,a Andrea Camattari,b Sandra Abad,b Anton Glieder,b Matthias Kittelmann,c
Stephan Lu
¨
tz,c Beat Wirza and Margit Winkler*b
Received 16th December 2011, Accepted 25th April 2012
DOI: 10.1039/c2cc17878h
A panel of human flavin monooxygenases were heterologously
expressed in E. coli to obtain ready-to-use biocatalysts for the
in vitro preparation of human drug metabolites. Moclobemide-
N-oxide (65 mg) was the first high-priced metabolite prepared
with recombinant hFMO3 on the multi-milligram scale.
as analytical references. Ideally, this can be done with readily
available i.e. heterologously expressed human enzymes
in vitro.
Generally, the over-expression of membrane associated
enzymes is still a bottleneck in biotechnology. The membranes’
capacities are limited and the presence of excessive amounts
of a particular recombinant protein may destroy membrane
functionality, and therefore cell viability. Nevertheless, human
FMO3 and 5 were functionally expressed in E. coli.9 hFMO1
and 2, by contrast, were expressed in the Baculovirus expres-
sion system10 and are commercially available as ‘supersomes’
(BD biosciences). Attempts towards high yielding expression of
soluble hFMOs in E. coli have also been made for e.g. hFMO3
by fusion with maltose binding protein.11
Oxidative biotransformations of drugs are often catalyzed by
cytochrome P450 (CYP) enzymes and for this reason extensive
efforts have been made to employ recombinant CYPs as
biocatalysts for the preparation of drug metabolites.1,2 Less
attention has so far been paid to flavin monooxygenases
(FMOs), even though a large number of drug substances are
known to be FMO substrates, for example tamoxifen3 and
cimetidine.4 FMOs are membrane associated proteins that
oxidize substrates containing soft nucleophiles such as nitrogen
or sulphur at the expense of reducing equivalents (NADPH)
and molecular oxygen.5 Six human FMO isoforms (hFMO)
have been described to date (hFMO1–hFMO6). hFMO3 is the
most abundant non-CYP metabolizing enzyme variant in
adult human liver whereas hFMO1 is found in fetal liver.6
The majority of hFMO2 alleles yield inactive protein due to a
premature stop codon and FMO6 genes appear to be non-
functionally transcribed.7
In order to get a full spectrum of hFMOs in a convenient
host, the complete set of genes (native and codon optimized)
was cloned into the pEamTA or its parental pMS470 plasmid.12
All sequences were verified by sequence analysis (LGC
Genomics). For expression experiments, E. coli BL21 DE3
Gold were transformed with the corresponding plasmids.
A standard cultivation protocol was applied and the membrane
fraction was isolated as described by Lawton and Philpot.13
A wild-type strain harbouring a stuffer fragment instead of
FMO insert was cultivated and measured in parallel as a
negative control. Activity of the E. coli membrane prepara-
tion was measured using an NADPH depletion assay with
methimazole as a substrate.5 FMO expression levels were
visualized by NuPAGE (Fig. 1). Codon optimization did not
lead to improvement of expression levels except for FMO1:
however unexpected, it could be addressed as a lack of
performance of the optimization algorithm for this class of
proteins. Both the native and codon optimized hFMO3 var-
iant displayed highest oxidation activity, followed by hFMO1,
which represents the first report of functional hFMO1 expres-
sion in E. coli. hFMO1 expression was significantly lower than
hFMO3 expression, which explains its lower overall activity
(Fig. 1). Interestingly, the expression level of hFMO5 –
especially the native variant – was very high. This was not
reflected directly in its activity for methimazole oxidation
which is in accordance with a previous report that hFMO5
activity for methimazole is 1/5000 lower than hFMO3’s.14
Similar to previous reports, full length hFMO4 was not
expressed in our hands.15
New active pharmaceutical ingredients (APIs) are frequently
tested with model species such as rat, mouse or monkey in the
preclinical phase to predict effectiveness, metabolism and
toxicology of the drug candidate and its metabolites. The
accuracy of predictions based on animal models is, however,
limited because the corresponding metabolic enzymes are –
although similar – not identical to hFMOs8 and, moreover,
not expressed to the same extent in the same tissue.3 In course
of the drug development process it is routinely necessary to
prepare authentic metabolites for structure elucidation and
a F. Hoffmann-La Roche Ltd., 4070 Basel, Switzerland.
Fax: +41 61 688 1673
b acib GmbH c/o Institute of Molecular Biotechnology,
Graz University of Technology, Petersgasse 14, 8010 Graz, Austria.
E-mail: margit.winkler@acib.at; Fax: +43 316 873 9302;
Tel: +43 316 873 9333
c Novartis Pharma AG, 4070 Basel, Switzerland.
