Glutamate-haem ester bond formation is disfavoured in flavocytochrome P450 BM3: characterization of glutamate substitution mutants at the haem site of P450 BM3.

Abstract

Bacillus megaterium flavocytochrome P450 BM3 (CYP102A1) is a biotechnologically important cytochrome P450/P450 reductase fusion enzyme. Mutants I401E, F261E and L86E were engineered near the haem 5-methyl group, to explore the ability of the glutamate carboxylates to form ester linkages with the methyl group, as observed for eukaryotic CYP4 relatives. Although no covalent linkage was detected, mutants displayed marked alterations in substrate/inhibitor affinity, with L86E and I401E mutants having lower Kd values for arachidonic acid and dodecanoic (lauric) acid than WT (wild-type) BM3. All mutations induced positive shifts in haem Fe(III)/Fe(II) potential, with substrate-free I401E (-219 mV) being >170 mV more positive than WT BM3. The elevated potential stimulated FMN-to-haem electron transfer ~2-fold (to 473 s-1) in I401E, and resulted in stabilization of Fe(II)O2 complexes in the I401E and L86E P450s. EPR demonstrated some iron co-ordination by glutamate carboxylate in L86E and F261E mutants, indicating structural plasticity in the haem domains. The Fe(II)O2 complex is EPR-silent, probably resulting from antiferromagnetic coupling between Fe(III) and bound superoxide in a ferric superoxo species. Structural analysis of mutant haem domains revealed modest rearrangements, including altered haem propionate interactions that may underlie the thermodynamic perturbations observed. The mutant flavocytochromes demonstrated WT-like hydroxylation of dodecanoic acid, but regioselectivity was skewed towards omega-3 hydroxydodecanoate formation in F261E and towards omega-1 hydroxydodecanoate production in I401E. Our data point strongly to a likelihood that glutamate-haem linkages are disfavoured in this most catalytically efficient P450, possibly due to the absence of a methylene radical species during catalysis.

Full text

Page 1

Structure and catalysis in P450 BM3 mutants
1
Glutamate-heme ester bond formation is disfavoured in
flavocytochrome P450 BM3

Characterization of glutamate substitution mutants at the heme site of P450 BM3


1Hazel M. Girvan, 1Colin W. Levy, 1Paul Williams, 1Karl Fisher 2Myles R. Cheesman,
Accepted Manuscript
1Stephen E. J. Rigby, 1David Leys, and 1Andrew W. Munro
1Faculty of Life Sciences, Manchester Interdisciplinary Biocentre, University of Manchester,
131 Princess Street, Manchester M1 7DN, UK. 2School of Chemical Sciences, University of
East Anglia, Norwich NR4 7TJ, UK.
Running title: Structure and catalysis in P450 BM3 mutants
Address correspondence to: A. W. Munro or H. M. Girvan. Tel: +44 161 3065151; Fax: +44
161 3068918; E-mail: Andrew.Munro@Manchester.ac.uk or
Hazel.Girvan@Manchester.ac.uk




Running Title: Structure and catalysis in P450 BM3 mutants
Biochemical Journal Immediate Publication. Published on 24 Feb 2010 as manuscript BJ20091603
THIS IS NOT THE VERSION OF RECORD - see doi:10.1042/BJ20091603
Licenced copy. Copying is not permitted, except with prior permission and as allowed by law.
© 2010 The Authors Journal compilation © 2010 Portland Press Limited

Page 2

Structure and catalysis in P450 BM3 mutants

2
Synopsis
Bacillus megaterium flavocytochrome P450 BM3 (BM3, CYP102A1) is a biotechnologically
important cytochrome P450/P450 reductase fusion enzyme. Mutants I401E, F261E and L86E
were engineered near the heme 5-methyl group, to explore ability of the glutamate carboxylates
to form ester linkages to the methyl group, as observed for eukaryotic CYP4 relatives. Although
no covalent linkage was detected, mutants displayed marked alterations in substrate/inhibitor
affinity, with L86E and I401E mutants having lower Kd values for arachidonic acid and lauric
acid than wild-type (WT) BM3. All mutations induced positive shifts in heme Fe(III)/Fe(II)
potential, with substrate-free I401E (-219 mV) being >170 mV more positive than WT BM3. The
elevated potential stimulated FMN-to-heme electron transfer ~2-fold (to 473 s-1) in I401E, and
resulted in stabilization of Fe(II)O2 complexes in the I401E/L86E P450s. EPR demonstrated
some iron coordination by glutamate carboxylate in L86E/F261E mutants, indicating structural
plasticity in the heme domains. The Fe(II)O2 complex is EPR silent, likely resulting from
antiferromagnetic coupling between Fe(III) and bound superoxide in a ferric superoxo species.
Structural analysis of mutant heme domains revealed modest rearrangements, including altered
heme propionate interactions that may underlie thermodynamic perturbations observed. The
mutant flavocytochromes demonstrated WT-like hydroxylation of lauric acid, but regioselectivity
was skewed towards ω-3 hydroxylaurate formation in F261E and towards ω-1 hydroxylaurate
production in I401E. Our data point strongly to a likelihood that Glu-heme linkages are
disfavoured in this most catalytically efficient P450, possibly due to the absence of a methylene
radical species during catalysis.







Keywords: cytochrome P450, electron paramagnetic resonance, ferrous-oxy complex, electron
transfer, substrate oxidation, potentiometry.

Abbreviations: BM3 – Bacillus megaterium flavocytochrome P450 BM3; CN – sodium cyanide;
CYP – cytochrome P450; KPi – potassium phosphate; NPG – N-palmitoylglycine; PDA –
photodiode array; PEG – polyethylene glycol; P450 – cytochrome P450; WT – wild-type; 4-PIM
– 4-phenylimidazole.
Biochemical Journal Immediate Publication. Published on 24 Feb 2010 as manuscript BJ20091603
THIS IS NOT THE VERSION OF RECORD - see doi:10.1042/BJ20091603
Accepted Manuscript
Licenced copy. Copying is not permitted, except with prior permission and as allowed by law.
© 2010 The Authors Journal compilation © 2010 Portland Press Limited

Page 3

Structure and catalysis in P450 BM3 mutants

3
Introduction
The cytochromes P450 (P450s) are a diverse enzyme superfamily, most of whose members
catalyse reductive scission of molecular oxygen and the introduction of an oxygen atom into an
array of organic substrates [1]. The 57 human P450s are involved in e.g. xenobiotic and
prescribed drug metabolism, steroid and lipid metabolism [2], while bacterial isoforms have roles
in e.g. synthesis of polyketides, biotin biogenesis, degradation of organic molecules for energy
and modification of various lipids [3-6]. Activation of molecular oxygen is achieved at a b-type
heme iron that is proximally coordinated to the protein via a cysteine ligand in the thiolate form
[7]. The P450 heme iron is ferric in the resting state, but is reduced to ferrous by a redox partner
(usually containing iron-sulfur or flavin redox cofactors) [8]. Oxygen binds to ferrous iron and
further reduction/protonation of the ferrous-oxy complex leads to the formation of reactive
intermediates, and ultimately to a ferryl-oxo species that catalyses oxygen insertion into a
substrate bound in the P450 active site [1,9].

