Flavocytochrome P450 BM3 mutant W1046A is a NADH-dependent fatty acid hydroxylase: Implications for the mechanism of electron transfer in the P450 BM3 dimer

Article abstract of DOI:10.1016/j.abb.2010.09.014

Bacillus megaterium P450 BM3 (BM3) is a P450/P450 reductase fusion enzyme, where the dimer is considered the active form in NADPH-dependent fatty acid hydroxylation. The BM3 W1046A mutant was generated, removing an aromatic "shield" from its FAD isoalloxazine ring. W1046A BM3 is a catalytically active NADH-dependent lauric acid hydroxylase, with product formation slightly superior to the NADPH-driven enzyme. The W1046A BM3 K _m for NADH is 20-fold lower than wild-type BM3, and catalytic efficiency of W1046A BM3 with NADH and NADPH are similar in lauric acid oxidation. Wild-type BM3 also catalyzes NADH-dependent lauric acid hydroxylation, but less efficiently than W1046A BM3. A hypothesis that W1046A BM3 is inactive [15] helped underpin a model of electron transfer from FAD in one BM3 monomer to FMN in the other in order to drive fatty acid hydroxylation in native BM3. Our data showing W1046A BM3 is a functional fatty acid hydroxylase are consistent instead with a BM3 catalytic model involving electron transfer within a reductase monomer, and from FMN of one monomer to heme of the other [12]. W1046A BM3 is an efficient NADH-utilizing fatty acid hydroxylase with potential biotechnological applications.

Full text of DOI:10.1016/j.abb.2010.09.014

Archives of Biochemistry and Biophysics 507 (2011) 75–85
Contents lists available at ScienceDirect
Archives of Biochemistry and Biophysics
journal homepage: www.elsevier.com/locate/yabbi
Flavocytochrome P450 BM3 mutant W1046A is a NADH-dependent fatty acid
hydroxylase: Implications for the mechanism of electron transfer in the P450
BM3 dimer
a
a
a
b
a
Hazel M. Girvan , Adrian J. Dunford , Rajasekhar Neeli , Idorenyin S. Ekanem , Timothy N. Waltham ,
c
a
b
a
a
d
M. Gordon Joyce , David Leys , Robin A. Curtis , Paul Williams , Karl Fisher , Michael W. Voice ,
a,
⇑
Andrew W. Munro
a
Manchester Interdisciplinary Biocentre, Faculty of Life Sciences, University of Manchester, 131 Princess Street, Manchester M1 7DN, UK
School of Chemical Engineering and Analytical Science, University of Manchester, 131 Princess Street, Manchester M1 7DN, UK
Department of Biochemistry, University of Leicester, Henry Wellcome Building, Lancaster Road, Leicester LE1 9HN, UK
Cypex Ltd., 6 Tom McDonald Avenue, Dundee DD2 1NH, UK
b
c
d
a r t i c l e i n f o
a b s t r a c t
Article history:
Available online 22 September 2010
Bacillus megaterium P450 BM3 (BM3) is a P450/P450 reductase fusion enzyme, where the dimer is con-
sidered the active form in NADPH-dependent fatty acid hydroxylation. The BM3 W1046A mutant was
generated, removing an aromatic ‘‘shield” from its FAD isoalloxazine ring. W1046A BM3 is a catalytically
active NADH-dependent lauric acid hydroxylase, with product formation slightly superior to the NADPH-
Keywords:
Cytochrome P450
Flavocytochrome P450 BM3
Electron transfer
FMN dissociation
Fatty acid hydroxylation
Coenzyme selectivity
m
driven enzyme. The W1046A BM3 K for NADH is 20-fold lower than wild-type BM3, and catalytic effi-
ciency of W1046A BM3 with NADH and NADPH are similar in lauric acid oxidation. Wild-type BM3 also
catalyzes NADH-dependent lauric acid hydroxylation, but less efficiently than W1046A BM3. A hypoth-
esis that W1046A BM3 is inactive [15] helped underpin a model of electron transfer from FAD in one BM3
monomer to FMN in the other in order to drive fatty acid hydroxylation in native BM3. Our data showing
W1046A BM3 is a functional fatty acid hydroxylase are consistent instead with a BM3 catalytic model
involving electron transfer within a reductase monomer, and from FMN of one monomer to heme of
the other [12]. W1046A BM3 is an efficient NADH-utilizing fatty acid hydroxylase with potential biotech-
nological applications.
Ó 2010 Elsevier Inc. All rights reserved.
Introduction
membrane anchor region [5]. This has meant that the prokaryotic
P450s have historically proven more amenable to high level
The cytochromes P450 (P450s) are heme b-containing enzymes,
the vast majority of which catalyze the reductive activation of
dioxygen bound to the heme iron, and the insertion of a single
atom of oxygen into a substrate molecule bound close to the heme
in the P450 active site. A cysteine thiolate proximal ligand is found
in all P450s, with a water (or hydroxide ion) usually located as the
distal ligand in the resting ferric form, and with this being replaced
by dioxygen on heme reduction and catalytic cycle initiation [1,2].
The P450s were the original enzyme ‘‘superfamily” and their num-
bers continue to expand with ongoing genome sequencing efforts
expression and purification, leading to structural characterization
by X-ray crystallography. As a consequence, selected bacterial
P450s are among the most intensively analyzed P450 isoforms,
with the camphor hydroxylase P450cam (CYP101A1) from Pseudo-
monas putida and the fatty acid hydroxylase P450 BM31
(CYP102A1) from Bacillus megaterium being arguably the best stud-
ied of all the P450s [6,7]. P450 BM3 (BM3) has a particularly impor-
tant position in the P450 superfamily as the first example of a
bacterial P450 that uses a eukaryotic-like redox partner (the
FAD- and FMN-binding enzyme NADPH-cytochrome P450 reduc-
tase, CPR) and the first P450 enzyme found fused to its redox part-
ner [7,8]. BM3 is a 119 kDa P450-CPR fusion that has the highest
(
e.g. 20 P450s in Mycobacterium tuberculosis and >250 P450s in Ara-
bidopsis thaliana) [3,4].
A schism in the P450 superfamily occurs between eukaryotic
and prokaryotic P450s, with the former being soluble enzymes
and the latter being membrane associated via a N-terminal trans-
1
Abbreviations used: BM3 or P450 BM3, flavocytochrome P450 BM3 from Bacillus
megaterium; BSTFA, N,O-bis(trimethylsilyl)trifluoroacetamide; CO, carbon monoxide;
CPR, NADPH-cytochrome P450 reductase; KPi, potassium phosphate; NAD(P)H, either
NADH or NADPH; NO, nitric oxide; NOS, nitric oxide synthase; P450, cytochrome
P450; TMCS, trimethylchlorosilane; WT, wild-type.
⇑
Corresponding author. Fax: +44 161 306 4192.
E-mail address: Andrew.Munro@Manchester.ac.uk (A.W. Munro).
0
003-9861/$ - see front matter Ó 2010 Elsevier Inc. All rights reserved.
doi:10.1016/j.abb.2010.09.014

7
6
H.M. Girvan et al. / Archives of Biochemistry and Biophysics 507 (2011) 75–85
turnover number reported for a P450 oxygenase enzyme (ca
the presence of NPG substrate and NADPH (but not NADH),
the oxygen consumption rates in the F87Y and W1046A mu-
tants were substantially greater than for their respective double
mutants in combination with G570D [15]. The G570D point
mutant of P450 BM3 was shown to have no measurable FMN
content or myristic acid hydroxylase activity in studies of Klein
and Fulco [16], and thus there appeared a likelihood that fatty
acid oxidase activity might be retained in one or both of the
W1046A and F87Y BM3 mutants.
ꢀ1
2
80 s with arachidonic acid), and the wild-type (WT) BM3 cata-
lyzes hydroxylation of long chain fatty acids, predominantly at the
-1 to -3 positions [9,10].
