Evaluation of coumarin-based fluorogenic P450 BM3 substrates and prospects for competitive inhibition screenings

Article abstract of DOI:10.1016/j.ab.2014.03.022

Fluorescence-based assays for the cytochrome P450 BM3 monooxygenase from Bacillus megaterium address an attractive biotechnological challenge by facilitating enzyme engineering and the identification of potential substrates of this highly promising biocatalyst. In the current study, we used the scarcity of corresponding screening systems as an opportunity to evaluate a novel and continuous high-throughput assay for this unique enzyme. A set of nine catalytically diverse P450 BM3 variants was constructed and tested toward the native substrate-inspired fluorogenic substrate 12-(4-trifluoromethylcoumarin-7- yloxy)dodecanoic acid. Particularly high enzyme-mediated O-dealkylation yielding the fluorescent product 7-hydroxy-4-trifluoromethylcoumarin was observed with mutants containing the F87V substitution, with A74G/F87V showing the highest catalytic efficiency (0.458 min^-1 μM^-1). To simplify the assay procedure and show its versatility, different modes of application were successfully demonstrated, including (i) the direct use of NADPH or its oxidized form NADP⁺ along with diverse NADPH recycling systems for electron supply, (ii) the use of cell-free lysates and whole-cell preparations as the biocatalyst source, and (iii) its use for competitive inhibition screens to identify or characterize substrates and inhibitors. A detailed comparison with known, fluorescence-based P450 BM3 assays finally emphasizes the relevance of our contribution to the ongoing research.

Full text of DOI:10.1016/j.ab.2014.03.022

Accepted Manuscript
Evaluation of coumarin-based, fluorogenic P450 BM3 substrates and prospects
for competitive inhibition screenings
Katharina Neufeld, Sonja Meyer zu Berstenhorst, Jörg Pietruszka
PII:
S0003-2697(14)00121-3
DOI:
http://dx.doi.org/10.1016/j.ab.2014.03.022
YABIO 11695
Reference:
To appear in:
Analytical Biochemistry
Received Date:
Revised Date:
Accepted Date:
10 February 2014
25 March 2014
28 March 2014
Please cite this article as: K. Neufeld, S. Meyer zu Berstenhorst, J. Pietruszka, Evaluation of coumarin-based,
fluorogenic P450 BM3 substrates and prospects for competitive inhibition screenings, Analytical Biochemistry
(
2014), doi: http://dx.doi.org/10.1016/j.ab.2014.03.022
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Title
Evaluation of coumarin-based, fluorogenic P450 BM3 substrates and prospects for
1
competitive inhibition screenings
Short title
Coumarin-based, fluorogenic P450 BM3 substrates
Authors
a
Katharina Neufeld* , Sonja Meyer zu Berstenhorst and Jörg Pietruszka
a
a,b
a
Institut für Bioorganische Chemie der Heinrich-Heine-Universität Düsseldorf im
Forschungszentrum Jülich, Stetternicher Forst, Geb. 15.8, 52426 Jülich
(
Deutschland).
E-Mail: j.pietruszka@fz-juelich.de; Fax: +49-2461-616196; Tel. +49-2461-614158.
b
Institut für Bio- und Geowissenschaften (IBG-1: Biotechnologie),
Forschungszentrum Jülich, Wilhelm-Johnen-Straße, 52425 Jülich (Deutschland).
Subject category for the table of contents
enzymatic assays and analyses
1

Abstract
Fluorescence-based assays for the cytochrome P450 BM3 monooxygenase from
Bacillus megaterium address an attractive biotechnological challenge by facilitating
enzyme engineering and the identification of potential substrates of this highly
promising biocatalyst. In the present study, we used the scarcity of corresponding
screening systems as an opportunity to evaluate a novel and continuous high-
throughput assay for this unique enzyme. A set of nine catalytically-diverse P450
BM3 variants was constructed and tested toward the native-substrate-inspired fluoro-
genic substrate 12-(4-trifluoromethylcoumarin-7-yloxy)dodecanoic acid. Particularly
high enzyme-mediated O-dealkylation yielding the fluorescent product 7-hydroxy-4-
trifluoromethylcoumarin was observed with mutants containing the F87V substitution,
-1
-1
with A74G/F87V showing the highest catalytic efficiency (0.458 min µM ). To
simplify the assay procedure and show its versatility, different modes of application
were successfully demonstrated including i) the direct use of NADPH or its oxidized
+
form NADP along with diverse NADPH-recycling-systems for electron supply, ii) the
use of cell-free lysates and whole-cell preparations as the biocatalyst source, and iii)
its use for competitive inhibition screens to identify or characterize substrates and
inhibitors. A detailed comparison with known, fluorescence-based P450 BM3 assays
finally emphasizes the relevance of our contribution to the ongoing research.
Keywords: cytochrome P450 BM3, fluorescence-based assay, microtiter plate, high-
throughput screening, P450 inhibition
2

Introduction
2
Cytochrome P450 monooxygenases (P450s) form a diverse superfamily of heme-
enzymes that are highly relevant for pharmaceutical, environmental and biocatalytic
applications due to their ability to degrade toxic xenobiotics, metabolize drugs and
catalyze the production of high-value fine chemicals [for selected reviews, see 1-7].
In this regard, the bacterial P450 BM3 monooxygenase from Bacillus megaterium
(
alias CYP102A1; EC 1.14.14.1) is one of the best studied and most widely used
enzymes because of its advantageous properties plus the availability of large-scale
production and purification protocols [8-10]. BM3 is a soluble and self-sufficient
fusion protein, which was originally classified as a fatty acid hydroxylase [11]. Its
published mutants are able to catalyze the oxygenation, mainly hydroxylation or
2
epoxidation, of a variety of organic compounds in the presence of O and NADPH
with high efficiencies and unique selectivities, thereby illustrating the high potential
and versatility of this biocatalyst [12,13]. Successful preparative scale applications
further emphasize the superior catalytic features of this enzyme compared to other
P450s [14-19]. Besides the production of fine chemicals, which is repeatedly men-
tioned as the major field of application for P450 BM3, this biocatalyst has attracted
growing interest for the synthesis of human metabolites (for the approval of new
therapeutics through toxicological characterization of metabolic profiles of potential
drugs) and lead structure diversification in the last decade [20-24]. This interest can
be ascribed to the high sequential, structural and functional similarities of the bacte-
rial enzyme with eukaryotic counterparts and the high consensus among the product
profiles of P450 BM3 mutants and mammalian P450s along with comparatively
higher activity and stability as well as advantageous handling of the first-mentioned
due to its bacterial origin [25-28]. In the past, identification of beneficial variants
3

possessing highly expanded substrate spectra and improved catalytic properties has
been the result of successful protein engineering using predominantly colorimetric
assays. The majority of these are indirect or end-point assays, so that the reaction
rates are either inferred from the decrease or increase of a participating reagent or
product formation is retrospectively ascertained upon a color-forming transformation
[29]. The first direct and continuous assay for P450 BM3 was established by
Schwaneberg et al. in 1999 and uses p-nitrophenol ethers as chromogenic
substrates; O-dealkylation proceeds through P450 BM3-catalyzed ω-hydroxylation
yielding an instable hemiacetal, which decomposes into an aldehyde and the
spectrophotometrically detectable chromophore p-nitrophenolate [30]. Fluorescence-
based assays are considered to be superior compared to their colorimetric alter-
natives due to higher sensitivity and lower susceptibility to interference by contami-
nants and assay components [31]. Thus, for assaying metagenomes or human
P450s - both challenging applications due to low enzyme contents and activities –
the state-of-the-art are ethers of fluorescent phenols such as resorufin or umbelli-
ferone which follow the same mode of action [32-34]. Furthermore, fluorescence-
based assays can be used for competitive inhibition screens, enabling the rapid
identification of potential enzyme substrates or inhibitors, as these compounds com-
pete with the fluorogenic substrate in the P450 reaction and thereby decelerate its
output [35-38]. The first fluorescence-based assay for P450 BM3 using alkoxyreso-
rufins as drug-like substrates was described by Lussenburg et al. in 2005 and
extended to various resorufin derivatives [39-42]. Moreover, Kim et al. studied
alkoxycoumarins as promising P450 BM3 substrates and Ruff et al. reported of 7-
benzyloxy-3-carboxycoumarin ethyl ester (BCCE) as a surrogate substrate [43-44].
Despite the many benefits of fluorogenic substrates, the small number of known
4