While c-type heme macrocycles are covalently linked to their apoproteins by thioether bonds to
heme vinyl groups, involving (usually) cysteine side chains, most b-type hemes are not
covalently attached to the protein by linkages other than axial ligands to the heme iron [10]. This
was thought to be exclusively the case for the P450s until studies revealed covalent binding of the
heme macrocycle in mammalian members of the CYP4 family [11,12]. These bonds formed
through turnover-dependent linkage of a conserved acidic residue (a glutamate) with a heme
methyl group [13]. The possibility of stabilizing heme binding (or avoiding perturbing heme iron
thiolate coordination and inducing inactive P420 formation) in other P450s was investigated by
rationally engineering a glutamate residue close to a heme methyl in the Pseudomonas putida
P450cam (CYP101A1) camphor hydroxylase and the Bacillus megaterium P450 BM3
(CYP102A1) fatty acid hydroxylase [14,15]. P450 BM3 (BM3) is more closely
structurally/functionally related to eukaryotic CYP4s than is P450cam. It is a soluble enzyme
with its redox partner (cytochrome P450 reductase or CPR) fused to the P450 in a single
polypeptide, and has potential biotechnological applications [6]. For the G248E P450cam mutant,
extensive turnover with camphor produced ~40% covalent ligation of the heme group, but
inactivated the enzyme [14]. No significant amount of covalent binding of the heme group
occurred for the BM3 A264E mutant [15]. However, the BM3 A264E mutant instead formed a
distal glutamate ligation to the heme iron [15]. While this finding was exploited to generate
several other novel heme ligand sets through further mutagenesis (e.g. Lys/Met/His distal ligands
in A264K/M/H mutants) [16,17], it remained unclear whether linkages between the BM3 heme 5-
methyl group and P450 glutamate residues could be formed if glutamates were engineered at
alternative positions in the heme binding pocket. To investigate further, we engineered glutamate
residues as close as possible to the heme 5-methyl group with minimal disruption to protein
structure and function. Certain residues close to heme methyls (e.g. Thr268, Phe393 and Cys400)
were not selected for mutation, as variants at these positions would perturb function by affecting
catalytic proton relay (Thr268), regulation of heme iron potential and reactivity (Phe393) and
proximal ligation of the heme iron (Cys400) [18-20]. Residues Leu86, Phe261, and Ile401 were
found to fulfill the proximity criteria, and there was not other evidence for their essentiality for
e.g. folding or catalysis. Indeed, recent studies of a BM3 I401P mutant indicated that this enzyme
had enhanced affinity (Km) for and catalytic activity (kcat) towards the lipid substrate lauric acid
[21].

In this paper we report spectroscopic, structural and thermodynamic properties of the L86E,
F261E and I401E mutants of the intact flavocytochrome P450 BM3 enzyme (where the P450 is
fused to its FAD- and FMN-containing CPR partner) and of its heme (P450) domain. Each
mutation was generated due to predicted proximity of the introduced glutamate residue to the
heme 5-methyl group, and in view of their distinctive predicted orientations towards the methyl
Biochemical Journal Immediate Publication. Published on 24 Feb 2010 as manuscript BJ20091603
THIS IS NOT THE VERSION OF RECORD - see doi:10.1042/BJ20091603
Accepted Manuscript
Licenced copy. Copying is not permitted, except with prior permission and as allowed by law.
© 2010 The Authors Journal compilation © 2010 Portland Press Limited

Page 4

Structure and catalysis in P450 BM3 mutants

4
group. The results demonstrate ineffective covalent modification of the heme macrocycle in these
BM3 mutants by comparison with the previously studied mammalian CYP4s. However, the
mutations at distinct positions around the BM3 heme have major effects on thermodynamic and
catalytic properties of BM3, and provide important information on conformational regulation and
on determinants of stabilization of the ferrous iron oxygen complex.


Experimental

Site directed mutagenesis. Point mutants F261E, L86E and I401E were made in both the heme
domain and intact flavocytochrome plasmid constructs (pBM20 and pBM25, respectively
[22,23]), using the Stratagene QuikChange mutagenesis system. Oligonucleotide primers
L86EFor (CAGGAGACGGGGAATTCACAAGCT)
(AGCTTGTGAATTCCCCGTCTCCTG) were used to generate the L86E mutants. The mutated
codon is in bold and an EcoRI restriction site added by silent mutation is underlined. Primers
F261EFor (CGCTATCAAATTATTCACGAATTAATTGCGGG)
(CCCGCAATTAATTCTGTAATAATTTGATAGCG) were used for F261E mutants (mutated
codon in bold). I401E mutants were
(CGGTCAGAGGGCCTGTGAAGGTCAGCAGTTCGC)
(GCGAACTGCTGACCTTCACAGGCCCTCTGACCG) which contain an Eco0109 I restriction
enzyme site introduced by silent mutations, shown underlined (mutated codon in bold).

Protein expression and purification. Both wild-type (WT) and mutant heme domain and
flavocytochrome constructs were expressed in E. coli TG1 cells. Cells were harvested and lysed
as described previously [16,23]. Purification was by ammonium sulfate precipitation, DEAE and
Q-Sepharose anion exchange and hydroxyapatite chromatography, with a final Sephacryl 200 gel
filtration stage to obtain a high level of purity for crystallography, as described elsewhere [16,17].
After each stage of purification the purity was checked by determining the total P450 heme (A418)
to (A280) protein to ratio, or Rz (Reinheitzahl) value. At each stage, the most pure fractions by this
criterion were pooled and taken forward. Purity was verified by SDS-PAGE and by Rz values of
>1.7 and >0.9 for heme domain and flavocytochrome P450 BM3 enzymes, respectively.

Assessment of covalent heme macrocycle ligation. Covalent ligation of the heme macrocycle to
the protein was investigated by heme staining on SDS-PAGE. P450 samples were resolved on
12% and 8% gels for heme domain and flavocytochrome enzymes, respectively. Samples were
incubated (for 45 minutes) in the presence of (i) substrate (arachidonate, 1 mM), (ii) sodium
dithionite (10-fold molar excess), (iii) substrate together with sodium dithionite, (iv) substrate and
NADPH (2 mM), (v) hydrogen peroxide (H2O2, 0.67% v/v) or (vi) H2O2 together with substrate,
prior to electrophoresis. A positive control of horse heart cytochrome c was included on each gel.
Following heme staining [15], all gels were also stained with coomassie blue to check for suitable
quantities of protein and positions of proteins on the gel.