Fax: +41 61 324 2103
w Electronic supplementary information (ESI) available. See DOI:
10.1039/c2cc17878h
c
This journal is The Royal Society of Chemistry 2012
Chem. Commun., 2012, 48, 6001–6003 6001

View Article Online
The plates were incubated at 37 1C overnight. A loop of the
resulting cells was used to inoculate 500 mL baffled flasks
containing 100 mL LB media supplemented with ampicillin
(100 mg LÀ1). Following incubation for 18 hours at 37 1C with
shaking at 90 rpm (orbital shaker, 5 cm radius), 100 mL of the
pre-culture was used to inoculate 100 L of LB–ampicillin
(100 mg LÀ1) and Aseol antifoam (0.01% v/v) in a Braun
Biotech 150 L fermentation vessel. Cultivation parameters
were as follows: temperature 37 1C, stirring 120 rpm, airflow
10 L minÀ1. No pH regulation was employed. When OD600
reached 0.6–0.8, filter sterilized IPTG was added to a final
concentration of 1 mM. After approx. 24 h cultivation the
biomass was harvested by continuous flow centrifugation at
13 000 rpm in a Heraeus Contifuge 20 RS at 4 1C. The
resulting cell paste (230 g) was then shock frozen in dry ice
before storage at À80 1C.
Biotransformations with whole cells as biocatalysts have
several advantages and were therefore chosen for moclobemide
oxidation: (1) the recombinant enzymes are protected against
physical and chemical stress and are in many cases more stable
than in their isolated form, which leads to higher productivities;
(2) when cofactors are needed for catalysis, the cells provide
the machinery for cofactor recycling at the expense of cheap
sacrificial co-substrates; (3) removal of the biocatalyst is
simple and facilitates downstream processing.
Fig. 1 Relative activity and NuPAGE Gel. Top: comparison of
methimazole oxidation activity of E. coli membrane fraction
(0.5 mg mLÀ1) containing hFMO encoded by its native or E. coli
optimized sequence (native and Ec, resp.); bottom: membrane fraction
was separated on 4–12% Bis-Tris-Gel using MES buffer (both Invitrogen)
and stained with CoomassieBlue. Lane 1 and 13: see Blue Plus2
PreStained Standard (M: Myosin 188 kDa, P: Phosphorylase 98 kDa,
B: BSA 62 kDa, G: Glutamic Dehydrogenase 49 kDa, A: Alcohol
Dehydrogenase 38 kDa, C: Carbonic Anhydrase 28 kDa). Lane 2:
vector control with stuffer fragment (C–). Lane 3–12, respectively:
hFMOs, from 1 to 5, E. coli optimized (Ec) and native codon usage (n).
The choice of a reaction buffer for use with ‘intact’ frozen
and thawed cells was based on experience with recombinant
human CYP450 enzymes for which citrate and NADP+ were
used to support cofactor regeneration.19 The effects of addi-
tion of co-substrates for cofactor regeneration are shown in
Table 1 and demonstrate that both citrate and NADP+ are
required for full FMO3 activity.
Further optimization was achieved by increasing the pH of
the buffer (see ESIw).
Moclobemide is a monoamine oxidase A (MAO-A) inhibitor
which is routinely prescribed for depression.16 Moclobemide’s
major metabolic pathways include aromatic hydroxylation,
C-oxidation, deamination and N-oxidation.17 The latter was
ascribed to FMO activity (Scheme 1) and hFMO3 was postu-
lated to be the most likely isoform responsible due to
strong correlation with benzydamine oxidation activity in
human liver microsomes.18 A screening for moclobemide
oxidation of the above described strains indeed revealed
hFMO3 as the most promising strain for moclobemide-
N-oxide preparation.
For preparative scale oxidation, frozen cell paste was
thawed and re-suspended in 0.1 M potassium phosphate buffer
to 20% w/v. The suspension was held on ice until required.
The biotransformations were carried out in a total volume of
1 L divided in two 2 L baffled flasks each containing 25 g wet wt
E. coli cells and 0.1 M phosphate buffer pH 8.5, 50 mM
trisodium citrate, 10 mM MgCl2, 50 mM NADP+ and 50 mg
moclobemide (final concentration 100 mg LÀ1) in 2.5 mL of
MeOH (final concentration 0.5% v/v). The flasks were incubated
at 27 1C with agitation at 120 rpm on an orbital shaker. At various
time points aliquots were removed and mixed with an equal
volume of acetonitrile (ACN) before centrifuging at 13 200 rpm
for 2 min. The supernatants were analyzed by HPLC using an
Agilent 1200 Rapid Resolution HPLC system equipped with an
Agilent Zorbax SB-C18 column (4.6 Â 50 mm, 1.8 microns).