compounds highlights the importance of developing new fluorescence-based
screening systems for this biotechnologically important enzyme. This need is further
encouraged by the practical advantages of corresponding inhibition assays over
available alternatives for the structure- and product-independent identification of
+
P450 BM3 substrates such as i) NADP /NADPH- or O
2
-based detection systems, ii)
the spectroscopic proof of substrate binding via detecting the low-to-high spin state
transition of the heme iron, iii) the covalent labeling of P450 BM3 with a fluorophore
functioning as a reporter for ligand screening, or iv) in silico approaches, all of which
suffer from high costs, low sensitivity, are laborious, require purified enzymes and
involve false positives or negatives [29,45-48].
In this study, we describe a continuous, fluorescence-based, high-throughput assay
for detecting P450 BM3 hydroxylation activity by employing a set of catalytically-
diverse P450 BM3 mutants and a novel, native-substrates-inspired, coumarin-based
surrogate substrate. To estimate the performance of this compound in a wider
sense, the assay was further simplified toward the application with diverse NADPH-
recycling-systems plus whole-cell samples and applied to quantitative and qualitative
competitive inhibition screenings. A comparison with existing assays highlights the
relevance of the presented approach.
5

Materials and Methods
Reagents, enzymes and strains
All chemicals were purchased from standard suppliers and used without further puri-
fication except for petroleum ether (PE) and EtOAc, which were distilled prior to use.
Preparative column chromatography was performed using silica gel 60 (particle size
2
0.040-0.063 mm, 230-240 mesh). NADP-Na was kindly provided by Jülich Fine
Chemicals. The P450 BM3-containing pET28a(+) plasmids were available as
previously mentioned [49]; for construction of P450 BM3 variants see Sect. 5 in the
Supporting Information. Enzymes for DNA manipulation and dNTPs were procured
from Fermentas. Pfu DNA polymerase was a kind gift from Prof. K.-E. Jaeger (IMET,
HHU Düsseldorf, Germany). GDH and ADHLK were generously provided by Prof. W.
Hummel (IMET, HHU Düsseldorf, Germany; glucose dehydrogenase, alcohol
dehydrogenase from Lactobacillus kefir) and FDH (Jülich Fine Chemicals; formate
dehydrogenase) was a gift from Prof. M. Pohl (IBG-1: Biotechnology, FZ Jülich, Ger-
many); for determination of enzyme activities see Sect. 4 in the Supporting
Information. Lysozyme from chicken egg white (#62971), CalB (Candida antarctica
lipase B immobilized on acrylic resin, #L4777), RedTaq-Mix (#R2523), and oligonuc-
leotides were purchased at Sigma-Aldrich. E. coli strains DH5α or BL21 (DE3) were
used for DNA manipulations and heterologous protein expression, respectively.
Analytical methods
®
Analytical TLC was carried out on Polygram SIL G/UC245 plastic sheets (Macherey
Nagel). Aliphatic compounds were detected by oxidative staining (applying a mixture
of 300 mL AcOH, 6 mL conc. H
2 4
SO and 3 mL p-anisaldehyde and color develop-
ment with a hot-air blow dryer) and aromatic compounds were visualized by UV irra-
6

1
13
19
diation at 254 nm or observed with the naked eye. H, C and F NMR spectra
were recorded on an Advance/DRX 600 nuclear magnetic resonance spectrophoto-
meter (Bruker) at rt and 600, 151 and 565 MHz, respectively. Chemical shifts (δ) are
given in ppm relative to tetramethylsilane as an internal standard or relative to the
1 13
4
solvent [ H: δ(SiMe ) = 0.00 ppm, δ(DMSO) = 2.50 ppm, δ(pyridine) = 8.74 ppm; C:
δ(CDCl ) = 77.16 ppm, δ(DMSO) = 39.52 ppm]. Assignment was performed by
3
means of H-COSY-, HSQC- and HMBC-experiments. IR measurements were
conducted on a SpectrumOne spectrometer (PerkinElmer). Melting points were
determined using a Melting Point B-450 instrument (Büchi). ESI-MS data were recor-
ded using a MAT LC-Q mass spectrometer (Finnigan). Combustion analyses and
HRMS measurements were carried out by the ZEA-3 of FZ Jülich.
Synthesis
7-Hydroxy-4-trifluoromethylcoumarin (1) was synthesized according to a known pro-
cedure [50]. Resorcinol (2) (440 mg, 4.00 mmol), ethyl 4,4,4- trifluoroaceto-
acetate (3) (736 mg, 585 µL, 4.00 mmol) and trifluoroacetic acid (2 mL) were heated
in an oven-dried 10 mL heavy-walled microwave reaction vessel equipped with a
cross-shaped, magnetic stirring bar and closed with a snap cap to 100 °C for 30 min
in a Discovery microwave (CEM). After cooling to rt, the reaction mixture was slowly
added to an ice:water-mixture (120 mL, 1:1) under vigorous stirring. The pink precipi-
tate was filtered off, washed with cooled dest. water and dried under reduced
pressure. Chromatography on silica gel with PE:EtOAc (8:2) provided the fluorescent
dye 1 (497 mg, 2.16 mmol, 54%) as a colorless solid. The analytical data were gene-
rally consistent with those previously reported [50,51]. TLC: R
f
0.58 (PE:EtOAc, 1:1);
mp: 182 °C; IR (ATR, film): 3395 (O-H), 3106, 1893, 1766, 1709 (C=O), 1630, 1607,
1575, 1519, 1494, 1441, 1405, 1362, 1337, 1284, 1214, 1190, 1158, 1120, 1008,
7