Spectroscopic analysis. UV-visible spectroscopy. All UV-visible absorption spectra were
collected on a Cary 50 scanning spectrophotometer (Varian) using a 1 cm path length quartz
cuvette. Concentrations of purified proteins were calculated by the method of Omura and Sato
[24] using an extinction coefficient of ∆ε450-490 = 91 mM-1 cm-1 from the difference spectrum
generated by subtracting the absolute spectrum of the reduced P450 from that for the reduced CO-
bound species form. This process also verified that cysteine thiolate coordination was maintained
in each mutant. P450 concentrations were also verified using the coefficients ε418 = 95 mM-1cm-1
for the heme domains and ε418 = 105 mM-1cm-1 for the intact flavocytochromes [23].

and L86ERev
and F261ERev
made using primers I401EFor
I401ERev and
Biochemical Journal Immediate Publication. Published on 24 Feb 2010 as manuscript BJ20091603
THIS IS NOT THE VERSION OF RECORD - see doi:10.1042/BJ20091603
Accepted Manuscript
Licenced copy. Copying is not permitted, except with prior permission and as allowed by law.
© 2010 The Authors Journal compilation © 2010 Portland Press Limited

Page 5

Structure and catalysis in P450 BM3 mutants

5
Substrate and inhibitor binding analysis. Determination of the binding constants (Kd values) for
the substrates arachidonic acid and lauric acid, and for the inhibitors 4-phenylimidazole (4-PIM)
and sodium cyanide (CN) was done by optical titration of substrates/ligands against the WT and
mutant BM3 heme domains, as described [16,23]. Kd values were determined from plots of
induced absorption changes versus ligand concentration, with data fitted using a standard
hyperbolic function or (where the Kd value was not ≥5x the P450 concentration) to a quadratic
function designed for tight-binding ligands, as described previously [25]. Data were fitted using
Origin software (OriginLab, Northampton, MA).
Accepted Manuscript

( ) (
×
)()()
()
(
)
5 . 0
2
42
EtSKEt

SK EtS EtAA
dd max obs
××−++−++=

Equation 1
In Equation 1, Aobs is the observed absorbance change at substrate/ligand concentration S, Amax is
the absorbance change at substrate/ligand saturation, Et is the P450 BM3 enzyme concentration
and Kd is the dissociation constant for the P450 BM3-substrate/ligand complex.

EPR spectroscopy. EPR spectra for ferric samples of the WT and L86E, F261E and I401E heme
domains were collected using an EPR spectrometer comprising an ER200D electromagnet and
microwave bridge interfaced to a EMX control system (Bruker Spectrospin), and fitted with a
liquid helium flow cryostat (ESR-9, Oxford Instruments) and a dual-mode X-band cavity (Bruker
type ER4116DM). Spectra were collected for WT and mutant heme domains (~200 µM) in assay
buffer (100 mM potassium phosphate, pH 7.0) at 10K, with a microwave power of 2.08 mW and
a modulation amplitude of 10 G. To obtain EPR spectra of oxyferrous forms of WT and mutant
BM3 heme domains, proteins were reduced anaerobically by addition of a stoichiometric amount
of sodium dithionite. P450 reduction was verified spectrophotometrically, and samples were then
transferred directly to an EPR tube. A few bubbles of air were introduced into samples, which
were then either frozen immediately using liquid nitrogen, or after a one minute incubation. EPR
spectra of the oxyferrous samples were obtained using a Bruker E500 ELEXSYS instrument
operating at X-band. Temperature was controlled using an Oxford Instruments ESR900 liquid
helium cryostat together with a ITC503 temperature controller. EPR spectra were obtained at 10
K using 0.5 mW microwave power, a modulation frequency of 100 KHz, and a modulation
amplitude of 5 G. EPR spectra of samples of ferric WT/mutant BM3 heme domains were also
collected at the same protein concentration as for the oxyferrous samples.

Resonance Raman spectroscopy. Resonance Raman (RR) spectra were collected using a 15 mW
406.7 nm radiation source at the sample delivered from a Coherent Innova 300 krypton ion laser,
with spectral acquisition by a Renishaw micro-Raman system 1000 spectrometer. Spectra of WT
and mutant heme domains (50 µM) were collected in both the presence and absence of
arachidonate at ambient temperature, and extended scans obtained from 200-1700 cm-1. Samples
were held in a capillary under a microscope and each was subjected to 5 x 15 s exposures. Data
processing, curve fitting, and band assignment was done using GRAMS/32 software (Thermo
Scientific).

Steady-state kinetic and product characterization studies. Steady-state measurements of the
substrate-dependent (lauric acid and arachidonic acid) NADPH oxidation were made for the
mutant flavocytochromes, as described previously [16,23]. Products of lauric acid oxidation were
investigated for WT/mutant flavocytochromes by incubating lauric acid (800 µM) with NADPH
(2 mM) and 0.5 µM enzyme for 1 hour at 25ºC. Reactions were halted by addition of HCl to
acidify the mixtures to pH ~2. Lauric acid substrate and hydroxylated products were isolated
from the mixtures by binding to SPE columns (Phenomenex, Macclesfield UK) and with elution
using methanol. A small amount of phenylacetic acid dissolved in methanol (20 µL of a 70 µM
Biochemical Journal Immediate Publication. Published on 24 Feb 2010 as manuscript BJ20091603
THIS IS NOT THE VERSION OF RECORD - see doi:10.1042/BJ20091603
Licenced copy. Copying is not permitted, except with prior permission and as allowed by law.
© 2010 The Authors Journal compilation © 2010 Portland Press Limited

Page 6

Structure and catalysis in P450 BM3 mutants

6
stock), which was used as an internal standard, was added to the isolated lipids before
evaporating them to dryness. The residue was reconstituted with 150 µL N,O-
Bis(trimethylsilyl)trifluoroacetamide (BSTFA), 450 µL Trimethylchlorosilane (TMCS) was
added and the mixture was incubated at 60ºC for 60 min. An aliquot (1 µL) of the derivatized
extract was analyzed by GC-MS. Quantitative analyses were performed on a Finnigan PolarisQ
GC/MSn Benchtop Ion Trap Mass Spectrometer fitted with a 30 m Phenomenex ZB5-MS column
(ID 0.32 mm : phase thickness 320 nm) and a liquid autosampler. High purity helium was used as
the carrier gas at a flow rate of 1.5 mL/min in constant flow mode. Samples were injected into a
Progammable Temperature Vapouriser (PTV) in Constant Temperature (CT) mode with a split
ratio of 50:1, and the injection port maintained at 250ºC. The Gas Chromatograph to Mass
spectrometer interface temperature was 240ºC and the Ion trap temperature was maintained at
100ºC throughout. The initial oven temperature was 80ºC, held for two minutes after sample
injection, programmed to 275ºC at 15ºC/min, and held for 4 min. The mass spectrometer scanned
from 45-450 Daltons at 3 scans/second in MS1 mode. Retention times for ω-1, ω-2 and ω-3
hydroxylaurate products were 13.27, 13.16 and 12.97 minutes, respectively.

Potentiometric analysis. Spectroelectrochemical titrations to determine heme iron redox potential
(Fe3+/Fe2+ couple) were done as described previously [19,26,27]. Measurements were made for
WT/mutant heme domains at a concentration of 6–10 µM, both in the presence and absence of
saturating arachidonate (ca 100 µM), in redox buffer (100 mM potassium phosphate (KPi), pH
7.0, 10 % glycerol) at 25ºC. All measurements were made in a Belle technology glove box under
a nitrogen atmosphere, and with oxygen concentration maintained at less than 2 ppm. Spectra
were collected on a Cary 50 scanning spectrophotometer (Varian) external to the glove box, with
a fibre-optic UV-visible probe running from the spectrophotometer to the enzyme sample in the
anaerobic box. Potentials were measured with a Thermo Russell calomel electrode attached to a
Hanna pH211 microprocessor meter.