Studies by Black and co-workers highlighted that BM3 had a
x
x
quaternary structure, unlike any previously characterized bacte-
rial P450 enzyme [11]. In their study, gel filtration analysis indi-
cated that dimer as well as a number of higher oligomeric
species could be observed. Neeli et al. provided initial evidence
that the dimeric form of BM3 was the catalytically relevant fatty
acid hydroxylase species, with studies that included the demon-
stration that catalytically inactive point mutant enzymes A264H
WT P450 BM3, like CPR and other members of the diflavin
reductase enzyme family, shows a strong selectivity for NADPH
over NADH as its reducing coenzyme. The specificity constant
ꢀ
ꢀ
1
1
ꢀ1
ꢀ1
(
in which the His264 coordinates the heme iron to occlude the
(kcat/K
m
) in cytochrome c reduction assays is ꢁ262
l
l
M
M
min
min
oxygen binding site) and G570D (in which FMN binding is abol-
ished) could be reconstituted to regenerate fatty acid hydroxy-
lase activity [12]. The conclusion from the mutant studies was
that electron transfer should occur from NADPH to FAD and
then FMN within one monomer (A264H), and then from this
FMN across to the heme in the second monomer (G570D) with-
in the dimer (Fig. 1). This model of electron transfer within a
flavocytochrome dimer is the same as that proposed to occur
in nitric oxide synthase (NOS), which is also formed from the
fusion of a CPR-like module to a heme thiolate protein that cat-
with NADPH as the reducing coenzyme and ꢁ0.06
with NADH in WT human CPR (i.e. an approximately 4370-fold dif-
ference in favor of NADPH) [17]. In previous studies, Döhr and co-
workers generated the W676H and W676A mutants of human CPR,
where this tryptophan occupies the same position as W1046 in
P450 BM3 reductase with its aromatic side chain covering the re-
face of the FAD isoalloxazine ring. Importantly, Döhr et al. showed
that the W676A mutation effected a considerable (ꢁ1090-fold)
switch in coenzyme specificity from NADPH towards NADH, and
enabled the W676A CPR enzyme not only to function efficiently
as a reductase of ferricyanide and cytochrome c, but also to support
catalytic function of the human CYP1A2 enzyme [17]. While Kitaz-
ume et al. considered that the P450 BM3 W1046A mutant ‘‘should
lack NADPH-dependent activity”, evidence for absence of fatty acid
hydroxylation was not presented. Our own previous studies on the
W1046A mutant in the CPR domain of BM3 (residues 473–1048 of
the 1048 amino acid protein) and in the smaller FAD/NADPH-bind-
ing (ferredoxin reductase-like) domain of BM3 (residues 664–
1048) showed that the W1046A mutation (as did the W676A mu-
tant of human CPR) induced a substantial switch in coenzyme
selectivity towards NADH in both the BM3 W1046A CPR and
FAD/NADPH domain enzymes [18]. The shifts in catalytic efficiency
alyzes successive oxidations of L-arginine to generate L-citrulline
and nitric oxide (NO) as a signaling molecule [13,14]. Subse-
quently, Kitazume et al. also concluded that the BM3 dimer
was the functionally relevant species, but suggested that elec-
tron transfer could occur from the FAD of monomer 1 to the
FMN of monomer 2, and then either back to the heme of mono-
mer 1 or on to the heme of monomer 2 within the dimer [15].
This conclusion was also based on studies of mutant comple-
mentation to restore activity in heterodimeric enzymes, with
the F87Y mutant in the heme domain considered as inactive
in fatty acid oxidation (through disruption of P450 active site
structure) and the W1046A mutant considered ‘‘inactive” due
to alterations in the vicinity of the FAD cofactor. However, in
towards NADH (measured as the ratio of specificity constants, kcat
/
K
m
values, for NADH-dependent versus NADPH-dependent activi-
ties) were ꢁ1710-fold relative to WT for the W1046A reductase
in cytochrome c reduction and >5700-fold in ferricyanide reduc-
tion, even greater coenzyme specificity switches than were re-
ported for the comparable W676A mutant of human CPR [17,18].
In view of our preceding data and the absence of specific evi-
dence for the inactivity (or otherwise) of the W1046A flavocyto-
chrome P450 BM3 in fatty acid hydroxylation, we decided to
characterize this mutant of the full length P450 BM3 in order to
establish its spectroscopic and kinetic properties, and its compe-
tence as a NADPH- and NADH-dependent fatty acid hydroxylase.
The data presented in this paper demonstrate that W1046A P450
BM3 is a functional fatty acid oxidase with either NADH or NADPH
(
NAD(P)H) as reductant, providing further support to our previous
model of NOS-like FMN (monomer 1)-to-heme (monomer 2) elec-
tron transfer to support catalytic function in the catalytically rele-
vant P450 BM3 dimer. We also present data demonstrating that
FMN dissociation occurs in P450 BM3 and its reductase domain
as a consequence of its extended incubation at low concentrations,
and that this flavin dissociation results in loss of cytochrome c
reductase activity. These data also have important ramifications
for the development of a model of electron transfer in the flavocy-
tochrome P450 BM3 dimer, since previous studies by Kitazume et
al. resulted in the hypothesis that loss of cytochrome c reductase
activity during such incubations may instead be purely a conse-
quence of BM3 dimer dissociation [15]. Collectively, the data pre-
sented are consistent with our earlier studies on the BM3 dimer
and inter-monomer electron transfer, and simultaneously high-
light the production of an efficient, NADH-specific variant of the
P450 BM3 fatty acid hydroxylase enzyme.
Fig. 1. Schematic of putative electron transfer pathways in flavocytochrome P450
BM3. The dimeric form of P450 BM3 is required for efficient fatty acid hydroxylase
activity. The left side panel illustrates the organization of FAD and FMN cofactors (in
the CPR domain) and arrows indicate possible directions of NAD(P)H-derived
electron transfer through FAD and FMN to the heme iron in the P450 domain. The
most plausible electron transfer routes are either directly through cofactors within
each monomer (dotted arrows) or from the FMN cofactor in one monomer (FMN
1
)
to the heme iron in the other (heme ) (solid arrows), as appears to be the case in the
2
eukaryotic nitric oxide synthase (NOS) flavocytochromes [13]. The central panel
shows the NOS-like electron transfer pathway predicted in the heterodimer of
G570D and A264H mutants of P450 BM3. The former mutation prevents FMN
binding while the latter has His264 as a 6th ligand to the P450 heme iron. Both
mutants are inactive as fatty acid hydroxylases, but activity is recovered in the
1 2
heterodimer, consistent with FMN -to-heme electron transfer [12,38]. The right
side panel shows WT-like potential electron transfer pathways in the W1046A BM3
mutant. Data presented in this study demonstrate that this mutant enzyme is active
in fatty acid hydroxylation, contrary to conclusions made in a previous study [15].
In light of previous studies of the G570D/A264H mutants, the FMN
electron transfer model is also favoured for the W1046A BM3 mutant.
1 2
-to-heme

H.M. Girvan et al. / Archives of Biochemistry and Biophysics 507 (2011) 75–85
77
Materials and methods
Stopped-flow analysis of heme iron reduction in WT and W1046A
flavocytochromes P450 BM3
Generation of the W1046A mutant in flavocytochrome P450 BM3
Apparent flavin-to-heme electron transfer rates were measured
by stopped-flow absorbance spectroscopy using an Applied Photo-
physics SX18 MVR stopped-flow spectrophotometer contained
within an anaerobic glove box (Belle technology) to maintain an
oxygen free environment. To measure the rate of flavin (FMN)-
to-heme electron transfer, the accumulation of the Fe2+–CO com-
plex of the BM3 heme iron was monitored at 450 nm. All buffers
used were degassed and saturated with CO. Enzyme (WT or
The production of the W1046A variant of the P450 BM3 reduc-
tase domain was described in our previous publication [18]. Site-
directed mutagenesis to create the W1046A mutant of intact flavo-
cytochrome P450 BM3 was carried out as described previously,
using the QuikChange method (Stratagene) and the aforemen-
tioned primers [18]. The WT flavocytochrome P450 BM3
(CYP102A1) gene was used as the template in the expression plas-
mid pBM25 [19]. Introduction of the required mutation and the ab-
sence of undesired secondary mutations were verified by DNA
sequencing of the full gene (GeneService, London, UK).
W1046A BM3, 3
first syringe were then mixed against NAD(P)H (200
stopped-flow, and the 450 absorption transients collected were
l
M) and lauric acid (100
l
M) pre-mixed in the
lM) in the
D
A
fitted using either a monophasic or a biphasic exponential process
using the Applied Photophysics software. To examine the effect of
pre-incubation of WT and W1046A enzymes with NAD(P)H,
sequential mixing experiments were done. Pre-mixed lauric acid
Expression and purification of W1046A and WT flavocytochromes
P450 BM3, and the WT BM3 reductase domain
(
(
100
200
l
M) and enzyme (3
lM) were first mixed with NAD(P)H
The WT and W1046A flavocytochromes P450 BM3 were ex-
pressed in Escherichia coli strain TG1. Cells were grown in Terrific
Broth at 37 °C, harvested and lysed as described previously
lM) and incubated for either 100 ms or 60 s, prior to a second
mix with an equal volume of CO-saturated anaerobic buffer. The
450 transient data were then fitted using a biphasic exponential
D
A
[
9,20]. Enzymes were purified by successive chromatographic
function to determine the apparent rate constants for FMN-to-
heme electron transfer (kred).
steps using DEAE- and Q-Sepharose, followed by hydroxyapatite
chromatography and a final gel filtration step on Sephacryl S-
2
00, as previously [20,21]. The WT BM3 reductase domain was ex-
Analysis of flavin fluorescence and cytochrome c reductase activity in
diluted samples of WT flavocytochrome P450 BM3 and its CPR domain
pressed and purified as before [18]. EDTA (1 mM) and the protease
inhibitors phenylmethanesulfonyl fluoride (PMSF, 1 mM) and ben-
zamidine hydrochloride (1 mM) were kept in all buffers during
purification steps for the proteins. Concentrations for the oxidized
forms of purified WT and W1046A flavocytochromes P450 BM3
WT P450 BM3 and its CPR domain were incubated in either
5
mM KPi (pH 7.4) or 50 mM KPi (pH 7.4) in stock solutions at var-
ious concentrations in the range between 25 nM and 3 M at 25 °C
l
ꢀ1
ꢀ1
were determined using the coefficient
for the oxidized form of the WT reductase domain using
e418 = 105 mM cm , and
for a period of 1 h, and as described by Kitazume et al. [15]. There-
after, samples were taken from the incubations and assayed di-
rectly (in the case of the 25 nM sample) or diluted to a final
concentration of 25 nM from the other samples into the same buf-
fer. They were then assayed for cytochrome c reductase activity at
ꢀ1
ꢀ1
e
456 = 21.2 mM cm , as described previously [9,19,22,23].