-
1
1
63, 891, 856, 821, 798, 750, 743, 721, 710, 670, 657 cm ; H NMR (DMSO-d
9
6
):
δ = 6.76 (s, 1 H, 3-H), 6.83 (d, J = 2.4 Hz, 1 H, 8-H), 6.91 (dd, J = 8.9, 2.4 Hz, 1 H, 6-
13
H), 7.56 (dd, J = 8.9, 1.6 Hz, 1 H, 5-H), 10.95 (brs, 1 H, OH) ppm; C NMR (DMSO-
): δ = 103.2 (C-8), 105.3 (C-8a), 112.0 (q, JC-F = 5.8 Hz, C-3), 114.1 (C-6), 121.8
q, JC-F = 275.8 Hz, C-4), 126.2 (C-5), 139.8 (q, JC-F = 31.9 Hz, C-4a), 156.0 (CF ),
): δ = -63.6 ppm; HRMS (ESI):
] ([M+H] ): 231.02636; found: 231.02635; Anal.: calcd. for
d
6
(
3
1
9
1
58.9 (C-2), 162.2 (C-7) ppm; F NMR (DMSO-d
6
+
+
6 3 3
calcd. for [C10H F O
10 5 3 3
C H F O : C, 52.19; H, 2.19; found: C, 52.03 ± 0.09; H, 2.23 ± 0.03.
Methyl 12-bromododecanoate (7) was synthesized via a standard procedure by
dissolving 12-bromododecanoic acid (6) (1.0 g, 3.59 mmol) in dry MeOH (10 mL)
containing conc. H
5 h. MeOH was removed under reduced pressure, the residual diluted with EtOAc
100 mL), washed with a saturated NaHCO -solution (3 x 20 mL) and brine (2 x
0 mL), dried with MgSO and the solvent was removed under reduced pressure.
Chromatography on silica gel with PE:EtOAc (97.5:2.5) provided the desired ester 7
974 mg, 3.32 mmol, 93%) as colorless oil. The analytical data were generally con-
sistent with those previously reported [52-54]. TLC: R 0.45 (PE:EtOAc, 9:1); IR
ATR, film): 2926, 2854, 1740 (C=O), 1459, 1436, 1362, 1252, 1242, 1197, 1171,
2 4
SO (200 µL) under argon atmosphere and heating to reflux for
1
(
3
2
4
(
f
(
-
1 1
1
3
120, 1103, 1012, 881, 723 cm ; H NMR (CDCl ): δ = 1.23-1.35 (m, 12 H, 4-H to 9-
H), 1.42 (tt, J = 7.4, 7.2 Hz, 2 H, 10-H), 1.62 (tt, J = 7.6, 7.4 Hz, 2 H, 3-H), 1.85 (tt,
J = 7.4, 6.9 Hz, 2 H, 11-H), 2.30 (t, J = 7.6 Hz, 2 H, 2-H), 3.41 (t, J = 6.9 Hz, 2 H, 12-
13
H), 3.67 (s, 3 H, OCH
3 3
) ppm; C NMR (CDCl ): δ = 25.1 (C-3), 28.3 (C-10), 28.9,
29.3, 29.4, 29.5, 29.5, 29.6 (C-4 to C-9), 33.0 (C-11), 34.2 (C-12), 34.3 (C-2), 51.6
7
9
81
+
+
(
OCH ), 174.5 (C-1) ppm; HRMS (ESI): calcd. for [C13
3
26 2
H O Br/ Br] ([M+H] ):
293.11107/ 295.10902; found: 293.11113/ 295.10908.
8

Methyl 12-(4-trifluoromethylcoumarin-7-yloxy)dodecanoate (5) was synthesized
according to a known procedure [55]. Dye 1 (549 mg, 2.39 mmol) and anhydrous
2 3
K CO (363 mg, 2.63 mmol) were incubated in DMF (10 mL) under argon atmos-
phere for 2 h at rt. Subsequently, bromide 7 (700 mg, 2.39 mmol) was added and the
reaction was heated to 90 °C for 15 h. After cooling to rt, the reaction mixture was
diluted with EtOAc (100 mL), washed with saturated NaHCO
and brine (2 x 30 mL), dried with MgSO and the solvent was removed under redu-
ced pressure. Chromatography on silica gel with PE:EtOAc (95:5) provided the fluo-
rogenic substrate 5 (1.01 g, 2.28 mmol, 95%) as a colorless solid. TLC: R 0.76
PE:EtOAc, 7:3); mp: 85 °C; IR (ATR, film): 3096, 2932, 2918, 2853, 1736 (C=O),
3
-solution (2 x 30 mL)
4
f
(
1
1
9
722 (C=O), 1611, 1562, 1517, 1473, 1440, 1427, 1418, 1408, 1395, 1378, 1348,
330, 1278, 1270, 1209, 1193, 1160, 1129, 1080, 1070, 1057, 1025, 1008, 982, 970,
-
1
1
58, 940, 884, 871, 847, 835, 827, 802, 752, 742, 722, 710, 668 cm ; H NMR
): δ = 1.24-1.40 (m, 12 H, 4-H to 9-H), 1.47 (m , 2 H, 10-H), 1.62 (m , 2 H, 3-
H), 1.82 (m , 2 H, 11-H), 2.30 (t, J = 7.6 Hz, 2 H, 2-H), 3.67 (s, 3 H, CH ), 4.04 (t,
J = 6.5 Hz, 2 H, 12-H), 6.60 (s, 1 H, 3-HAr), 6.86 (d, J = 2.5 Hz, 1 H, 8-HAr), 6.91 (dd,
(
CDCl
3
c
c
c
3
1
J = 9.0, 2.5 Hz, 1 H, 6-HAr), 7.61 (dd, J = 9.0, 1.7 Hz, 1 H, 5-HAr) ppm; C NMR
3
(
(
(
CDCl
C-4 to C-9), 34.2 (C-2), 51.6 (CH
q, JC-F = 5.8 Hz, CAr-3), 113.9 (CAr-6), 121.7 (q, JC-F = 276.3 Hz, CAr-4), 126.4 (CAr
3
): δ = 25.1 (C-3), 26.0 (C-10), 29.0 (C-11), 29.6, 29.6, 29.5, 29.4, 29.4, 29.3
3
), 69.0 (C-12), 102.0 (CAr-8), 106.9 (C-8a), 112.1
-
5), 141.7 (q, JC-F = 32.7 Hz, C-4a), 156.5 (CF
3
), 159.6 (CAr-2), 163.3 (CAr-7), 174.4
1
C-1) ppm; F NMR (CDCl
9
+
(
(
3
30 3 5
): δ = -64.7 ppm; HRMS (ESI): calcd for [C23H F O ]
+
[M+H] ): 443.20399; found: 443.20399; Anal.: calcd. for C23
H
29
F
O
3 5
: C, 62.43; H,
6.61; found: C, 62.49 ± 0.06; H, 6.63 ± 0.02.
9

1
2-(4-Trifluoromethylcoumarin-7-yloxy)dodecanoic acid (4) was synthesized via a
standard procedure by dissolving ester 5 (364 mg, 0.82 mmol) in THF (10 mL). KOH
346 mg, 6.17 mmol) dissolved in 10 mL H O was slowly added at 0 °C, the reaction
(
2
was incubated for 2 h on ice and subsequently warmed to rt. Full conversion was
confirmed by TLC after additional 5 h. The reaction mixture was diluted with ice-cold
1
M aqueous HCl (40 mL) and extracted with CH
organic phases were dried with MgSO and the solvent was removed under reduced
pressure. Chromatography on silica gel with PE:EtOAc (7:3) provided the desired
acid 4 (350 mg, 0.82 mmol, quant.) as a colorless solid. TLC: R 0.43 (PE:EtOAc,
:3); mp: 90 °C; IR (ATR, film): 3085, 2921, 2852, 1757, 1733 (C=O), 1698 (C=O),
2 2
Cl (4 x 30 mL). The combined
4
f
7
1
1
9
611, 1557, 1517, 1474, 1432, 1412, 1402, 1364, 1348, 1325, 1305, 1296, 1276,
244, 1213, 1189, 1168, 1138, 1117, 1084, 1039, 1029, 1017, 1002, 984, 976, 958,
-
1
1
42, 868, 857, 833, 799, 785, 752, 735, 719, 711, 685 cm ; H NMR (CDCl
, 2 H, 10-H), 1.63 (m , 2 H, 3-H), 1.82
, 2 H, 11-H), 2.35 (t, J = 7.5 Hz, 2 H, 2-H), 4.04 (t, J = 6.5 Hz, 2 H, 12-H), 6.61 (s,
H, 3-HAr), 6.86 (d, J = 2.5 Hz, 1 H, 8-HAr), 6.91 (dd, J = 9.0, 2.5 Hz, 1 H, 6-HAr),
3
):
δ = 1.24-1.40 (m, 12 H, 4-H to 9-H), 1.47 (m
c
c
(m
c
1
7
1
1
1
3
3
.61 (dd, J = 9.0, 1.7 Hz, 1 H, 5-HAr) ppm; C NMR (CDCl ): δ = 24.8 (C-3), 26.0 (C-
0), 29.0 (C-11), 29.6, 29.6, 29.5, 29.4, 29.3, 29.2 (C-4 to C-9), 34.1 (C-2), 69.0 (C-
2), 102.0 (CAr-8), 107.0 (C-8a), 112.2 (q, JC-F = 5.8 Hz, CAr-3), 113.9 (CAr-6), 121.7
(
(
q, JC-F = 276.0 Hz, CAr-4), 126.4 (CAr-5), 141.8 (q, JC-F = 32.9 Hz, C-4a), 156.5
19
CF
3
), 159.6 (CAr-2), 163.3 (CAr-7), 174.8 (C-1) ppm; F NMR (CDCl
3
): δ = -
+
4.7 ppm; MS (ESI): m/z (%) = 429 [(M+H) ]; Anal.: calcd. for C22
6
H
27
F
3 5
O : C, 61.67;
H, 6.35; found: C, 61.87 ± 0.05; H, 6.42 ± 0.01.
Methyl 12-(3H-phenoxazin-3-one-7-yloxy)dodecanoate (19) was prepared as de-
scribed for ester 5 except that higher solvent amounts and longer reaction time were
10