Crystallography. Each BM3 heme domain mutant was crystallized by the sitting drop technique
at 4ºC. Drops were prepared by addition of 2 µl of 12 mg/ml mother liquor to 2 µl 12 mg/ml
heme domain in 10 mM Tris.HCl pH 7.4 (plus or minus 500 µM N-palmitolylglycine [NPG]).
I401E mutant crystals were obtained with mother liquor containing 100 mM cacodylic acid, pH
6.0, containing 18% PEG 3350 and 140 mM MgCl2. Mother liquor containing 100 mM
cacodylic acid, pH 6.0, 13% PEG 3350, and 140 mM MgCl2 was used to obtain F261E crystals.
L86E crystals were obtained with mother liquor containing 100 mM cacodylic acid, pH 6.0,
16% PEG 3350, and 140 mM MgCl2. Crystals were flash frozen in liquid nitrogen using 10%
PEG 200 as cryoprotectant. Data for all crystal structures were collected to respective
resolutions from a single cryofrozen crystal at ESRF Grenoble, France or Diamond, Harwell,
UK beamlines. The data were scaled and integrated using the XDS package [28] and
subsequently handled using the CCP4 suite [29]. All structures were solved using difference
Fourier methods. Refinement and model building were carried out using Refmac 5 [30] and
COOT [31]. Data and final refinement statistics are in Supplementary data Table 1.
Stopped-flow kinetic studies. Stopped-flow absorption measurements were made using an
Applied Photophysics SX18 MVR stopped-flow spectrophotometer in an anaerobic environment
(Belle Technology glove box) to maintain oxygen levels at <2 ppm. Both single wavelength and
multiple wavelength (spectral acquisition) data were collected using a photodiode array (PDA)
detector and XSCAN software, with the first spectrum recorded 1.28 ms after mixing. P450 BM3
flavin (FMN)-to-heme electron transfer rates were measured for the WT and mutant
flavocytochromes at 25ºC by monitoring formation of the ferrous-CO complex at 450 nm. All
experiments were done with anaerobic assay buffer pre-saturated by extensive bubbling with CO
gas (~975 µM). Reactions were initiated by mixing a 200 µM NADPH solution with mutant and
Biochemical Journal Immediate Publication. Published on 24 Feb 2010 as manuscript BJ20091603
THIS IS NOT THE VERSION OF RECORD - see doi:10.1042/BJ20091603
Accepted Manuscript
Licenced copy. Copying is not permitted, except with prior permission and as allowed by law.
© 2010 The Authors Journal compilation © 2010 Portland Press Limited

Page 7

Structure and catalysis in P450 BM3 mutants

7
WT flavocytochromes (3 µM) containing varying concentrations of arachidonic acid substrate (0-
500 µM). P450 heme oxyferrous complex formation and stability was measured at 20ºC. An
anaerobic sample of mutant or wild-type heme domain was reduced by titration with dithionite
prior to stopped-flow mixing with oxygenated assay buffer. PDA measurements (from 350 to 700
nm) were made for periods up to 200 seconds to observe spectral evolution and decay of the
complex. Reactions were performed both for the substrate-free heme domains and for the
substrate-bound forms (NPG at 50 µM).


Results

UV-visible absorption spectroscopy studies
The heme cofactor of P450s produces a distinctive spectral signature in the UV-visible region
that can be highly informative with respect to the heme iron oxidation- and spin-state, as well as
highlighting the presence of ligands interacting with the heme iron at its distal face, as seen in our
previous studies of BM3 mutants [15-17,32]. Absorption spectra suggested that both the purified
F261E and I401E enzymes (heme domains and flavocytochromes) were extensively low-spin
(LS) with Soret maxima at 419 nm for both heme domains. However, for L86E there was a
spectral feature evident at 392 nm (alongside the 418 nm LS maximum) indicating a small
proportion (ca 10 %) of high-spin (HS) ferric heme iron in this heme domain. The Soret peak for
WT BM3 heme domain was at 419 nm. In solution at room temperature, there is no indication of
distal heme iron coordination by the introduced glutamate residues in any of the mutant proteins
(i.e. a red shifted Soret feature at ~425 nm, as seen in the A264E mutant [15]).

The binding of fatty acid substrates to BM3 leads to displacement of a water ligand as the 6th
(axial) ligand to the heme iron, conversion of heme from LS to HS, and (from crystal structures)
the stabilization of a different conformational state to that seen for the substrate-free heme
domain [20,32,33]. Arachidonic acid is an excellent BM3 substrate, inducing an extensive
conversion to the HS state. Optical studies of the I401E heme domain indicated a near complete
arachidonate-induced conversion to HS, similar to WT BM3 (Figure 1A and 1B). By contrast, the
F261E/L86E heme domains exhibited less complete arachidonate-induced HS conversions
(~70%/65% the extent of the WT HS shift, Figure 1B). As for WT BM3, there were smaller shifts
to HS with lauric acid (dodecanoic acid) as substrate than with arachidonate in all mutants. I401E
gave a much lower laurate-induced conversion to HS (~50%) than did WT heme domain, while
L86E gave a conversion comparable to WT, suggesting altered binding modes for the different
substrate types between mutants. The Kd data (Table 1) show that arachidonate binding remains
tight in all mutants, with the L86E mutant exhibiting very tight binding (Kd ~20 nM) despite a
lower extent of HS shift than seen for WT/I401E P450s. For the F261E mutant, the extent of
optical change mediated by lauric acid binding was insufficient to enable accurate Kd
determination. An estimate of >1 mM was made, but only ~10 % HS heme iron accumulated in
the F261E heme domain at saturating levels of lauric acid. However, Kd values for lauric acid
were lower for both I401E/L86E heme domains (11.9/1.0 µM) than for WT heme domain (89
µM), with laurate binding for the L86E mutant improved by almost 2 orders of magnitude.

Comparative studies of WT/mutant heme domains were done for binding of the heme
coordinating ligands cyanide (CN) and 4-phenylimidazole (4-PIM). WT and all mutants gave
type II optical shifts to hexacoordinated forms with the exogenous ligands replacing the distal
aqua ligand, and with Soret maxima shifted to ~440 nm (CN) and 424 nm (4-PIM) (Figure 2). 4-
PIM bound substantially tigher than did the polar CN ion in all cases. The Kd values are shown in
Table 1. For I401E, the Kd for 4-PIM was markedly lower than for WT (0.1 µM versus 0.85 µM).
For WT/L86E mutant heme domains, plots of induced heme optical change versus cyanide
Biochemical Journal Immediate Publication. Published on 24 Feb 2010 as manuscript BJ20091603
THIS IS NOT THE VERSION OF RECORD - see doi:10.1042/BJ20091603
Accepted Manuscript
Licenced copy. Copying is not permitted, except with prior permission and as allowed by law.
© 2010 The Authors Journal compilation © 2010 Portland Press Limited

Page 8

Structure and catalysis in P450 BM3 mutants

8
concentration were hyperbolic, suggesting a single, saturable binding site on the heme iron.
Affinity for cyanide was ~9-fold weaker in L86E than in WT BM3, likely due to the
electrostatic/steric influence of the Glu86 side chain close to the heme. However, the cyanide
binding plots were sigmoidal for both the F261E/I401E heme domains, and apparent Kd values
(i.e. the points of inflexion of the sigmoidal fits) were ~40-50 mM in both cases (Table 1), again
indicating weaker binding of CN than for the WT heme domain. For F261E, repulsive
electrostatic interactions of the E261 carboxylate with CN may occur, but in the I401E mutant the
mutation is on the heme proximal side, and thus unlikely to diminish CN affinity by direct
electrostatic repulsion. The basis for apparent cooperative binding of CN to the I401E/F261E
mutants remains uncertain, although it is of note that the F261E mutant crystallizes in a
conformational state only rarely observed previously (see Structural studies on P450 BM3
mutants section).