UV–visible absorbance and substrate binding analysis of P450 enzymes
2
5 °C by addition of cytochrome c (35
l
M) and NADPH (200
l
)
M),
on
ꢀ1
ꢀ1
and by following A550 change
(
D
e
550 = 22.64 mM cm
All UV–visible absorbance measurements to analyze spectral
properties of the WT and W1046A variant of flavocytochrome
P450 BM3, and the WT BM3 CPR domain were done using a Cary
reduction of the cytochrome c heme iron from ferric to ferrous
9]. In parallel, the flavin fluorescence was measured in the final
[
2
4
5 nM samples (with no further additions) with excitation at
50 nm and emission spectra collected between 490 and 650 nm.
5
0 UV–visible scanning spectrophotometer (Varian) and 1 cm
pathlength quartz cuvettes. Analysis of the binding of arachidonic
acid, lauric acid, and N-palmitoylglycine (NPG) to the W1046A
BM3 enzyme as well as NAD(P)H-dependent reduction and sodium
dithionite reduction/CO (carbon monoxide)-binding assays were
done as previously [9,18]. Unless otherwise stated, assays were
done in standard assay buffer (50 mM potassium phosphate
Fluorescence spectra were all corrected for Raman scattering by
subtraction of a buffer-only spectrum in each case. Samples were
then boiled for 5 min, centrifuged in a microfuge at full speed for
1
min, and fluorescence spectra again recorded for the superna-
tants. Fluorescence spectra were all recorded for samples at
5 °C. UV–visible spectroscopic assays were done on a Cary 50
2
[
KPi], pH 7.0) at 25 °C. Spectral data fitting was done as previously
spectrophotometer (as described above). Fluorescence spectra
were recorded on a Cary Eclipse fluorimeter (Varian) using a
described, and using Origin software (OriginLab, Northampton,
MA).
1
cm pathlength quartz fluorescence cuvette and with excitation
and emission slits both set to 20 nm.
Steady-state fatty acid turnover assays with WT and W1046A
flavocytochromes P450 BM3
Fatty acid product characterization from lauric acid oxidation by WT
and W1046A P450 BM3 enzymes
The steady-state turnover of selected fatty acids (lauric acid and
N-palmitoylglycine [NPG]) was monitored indirectly by oxidation
of NADPH and NADH coenzymes on a Cary 50 spectrophotometer,
Analysis of oxygenated products from NAD(P)H-dependent
turnover of lauric acid by both WT and W1046A flavocytochromes
P450 BM3 was done both using LC with radiometric detection of
C labeled lauric acid and by GC–MS on unlabeled lauric acid fol-
lowing derivatization of reaction products using BSTFA [N,O-
ꢀ
1
ꢀ1
and using
determined with flavocytochrome enzymes (WT and W1046A)
typically at 450 nM, fatty acids (NPG at 50 M, lauric acid at
00
M) and NAD(P)H varied across a range up to ꢁ1.5 mM. Plots
of specific rates at different NAD(P)H concentrations were fitted to
the Michaelis–Menten equation to define apparent kcat and K
m
De340 = 6.21 mM cm . The kcat and K values were
1
4
l
5
l
bis(trimethylsilyl)trifluoroacetamide] and TMCS (trimethylchloro-
silane). For the radiolabelled lauric acid assays, 1
W1046A P450 BM3 (2 l volumes) was added to 158
mixed (873 M lauric acid, 12.7
potassium phosphate, pH 7.4. This was incubated at 37 °C for
l
M WT or
l of pre-
C lauric acid) in 100 mM
m
l
l
1
4
parameters in each case, using Origin software. Assays were done
in assay buffer at 25 °C.
l
lM

7
8
H.M. Girvan et al. / Archives of Biochemistry and Biophysics 507 (2011) 75–85
1
min prior to addition of coenzyme (40 ll of 5 mM NAD(P)H or
+
NADP as control). A control reaction was also done using the
FMN-free G570D mutant. Incubations were continued for various
points up to 2 min, and then stopped by addition of 25 ll of 1 M
HCl. The samples were centrifuged in a microfuge at 13,000 rpm
for 10 min, and supernatants analyzed by HPLC. Samples were re-
solved using an Agilent 1100 system on a Phenomenex hyperclone
3
l
m C18 ODS column (150 ꢂ 3.2 mm) with radiometric detection
using solid scintillant. Reactions for GC–MS analysis were done
using 1 M enzyme and 700 M sodium dodecanoate (lauric acid).
Reactions were also done with 300 M n-palmitoylglycine (NPG),
l
l
l
along with negative controls run in absence of added lipid sub-
strate or coenzyme. Reactions were in assay buffer and started
by addition of NAD(P)H (1 mM) with stirring at 25 °C. Reactions
were terminated at 10 s, 30 s, or 2 min by acidification with HCl,
as above. Thereafter, oxidized samples were isolated, derivatized,
and resolved as described in our previous work [24]. Total abun-
dance of different products (x-1, x-2, and x-3 hydroxy fatty acids)
was determined by integration of product peaks using system
software.
Materials
14C-labeled lauric acid was from Cypex (Dundee, UK). Bacterial
growth medium (Tryptone, yeast extract) was from Melford Labo-
ratories (Ipswich, Suffolk, UK). Unless otherwise state, all other re-
agents were from Sigma–Aldrich and were of the highest grade
available.
Results
Expression, purification, and optical properties of the P450 BM3
W1046A mutant
Fig. 2. UV–visible spectroscopic properties of the P450 BM3 W1046A mutant. (A)
Absorbance spectra for oxidized (thick solid line), NADH-reduced (75 M, thin solid
line), and sodium dithionite-reduced and CO-bound (dashed line) forms of the BM3
The W1046A mutation was successfully introduced into the in-
tact flavocytochrome P450 BM3 enzyme, and the mutant enzyme
isolated according to previously described methods [17,19,20].
The purified enzyme exhibited optical properties near-identical
to those for the WT BM3. The P450 is extensively low-spin (LS)
with its Soret band at 418.5 nm, and the absence of any significant
high-spin (HS) feature at ꢁ390 nm. The alpha and beta bands of the
W1046A BM3 heme are located at 569 and 535 nm, respectively.
The presence of flavin cofactors (FAD and FMN) is evident from
the substantial spectral contribution in the region between 450
and 540 nm, consistent with the spectral properties of the oxidized
WT P450 BM3 [19] (Fig. 2A). Previous studies of the W1046A CPR
and FAD/NADPH binding domain indicated that the FAD and FMN
cofactors were incorporated to the same levels as the WT proteins,
and also that these mutant enzymes had high activity in both cyto-
chrome c reductase and potassium ferricyanide reductase assays
l
W1046A enzyme (5.5 lM). There is no significant NAD(P)H-dependent formation of
the Fe2 –CO complex of the BM3 W1046A enzyme in the absence of fatty acid
+
substrate. (B) Absorbance spectrum for oxidized, substrate-free W1046A P450 BM3
(
5.9
spectra from the titration of the enzyme with arachidonic acid. The final spectrum
shown (8.25 M arachidonic acid) has absorbance maximum at 394 nm. Arrows
lM, spectrum with greatest intensity at 418.5 nm), accompanied by selected
l
indicate directions of absorbance change on arachidonic acid addition at selected
points in the W1046A BM3 UV–visible spectrum.
flavocytochrome (Fig. 2A). In common with the WT P450 BM3 en-
2
+
zyme, there is no significant formation of the Fe –CO complex on
addition of carbon monoxide to the NAD(P)H-reduced W1046A
P450 BM3 enzyme, demonstrating that these coenzymes reduce
the flavin cofactors, but not the heme iron. Conversion to a Fe2+
–
[
12]. Thus, preceding data were consistent with efficient
CO species with spectral maximum at 448 nm is achieved using so-
dium dithionite as a reductant. The spectrum of the W1046A CO
complex demonstrates retention of cysteine thiolate as the proxi-
mal ligand to the heme iron in this variant BM3 enzyme, as might
be expected for a mutation distant from the heme binding site
(Fig. 2A).
To demonstrate typical substrate-induced optical changes in
the W1046A BM3 heme, binding of various BM3 fatty acid sub-
strates (e.g. lauric acid, NPG, arachidonic acid) was done, and in
each case W1046A heme spectral changes occurred that were con-
sistent with a shift in ferric heme iron spin-state equilibrium from
low-spin to high-spin. Fig. 2B shows results from a spectral titra-
tion of the BM3 W1046A enzyme with arachidonic acid. There is
a progressive shift in the Soret maximum from 418.5 to 394 nm
as arachidonic acid binds, along with the development of a charge
NAD(P)H-dependent electron transfer to substrates that receive
electrons mainly from either the FAD (ferricyanide) and the FMN
(
cytochrome c) cofactors. Addition of either NADPH or NADH to
the W1046A flavocytochrome resulted in the same extent of spec-
tral bleaching in the 450–540 nm region, consistent with substan-
tial reduction of the flavin cofactors (Fig. 2A). The absence of any
significant change in absorbance in the 600 nm region on P450
BM3 W1046A reduction by NAD(P)H is indicative of a lack of accu-
mulation of the neutral (blue) semiquinone form of the FAD. Previ-
ous studies of the W1046A CPR and FAD/NADPH binding domains
showed that this mutation destabilized the FAD blue semiquinone
(
formed in the WT enzymes) in favor of the 2-electron reduced
hydroquinone FAD form [18]. Spectral data for the W1046A P450
BM3 enzyme indicate that this property is retained in the intact

H.M. Girvan et al. / Archives of Biochemistry and Biophysics 507 (2011) 75–85
79
transfer band at 648 nm that is typical for substrate-bound P450s.