needed due to lower solubility. The reaction mixture containing resorufin (18)
(
369 mg, 1.73 mmol), anhydrous K
507 mg, 1.73 mmol) in DMF (24 mL) was heated for 2 d. The dilution ensued with
L EtOAc and each wash step was performed with 500 mL. Chromatography on sili-
ca gel with PE:EtOAc (7:3→EtOAc) provided the fluorogenic substrate 19 (452 mg,
.06 mmol, 61%) as an orange solid. TLC: R 0.64 (PE:EtOAc, 1:1); mp: 128 °C; IR
ATR, film): 3083, 3053, 2995, 2920, 2850, 1730 (C=O), 1644, 1604, 1589, 1564,
2 3
CO (264 mg, 1.90 mmol) and bromide 7
(
2
1
f
(
1
1
8
1
7
3
6
505, 1483, 1472, 1448, 1428, 1415, 1394, 1369, 1339, 1317, 1271, 1233, 1213,
203, 1183, 1169, 1159, 1120, 1102, 1036, 1024, 1011, 995, 979, 958, 924, 905,
-
1 1
3
88, 862, 836, 818, 780, 766, 746, 734, 721, 711 cm ; H NMR (CDCl ): δ = 1.24-
.40 (m, 4-H to 9-H, 12 H), 1.47 (tt, J = 7.3, 7.3 Hz, 2 H, 10-H), 1.62 (tt, J = 7.6,
.6 Hz, 2 H, 3-H), 1.84 (tt, J = 7.3, 6.5 Hz, 2 H, 11-H), 2.30 (t, J = 7.5 Hz, 2 H, 2-H),
3
.67 (s, 3 H, CH ), 4.06 (t, J = 6.5 Hz, 2 H, 12-H), 6.32 (d, J = 2.6 Hz, 1 H, 8-HAr),
.80 (d, J = 2.0 Hz, 1 H, 10-HAr), 6.83 (dd, J = 9.8, 2.0 Hz, 1 H, 2-HAr), 6.93 (dd,
J = 8.9, 2.6 Hz, 1 H, 6-HAr), 7.42 (d, J = 9.8 Hz, 1 H, 3-HAr), 7.70 (d, J = 8.9 Hz, 1 H,
1
3
5
2
1
-HAr) ppm; C NMR (CDCl
3
): δ = 25.1 (C-3), 26.1 (C-10), 29.1 (C-11), 29.6, 29.6,
9.5, 29.4, 29.4, 29.3 (C-4 to C9), 34.3 (C-2), 51.6 (CH
3
), 69.3 (C-12), 100.6 (CAr
-
0), 106.8 (CAr-8), 114.3 (CAr-6), 128.4 (C-4a), 131.7 (CAr-5), 134.3 (CAr-2), 134.8
Ar-3), 145.5 (C-8a), 145.9 (C-3a), 150.0 (C-9a), 163.5 (CAr-7), 174.5 (C-1), 186.4
(
(
C
+
+
C
Ar-1) ppm; HRMS (ESI): calcd. for [C25
26.22736; Anal.: calcd. for C25 : C, 70.57; H, 7.34; N, 3.29; found: C,
0.40 ± 0.04; H, 7.32 ± 0.02; N, 3.26 ± 0.02.
5
H32NO ] ([M+H] ): 426.22750; found:
4
7
H31NO
5
12-(3H-Phenoxazin-3-one-7-yloxy)dodecanoic acid (20) was synthesized according
to a known procedure [30]. Ester 19 (156 mg, 0.37 mmol) was dissolved in 500 mL
acetone/buffer (100 mM KPi-buffer, pH 8.0, 80% acetone), CalB (2 kU) was added
11

and the reaction was incubated for 2 d at 37 °C and 140 rpm under exclusion of light.
The catalyst was removed by filtration, acetone was evaporated under reduced
pressure, 1 M aqueous HCl was added to adjust a pH of 1-2 and the precipitated
product was filtered off. Washing with cooled dest. water and drying under reduced
pressure gave the desired acid 20 (133 mg, 0.32 mmol, 88%) as an orange solid.
TLC: R
f
0.33 (PE:EtOAc, 1:1); mp: 200 °C; IR (ATR, film): 3079, 2939, 2916, 2850,
717 (C=O), 1646, 1607, 1577, 1554, 1504, 1487, 1470, 1435, 1409, 1367, 1341,
331, 1289, 1267, 1233, 1215, 1202, 1175, 1164, 1123, 1110, 1085, 1068, 1054,
1
1
1
-1 1
019, 988, 973, 935, 911, 889, 862, 849, 837, 793, 762, 730, 716, 654 cm ; H NMR
(
pyridine-d
5
): δ = 1.19-1.36 (m, 5-H to 9-H, 10 H), 1.37-1.51 (m, 4 H, 4-H, 10-H),
1
.73-1.87 (m, 4 H, 3-H, 11-H), 2.55 (t, J = 7.3 Hz, 2 H, 2-H), 4.10 (t, J = 6.4 Hz, 2 H,
2-H), 6.48 (d, J = 1.7 Hz, 1 H, HAr), 6.97 (dd, J = 9.8, 1.7 Hz, 1 H, HAr), 7.01 (s, 1 H,
1
H
1
Ar), 7.11 (d, J = 8.8 Hz, 1 H, HAr), 7.51 (d, J = 9.8 Hz, 1 H, HAr), 7.85 (d, J = 8.8 Hz,
+
+
H, HAr) ppm; HRMS (ESI): calcd. for [C24
: C, 70.05; H, 7.10; N, 3.40; found: C,
9.72 ± 0.14; H, 7.10 ± 0.04; N, 3.40 ± 0.01.
5
H30NO ] ([M+H] ): 412.21185; found:
412.21185; Anal.: calcd. for C24H29NO
5
6
Absorption and fluorescence measurements in MTPs
Absorption and fluorescence measurements were carried out under aerobic condi-
®
tions using an Infinite M1000 spectrophotometer plate reader (Tecan) equipped
TM
with the i-control
software in colorless, flat-bottomed 96-well and black, flat-
bottomed 384-well microtiter plates (MTPs; Grainer Bio-One) at 30 °C applying a
final volume of 200 and 100 µL/well, respectively. The excitation and emission wave-
lengths used were λex: 420 nm/ λem: 500 nm for 1 and λex: 530 nm/ λem: 595 nm for 18
(
bandwidth 5 nm). All experiments were performed in a KPi/Tris-buffer-mixture (BM3
buffer, pH 7.8, 50 mM KPi, 50 mM Tris/HCl). Stock solutions and dilutions of sub-
12