To generate sufficient biomass for optimization studies and
preparative biotransformation, a large scale cultivation of
recombinant E. coli was carried out as follows: cells of
E. coli BL21 (DE3) Gold harbouring the native hFMO3 gene
were taken from vials stored in liquid nitrogen and streaked
on Luria Bertani (LB) agar plates containing ampicillin.
Table 1 Effect of reaction buffer composition on oxidation of
moclobemide by recombinant hFMO3. Incubations were carried out
for 24 h in 24-low well plates containing 1 mL phosphate buffer,
pH 7.4, containing E. coli cell paste (0.05 g), MgCl2 (10 mM) and
moclobemide (0.2 mM)
Entry NADP+ Citrate Moclobemide/mg LÀ1 N-Oxide (%)
1
2
3
+
À
À
135
84.8
49.2
2.9
32.7
70.4
+
+
+
Scheme 1 FMO catalyzed N-oxidation of moclobemide.
6002 Chem. Commun., 2012, 48, 6001–6003
c
This journal is The Royal Society of Chemistry 2012

View Article Online
In conclusion, we demonstrated the first preparation of
multi-milligram amounts of a drug metabolite using a recom-
binant FMO enzyme. These enzymes should be generally
applicable in the biocatalytic oxidation of drug candidates.
Especially when the substrates contain more than one soft-
nucleophile the enzymes are expected to exhibit selectivity, in
contrast to chemical oxidation methods.
Monika Muller (DSM Pharmaceutical Products, Geleen, NL)
¨
is kindly acknowledged for providing gene optimizations,
Thorsten Bachler for the molecular biology work, M.-O.
Grieneisen and M. Birrer for cultivation and biotrans-
formation as well as Fabian Eggimann for the purification
of moclobemide-N-oxide. This work has been supported
by the Austrian BMWFJ, BMVIT, SFG, Standortagentur
Tirol and ZIT through the Austrian FFG-COMET-Funding
Program.
Fig. 2 Moclobemide oxidation by recombinant hFMO3 containing
E. coli on 1 L scale. Reaction in 1 L 0.1 M phosphate, pH 8.5,
containing E. coli cell paste (50 g), trisodium citrate (50 mM), MgCl2
(10 mM), NADP+ (0.05 mM) and moclobemide (100 mg LÀ1).
E: Moclobemide, ’: Moclobemide-N-oxide.
Detection was at 240 nm and mobile phases consisted of
A: 0.1% formic acid in water, B: ACN with a gradient of
0–50% B in 2 min. The flow rate was 1.5 mL minÀ1 and the
column oven temperature 50 1C. The retention times of
moclobemide and the N-oxide were 1.35 and 1.40 min, respec-
tively. After 24 h incubation, the N-oxide titer reached 86 mg LÀ1
(Fig. 2) and the broth was centrifuged at 4000 rpm for 45 min at
4 1C. The supernatant was then shock frozen in dry ice and
stored at À20 1C prior to product purification.
Notes and references
1 K. Schroer, M. Kittelmann and S. Lutz, Biotechnol. Bioeng., 2010,
¨
106, 699.
2 R. Weis, M. Winkler, M. Schittmayer, S. Kambourakis, M. Vink,
J. D. Rozzell and A. Glieder, Adv. Synth. Catal., 2009, 351, 2140.
3 C. Mani, E. Hodgson and D. Kupfer, Drug Metab. Dispos., 1993,
21, 657.
4 J. R. Cashman, Curr. Drug Metab., 2000, 1, 181.
5 E. T. Tynes and E. Hodgson, Arch. Biochem. Biophys., 1985,
240, 77.
After thawing in a 50 1C water bath, the supernatant of the
biotransformation was mixed with 5 mL of aqueous ammonia
(25%) and 100 g of XAD-16 (Rohm and Haas, The Dow
Chemical Company, Germany) at room temperature for 30 min.