A data fit for the arachidonic acid-induced change in the heme
(Fig. 2A), in addition to evidence provided from our previous stud-
ies on the WT and W1046A CPR and FAD/NADPH binding domains
[18]. There are alterations in the relative potentials of the oxidized/
semiquinone (ox/sq, 0- to 1-electron reduced) and semiquinone/
hydroquinone (sq/hq, 1- to 2-electron reduced) couples of the
FAD cofactor in the W1046A mutant as a consequence of the re-
moval of the ‘‘shielding” tryptophan 1046 side chain over the
FAD isoalloxazine ring. As a result, the equilibrium level of
NAD(P)H-dependent FAD reduction is predominantly to the semi-
quinone (1-electron reduced) in the WT, and to the hydroquinone
in the W1046A mutant [18]. To examine the influence of the
W1046A mutation on the apparent rate of NAD(P)H-dependent
electron transfer to the BM3 heme iron, we measured the rate of
formation of the heme Fe –CO complex following reduction of
lauric acid-bound WT and W1046A BM3 enzymes with NAD(P)H
in CO-saturated, anaerobic buffer, and as described in Materials
and methods. Studies of the influence of brief (100 ms) and ex-
tended (60 s) pre-incubation of enzymes with NAD(P)H were also
done, in light of the kinetic preference for BM3 heme iron reduc-
tion by the short-lived FMN anionic radical species, and the known
accumulation of the FMN hydroquinone (a much poorer reductant
of the heme) on incubation of WT BM3 (or its CPR domain) with
NADPH in the absence of fatty acid substrate or other electron
acceptor, such as cytochrome c [28,29]. Data from these experi-
ments are shown in Table 2.
spectrum produces a K
previously reported values for the WT P450 BM3 enzyme (e.g.
.55 M) [9,24]. The W1046A BM3 K values for lauric acid and
NPG were 216 ± 14 M and 0.13 ± 0.01 M, respectively. These
d
value of 0.24 ± 0.07 lM, consistent with
0
l
d
l
l
are similar to values reported previously for the binding of these
lipids to WT P450 BM3, suggesting that active structure in the
BM3 heme domain is not influenced substantially by the
W1046A mutation in the FAD/NADPH binding domain [25,26].
Overall, W1046A BM3 spectral data indicate the enzyme has full
heme and flavin cofactor content, and that it undergoes coenzyme
reduction, substrate binding, and CO complexation reactions typi-
cal of the WT enzyme.
2
+
Steady-state kinetic analysis of the W1046A flavocytochrome P450
BM3 enzyme
To determine the apparent kinetic parameters for substrate
turnover by the W1046A flavocytochrome P450 BM3, the sub-
strates lauric acid and NPG were analyzed. Substrates were kept
at near-saturating concentrations, and substrate-dependent oxida-
tion of NADPH and NADH coenzymes was measured at 340 nm. Ta-
ble 1 presents the apparent kcat and K
studies. For the WT P450 BM3 enzyme, the kcat values are an order
of magnitude greater with NADPH than with NADH, and the K
values for the coenzymes are substantially in favor of NADPH.
For instance, with NPG as substrate the K is >180-fold lower for
WT BM3 with NADPH (3.4 M) than with NADH (625 M). How-
m
parameters from these
The results show that electron transfer occurs from both NADH
and NADPH coenzymes through the FAD and FMN cofactors, and to
the heme iron in both the WT and W1046A flavocytochromes,
leading to CO binding to the ferrous heme iron and accumulation
of absorbance at 450 nm associated with the characteristic P450
form of this cysteine thiolate coordinated enzyme [30]. In all cases,
m
m
l
l
ever, there are striking differences in the catalytic parameters with
the W1046A BM3 mutant. While the W1046A kcat values with
NADPH as reducing coenzyme are lower than those for WT P450
BM3 (ꢁ12.5-fold and 8-fold with lauric acid and NPG, respec-
tively), the kcat value with NADH is comparable to that for WT with
lauric acid as substrate, and ꢁ9-fold higher with NPG. Moreover,
2
+
near-complete accumulation of the P450 Fe –CO complex oc-
curred, albeit with widely varying rate constants according to en-
zyme, reductant and pre-incubation with coenzyme (Table 2). In
all but one case, reaction transients fitted best to biphasic expo-
nential processes, with amplitudes of absorbance similar in both
the fast and slow phases of the reactions. In all cases, observed rate
constants (kred) reported are the averages of at least three individ-
ual transients that varied by <5% from the averages reported in Ta-
ble 2. In recent fast reaction studies using laser excitation methods,
we demonstrated that CO binding to ferrous P450 BM3 heme iron
the W1046A K
M with NPG) are substantially lower than those for the WT
BM3 with the same substrates (20-fold and 10-fold, respectively)
Table 1). Using the specificity constant (kcat/K ) as a measure of
m
values for NADH (11.5 lM with lauric acid,
62 l
(
m
relevant efficiency of the WT and W1046A BM3 enzymes with
NADPH, there are apparent ꢁ16.7-fold and 11.3-fold decreases in
efficiency of the W1046A mutant (cf. WT) with lauric acid and
NPG, respectively. However, with NADH as the reducing coenzyme
there are ꢁ13.0-fold and 91.4-fold increases in efficiency (cf. WT)
with lauric acid and NPG, respectively.
ꢀ1
occurs with a rate constant in excess of 1000 s at saturating CO
conditions in solution. This is much faster than BM3 heme iron
reduction by electron transfer from the FMN cofactor and thus un-
likely to be a rate-limiting factor in the CO complex formation [31].
In previous stopped-flow studies of WT P450 BM3, we reported
that the observed rate constant for CO complex formation is
Stopped-flow absorbance analysis of heme iron reduction in WT and
W1046A flavocytochromes P450 BM3
ꢀ
1
ꢁ
224 s for myristic acid-bound P450 BM3 reduced by near-satu-
rating NADPH (500 M). We concluded that the first FMN-to-heme
electron transfer was a major rate-limiting step in the P450 BM3
l
Early studies on the P450 BM3 system from Fulco’s group dem-
onstrated that NADPH was the preferred coenzyme to support
BM3’s fatty acid hydroxylase activity [27]. However, both NADPH
and NADH are able to transfer electrons to P450 BM3, as shown
by the ability of these coenzymes to effect a similar level of reduc-
tion of the WT and W1046A BM3 flavin cofactors at equilibrium
ꢀ1
catalytic cycle, since 224 s is close to the apparent kcat for arachi-
ꢀ1
donic acid oxidation (285 s ) under similar conditions [9,23]. In
the current study (Table 2), data are again consistent with this con-
clusion, with the fast phase of NADPH-dependent heme reduction
(kred 1) for WT BM3 (without coenzyme pre-incubation) being
Table 1
Steady-state kinetic parameters for substrate-dependent NAD(P)H oxidation in WT and W1046A flavocytochrome P450 BM3 enzymes. Experiments were done as detailed in
Materials and methods.
Substrate
P450 BM3 WT
NADPH
P450 BM3 W1046A
NADH
cat (min
NADPH
NADH
cat (min
69.5 ± 4
726 ± 59
cat (minꢀ1)
K
(lM)
k
ꢀ1
)
K
(
lM)
kcat (min
ꢀ1
)
K
(
l
M)
k
ꢀ1
)
k
m
m
m
m
K (lM)
Lauric acid
NPG
1040 ± 96
1343 ± 97
6.5 ± 2.5
3.4 ± 1.1
106 ± 24
80 ± 15
228 ± 107
625 ± 170
83.5 ± 18
172 ± 7
8.7 ± 0.6
4.9 ± 1.0
11.5 ± 2.5
62 ± 13

8
0
H.M. Girvan et al. / Archives of Biochemistry and Biophysics 507 (2011) 75–85
Table 2
Stopped-flow kinetic data for NADPH- and NADH-dependent heme iron reduction in WT and W1046A flavocytochrome P450 BM3 enzymes. The parameters kred 1 and kred
relate to rate constants for the fast and slow phases from fits of 450 reaction progress curves with a double exponential function. In the case of the W1046A BM3 mutant with
2
D
A
NADPH (and in absence of pre-incubation with coenzyme), reaction transients were monophasic. The absorbance increase at 450 nm relates to electron transfer to the heme iron
and the rapid binding of CO to the ferrous heme to form the archetypal cysteine thiolate-coordinated Fe2 –CO P450 complex. Reactions were initiated either by (i) direct mixing of
lauric acid-bound BM3 enzymes with NAD(P)H in CO-saturated buffer, or by (ii) pre-mixing lauric acid-bound BM3 enzymes with NAD(P)H in degassed buffer for 100 ms or 60 s
prior to a second stopped-flow mix with CO-saturated buffer, as described in Materials and methods.