strates, fluorophores and inhibitors were prepared in DMSO (heating to 80 °C facili-
tated compound solution). Fluorescence intensities were converted to product con-
centrations using calibration curves measured for each experiment individually by
addition of 1 or 18 (serial 2:3 dilutions plus blank values, max. final concentration
(
f. c.) 10 µM) instead of substrate to the assay mixtures. All measurements were per-
formed in triplicate.
Determination of substrate 4 O-dealkylation: Protein expression in deep-well plates
and preparation of cell-free crude lysates (CCLs) for assaying O-dealkylation activi-
ties of the constructed P450 BM3 variants toward substrate 4 were performed as
previously described [49]
.
Reactions composed of 70 µL BM3 buffer, 10 µL CCL and
10 µL substrate 4 (100 µM) were initialized by addition of 10 µL NADPH (3 mM in
H
2
O). Formation of 1 was monitored for 5 min and evaluation of data was conducted
in end-point mode by subtracting the initial fluorescence intensity from those ob-
tained after a 5-min reaction and normalizing the values according to the empty
vector negative control.
To examine whether NADPH-recycling-systems are likewise suitable to support
P450 BM3-mediated O-dealkylation, the same conditions were used and 10 µL of
+
NADP -coenzyme-cosubstrate-mixtures were added to initialize the reactions: GDH-
2
samples contained 5 µL glucose (200 mM in H O) and 12.8 mU GDH, FDH-samples
5
µL sodium formate (2 M in H O) and 3.3 mU FDH and ADH-samples 5 µL isopro-
2
2 2
panol (200 mM in H O supplemented with 20 mM MgCl ) and 88 mU ADH, respecti-
+
vely, along with 0.2 µL NADP (50 mM in H
2
O) in BM3 buffer (add to 10 µL).
To examine whether CCLs can be substituted with E. coli whole-cells plus lysozyme
or polymyxin B treated samples, the previous experiment was repeated using the
13

GDH-recycling-system. Intact whole-cell samples were prepared by resuspending
the cell pellets in 240 µL 50 mM KPi buffer (pH 7.5). Lysozyme-treated samples were
-1
obtained in the same manner except that the buffer contained 5 mg mL lysozyme
and the samples were incubated at 37 °C and 900 rpm for 1 h before use. Poly-
myxin B-treated samples were generated from whole-cell samples by addition of
10 µL polymyxin B (0.75 mM in buffer) to 10 µL cell suspension and 60 µL BM3
buffer plus incubation for 15 min. NADPH-supply was a prerequisite to observe acti-
vity in all cases.
Determination of kinetic parameters: Flask expressions, protein purification and
quantification of P450 BM3 contents were performed as previously described [49].
Based on pre-test results, samples for kinetic measurements were composed of
5.3 pmol GV, 15.8 pmol F87V, 29.2 pmol GQ and 17.0 pmol F87A for substrate 4,
53.0 pmol GV and 31.6 pmol F87V for substrate 5 and 117.8 pmol GQ for substrate
19 in 80 µL BM3 buffer, respectively (see Fig. S5 in the Supporting Information).
Hereto, 10 µL of substrate (serial 2:3 dilutions plus blank values; max. f. c. for 4 was
33 µM, for 5 70 µM and for 19 9.7 µM, respectively) were added and the mixtures
1
were pre-incubated for 5 min at 900 rpm and rt before reactions were initialized by
addition of 10 µL ADH-recycling-system. Formation of 1 was monitored for 2-10 min.
Reaction rates were calculated based on the linear portions of curves and plotted
against the substrate concentrations. Using the non-linear regression tool of
OriginPro 9.0G, data showing sigmoidal kinetics were fitted using the Hill equation
and data indicating excess-substrate inhibition were fitted using a modified Hill equa-
2
tion: y = vmax • x/(x /K
i
+ K
M
+ x) [56].
14

Qualitative inhibition screening: The ability of the established assay to identify
ligands of P450 BM3 was tested with known non-substrates (caffeine 16 and iso-
octane 17), substrates (myristic acid 10, naphthalene 11 and methyl 3-methoxy-2-
methylbenzoate 12) and inhibitors (ketoconazole 13, imidazole 14 and cyanide 15).
Flask expression and preparation of CCLs was performed as described elsewhere
except that the BM3 buffer was used [49]. Samples containing 27.6 pmol GV or
48.4 pmol F87V in 80 µL BM3 buffer, 5 µL substrate 4 (200 µM) and 5 µL compound
of interest (test samples with varying concentrations: 100, 20 and 2 mM for 16, 17,
11, 12 and 14; 4, 2 and 0.4 µM for 10; 100, 20 and 4 µM for 13 and 100, 20, 2 and
0.4 µM for 15) or 5 µL DMSO (reference samples) were pre-incubated for 5 min at
900 rpm and rt before reactions were initialized by addition of 10 µL GDH-recycling-
system. Formation of 1 was monitored for 5 min and evaluation was conducted in
end-point mode as described above. The results were normalized relative to the
activity obtained in absence of a probe compound and a threshold value of 90% was
defined upon which enzyme inhibition was stated.
Quantitative inhibition screening: The ability of the established assay to provide IC50
values was tested with the known non-native substrate 12 and purified P450 BM3
F87A. Samples containing 54.7 pmol F87A in 80 µL BM3 buffer, 5 µL substrate 4
(
300 µM) and 5 µL compound 12 (test samples: serial 2:3 dilution; max. f. c. 1 M) or
µL DMSO (reference sample) were pre-incubated for 5 min at 900 rpm and rt
5
before reactions were initialized by addition of 10 µL ADH-recycling-system. Forma-
tion of 1 was monitored for 10 min and evaluation was conducted in end-point mode
as described above. The obtained inhibition values were plotted against the concen-
tration of 12 and data were fitted using the Michaelis-Menten equation.
15

Results and Discussion
Assay-related fluorophore selection and synthesis
Coumarin derivatives have proven to be excellent fluorescence markers and are
nowadays routinely applied for the generation of diverse surrogate substrates
[57,58]. In the present approach, 7-hydroxy-4-trifluoromethylcoumarin (1) (HTC) was
favored due to its advantageous features related to P450 catalysis. In contrast to
other commonly used fluorophores, HTC 1 ensures lower background signals and
higher signal intensities as its excitation and emission spectra are shifted to higher
wavelengths. Thus, the overlap with the optical characteristics of assay components
is minimized and interference with NADPH can be eliminated [59]. Furthermore, the
a
comparatively low pK value of 6.8 enables screening at slightly basic pH, which is
beneficial as P450 BM3 shows its highest activity around pH 7 [60]. In the present
study, λex/ λem were set to 420/ 500 nm and a pH of 7.8 was selected as a compro-
mise between enzyme activity and assay sensitivity (for a detailed illustration of the
advantageous features of HTC 1, see Sect. 1 in the Supporting Information). More-
over, while studying the P450 BM3-catalyzed oxygenation of 7-ethoxycoumarin, Kim
et al. found most of the highly active P450 BM3 mutants to favor the 3-hydroxylation
over the O-deethylation reaction [43]. We assumed that the trifluoromethyl moiety in
1
could shift the enzymeꞌs preference toward O-dealkylation through steric effects,
thereby not providing a P450 BM3 target since C-F bonds are considered to be inert
61-64]. Compound 1 was initially provided via a Pechmann reaction from resorci-
[
nol (2) and ethyl 4,4,4-trifluoroacetoacetate (3) according to a procedure of Katkevičs
et al. in 54% yield (Scheme 1) [50]. However, due to ongoing efforts in assay re-
search, this dye has recently become commercially affordable.
16