Reduction of the WT and mutant heme domains showed that all were converted to the ferrous
form, with their Soret bands shifted to 409 ± 1 nm. The Soret blue shift is consistent with
retention of cysteine (Cys400) thiolate as the proximal heme ligand in all cases for the ferrous
hemoproteins [34,35]. Consistent with this conclusion, all ferrous proteins bound carbon
monoxide (CO) to produce Soret shifts to 448 nm (WT), 450 nm (I401E), 448 nm (F261E) and
450 nm (L86E), respectively. Thus, introduction of acidic residues on proximal and distal sides of
the heme did not disrupt thiolate coordination, and all Fe(II)CO complexes were of the
characteristic P450 spectral type, with minimal amounts of P420 that might indicate proximal
thiol ligation. An example is shown for the F261E heme domain in Figure 2.

Covalent heme macrocycle ligation
Heme staining of WT/mutant BM3 enzymes (resolved by SDS PAGE) provided no evidence for
any significant extent of covalent linkage of the heme macrocycle to the protein, even after
extended turnover of WT and mutant flavocytochromes in presence of arachidonic acid substrate
and NADPH, or after incubation with H2O2 (either plus or minus substrate).

Spectroscopic analysis of P450 BM3 mutants
EPR. Low temperature EPR studies of ferric P450s are informative in relation to the heme iron
spin-state and ligation state. EPR spectra were collected for WT and all mutant heme domains. In
all cases, spectra were consistent with a near-exclusively LS ferric heme (as expected for
substrate-free BM3 at cryogenic temperatures). A typical P450 rhombic spectrum was observed
for WT and mutant proteins. The major set of g-values was near-identical in all cases, with gx =
2.42, gy = 2.26, gz = 1.92 for WT BM3, and values of 2.42/2.26/1.93 for each of the L86E, I401E
and F261E mutants. The positions of these features confirm retention of heme iron cysteinate
coordination in all these P450s. For the F261E mutant, small gz features at 2.63/2.52 and gx
features at 1.90/1.84 are consistent with the presence of alternative LS species. In the L86E
mutant there are similar (and slightly more pronounced) features at 2.58/2.53/2.48 and 1.90/1.87
(Figure 3). These minor sets of g-values are similar to those we observed for the A264E mutant
of P450 BM3, where there was also crystal structure evidence for distal coordination of heme
iron by Glu264 [15]. Some heterogeneity in the g-values for the Glu264-ligated form of the
A264E mutant were also observed, and assigned to signals that could arise from different
orientations of the Glu264 ligand, ferric iron interactions with different oxygen atoms of the
carboxylate, or structural influences of active site residues (e.g. Phe87). It thus appears likely that
the minor sets of g-values in the L86E/F261E mutants also originate from sub-populations in
which Glu86/Glu261 ligate distally to the ferric heme iron. There is not firm evidence for such
coordination at ambient temperature (or in crystals), but since EPR indicates that such Glu-
coordinated forms are minor species, any putative red shift of the Soret band induced might not
influence the optical spectra discernibly. However, given the proximity of both Glu261 and
Biochemical Journal Immediate Publication. Published on 24 Feb 2010 as manuscript BJ20091603
THIS IS NOT THE VERSION OF RECORD - see doi:10.1042/BJ20091603
Accepted Manuscript
Licenced copy. Copying is not permitted, except with prior permission and as allowed by law.
© 2010 The Authors Journal compilation © 2010 Portland Press Limited

Page 9

Structure and catalysis in P450 BM3 mutants

9
Glu86 to the iron on the distal face of the heme (and since there are no such novel EPR signals in
the I401E mutant on the proximal face of the heme where cysteinate ligation occurs) it appears
that freezing of these mutants to 10K is able to capture conformational states in which glutamate
ligation to heme iron occurs to a small extent. In addition to the known BM3 substrate-free and
substrate-bound heme domain conformations, we have revealed crystallographically a third
conformation of the BM3 heme domain (in WT and an A264H mutant) and it is thus clear that
different conformational states of the enzyme are accessible in solution [16]. EPR spectra were
also recorded for ferrous WT and mutant heme domains that had been reacted with oxygen, as
described in the Analysis of the formation and decay of the ferrous-oxy complex section.

Resonance Raman. Resonance Raman (RR) was applied to the BM3 heme domain mutants to
probe for perturbations to the electronic properties and to the geometry of the heme and its
peripheral groups. RR data show that WT and all mutant heme domains have an oxidation state
marker (ν4) at 1371 cm-1, confirming the ferric state. All are predominantly LS, with the major ν3
spin-state marker band at 1500, 1499 and 1500 cm-1 for the L86E, F261E and I401E mutants
[36]. WT and mutant heme domains retain ferric heme iron on substrate addition (arachidonate),
but changes in heme iron spin state equilibrium occur, with increased HS heme iron in all
mutants, shown by the increased intensity of the ν3 band at ~1485 cm-1. The F261E mutant has a
somewhat smaller ν11 band at 1568 cm-1 than WT BM3. This band reports on electronic
conjugation of porphyrin and vinyl groups, and hence on the in-plane asymmetry of the heme
ring [37]. It is also affected by a 6th ligand, and thus the small amount of glutamate coordination
predicted by EPR studies of the F261E heme domain are consistent with its weaker ν11 signal.

Redox Potentiometry
Despite the apparent lack of covalent bonding between the introduced glutamate side chains and
the heme methyl groups (or other parts of the macrocycle), the introduction of charged residues in
the heme vicinity might affect its thermodynamic and/or catalytic properties. Thus, we analysed
the heme iron redox potential in all mutants, and compared these to values for the WT heme
domain plus/minus substrate (arachidonate). Redox potential values for WT and mutant BM3
heme domains are shown in Table 2. All mutations increased the potential for the Fe3+/Fe2+ heme
iron transition. In all cases, binding of arachidonic acid induced spectral changes consistent with
the accumulation of HS heme iron (see UV-visible absorption spectroscopy studies), and induced
further positive shifts in heme iron potential. The substrate-free F261E heme domain has the
closest potential to the WT heme domain (-367 ± 8 mV versus -395 ± 4 mV for WT), while L86E
(-307 ± 5 mV) and I401E (-219 ± 4 mV) were much more positive. Previous studies of both BM3
and P450cam showed that the near-complete conversion of heme iron spin-state to HS on
substrate binding was accompanied by elevation of heme iron potential by >100 mV. A similar
extent of heme potential shift was observed for arachidonate-bound L86E (131 mV to -176 ± 6
mV) and F261E (170 mV to -197 ± 6 mV) heme domains, although the potential shift for I401E
was less substantial (53 mV to -166 ± 7 mV). Figure 4A shows exemplary spectra collected
during the I401E heme domain redox titration, while Figure 4B shows overlaid Nernst function
fits of heme absorption (% heme reduced) versus applied potential for both WT and I401E heme
domains in their substrate-free and arachidonate-bound forms. Previous studies of BM3 F393A/H
mutants that (like I401E) are located near the heme proximal ligand (Cys400) showed that these
had more positive heme iron potentials than WT BM3 and stabilized the ferrous-oxy complex of
the P450 [19,38]. In view of this, we analysed the transient and steady-state kinetic properties of
the thermodynamically perturbed I401E, F261E and L86E flavocytochromes P450 BM3. This
was done to establish effects on turnover and reductase FMN-to-P450 heme electron transfer, and
also to assess capacity of these mutants to stabilize a ferrous-oxy form that is barely detectable
for WT BM3 at ambient temperature.