+
Coenzyme pre-incubation time
Coenzyme
P450 BM3 WT
P450 BM3 W1046A
red 1 (s 1
ꢀ
)
kred 2 (s
ꢀ1
)
kred 1 (s
ꢀ1
)
kred 2 (s
ꢀ1
)
k
0
NADH
NADPH
NADH
NADPH
NADH
NADPH
52.3
12.8
33.0
4.0
6.1
0.003
0.003
18.1
4.5
14.7
16.0
0.0075
0.0105
3.8
–
1.8
3.9
0.001
0.0015
291
35.9
61.4
0.012
0.013
1
6
00 ms
0 s
ꢀ
ꢀ
1
2
5
91 s . The comparable WT rate with NADH reductant is lower at
W1046A BM3 that precedes formal FAD reduction and oxidized
coenzyme release. Such phenomena were observed previously in
stopped-flow reduction of W1046A FAD/NADPH and CPR domains
[18]. The movement of the indole side chain of Trp1046 is required
to enable binding of the nicotinamide portion of NAD(P)H close to
the FAD isoalloxazine ring and in order for electron (formally hy-
dride ion) transfer, as was shown in studies of the role of Trp676
in human CPR, and further confirmed in studies of the W1046A
FAD/NADPH and CPR domain mutants [18,32]. The conclusion that
the Trp1046 side chain is also important for displacement of
1
2.3 s , but clearly demonstrates that reduction of the fatty acid-
bound BM3 heme is feasible with NADH (Fig. 3). Rates are slower
again in the W1046A mutant, but reduction is apparently faster
with NADH (18.1 s ) than with NADPH (4.5 s ), and the
transients are also essentially monophasic in W1046A with NADPH
as reductant. Following a short (100 ms) pre-incubation with
reducing coenzyme, the WT BM3 heme reduction rates are slowed
to 61.4 s (NADPH) and 35.9 s (NADH), which would be consis-
tent with accumulation of the FMN hydroquinone in this timescale.
For W1046A BM3, a 100 ms pre-incubation with NADH decreases
ꢀ
1
ꢀ1
DA
450
ꢀ1
ꢀ1
+
NAD(P) was apparent from the development of CT species in these
ꢀ1
the heme reduction rate by ꢁ20% to 14.7 s . However, with
W1046A domains [18]. The presence of other side chains interact-
ꢀ1
0
NADPH the apparent rate is increased (16.0 s ) and reaction tran-
sients become biphasic again. Closer inspection of the data re-
corded for NADPH-dependent W1046A heme reduction reveals a
short lag phase for reaction transients in absence of pre-incubation
with coenzyme (20–30 ms) that is much less pronounced in the
other comparable transients in absence of coenzyme pre-mixing
ing with the 2 -phosphate group of NADP(H) (which is absent in
NAD(H)) may offer some additional stabilization to the binding
+
+
of NADP (compared to NAD ) to explain the brief lag phase in
W1046A BM3 heme reduction transients that occur in the absence
of pre-incubation with NADPH.
For the WT and W1046A mutants of BM3, extended (60 s) pre-
incubation with NAD(P)H leads to a dramatic decrease in heme
(
Fig. 3) and barely discernible for any of the reaction transients fol-
lowing the 100 ms pre-incubation with NADPH. Possibly, the
apparent lag phase and slower electron transfer rate in the non-
pre-incubated reaction is due to the formation of a charge transfer
CT) complex between NADP and hydroquinone FAD (FADH
ꢀ1
ꢀ1
reduction rate to ꢁ0.01 s (ꢁ6 min ) in each case. In the anaero-
bic solutions used for these experiments, there is no possibility of
any substantial oxidation of coenzyme in the pre-incubation phase,
+
(
2
) in
and this conclusion is also borne out by the fact that the
DA
450
transients are of similar final intensity in all of the experimental
2
+
sets (i.e. near-complete P450 Fe –CO complex formation is
achieved). The logical conclusion drawn is that the 60 s pre-incu-
bation with NAD(P)H enables the CPR domains of both WT and
W1046A flavocytochromes P450 BM3 to become fully loaded with
electrons in accordance with the relative reduction potentials of
NAD(P)H and the FAD/FMN flavins. Both the FMN ox/sq and sq/
hq redox couples and the FAD ox/sq couple are considerably more
+
positive than that for NAD(P)H/NAD(P) (ꢀ320 mV versus the nor-
mal hydrogen electrode), while the FAD sq/hq couple is slightly
more negative in the case of WT BM3, but more positive in the
W1046A enzyme [18,28,33]. As a consequence, these equilibrium
forms are ones in which the BM3 FAD either populates a mixture
of semiquinone and hydroquinone (WT) or is near-fully hydroqui-
none (W1046A), whereas the FMN is effectively completely in the
hydroquinone state in both WT and the W1046A mutant. The FMN
hydroquinone is an inferior reductant of the heme (compared to
the semiquinone), leading to the much lower rates of Fe2+–CO
complex formation observed (Table 2) and converting P450 BM3
to a low activity fatty acid hydroxylase form [8].
Fig. 3. Stopped-flow absorbance analysis of Fe2+–CO complex formation in WT and
W1046A P450 BM3 enzymes. Stopped-flow transients at 450 nm are shown,
reporting on heme iron reduction in WT and W1046A P450 BM3, followed by rapid
CO binding to form the P450 Fe –CO complex. Reactions were done as described in
Materials and methods, and without pre-incubation of lauric acid-bound enzymes
with NAD(P)H. The thick solid line represents the reduction of WT BM3 with
NADPH, and the thin solid line is WT reduced with NADH. The dotted line is
W1046A BM3 reduced with NADH and the dashed line is W1046A reduced with
NADPH. Rate constants derived from these data are presented in Table 2.
Demonstration of NADPH- and NADH-dependent fatty acid
hydroxylation in WT and W1046A flavocytochromes P450 BM3
2
+
Following the demonstration of NAD(P)H-dependent reduction
of the heme iron in both WT and W1046A P450 BM3, we analyzed
the capacity of these enzymes to catalyze hydroxylation of lauric

H.M. Girvan et al. / Archives of Biochemistry and Biophysics 507 (2011) 75–85
81
acid using both NADH and NADPH coenzymes as reductants. In
preliminary work we examined the enzyme-mediated oxidation
(
x
-2 hydroxylauric acid), and 13.29 min (
x
-1 hydroxylauric acid).
Fig. 4B shows the mass spectrum of the product species at
12.99 min, corresponding to the TMS-derivative of -3 hydroxyl-
auric acid. Fig. 5 is a scheme showing the proportions of the vari-
ous -1 to -3 hydroxylauric acid products generated from the
1
4
of C labeled lauric acid by HPLC, as detailed in the Materials
and methods section. Using the system described, unconverted
lauric acid substrate eluted at ꢁ14.6 min, with three products
x
x
x
(
i.e. the
x
-1,
x
-2, and
x
-3 hydroxylauric acid metabolites) eluting
various enzyme/coenzyme combinations in reactions over both
30 s and 2 min. The data here are consistent with those from the
HPLC studies reported above, confirming that the WT BM3 with
NADPH is the most efficient combination, converting >90% of lauric
acid to hydroxylated products in 30 s. These data thus contradict
previous suggestions that the W1046A flavocytochrome BM3 mu-
tant is inactive as a fatty acid hydroxylase [15]. In fact, W1046A
BM3 is active with both NADPH and NADH as reductants, and is
rather more efficient with NADH as the coenzyme. In the 30 s incu-
bation with NADH, the W1046A BM3 enzyme converts ꢁ69% of
in a closely spaced group at ꢁ6.5–7.7 min. Products were observed
for both WT and W1046A P450 BM3 and with both NADH and
NADPH as reductants. While quantification of individual metabo-
lites was difficult due to a lack of resolution in their retention
times, what was clear was that the WT BM3 with NADPH reductant
was the most efficient hydroxylation system, with ꢁ90% of the
substrate converted to hydroxylated products within 30 s. In addi-
tion, the W1046A BM3 mutant hydroxylated a greater proportion
of lauric acid with NADH coenzyme than did the WT BM3 enzyme
over the same time period. In control experiments, it was also
established that no oxidized lauric acid products were generated
in the absence of enzyme or coenzymes, and that the FMN-free
G570D mutant of flavocytochrome P450 BM3 was completely inac-
tive in lauric acid hydroxylation, entirely consistent with our pre-
vious studies of BM3 dimer formation [12].
lauric acid to the
x-1 to x-3 hydroxylauric acid products, and this
proportion rises to 88% following a 2 min incubation. Also of note is
that the WT BM3 enzyme is clearly capable of NADH-dependent
lauric acid hydroxylation, albeit less efficiently than the W1046A
enzyme driven by NAD(P)H. In turnover driven by NADH, WT
BM3 converts ꢁ20% of lauric acid to hydroxylated products in
30 s, increasing to ꢁ52% product formation over 2 min (Fig. 5). In
related experiments, we could also demonstrate the NAD(P)H-
dependent oxidation of NPG substrate by both WT and W1046A
P450 BM3 enzymes. However, in addition to the three products ex-
In order to achieve better resolution of the different hydroxyl-
ated products and to obtain a more quantitative analysis of their
individual abundance, we used GC–MS methods, as described in
Materials and methods, and in our previous studies of P450 BM3
[
24]. Data from these analyses further confirmed the production
pected (the TMS derivatives of x-1 to x-3 NPG), GC–MS revealed a
of -1, -2, and -3 hydroxylauric acid metabolites from both
x
x
x
further three products that likely resulted from fragmentation of
the former trio of oxidized products at the GC–MS analysis stage,
or during derivatization of the products (data not shown). Given
that WT BM3 is less efficient in NADH-dependent lauric acid
hydroxylation than is W1046A BM3 with NAD(P)H, it appears to
be the case that the apparent rates of P450 heme iron reduction
(where NADH-dependent heme iron reduction of WT BM3 was fas-
ter than NAD(P)H-dependent heme iron reduction in W1046A
WT/W1046A BM3 enzymes, and using both NADPH and NADH
coenzymes. Fig. 4A shows a typical gas chromatogram of reaction
products – in this case from the turnover of lauric acid by the
W1046A BM3 enzyme with NADH as the reducing coenzyme.