Synthesis of novel, fluorogenic BM3 substrates
Referring to the idea of Schwaneberg et al. for the p-nitrophenol-based, colorimetric
P450 BM3 assay, in the present approach a fluorogenic P450 BM3 substrate bearing
structural similarity to the native fatty acid substrates was envisaged [30]. Dodeca-
noic acid was selected as the substrate synthon and the HTC-labeled compound
methyl 12-(4-trifluoromethylcoumarin-7-yloxy)dodecanoate (4) could be obtained
from 12-bromododecanoic acid (6) and fluorophore 1 in an overall yield of 88% in
three steps (Scheme 2). Bromide 6 was first protected to the corresponding methyl
ester 7 in 93% yield, which was subsequently applied for the alkylation of 1 to give
ester 5 in 95% yield; saponification under basic conditions provided the desired
acid 4 quantitatively. Subsequently, the optical properties of substrates 4 and 5 and
their solubility under assay conditions were investigated. Fig. 1 illustrates a shift of
the absorption, excitation and emission spectra of these compounds to lower wave-
lengths compared to those of 1, the optimal wavelengths being λabs/λex: 335 nm and
λ
λ
em: 425 nm for 4, λabs: 340 nm, λex: 330 nm, and λem: 410 nm for 5 plus
abs: 390 nm, λex: 380 nm, and λem: 500 nm for 1. This shift ensures selective excita-
tion and detection of HTC 1 in the presence of substrates 4 and 5 and proves the
latter to be suitable fluorogenic substrates. Further, the ethers exhibit minor fluores-
cence intensities compared to HTC 1 by a factor of 6 and 11 for 4 and 5, respectively
(λex/λem: 1: 4164 a.u.; 4: 698 a.u.; 5: 371 a.u.), which is consistent with the ꞌblockedꞌ
character of 1 within these compounds. Solubility was examined by preparing serial
dilutions of ethers, addition of buffer and measuring absorption at 600 nm to verify
precipitation. Limits were determined to be 197 µM for ether 4 and 74 µM for ether 5,
respectively (see Sect. 2 in the Supporting Information).
17

Construction of the enzyme library
A small enzyme library consisting of nine P450 BM3 mutants was constructed to
provide a versatile catalyst pool for the evaluation of the planned fluorescence-based
assay. Literature-known mutants noted for wide substrate spectra, distinguishable
product profiles and the ability to oxygenate sterically demanding compounds were
assumed to be particularly suitable regarding surrogate substrate 4 and thus, seven
of the selected variants contained the A74G (G), F87V (V) or L188Q (Q) substitution
and combinations thereof. Especially the F87V single and the GVQ triple mutant
have been repeatedly examined in the context of non-natural substrates in general
and sterically demanding compounds in particular [65-74]. The F87A variant was
also chosen due to its unique catalytic efficiency, which gave rise to iterated and ex-
tensive studies, and the WT enzyme was included for the sake of completeness [75-
77]. Generation of the different variants was achieved by site-saturation mutagene-
sis.
Determination of P450 BM3-mediated O-dealkylation
The generally accepted principle of the metabolism of fluorogenic ether substrates
into their highly fluorescent phenols by P450s is provided in Scheme 3 using the
example of substrate 4. O-dealkylation proceeds via enzyme-catalyzed ω-hydroxyla-
tion yielding the instable hemiacetal 8 as intermediate, which releases fluorophore 1
and aldehyde 9 upon spontaneous decomposition [30]. For testing of generated
P450 BM3 mutants regarding this reaction, enzyme expression was performed in 96-
deep-well plates to simulate screening conditions. Cell-free crude lysates were
obtained upon cell lysis using lysozyme and incubated in 384-well format with 10 µM
of 4 in the presence of NADPH. To our delight, formation of HTC 1 was observed
and the values shown in Fig. 2 indicate that particularly P450 BM3 variants con-
18

taining the F87V substitution exhibit high O-dealkylation ability. The highest value of
8.2 ± 4.3 pmol/5min was obtained for the double mutant GV and other P450 BM3
5
variants metabolized 4 only at low rates, which were clearly distinguishable from the
empty vector background. These results confirm the crucial role of the F87V muta-
tion in controlling the substrate specificity of P450 BM3 and prove the designed and
synthezised compound 4 to be a suitable fluorogenic substrate for this monooxy-
genase.
Establishment of simplified assay procedures
Regarding high-throughput applications direct utilization of NADPH as the electron
+
source is undesirable due to its high costs and low stability [78]. Use of NADP along
with an appropriate NADPH-recycling-system offers a solution to this problem and
high versatility of the established assay was shown by successful application of
different enzyme-coupled recycling-systems. A GDH/glucose-, a FDH/formate- and
an ADH/isopropanol-system were selected as these are the most commonly used
NADPH-recycling-systems for P450 catalysis [79]. In all cases the co-enzyme cata-
lyzes the oxidation of the co-substrate (glucose, formate or isopropanol) and simulta-
+
neous reduction of NADP to NADPH, providing the latter for the P450 BM3-
mediated O-dealkylation. Initial tests were performed to determine appropriate co-
enzyme and co-substrate concentrations, thereby ensuring the P450 BM3-catalyzed
reaction to be the rate-determining step within the assay (data not shown). Subse-
quent screening of the P450 BM3 library using these NADPH-recycling-mixtures re-
produced the previously observed activity pattern with the GV mutant giving the
highest activity in all measurement series, even though the total values varied
depending on the applied recycling-system (Table 1, upper part). To facilitate com-
parability, reaction rates were standardized to the GV results and the previous
19

experiment was included for comparison (Table 1, lower part). Observed differences
between the different measurement series result from the nature of the NADPH-
providing step and the P450 BM3 activity under the respective conditions. The GDH
and FDH reactions are assumed to be irreversible as glucose is converted to
gluconolactone, which decomposes to gluconic acid under assay conditions, and for-
2
mate is converted to CO , which can depart from the reaction mixture. In both cases
the pH of the assay solution is affected by formation of the co-product (decreasing
and increasing effect in case of GDH and FDH, respectively), which has a major
influence on the protonation state of HTC 1 and therefore, on the intensity of the fluo-
rescence signal. Thus, besides the ability of the P450 BM3 variants to mediate O-
dealkylation, their ratio of decoupling significantly contributes to the final screening
results (decoupling denotes the amount of consumed NADPH not used for substrate
oxidation). The ADH reaction reversibly yields acetone as the oxidized co-product. In
contrast to the previous systems an impact on the pH can be excluded, but ADH-
catalyzed oxidations of assay components are possible and a negative influence of
acetone or isopropanol on the solvent-sensitive P450 BM3 enzyme is conceivable in
common with a positive effect on substrate solubility in the presence of these co-
solvents [80]. Based on these considerations, the higher values obtained with GDH-
and ADH-recycling compared to those for the NADPH approach can be lead back to
higher cofactor amount under the first mentioned conditions. In this, the compara-
tively minor values using GDH-recycling arise from acidification of the sample
mixtures in the course of the reactions. Further, GDH- and ADH-promoted reactions
revealed better reproducibility across replicates compared to the NADPH approach.
On the contrary, the low values obtained with FDH-recycling indicate minor P450
BM3 activity in the presence of formate. To the best of our knowledge, this is the first
20