Biochemical Journal Immediate Publication. Published on 24 Feb 2010 as manuscript BJ20091603
THIS IS NOT THE VERSION OF RECORD - see doi:10.1042/BJ20091603
Accepted Manuscript
Licenced copy. Copying is not permitted, except with prior permission and as allowed by law.
© 2010 The Authors Journal compilation © 2010 Portland Press Limited

Page 10

Structure and catalysis in P450 BM3 mutants

10
Flavin-to-heme electron transfer in WT and mutant P450 BM3 enzymes
Stopped-flow studies followed the formation of the dead-end Fe(II)CO complex (with absorption
maximum at ~448 nm) to determine FMN-to-heme electron transfer kinetics [15,39]. CO binding
to ferrous BM3 heme iron is substantially faster than the inter-cofactor electron transfer rate
under the conditions used [40], and thus the stopped-flow method is useful for analysing the
influence of altered heme redox potentials on heme reduction rate in the glutamate substitution
mutants.

For the I401E mutant, the apparent limiting rate of electron transfer (klim) for the arachidonate-
bound enzyme is ~2-fold that for WT BM3 (473 s-1 versus 250 s-1), consistent with its more
positive heme potential. Substrate-free I401E is converted rapidly and completely to its Fe(II)CO
complex (klim = 60 s-1), in contrast to WT BM3 where there is negligible Fe(II)CO complex
formation for substrate-free enzyme on a stopped-flow time scale. Substrate-free I401E is
extensively low-spin in its resting form, but its potential (-219 mV) is more positive than that for
the arachidonate-bound form of WT BM3, indicating the importance of heme iron
thermodynamics in regulating rate and extent of heme reduction. The arachidonate-bound F261E
mutant exhibits only partial conversion to the HS form, and has a klim of 58 s-1 for Fe(II)CO
complex formation, indicating substantially slower FMN-to-heme electron transfer compared to
WT BM3. There is much less Fe(II)CO complex formation in the substrate-free F261E enzyme
(~20%) with a klim of 27 s-1. For the L86E mutant (that has a more positive potential than WT
BM3 in both substrate-free and arachidonate-bound forms, and some HS content in the ferric,
substrate-free enzyme), the klim for Fe(II)CO formation is 213 s-1 in the arachidonate-bound form
and 102 s-1 in the substrate-free form.

These data are consistent with a major role for heme potential in regulating FMN-to-heme
electron transfer. In the I401E/L86E mutants, the potential of the substrate-free heme iron (-219
mV/-307 mV, respectively) is much more positive than for the substrate-free WT BM3, and this
is certainly a factor underlying the efficient FMN-to-heme electron transport in these enzymes.
There is extensive formation of the Fe(II)CO complex in the substrate-free forms of I401E/L86E,
whereas this does not occur to any considerable extent for WT BM3 (or the F261E mutant). The
midpoint reduction potential of the (catalytically relevant) anionic semiquinone form of BM3’s
FMN cofactor (the oxidized/semiquinone couple) is -240 ± 10 mV, while that for the
semiquinone/hydroquinone couple is -160 ± 10 mV at pH 7.0 [41]. For each of the mutants
generated here, FMN-to-heme electron transfer is strongly thermodynamically favoured from the
FMN semiquinone to the substrate-bound ferric heme iron.

Analysis of the formation and decay of the ferrous-oxy complex
The more positive redox potential of the I401E and L86E mutants suggested that both might
stabilize the ferrous oxy (Fe(II)O2) complex of the P450 to a greater extent than in WT BM3. To
investigate further, we analysed reactivity of reduced WT and mutant NPG-bound heme domains
with oxygen, using the substrate-bound species to enable complete reduction with sodium
dithionite in all cases. For WT BM3 heme domain at 20°C, mixing of the ferrous NPG-bound
protein (in absence of excess dithionite) with aerobic oxygen results in negligible formation of a
Fe(II)O2 complex, and the reoxidized (ferric) form is near-completely reformed within one
second. The F261E heme domain behaved similarly to WT BM3. These data for WT BM3 heme
domain are consistent with previous studies, which estimated a rate of conversion (kox) of the
Fe(II)O2 complex to Fe(III) and superoxide at 0.22 s-1 at 20°C [42]. A more complete conversion
to the Fe(II)O2 form is seen for WT heme domain at -25°C, with kox = 0.1 min-1 (0.0017 s-1) [43].
Ost et al. reported that Fe(II)O2 complexes of the F393A/H BM3 heme domains were
considerably stabilized compared to WT, decaying with a half-life of ~30 s (kox ~0.023 s-1) at
15°C [19]. By comparison with the WT heme domain, the I401E and L86E mutants showed
Biochemical Journal Immediate Publication. Published on 24 Feb 2010 as manuscript BJ20091603
THIS IS NOT THE VERSION OF RECORD - see doi:10.1042/BJ20091603
Accepted Manuscript
Licenced copy. Copying is not permitted, except with prior permission and as allowed by law.
© 2010 The Authors Journal compilation © 2010 Portland Press Limited

Page 11

Structure and catalysis in P450 BM3 mutants

11
enhanced Fe(II)O2 complex stabilization. Spectral signatures of the NPG-bound Fe(II)O2 forms
of these mutants are shown in Figure 5. For I401E at 20°C, there is a spectral shift to 419 nm
(with a shoulder at 442 nm) at maximal population of the Fe(II)O2 species, prior to its collapsing
back to the ferric Soret at 395 nm with a kox = 0.0574 s-1 (half life ~12.1 s). For the L86E heme
domain, the Fe(II)O2 absorbs maximally at 423 nm, prior to decaying to the ferric form (again at
395 nm) with a kox = 0.1102 s-1 (half life ~6.3 s). To analyse the EPR properties of WT and
mutant BM3 heme domains in their ferrous-oxy forms, samples were prepared as described in the
Experimental section, and rapidly frozen. EPR revealed a complete absence of signals that could
be assigned to a ferric superoxo form of the proteins, even in the cases of the I401E and L86E
mutants where considerable formation of the oxy complex is seen spectrophotometrically (Figure
5). Minor signals (<20%) assigned to the reoxidized (ferric) forms of the complexes were present.
The absence of EPR signals for the ferrous-oxy forms is consistent with previous work on e.g.
P450cam and the human aromatase (CYP19) P450 [44,45]. We conclude that the oxygen-bound
WT BM3 and mutant heme domains are most likely in the ferric superoxo form, and are EPR
silent as a consequence of the iron and superoxide being antiferromagnetically coupled.