Unconverted lauric acid has a retention time of 11.25 min, and
the three expected metabolites of BM3-mediated substrate oxida-
tion are seen at 12.99 min (x-3 hydroxylauric acid), 13.18 min
Fig. 4. NADH-dependent hydroxylation of lauric acid catalyzed by the W1046A mutant of flavocytochrome P450 BM3. Panel A shows a gas chromatogram demonstrating
formation of -1, -2, and -3 hydroxylauric acid products from lauric acid substrate catalyzed by the W1046A P450 BM3 mutant and with NADH as the reducing coenzyme.
x
x
x
Reaction conditions were as described in Materials and methods, and the reaction was stopped after 10 s. Products and unconverted substrate were treated to form TMS
derivatives and resolved by gas chromatography, as described previously [24]. The identities of substrate/product peaks and their retention times are indicated. The minor
species with retention time of 13.07 min is a contaminant (possibly a plasticizer) that is also observed in negative control reactions where no other products are formed. Panel
B shows the mass spectrum of the product species with GC peak retention time of 12.99 min. This product corresponds to TMS derived
x-3 hydroxylauric acid.

8
2
H.M. Girvan et al. / Archives of Biochemistry and Biophysics 507 (2011) 75–85
Fig. 5. Proportions of products formed by WT and W1046A BM3-catalyzed lauric acid hydroxylation. The scheme shows the proportions of
x-1, x-2, and x-3 hydroxylauric
acid products (and unconverted lauric acid substrate) in reactions with WT and W1046A BM3 enzymes using NAD(P)H as reductants. Reactions were done as described in
Materials and methods, and stopped after incubation periods of either 30 s or 2 min. The data indicate WT BM3 with NADPH reductant is the most efficient lauric acid
hydroxylase, but that the W1046A BM3 mutant with NADH reductant is also an efficient lauric acid hydroxylase, and that WT P450 BM3 also has NADH-dependent fatty acid
oxidation activity.
BM3) need not correlate well with fatty acid hydroxylation out-
comes (Fig. 3). Thus, WT BM3 is likely to show a greater level of
uncoupling of NADH oxidation from hydroxylated product forma-
tion than does W1046A BM3 with NAD(P)H coenzymes.
Collectively, this data set demonstrates that (a) WT P450 BM3 is
a catalytically functional fatty acid hydroxylase with both NADH
and NADPH coenzymes (although NADPH is the preferred reduc-
tant); (b) W1046A P450 BM3 is also catalytically active in fatty acid
hydroxylation, using both NADPH and NADH, but with a modest
preference for NADH; and (c) W1046A P450 BM3 is markedly
superior to WT P450 BM3 in NADH-driven fatty acid
hydroxylation.
highlighted that the FMN cofactor is relatively weakly bound to
P450 BM3, and performed supplementation of the assay mixtures
with FMN (250 nM) in attempts to ensure that enzyme remained
replete with FMN, finding that this treatment did not result in
recovery of BM3 fatty acid hydroxylase activity. The experimental
approach of Kitazume et al. also varied from the one reported in
Neeli et al. in terms of the extensive incubation period for the di-
luted (10 nM to 3 lM) P450 BM3 of 1 h prior to assay, and the
use of a weak buffer system for the incubation step and enzyme as-
say (5 mM KPi, pH 7.4) [12,15]. In view of these differences in ap-
proach, we considered that dissociation of the FMN cofactor from
P450 BM3 could occur during the prolonged incubations of highly
dilute enzyme samples performed prior to the cytochrome c reduc-
tion experiments by Kitazume et al., and that this could lead to loss
of the activity as a consequence of the bound FMN being an obli-
gate requirement for efficient electron transfer to cytochrome c
FMN dissociation from P450 BM3 and its CPR domain following
extended incubation at low enzyme concentration
[
15]. Thus, we replicated the conditions of Kitazume et al. and
The data provided here demonstrating the catalytic competence
of the W1046A P450 BM3 enzyme contradict earlier conclusions
that this enzyme should be non-functional in fatty acid hydroxyl-
ation [15]. In turn, this leads to the requirement for a re-evaluation
of the models of electron transport within the P450 BM3 dimer
examined the flavin fluorescence in samples of P450 BM3 and its
CPR domain incubated for 1 h at concentrations between 25 nM
to 3 lM, with all samples then diluted to 25 nM and fluorescence
spectra read immediately.
The results in Fig. 6 (main panel) show the increase in flavin-
specific fluorescence for these 25 nM samples of WT P450 BM3
and its CPR domain, with data collected following the 1 h incuba-
tions at the indicated protein concentration. The data are corrected
to show the maximal percentage change in fluorescence (recorded
in the samples pre-incubated at 25 nM for both P450 BM3 and
CPR) as 100%, and relative to that for the lowest fluorescence signal
in each data set as 0%. In this way, the data for the P450 BM3 and
CPR domains can be directly overlaid, revealing the same pattern of
progressively increasing flavin fluorescence in those 25 nM en-
zyme samples derived from stock solutions of WT P450 BM3/CPR
domain incubated for 1 hour at 750 nM or less. The inset in Fig. 6
shows an example of the difference in flavin fluorescence between
(
see Discussion section). In the study of Kitazume et al., it was re-
ported that the specific rate of cytochrome c reduction catalyzed by
WT P450 BM3 (10–25 nM enzyme in assay) decreased progres-
sively in line with a concentration range (10 nM to 3 lM) in which
the enzyme was pre-incubated in 5 mM KPi (pH 7.4) for 1 h in ad-
vance of the assay. The authors concluded that the reason for this
apparent loss of cytochrome c reductase activity was that the di-
meric form of flavocytochrome P450 BM3 was required not only
for fatty acid hydroxylation (through the heme cofactor), but also
for cytochrome c reduction (via the FMN cofactor in the CPR do-
main) [15]. While we were able to replicate the cytochrome c activ-
ity loss results of Kitazume et al. using their assay conditions (and
also in 50 mM KPi, pH 7.4), there is a notable difference in the out-
come of this experiment by comparison to similar studies done
previously [12]. In our earlier work, Neeli et al. also investigated
the influence of dilution of P450 BM3 on the cytochrome c reduc-
tase specific activity, and found that there was only a small
decrease in the specific activity at the lowest enzyme concentra-
tions tested. In contrast, there were substantial decreases in P450
BM3’s specific activity for lauric acid-dependent NADPH oxidation
as the enzyme concentration was decreased below 20 nM in the
assay. Neeli et al. concluded that the most logical explanation for
P450 BM3 activity loss was the dissociation of its dimer into mono-
mers inactive in fatty acid hydroxylation in the concentration
range below 20 nM [12]. However, in this earlier paper, we also
samples of P450 BM3 pre-incubated at 2
lM (lower spectrum) and
2
5 nM (upper spectrum) for 1 h, prior to dilution and immediately
reading fluorescence emission at a protein concentration of 25 nM.
The increased fluorescence originates from free FMN that dissoci-
ates from the enzyme incubated at 25 nM, and as a consequence
of the K
the enzymes are pre-incubated. In related work, we have estab-
lished (from fluorescence quenching studies) that the FMN K for
d
for FMN binding being in the same range as that in which
d
its domain in P450 BM3 is ꢁ90 nM (Waltham, T.N., Ph.D. thesis,
University of Manchester, UK). The fluorescence data sets for both
the intact WT P450 BM3 and its CPR domain are highly similar, and
show a midpoint for fluorescence change of ꢁ100 nM, consistent

H.M. Girvan et al. / Archives of Biochemistry and Biophysics 507 (2011) 75–85
83
coenzyme [18]. Point mutations at residue Trp1046 (whose side
chain covers the re-face of the BM3 FAD isoalloxazine ring) had
spectacular effects on coenzyme selectivity in BM3, improving
the NADH K
the W1046A BM3 CPR domain (from 12.8 mM to ꢁ50
improving its catalytic efficiency (kcat/K
) with NADH by ꢁ2200-
m
>250-fold in cytochrome c reductase reactions by
l
M) and
m
fold [18]. This paper provides a detailed characterization of the cat-
alytic capacity of the W1046A mutant of the intact flavocyto-
chrome P450 BM3, and demonstrates that the enzyme is a
competent fatty acid hydroxylase using both NADPH and NADH
as reducing coenzymes, but with rather greater efficiency in lauric
acid hydroxylation using NADH. We also demonstrate here that the
WT P450 BM3 enzyme is active in lauric acid hydroxylation using
NADH, although the activity is weaker than that of the W1046A
mutant with NAD(P)H (Fig. 5). Nonetheless, this demonstrates
clearly that consecutive NADH-dependent electron transfers occur
between the CPR domain FMN cofactor and the BM3 heme iron (in
both WT and the W1046A mutant) in order to generate a reactive
ferryl-oxo species (compound 1) that catalyzes the hydroxylation
reaction. In addition, the regioselectivity of lauric acid hydroxyl-
Fig. 6. FMN dissociation from WT flavocytochrome P450 BM3 and its CPR domain
on extended incubation at low protein concentration. The main panel shows the
proportional increase in flavin fluorescence from samples of WT flavocytochrome
P450 BM3 (filled circles) and WT BM3 reductase (CPR) domain (open triangles) in
samples pre-incubated at the indicated protein concentration for 1 h in 5 mM KPi
buffer (pH 7.4), and prior to dilution to 25 nM in the same buffer for fluorescence
measurements. The sample with lowest flavin fluorescence in each group was set to
ation (x-1, x-2, and x-3 hydroxylauric acid products formed) is
retained with both NADH and NADPH as reductants and in both
WT/W1046A BM3 enzymes, confirming that this is a feature con-
trolled by the heme domain of BM3 itself.