time a direct comparison between different NADPH-recycling-systems with respect
to P450 BM3 activity is provided and all systems tested are superior over the
G6PDH-recycling, which requires expensive glucose-6-phosphate as co-substrate
and was exclusively applied in previous P450 BM3 fluorescence-based assays
(
G6PDH: glucose-6-phosphate dehydrogenase) [39,41,43].
To investigate whether the assay procedure can be further simplified by using turbid
lysozyme-lysed, polymyxin B-permeabilized (polymyxin B is pore-forming
a
membrane protein) or untreated whole-cells as biocatalyst sources instead of labo-
riously generated lysates, three additional measurement series were conducted [81].
Screening was performed using GDH-recycling and the obtained values are presen-
ted in Table 2; the previously obtained cell-free crude lysate results were included for
comparison. Lysozyme-lysed cells gave the highest consensus regarding the pre-
vious results, demonstrating that partition of insoluble cell components has only a
negligible benefit. Whole-cell samples revealed approximately 7-fold lower product
formation rates attributable to lower substrate accessibility to the enzyme as diffusion
or active transport of substrate through E. coli membranes is required. Nevertheless,
the expected activity pattern was fully reproduced, showing these samples to be like-
wise suited for the fluorescence-based assay. Interestingly, polymyxin B - although
its benefit was emphasized by Schwaneberg et al. in the course of the p-nitrophenol-
based P450 BM3 assay - had a negative effect by causing high background signals
compared to untreated whole-cells in the present assay [82].
Characterization of kinetic parameters
Kinetic parameters are important measures for the interaction of substrates with their
metabolizing enzymes. For characterization of surrogate substrate 4 and its ester
21

derivative 5, three different P450 BM3 variants were expressed, purified and used for
kinetic measurements. GV, F87V and GQ were chosen because they showed high,
medium and low O-dealkylation activities in the previous experiments. Enzyme con-
centrations ensuring linearity between enzyme amount and O-dealkylation activity
were determined in pre-tests for every enzyme-substrate-pair and the GQ-sub-
strate 5-pair was excluded from further investigations due to inappreciable activities.
Enzyme kinetics were analyzed by measuring initial product formation rates in the
presence of varied substrate amounts - the maximal value set according to pre-
viously determined solubility limits - using ADH-recycling (ensures pH stability).
Results for GV with substrate 4 and F87V with substrate 5 are representatively
shown in Fig. 3 and indicate inhibition of enzyme activity at high substrate concentra-
tions (for a full overview, see Fig. S6 and Table S1 in the Supporting Information).
The phenomenon of excess-substrate inhibition of P450s has repeatedly been
discussed in the literature and has mainly been explained through simultaneous bin-
ding of two substrate molecules to the enzyme blocking its activity [83,84]. However,
within the previous publications on fluorogenic P450 BM3 substrates, this behavior
was neither observed for 7-ethoxycoumarin or BCCE nor for resorufin ethers [39-44].
Correlation of the obtained data considering excess-substrate inhibition gave the
M i
kinetic parameters K , vmax, K and n as listed in Table 3. The best catalytic efficiency
-
1
-1
was 0.458 min µM for the GV mutant with substrate 4 and represents one of the
highest values ever observed for the P450 BM3-mediated O-dealkylation of fluoroge-
nic ethers [39-44]. In addition, the catalytic efficiencies for acid 4 exceeded those for
M
ester 5 by up to two orders of magnitude (even though the K values for the first
were considerably higher), which is attributable to the higher native-substrate-resem-
blance of substrate 4 over 5. The results further indicate a positive effect of the F87V
22

substitution on the P450 BM3 activity regarding both examined compounds, whereas
the A74G mutation seems to enhance enzyme activity exclusively for substrate 4.
Moreover, the high Hill coefficients obtained for acid 4 (n = 1.7 - 2.2) suggest positive
homotropic cooperativity, which is in accordance with the dimeric nature of the P450
BM3 enzyme, whereas cooperativity is significantly less distinct in case of ester 5
(
n = 1.4) [85].
Establishment of a competitive inhibition screening
The preceding survey provided enzyme-substrate-pairs for competitive inhibition
studies of P450 BM3 and the GV-substrate 4-pair was selected to validate the
assayꞌs ability for detecting inhibition caused by known P450 BM3 substrates, non-
substrates, and inhibitors. In detail, myristic acid (10) was selected as a native sub-
strate, naphthalene (11) and methyl 2-methoxy-3-methylbenzoate (12) representted
non-native substrates, ketoconazole (13), imidazole (14), and potassium
cyanide (15) were tested as inhibitors, and caffeine (16) plus isooctane (17) served
as non-substrates, as the latter are neither oxidized by P450 BM3 nor inhibit the
enzyme [39,49,86-89]. Fluorescence interference or quenching attributable to these
probe compounds was excluded a priori (data not shown). Moreover, K
M
values
serve as a guide for setting fluorogenic substrate concentrations for inhibition
assays, because high sensitivity toward inhibition along with high signal intensity is
assured at this concentration [59]. 10 µM of substrate 4 were used in the present
M
study according to the obtained K of 12.4 ± 2.5 µM. The inhibitory effect of probe
compounds on the O-dealkylation activity of GV toward 4 was quantified using
varying probe compound concentrations depending on respective inhibition capaci-
ties, the latter mirroring the compoundsꞌ affinities to the enzyme. The obtained
results are listed in Table 4 and attest to the inhibitory effect of all tested substrates
23

and inhibitors, whereas no significant influence on the reaction output was observed
for non-substrates. The presence of 1 µM of the highly potent inhibitor ketocona-
zole (13) lowered the reaction outcome to 65 ± 4% and at 5 µM the enzymeꞌs activity
was almost completely inhibited, whereas in case of the non-native substrate 12 con-
centrations ≥1 mM needed to be applied for categorical results. Nevertheless, prepa-
M
rative biotransformation of 12 was shown in a previous study (K : 1.8 ± 0.2 mM for
P450 BM3 mutant F87A/L188C, 0.6 mmol scale, 98% conversion, 73% yield [49])
and hence, the obtained results advise to use mM concentrations of non-natural sub-
strates in P450 BM3 inhibition assays as successful preparative transformations of
respective hits are likely, even though distinctly lower substance amounts are
common [39,41]. Furthermore, the F87V mutant was subjected to the same experi-
ment and the previously observed inhibition pattern was confirmed in the broadest
sense (Table 4). However, a closer look reveals varieties between the different P450
BM3 catalysts, as the extent of F87V inhibition was lower in case of imidazole (14)
and substrate 12 when compared to the GV results, whereas upon addition of
myristic acid (10) and naphthalene (11) higher inhibition was observed. These
M
differences are probably attributable to the different K values of both P450 BM3
mutants for the compounds tested and conversion studies indeed showed GV to be
the better catalyst regarding substrate 12 (data not shown). According to the
obtained results, the established assay provides a fast, valuable, universally appli-
cable and advantageous tool for substrate identification and allows limiting laborious
turnover studies to investigate P450 BM3-catalyzed oxygenation via GC, HPLC, MS
or NMR to those compounds that give positive results during this pre-screen. The
general applicability is further supported by the low failure rate of inhibition
screenings, which was reported to be below 0.8% (due to interference between
24