The resin was separated by filtration over gauze and eluted
twice, each time by mixing with 500 mL of ACN for 15 min
and filtering off the solvent. To the combined extracts, 10 g of
diatom granulate (Isolute HM-N, Separtis AG, Grellingen,
Switzerland) was added and the solvent was removed in a
rotary evaporator. The dry product coated Isolute HM-N was
filled into a 50 Â 25 mm precolumn. Preparative HPLC was
performed on a Macherey-Nagel Nucloeodur 100–10 C18 ec
(250 Â 20 mm) phase at RT. The mobile phase consisted of
solvent A: 10 mM ammonium formate, pH 6.5, and solvent B:
ACN with the following gradient: 0–5 min 5% B, 22.5 min
20% B, 25 min 38% B, 28 min 100% B, 30 min 5% B at a flow
rate of 20 mL minÀ1. Moclobemide-N-oxide was detected at
237 nm and eluted between 7 and 15% ACN (Rt = 8–17 min)
into 25 mL fractions. The product containing fractions were
combined, concentrated to about 100 mL and the solvent was
removed by lyophilization overnight. After dissolution of the
solid in 2 mL of ACN, addition of water (10 mL) and a second
lyophilization for removal of residual ammonium formate,
65 mg of moclobemide-N-oxide was obtained with >90%
purity (HPLC/full DAD), corresponding to an isolated yield
of 55%. Product identity was confirmed by comparison with
the authentic standard and NMR: 1H NMR (600 MHz,
DMSO-d6) d 3.27 (d, J = 11.71 Hz, 2H), 3.50 (t, J = 10.25 Hz,
2H), 3.62 (t, J = 4.80 Hz, 2H), 3.76 (d, J = 12.07 Hz, 2H),
3.79 (br s, 2H), 4.10 (t, J = 11.53 Hz, 2H), 7.55 (d, J = 8.30 Hz,
2H), 7.82 (d, J = 8.30 Hz, 2H), 9.86 (br s, 1H). 13C NMR
(600 MHz, DMSO-d6) d 33.3, 60.4, 62.7, 66.7, 128.2, 128.6,
133.4, 136.1, 165.0.
6 C. Dolphin, E. A. Shephard, S. Povey, C. N. A. Palmer,
D. M. Ziegler, R. Ayesh, R. L. Smith and I. R. Phillips, J. Biol.
Chem., 1991, 266, 12379.
7 R. N. Hines, K. A. Hopp, J. Franco, K. Saeian and F. P. Begun,
Mol. Pharmacol., 2002, 62, 320.
8 A. J. Lickteig, R. Riley, R. J. Melton, B. A. Reitz, H. D. Fischer
and J. C. Stevens, Drug Metab. Dispos., 2009, 37, 1987.
9 (a) N. Lomri, Z. Yang and J. R. Cashman, Chem. Res. Toxicol.,
1993, 6, 425; (b) L. H. Overby, A. R. Buckpitt, M. P. Lawton,
E. Atta-Asafo-Adjei, J. Schulze and R. M. Philpot, Arch. Biochem.
Biophys., 1995, 317, 275; (c) J. G. Falls, N. J. Cherrington,
K. M. Clements, R. M. Philpot, P. E. Levi, R. L. Rose and
E. Hodgson, Arch. Biochem. Biophys., 1997, 347, 9.
10 (a) C. T. Dolphin, D. J. Beckett, A. Janmohamed, T. E.
Cullingford, R. L. Smith, E. A. Shephard and I. R. Phillips,
J. Biol. Chem., 1998, 273, 30599; (b) C. K. Yeung, E. T. Adman
and A. E. Rettie, Arch. Biochem. Biophys., 2007, 464, 251–259;
(c) B. Furnes and D. Schlenk, Toxicol. Sci., 2004, 78, 196.
11 (a) A. Brunelle, Y.-A. Bi, J. Lin, B. Russell, L. Luy, C. Berkman
and J. Cashman, Drug Metab. Dispos., 1997, 25, 1001;
(b) V. Lattard, J. Zhang, Q. Tran, B. Furnes, D. Schlenk and
J. R. Cashman, Drug Metab. Dispos., 2003, 31, 854; (c) T. Borbas,
J. Zhang, M. A. Cerny, I. Liko and J. R. Cashman, Drug Metab.
Dispos., 2006, 34, 1995.
12 C. Reisinger, A. Kern, K. Fesko and H. Schwab, Appl. Microbiol.
Biotechnol., 2007, 77, 241.
13 M. P. Lawton and R. M. Philpot, J. Biol. Chem., 1993, 268, 5728.
14 B. Furnes, J. Feng, S. S. Sommer and D. Schlenk, Drug Metab.
Dispos., 2003, 31, 187.
15 K. Itagaki, G. T. Carver and T. M. Philpot, J. Biol. Chem., 1996,
271, 20102.
16 F. Lotufo-Neto, M. Trivedi and M. Thase, Neuropsychopharmacology,
1999, 20, 226.
17 R. Jauch, E. Griesser, G. Oesterhelt, W. Arnold, W. Meister,
W. H. Ziegler and T. W. Guentert, Acta Psychiatr. Scand., Suppl.,
1990, 360, 87.
18 J. Hoskins, G. Shenfield, M. Murray and A. Gross, Xenobiotica,
2001, 31, 387.
19 S. P. Hanlon, T. Friedberg, C. R. Wolf, O. Ghisalba and
M. Kittelmann, in Modern Biooxidation, ed. R. D. Schmid and
V. B. Urlacher, Wiley-VCH, Weinheim, 2007.
c
This journal is The Royal Society of Chemistry 2012
Chem. Commun., 2012, 48, 6001–6003 6003