0
% and the sample with highest fluorescence (the samples pre-incubated at 25 nM
While the proof of NADH-driven fatty acid hydroxylation and
enhancement of efficiency of this reaction in the W1046A BM3 en-
zyme are of interest in their own right and have possible biotech-
nological ramifications, these data also impact on our models of
the electron transfer mechanism in P450 BM3. It is now widely ac-
cepted that the dimeric state of the enzyme is the form active in
fatty acid hydroxylation [11,12,15]. In 2005, Neeli et al. [12] pub-
lished evidence for the functional BM3 dimer, concluding that a
in both cases) was set to 100%. This allowed the direct overlay of the data sets for
P450 BM3 and CPR domain samples. The specific flavin fluorescence was relatively
constant in the BM3 and CPR domain samples pre-incubated in the range from
ꢁ
750 nM to 3
concentrations, due to dissociation of FMN from the proteins. The midpoint (50%
value) occurs at ꢁ100 nM in both cases, consistent with the FMN K for the BM3
enzyme. The inset shows flavin fluorescence spectra for samples of WT P450 BM3
pre-incubated at protein concentrations of 2 M (dashed line) and 25 nM (solid
lM, but increased progressively in samples pre-incubated at lower
d
l
line), with fluorescence measurements made at an enzyme concentration of 25 nM
in both cases.
1 2
NOS-like electron transfer pathway from FMN -to-heme (and vice
versa) was likely, based on the restoration of lauric acid hydroxy-
lase activity in the heterodimer formed between inactive G570D
and A264H mutants (Fig. 1). Kitazume et al. then proposed a differ-
d
with the predicted FMN K . Subsequent to taking fluorescence
readings of the P450 BM3 and CPR domain samples, all were trans-
ferred to sealed Eppendorf tubes and boiled for 10 min, prior to
centrifugation to pellet debris and cooling to 25 °C prior to re-
determining flavin fluorescence as before. In all cases, flavin fluo-
rescence increased to approximately the same level (no more than
ent model in which electron transfer occurs from FAD
1 2
-to-FMN
(
and vice versa) with the possibility of electron transfer from
FMN to either heme in the dimer, with an assumption made that
the W1046A BM3 mutant is inactive in fatty acid hydroxylation
[
15]. However, their own data set did reveal substantial oxygen
1
0% greater) than that for the relevant intact protein samples
consumption by W1046A BM3 in presence of NPG and NADPH
whose fluorescence was measured after 1 h incubation at 25 nM.
This is consistent with both the complete removal of FMN by heat
denaturation of the protein samples, and with the much lower
fluorescence of FAD (ꢁ9-fold) due to quenching from its adenine
ring [34]. Thus, we conclude that loss of specific activity of P450
BM3 and its CPR domain in cytochrome c reduction likely arises
predominantly from dissociation of the crucial FMN cofactor on ex-
ꢀ1
(
specific rate of ꢁ75 min ), whereas the W1046A/G570D mutant
is essentially devoid of such activity [15]. Our studies reported
herein (under similar conditions, but using a substrate-dependent
NADPH oxidation assay) indicate quite similar kcat values of
ꢀ1
ꢀ1
8
3.5 min (lauric acid) and 172 min (NPG) to the rate reported
by Kitazume et al. (Table 1). Klein and Fulco demonstrated previ-
ously that the G570D BM3 mutant was completely depleted of
FMN and had no myristic acid hydroxylase activity [16], a result
confirmed by Neeli et al. in studies of lauric acid oxidation [12].
In combination with the results presented in this paper (demon-
strating catalytic properties of the W1046A BM3 mutant), it ap-
pears clear that the W1046A/G570D mutant is inactive purely as
a consequence of the G570D mutation and its effect of preventing
FMN binding. Thus, the conclusion that the FAD/NADPH binding
domain of P450 BM3 is inactivated in the W1046A mutant is inva-
lid, as shown both by (i) our earlier studies on the catalytic compe-
tence of the W1046A FAD/NADPH domain in ferricyanide
reduction and of the W1046A CPR domain in cytochrome c reduc-
tion (also demonstrating efficient FAD-to-FMN electron transfer
reactivity in the latter case) [18]; and (ii) by the results presented
in this paper which prove that a functional pathway of electron
transfer from NAD(P)H through FAD, FMN and onto heme iron oc-
curs in the W1046A BM3 enzyme to support lauric acid (and NPG)
hydroxylation. In view of our previous studies of the restoration of
tended enzyme incubation at concentrations less than ꢁ1
lM,
rather than from dissociation of a P450 BM3 (or CPR domain) di-
mer into monomers [15]. These data are discussed further in the
Discussion section below.
Discussion
The flavocytochrome P450 BM3 system has been studied for
more than 30 years following the recognition of a
x-2 palmitic acid
hydroxylase activity in cell extracts of B. megaterium [27]. The
activity was subsequently shown to reside on a 119 kDa enzyme
formed from the fusion of a soluble P450 hemoprotein to the first
recognized example of a soluble, bacterial CPR enzyme [8,35]. Like
eukaryotic CPR enzymes, BM3 shows a strong preference for
NADPH over NADH as its reducing coenzyme, although there has
been interest in engineering the BM3 CPR domain to facilitate tigh-
ter binding of NADH and to switch selectivity towards this cheaper

8
4
H.M. Girvan et al. / Archives of Biochemistry and Biophysics 507 (2011) 75–85
fatty acid hydroxylase activity in the A264H/G570D BM3 heterodi-
mer, this leads to a conclusion (see Fig. 1) that a likely electron
that the BM3 monomer should be inactive in cytochrome c reduc-
tion, given that the monomeric forms of eukaryotic CPR enzymes
are almost certainly proficient catalysts of cytochrome c reduction
and since our data herein are consistent with interflavin electron
transfer pathway in P450 BM3 should be FAD
as was postulated for the nitric oxide synthase flavocytochromes
13]. Kitazume et al. suggested that a functional A264H/G570D
BM3 heterodimer could be explained by the more complex elec-
tron transfer model of FAD -to-FMN -to-heme [15]. However,
1 1 2
-to-FMN -to-heme ,
[
transfer within a BM3 monomer (i.e. FAD
mation of monomers inactive in cytochrome c reduction would be
consistent with a model of obligate FAD -to-FMN electron trans-
1 1
-to-FMN ). However, for-
1
2
1
1
2
inherent in the development of this model was an assumption that
the W1046A mutation inactivates P450 BM3. In this paper we
show W1046A BM3 to be a functional fatty acid hydroxylase with
NAD(P)H coenzymes.
fer that could be peculiar to the BM3-type system. That said, an
alternative explanation for the loss of P450 BM3 cytochrome c
reductase activity following extended incubations of P450 BM3 en-
zyme at concentrations of 1 lM and lower would be that the more
However, Kitazume et al. also considered amino acid sequence
alignments of P450 BM3 and related flavocytochromes with those
of mammalian CPR enzymes, pointing out that the peptide regions
that likely form the ‘‘linker” between the FMN and FAD/NADPH do-
mains are shorter in the former group. The question was thus
raised as to whether a BM3-type linker was sufficiently long to en-
weakly bound flavin cofactor (the FMN) dissociates from the en-
zyme, disrupting the obligatory pathway of electron transfer from
protein-bound FMN to cytochrome c. The data shown in Fig. 6
show this to be the case, with enhanced flavin fluorescence in both
WT P450 BM3 (and BM3 CPR) samples incubated at concentrations
<750 nM occurring as a consequence of FMN dissociation from its
protein binding site and into the surrounding buffer. Thus, data
collected in this study suggest that loss of cytochrome c reductase
activity in highly diluted samples of P450 BM3 and its CPR domain
results primarily from loss of the FMN cofactor required as a con-
duit of electrons to the cytochrome c heme, and is probably unre-
lated to monomerization of the enzymes.
able intra-monomer electron transfer (FAD
ter-monomer electron transfer might be favoured (FAD
15]. The crystal structure of rat CPR provided a paradigm for this
1
-to-FMN
1
), whereas in-
1
-to-FMN )
2
[
family of diflavin reductase enzymes [36]. The modular structure
of CPR revealed the evolutionary adaptations made in this type
+
of ferredoxin NADP -reductase (FNR)/flavodoxin (FLD) fusion en-
zyme that enable the close docking of these domains to facilitate
the close approach of their FAD and FMN cofactors. In the CPR
structure of Wang et al., the FAD and FMN isoalloxazine rings are
separated by only ꢁ4 Å, entirely consistent with direct interflavin
electron transfer within a CPR monomer [36]. While different con-
formations of the domains are clearly required in rat CPR (and
other diflavin reductases) to enable the reduced FMN-binding do-
main to move away from its FAD/NADPH domain partner and to
communicate with a P450 protein, it is not immediately obvious
why the electron transfer system in P450 BM3 should be different
from that in rat CPR and instead require cross-interactions of flavin
Conclusion
The data presented in this manuscript demonstrate clearly that
the W1046A mutant of flavocytochrome P450 BM3 is a catalyti-
cally competent fatty acid hydroxylase exhibiting regioselectivity
of lauric acid hydroxylation the same as that observed for the
WT P450 BM3 enzyme. W1046A P450 BM3 is shown to be active
with both NADH and NADPH coenzymes, and to favor slightly
NADH in terms of quantities of hydroxylauric acid products formed
in unit time (Figs. 4 and 5). These data are consistent with our pre-
vious studies to engineer catalytic efficiency with NADH into the
W1046A BM3 reductase and FAD/NADPH domain enzymes [18],
and have potential biotechnological applications for the exploita-
tion of the P450 BM3 W1046A mutant in NADH-dependent sub-
strate oxidations. Importantly, the WT P450 BM3 enzyme was
also shown to catalyze NADH-dependent lauric acid hydroxylation,
albeit with much lower efficiency than WT BM3 with NADPH as
reductant, or than W1046A BM3 with NAD(P)H. These findings
lead to a re-evaluation of models of electron transport within the
P450 BM3 dimer, and are consistent with our previous model of
electron transfer from NAD(P)H occurring to FAD then FMN within
monomer 1 of the dimer, but then from this FMN to the heme iron
in monomer 2 of the dimer, as is also postulated to occur in NOS
enzymes.