Steady-state turnover of mutant P450 BM3 enzymes
To analyse catalytic properties of each flavocytochrome P450 BM3 mutant, steady-state analysis
of fatty acid substrate-dependent NADPH oxidation was carried out. The substrates laurate and
arachidonate were used. Data are summarized in Table 3. With the exception of L86E with
laurate as substrate (where kcat is slightly higher than for WT BM3), all mutants had lower kcat
values than WT BM3. For I401E, and despite a more rapid transfer of the first electron to the
heme iron, overall catalytic rate is decreased ~8-fold with arachidonic acid. A likely explanation
is that the elevated heme iron potential results in a decreased driving force for oxygen reduction
by the ferrous heme iron, as suggested previously for F393A/H variants [19]. As expected from
the high redox potential of the I401E heme iron and its ability to form the Fe(II)CO complex in
absence of fatty acid substrate, the I401E flavocytochrome is a much more efficient NADPH
oxidase in absence of fatty acids than is WT BM3. The rate of substrate-independent NADPH
oxidation is ~3.05 s-1 (I401E) compared to ~0.03 min-1 for WT P450 BM3 in aerobic buffer. The
comparable rates are 1.37/0.22 s-1 for the L86E/F261E mutants, respectively.

Analysis of hydroxylated products formed from oxidation of lauric acid
To verify that mutant flavocytochromes remained active in fatty acid oxidation, products were
isolated from turnover reactions with lauric acid and identified by GC-MS. Lauric acid was
converted completely to products for WT and all mutants. In all cases, there was evidence of the
formation of ω-1, ω-2 and ω-3 hydroxylaurate products, although the product ratios obtained
varied between the enzymes. For WT BM3, the hydroxylated products were approximately
equally distributed between ω-1, ω-2 and ω-3 positions (35%, 32% and 33%), consistent with
previous work [e.g. 46]. The data for the L86E mutant showed a similar ω-1, ω-2 and ω-3
product distribution to WT (42%, 28% and 30%), while F261E’s profile was biased towards ω-3
(17%, 35% and 48%) and I401E’s was towards ω-1 (57%, 19% and 24%). Thus, all mutants were
functional, but affected in regioselectivity of substrate oxidation, although not the positions of
oxidation. Of particular note was the influence of I401E in favouring ω-1 hydroxylation of
laurate, since this mutation resides on the proximal side of the heme cofactor and cannot directly
influence interactions with the substrate. Only monohydroxylated lauric acid products were
observed in these assays.

Structural studies on P450 BM3 mutants
Crystal structures of the BM3 I401E, L86E and F261E heme domains indicated no covalent
interaction with heme substituent (methyl or other) groups in all mutants, consistent with
preceding data. The I401E/F261E mutant heme domains were crystallized in the substrate-free
Biochemical Journal Immediate Publication. Published on 24 Feb 2010 as manuscript BJ20091603
THIS IS NOT THE VERSION OF RECORD - see doi:10.1042/BJ20091603
Accepted Manuscript
Licenced copy. Copying is not permitted, except with prior permission and as allowed by law.
© 2010 The Authors Journal compilation © 2010 Portland Press Limited

Page 12

Structure and catalysis in P450 BM3 mutants

12
form, while the L86E heme domain structure could only be resolved for the substrate (NPG)-
bound form.

L86E heme domain: The NPG-bound L86E heme domain crystals diffracted to a resolution of 1.8
Å. The L86E structure is nearly identical to the previously resolved WT heme domain NPG
complex (PDB code 1JPZ) [47]. In contrast to the 1JPZ structure, the L86E NPG complex
structure reveals partial occupancy of the distal water (modelled at 0.35 occupancy) in both active
sites in the asymmetric unit. This suggests that NPG binding is not necessarily incompatible with
retention of a distal water ligand to the heme iron and only partially drives conversion to the
penta-coordinate state under the crystal conditions. In contrast to the relatively large structural
changes induced by other point mutants in the heme vicinity (e.g. A264E) [15,32], the L86E
mutation has little effect on heme domain structure, with the obvious exception of the immediate
Glu86 environment. The Glu86 side chain is in hydrogen bonding contact with a heme propionate
group. This, in turn, leads to a small reorganization in the orientation of His100. The latter
residue is now in direct polar contact with both Glu86 and a heme propionate (Figure 6). These
novel interactions with heme peripheral substituent groups may explain the perturbation of the
L86E heme iron redox potential.

I401E heme domain: The crystals of the I401E heme domain were obtained in absence of
substrate and diffracted to 2.0 Å. The I401E structure is nearly identical to that of previously
resolved substrate-free heme domain structures (e.g. 1BU7) [48]. The Glu401 indirectly hydrogen
bonds via a water molecule to a heme propionate, while the direct environment of the Cys400
ligand appears unaltered (Figure 6). We again infer that new interactions affecting the propionate
group may be key to modulating heme iron potential in this mutant. Previous studies of Phe393
variants of BM3 also revealed large shifts in heme potential, despite apparent lack of structural
rearrangements in the vicinity of the cysteinate ligand from crystallographic analysis [34].

F261E heme domain: Crystals of F261E heme domain were obtained in absence of substrate and
diffracted to 1.7 Å. In contrast to I401E crystals, the F261E crystal form is similar to that
previously reported for the A264H heme domain (2IJ3), and which was also observed for the WT
heme domain (2IJ2) [16]. The introduction of the F261E mutation results in the largest structural
changes of the three mutants reported here. The Glu261 side chain occupies a space distinct from
that of Phe261 in the WT protein, and is located further from the heme moiety. This not only
leads to reorientation of Leu233, but also introduces novel water molecules close to the heme
porphyrin plane (and corresponding to the Phe261 position in the WT structure). Small
adaptations in the backbone positions of Cys156-Gly157 and Gln257-Ile259 are observed, likely
due to the altered positions of Leu233/Glu261 and the introduction of the new waters (Figure 6).
The structural rearrangement and introduction of water molecules in the heme vicinity for the
F261E mutant may underlie the smaller modulation of heme iron potential compared to the
I401E/L86E mutants (where novel heme propionate interactions are observed).

Discussion
Several mammalian CYP4 family P450s form a covalent linkage between a conserved glutamate
residue and the 5-methyl substituent on the heme in a turnover-dependent manner [10-13]. The
A264E mutation in P450 BM3 (mimicking the I helix glutamate in CYP4s) did not induce such
heme linkage, despite similarities in catalytic characteristics between BM3 and the CYP4 fatty
acid hydroxylases. BM3 has great biotechnological potential, and thus routes to improving its
operational stability are of interest [6,49]. With this in mind, we identified new sites and
generated BM3 mutants where engineered glutamates might enable interaction with the heme 5-
methyl group and lead to turnover-dependent covalent heme linkage. These were F261E (in the I
helix), L86E (in the BC loop between B’/C helices) and I401E (immediately following the
Biochemical Journal Immediate Publication. Published on 24 Feb 2010 as manuscript BJ20091603
THIS IS NOT THE VERSION OF RECORD - see doi:10.1042/BJ20091603
Accepted Manuscript
Licenced copy. Copying is not permitted, except with prior permission and as allowed by law.
© 2010 The Authors Journal compilation © 2010 Portland Press Limited

Page 13

Structure and catalysis in P450 BM3 mutants

13
proximal heme ligand, Cys400) (Figure 6). However, none of these mutants gave significant
levels of covalent heme linkage in either flavocytochrome or heme domain proteins. In the
mutant heme domain crystal structures, none of the engineered glutamates are in van der Waals
contact with the heme. This is not surprising, given their relative differences in polarity and in the
absence of any cross-link between the heme group and the glutamate side chains. However,
conformation(s) that would bring the glutamate carboxylates into direct contact with the heme
methyl groups are related by minor movements to the crystal structures, and are thus likely easily
accessible at room temperature. Indeed, our EPR studies indicate that conformations distinct from
those seen in the crystals can be trapped at cryogenic temperatures at least for the F261E/L86E
mutants, since in both these cases there is evidence for heme iron coordination by the introduced
glutamates (Figure 3).