probe compounds and the formed fluorophores) by Fowler and Zhang based on a
study of 5000 substances [90]. To the best of our knowledge, this is the first time
coumarin-based substrates were used for inhibition studies with P450 BM3, which so
far were limited to resorufin-based substrates. This progress was mainly enabled by
M
the low K values obtained for the designed substrates (substrate 4: 10.8 - 12.4 µM,
substrate 5: 3.1 - 3.4 µM), which are closer to those known for resorufin-substrates
than for coumarin-derivatives and mirror the high native-substrate-resemblance of
our compounds (e. g. 7-benzyloxyresorufin: 1.9 µM [39]; 7-ethoxyresorufin: 3.0 -
6
1
.3 µM [41] and 7-allyloxyresorufin: 7.9 - 14.7 µM [42] vs. 7-ethoxycoumarin: 45 -
220 µM [43] and BCCE: 25.2 - 34.3 µM [44]). On the contrary, the higher K values
M
of thus far known coumarin-based substrates aggravated inhibition assays by
necessitating high surrogate substrate amounts to assure sufficient activities. In
addition, BCCE is only partly suited for inhibition assays as a hydrolysis step is
required in addition to the P450 BM3-catalyzed hydroxylation to release the fluoro-
phore and it is unknown which of these reactions is the rate-determining one.
Application to IC value determination
5
0
Despite of qualitative inhibition studies as shown in the preceding section, competi-
tive inhibition assays also support quantitative screens for determination of IC50
values, which are measures for the affinity of compounds regarding the tested
M d
enzyme and can be considered as comparable with K or K values [59]. To prove
the suitability of the established system for this purpose and to demonstrate the
applicability of enzyme-substrate-pairs revealing low O-dealkylation activity for inhibi-
tion assays, the F87A mutant was used along with substrate 4 to determine the
concentration-dependent inhibitory effect of benzoate 12. Inhibition measurements
were performed as described beforehand, except that the reaction time was exten-
25

ded and the concentration of substrate 4 was increased due to low enzyme activity
for kinetic parameters of F87A for substrate 4, see Fig. S6 and Table S1 in the
(
Supporting Information). Concentrations of benzoate 12 were set according to former
studies and plotting against the respective inhibition values gave the hyperbolic
curve shown in Fig. 4 [49]. Non-linear fitting revealed an IC50 value of 3.5 ± 0.4 µM,
M
which correlates well with the previously determined K value of 3.7 ± 0.3 µM
obtained in the conventional NADPH-consumption assay, thereby proving the estab-
lished assay to be applicable for characterization of P450 BM3 mutants toward sub-
strates and inhibitors [49].
Synthesis and characterization of a resorufin-derived analog
Resorufin-based screenings must be assumed as the main alternative regarding the
established approach. Thus, the final step of the present study was the synthesis of
a resorufin-derived analog of acid 4 to compare the synthetic useability of HTC 1 and
resorufin (18) and their suitability as markers for generation of P450 BM3 substrates.
As shown in Scheme 4 and according to the synthetic strategy described before,
compound 7 was used for alkylation of phenol 18 giving the desired ester 19 in 61%
yield. Noteworthy, higher solvent amounts and a longer reaction time were needed
compared to the foregoing reaction due to low solubility of resorufin (18) and its deri-
vatives. The subsequent saponification using KOH base failed due to decomposition
of the starting material under alkaline conditions. Alternatively, the more pricy bio-
catalyst CalB enabled hydrolysis at almost neutral pH yielding acid 20 in 88%. The
difficulties faced during this synthesis and the inferior overall yield of substrate 20 of
50% (compared to 88% achieved for the HTC-analog 4) portend a lower suitability of
resorufin (18) compared to HTC 1 for surrogate substrate synthesis from the chemi-
cal point of view and indicate that generation of more sophisticated substrates using
26

fluorophore 18 might be hampered. Moreover, decomposition of ether 19 was obser-
ved upon storage in solution (detectable by TLC) and hence, for the following experi-
ments freshly prepared substrate samples were used. This instability of resorufin
ethers has already been described by Klotz et al. in 1984; according to their studies
7-ethoxyresorufin stock solutions degraded to multiple UV-active products within
24 h [91]. Investigation of the optical properties of ethers 19 and 20 revealed that - in
contrast to their HTC-analogs - these compounds are not fluorescent (optimal
absorption wavelengths being λabs: 375 nm for 19 and λabs: 475 nm for 20) and
fluorophore 18 shows a smaller Stokes shift compared to 1 (Fig. 5). The solubility
limits were determined to be 5.5 µM for ether 19 and 9.7 µM for ether 20 and hence,
are 13 and 20 times lower compared to analogs 5 and 4, respectively (see Fig. S4 in
the Supporting Information). Activity measurements aiming at the detection of P450
BM3-mediated O-dealkylation of acid 20 failed due to degradation of this compound
under assay conditions (proven via absorption of 20 at 475 nm, data not shown).
Instability or P450 BM3-catalyzed oxygenation at the marker moiety were assumed
as the most probable reasons for this observation, both rendering application of
substrate 20 for O-dealkylation assays impossible. Hence, comparison of equally
etherified HTC 1 and resorufin (18) was subsequently planned via the esters 5 and
19. Pre-tests indicated that sufficient O-dealkylation of substrate 19 was solely
mediated by the GQ double mutant, which showed the lowest activity for the HTC-
labeled substrates beforehand, and respective kinetic analysis results are presented
in Fig. 6. Interestingly, for this substrate no excess-substrate inhibition was obser-
ved, but - in contrast to 5 - the sigmoidal curve shape suggests positive homotropic
cooperativity, which was confirmed by a Hill coefficient of n = 2.1 ± 0.1 upon fitting of
M
the obtained data. In addition, a K value of 1.8 ± 0.1 µM and a catalytic efficiency of
27

-
.458⋅10 min µM were determined, the latter being 13 to 33 times lower com-
3
-1
-1
0
pared to the values for ester 5. The differences observed regarding the P450 BM3-
mediated O-dealkylation of substrates 19 and 5 suggest different binding modes of
these compounds within the enzyme, which might be of great interest for future
protein engineering approaches. However, considering the obtained results, HTC 1
emerged as the better marker for the generation of fluorogenic P450 BM3 substrates
as corresponding ethers showed higher stability, greater activity and better solubility
compared to their resorufin analogs and provided a broader substrate spectrum for
successful O-dealkylation screenings.
Conclusions
A novel, fluorescence-based and high-throughput-suited P450 BM3 assay was
proven to be a promising, fast, easy, and reliable tool to carry out directed evolution
of BM3 or identify potential substrates of this biocatalyst via inhibition studies. It is
amenable to automation and adds a new feature to the existing panel of assays by
accelerating screening efforts and increasing economic plus resource efficiency
through substantial reagent conservation via high substrate yields and miniaturized
formats. In detail, HTC 1 was proven to be a superior basis for the synthesis of P450
BM3 substrates compared to resorufin (18), since the corresponding ethers could be
obtained in better yields and showed higher stability, activity and solubility plus wider
applicability. The solubility aspect is especially interesting regarding multi-substrate
applications (e. g. pooling of various substrates for enzyme fingerprints or assaying
ꞌpseudoꞌ-stereoisomers), which are further supported by the availability of coumarin
derivatives with slightly different spectral properties. Moreover, the obtained results
suggest that fluorophore 1 allows access to a broader substrate spectrum than
fluorophore 18. The designed, native-substrate-resembling, fluorogenic compounds
28

4
and 5 were fortunately hydroxylated by the tested P450 BM3 mutants under
various conditions with high activities and led to the lowest K values observed to
M
date for coumarin-based P450 BM3 substrates [43,44]. Thus, these substrates offer
valuable tools for directed evolution campaigns targeting P450 BM3 features such as
solvent resistance or (thermo)stability. In addition, according to Yun et al. the
demand for highly active bacterial enzymes with human-like P450 activities will
constantly grow to meet the needs of pharmaceutical research and bioremediation
industries [25]. In this, P450 BM3 mutants are highly promising as they have been
shown to fulfill these requirements and could repeatedly be engineered toward non-
natural substrates. The established screening system allows studying P450 BM3
inhibition aiming at the identification of new substrates and hence, offers a new
access to the desired mutants.
Acknowledgements
We thank the Ministry of Innovation, Science and Research of the German federal
state of North-Rhine-Westphalia as well as the Heinrich Heine University Düsseldorf
(
scholarship within the CLIB-Graduate Cluster Industrial Biotechnology for K. N.) and
are grateful to Shawn Whitfield for carefully revising the manuscript.
29