1 2
domains across the dimer (FAD with FMN and vice versa). In re-
cent studies, we have determined the crystal structure of the BM3
FAD/NADPH binding domain (Joyce, M.G., Ph.D. thesis, University
of Leicester, UK) and we thus examined its interactions with the
FMN-binding domain of P450 BM3 [37]. We conclude that, in ab-
sence of specific knowledge on the relative docking orientations
of these BM3 domains and the precise domain boundaries, no ro-
bust conclusion can yet be drawn in favor of either an intra-mono-
1 1 1 2
mer (FAD -to-FMN ) or inter-monomer (FAD -to-FMN ) electron
transfer pathway. However, it is clear that the same type of inter-
domain interface as formed in rat CPR is also feasible between BM3
FAD/NADPH and FMN domains, placing the flavin isoalloxazine
dimethyl groups in close proximity. In absence of compelling
structural evidence to the contrary, we consider the CPR-like
intra-monomer electron transfer model most likely for P450 BM3
(
FAD
sistent with our data sets (and in light of the catalytic competence
of W1046A BM3) is FAD -to-FMN -to-heme (Fig. 1). While we
cannot rule out that an electron transfer pathway can occur
entirely within one monomer of the BM3 dimer (i.e. FAD -to-
FMN -to-heme ), we favor the NOS-like FAD -to-FMN -to-heme
1 1
-to-FMN ). Thus, a model of P450 BM3 electron transfer con-
Acknowledgment
1
1
2
This work was supported by research grant awards from the
Biotechnology and Biological Sciences Research Council, UK (grant
numbers BB/F00252/1 and BB/F00883X1).
1
1
1
1
1
2
scheme based on data presented in this paper and in our previous
study [12].
References
A further consideration with respect to rationalizing the func-
tional properties of P450 BM3 in its dimeric form also emerged
from earlier studies of Kitazume et al., where loss of cytochrome
c reductase activity was observed in P450 BM3 samples incubated
for 1 h at low protein concentrations. It was suggested that this
phenomenon may reflect the separation of the BM3 dimer. While
such changes in dimer–monomer equilibrium are likely to occur
at the lower end of the spectrum of protein concentrations used
in their study, we were uncertain of the validity of a conclusion
[
1] I.G. Denisov, T.M. Makris, S.G. Sligar, I. Schlichting, Chem. Rev. 105 (2005)
2253–2277.
[2] A.W. Munro, H.M. Girvan, K.J. McLean, Nat. Prod. Rep. 24 (2007) 585–609.
[
[
[
3] K.J. McLean, A.W. Munro, Drug Metab. Rev. 40 (2008) 427–446.
4] S.M. Paquette, K. Jensen, S. Bak, Phytochemistry 70 (2009) 1940–1947.
5] E.P. Neve, M. Ingelman-Sundberg, Curr. Opin. Drug Discov. Devel. 13 (2010)
78–85.
6] T.L. Poulos, Biochem. Biophys. Res. Commun. 312 (2003) 35–39.
7] A.W. Munro, D.G. Leys, K.J. McLean, K.R. Marshall, T.W. Ost, S. Daff, C.S. Miles,
S.K. Chapman, D.A. Lysek, C.C. Moser, C.C. Page, P.L. Dutton, Trends Biochem.
Sci. 27 (2002) 250–257.
[
[

H.M. Girvan et al. / Archives of Biochemistry and Biophysics 507 (2011) 75–85
85
[
[
8] L.O. Narhi, A.J. Fulco, J. Biol. Chem. 261 (1986) 7160–7169.
9] M.A. Noble, C.S. Miles, S.K. Chapman, D.A. Lysek, A.C. MacKay, G.A. Reid, R.P.
Hanzlik, A.W. Munro, Biochem. J. 339 (1999) 371–379.
[24] H.M. Girvan, C.W. Levy, P. Williams, K. Fisher, M.R. Cheesman, S.E.J. Rigby, D.
Leys, A.W. Munro, Biochem. J. 427 (2010) 455–466.
[25] T.W. Ost, C.S. Miles, J. Murdoch, Y. Cheung, G.A. Reid, S.K. Chapman, A.W.
Munro, FEBS Lett. 486 (2000) 173–177.
[26] S.S. Boddupalli, T. Oster, R.W. Estabrook, J.A. Peterson, J. Biol. Chem. 267 (1991)
10375–10380.
[27] Y. Miura, A.J. Fulco, J. Biol. Chem. 249 (1974) 1880–1888.
[28] S.C. Hanley, S. Daff, Biochem. Biophys. Res. Commun. 325 (2004) 1418–
1423.
[
[
[
10] Y. Miura, A.J. Fulco, Biochim. Biophys. Acta 388 (1975) 305–317.
11] S.D. Black, S.T. Martin, Biochemistry 33 (1994) 12056–12062.
12] R. Neeli, H.M. Girvan, A. Lawrence, M.J. Warren, D. Leys, N.S. Scrutton, A.W.
Munro, FEBS Lett. 579 (2005) 5582–5588.
13] U. Siddhanta, A. Presta, B. Fan, D. Wolan, D.L. Rousseau, D.J. Stuehr, J. Biol.
Chem. 273 (1998) 18950–18958.
[
[
[
14] D.J. Stuehr, J. Santolini, Z.Q. Wang, C.C. Wei, S. Adak, J. Biol. Chem. 279 (2004)
[29] M.B. Murataliev, M. Klein, A.J. Fulco, R. Feyereisen, Biochemistry 36 (1997)
8401–8412.
[30] T. Omura, R. Sato, J. Biol. Chem. 239 (1964) 2370–2378.
[31] H.M. Girvan, D.J. Heyes, N.S. Scrutton, A.W. Munro, J. Am. Chem. Soc. 129
(2007) 6647–6653.
3
6167–36170.
15] T. Kitazume, D.C. Haines, R.W. Estabrook, B. Chen, J.A. Peterson, Biochemistry
6 (2007) 11892–11901.
4
[
[
16] M.L. Klein, A.J. Fulco, J. Biol. Chem. 268 (1993) 7553–7561.
17] O. Döhr, M.J. Paine, T. Friedberg, G.C. Roberts, C.R. Wolf, Proc. Natl. Acad. Sci.
USA 98 (2001) 81–86.
[32] A. Gutierrez, O. Döhr, M. Paine, C.R. Wolf, N.S. Scrutton, G.C. Roberts,
Biochemistry 39 (2000) 15990–15999.
[
[
[
[
18] R. Neeli, O. Roitel, N.S. Scrutton, A.W. Munro, J. Biol. Chem. 280 (2005) 17634–
[33] S.N. Daff, S.K. Chapman, K.L. Turner, R.A. Holt, S. Govindaraj, T.L. Poulos, A.W.
Munro, Biochemistry 36 (1997) 13816–13823.
[34] A.W. Munro, M.A. Noble, Methods Mol. Biol. 131 (1999) 25–48.
[35] L.O. Narhi, A.J. Fulco, J. Biol. Chem. 262 (1987) 6683–6690.
[36] M. Wang, D.L. Roberts, R. Paschke, T.M. Shea, B.S. Masters, J.J. Kim, Proc. Natl.
Acad. Sci. USA 94 (1997) 8411–8416.
[37] I.F. Sevrioukova, H. Li, H. Zhang, J.A. Peterson, T.L. Poulos, Proc. Natl. Acad. Sci.
USA 96 (1999) 1863–1868.
[38] H.M. Girvan, H.E. Seward, H.S. Toogood, M.R. Cheesman, D. Leys, A.W. Munro, J.
Biol. Chem. 282 (2007) 564–572.
1
7644.
19] J.S. Miles, A.W. Munro, B.N. Rospendowski, W.E. Smith, J. McKnight, A.J.
Thomson, Biochem. J. 288 (1992) 503–509.
20] H.M. Girvan, K.R. Marshall, R.J. Lawson, D. Leys, M.G. Joyce, J. Clarkson, W.E.
Smith, M.R. Cheesman, A.W. Munro, J. Biol. Chem. 279 (2004) 23274–23286.
21] H.M. Girvan, H.S. Toogood, R.E. Littleford, H.E. Seward, W.E. Smith, I.S. Ekanem,
D. Leys, M.R. Cheesman, A.W. Munro, Biochem. J. 417 (2009) 65–76.
22] O. Roitel, N.S. Scrutton, A.W. Munro, Biochemistry 42 (2003) 10809–10821.
23] A.W. Munro, S. Daff, J.R. Coggins, J.G. Lindsay, S.K. Chapman, Eur. J. Biochem.
[
[
2
39 (1996) 403–409.