The absence of glutamate-heme methyl group ligation in the BM3 mutants may result from
mechanistic differences between mammalian CYP4 and BM3 enzymes. In the former, it is
postulated that production of compound 1 (cpd1, a highly reactive ferryl-oxo [FeIV=O] porphyrin
cation radical species considered as the active oxidant of P450 substrates) results in oxidation of
the glutamate to produce a carboxylic radical with reduction of cpd1 to cpd2. This radical then
abstracts a hydrogen atom from the heme 5-methyl group, forming a methylene radical which is
then converted to a carbocation via intra-heme electron transfer. The carbocation may then be
trapped by the glutamate carboxylate to form the ester linkage between protein and the heme
methyl [14]. BM3 is a substantially more efficient enzyme than the CYP4s, with electrons
delivered from its attached CPR domain, and between opposite CPR and P450 domains in a BM3
dimer [50]. Electron transfer rate between BM3 FMN and heme is >200 s-1 at ambient
temperature and overall turnover rate of the enzyme in steady-state is >15,000 min-1 (250 s-1)
with arachidonate [23,35]. These values are 1-2 orders of magnitude greater than for the CYP4
enzymes. We speculate that cpd1 may be more transient in BM3 than in CYP4s, in addition to
any considerations regarding non-optimal positioning of engineered glutamates in BM3
compared to CYP4s, where evolution has likely configured the relevant glutamates specifically
for formation of a covalent bridge to the heme methyl.

Notwithstanding the inability of P450 BM3 to covalently link various engineered glutamates to
the heme, profound effects are observed on the thermodynamic, substrate/ligand binding and
catalytic features of the L86E, F261E and I401E mutants, as well as on stabilization of the
ferrous-oxy forms of the L86E/I401E variants. Particularly notable is the substantially improved
affinity of the L86E mutant for both laurate/arachidonate substrates, with Kd values at least an
order of magnitude lower than for the WT BM3 (Table 1). This is consistent with the purification
of the L86E heme domain in a partially HS state, likely indicating co-purification of the enzyme
with some E. coli lipid bound. Previously, we co-purified the Bacillus subtilis P450 BioI
(CYP107H1) protein from E. coli bound to palmitic acid, which is a predominant lipid in E. coli
and a tight binding substrate for both P450 BioI and BM3 [51]. This enhanced fatty acid affinity
was also manifest in an improved Km for lauric acid with the L86E flavocytochrome,
accompanied by an ~1.5-fold improvement in kcat for this substrate. The structural origins of this
improved affinity are uncertain, since the NPG-bound L86E heme domain structure is highly
similar to the WT NPG complex. However, the substrate-free L86E heme domain was not readily
crystallized and substrate-free crystals obtained diffracted poorly, suggesting altered
conformational states and/or equilibria between these states in the L86E mutant. Enhanced
substrate affinity and altered structural conformations were also observed previously for the BM3
A264E mutant [32,33]. The I401E heme domain also displayed higher affinity for lauric and
arachidonic acids compared to WT BM3, with the Kd for laurate improved ~7-fold (Table 1).
However, despite substrate Km values comparable (or lower) to those for WT P450 BM3, the kcat
values were considerably lower. This is likely as a consequence of over-stabilization of the
Biochemical Journal Immediate Publication. Published on 24 Feb 2010 as manuscript BJ20091603
THIS IS NOT THE VERSION OF RECORD - see doi:10.1042/BJ20091603
Accepted Manuscript
Licenced copy. Copying is not permitted, except with prior permission and as allowed by law.
© 2010 The Authors Journal compilation © 2010 Portland Press Limited

Page 14

Structure and catalysis in P450 BM3 mutants

14
ferrous-oxy complex and retardation of subsequent electron transfer and oxygen activation steps
[52].

Potentiometry revealed positive shifts in heme iron potential (relative to WT) in all mutants, and
for both substrate-free and arachidonate-bound heme domains (Table 2). The potential of -219 ±
4 mV for the substrate-free I401E heme domain is the most positive yet reported for a substrate-
free BM3 mutant, and >100 mV more positive than for the substrate-free F393A/H mutants (-
312/-332 mV, respectively). Arachidonate binding increased the heme iron potentials of the
F393A/H heme domains to -151/-176 mV, respectively, similar to the value for the arachidonate-
bound I401E heme domain [19]. The mutant substrate-free heme potentials are closer to that for
the catalytically relevant form of the FMN cofactor (the oxidized/semiquinone couple at -240 mV
[41]) than is that for WT BM3. All mutant heme potentials are more positive than -240 mV in
their substrate-bound forms. This is consistent with the higher rates of substrate-independent
NADPH oxidation in the mutants. Table 2 reveals a reasonable correlation of FMN-to-heme
electron transfer rate with heme iron potential and the proportion of HS heme for the substrate-
free WT/mutant BM3 enzymes. In the substrate-bound enzymes the heme potentials are further
elevated, and FMN-to-heme electron transfer is accelerated. The correlation between rate and
heme potential/HS proportion is not absolute for the arachidonate-bound WT and mutants,
although heme reduction rate is fastest for I401E (klim = 473 s-1), consistent with its potential
being 119 mV more positive than arachidonate-bound WT BM3, and the L86E and F261E
mutants (that have a lower klim than WT) also show lower arachidonate-induced conversion to HS
heme iron. Crystal structures revealed new interactions between heme propionate and Glu401 in
I401E BM3, and between Glu86 and heme propionate/His100 in L86E BM3 (Figure 6).
Considerable structural rearrangements also occurred in the F261E heme domain, including
introduction of novel waters in the heme vicinity that may affect its electrostatic environment.
Mutant heme domain structures were also obtained only in the substrate-bound form (L86E) or in
an infrequently observed substrate-free conformation (F261E), pointing to structural
perturbations that could also impact on efficiency of FMN-to-heme electron transfer.

Wong and co-workers noted that a L358P mutation immediately following the cysteinate ligand
increased ethane/propane oxidation in P450cam [53]. They also showed that the BM3 I401P
mutant had a substrate-free heme potential of -303 mV (142 mV more positive than WT) and that
coupling of NADPH oxidation to lauric acid hydroxylation was similar in the I401P mutant
(53%) to WT BM3 (52%) [21]. The I401P flavocytochrome also displayed an ~16-fold increase
in substrate-independent NADPH oxidation. Our data for the I401E mutant are consistent with
these findings, although the substrate-free I401E potential (-219 mV) is ~84 mV more positive
than for the I401P mutant. The I401E, L86E and F261E enzymes all remain functional in fatty
acid oxidation, but with notable alterations in regioselectivity of laurate oxidation. The ω-1
hydroxylaurate product is favoured in the F261E mutant, with a substantial shift towards ω-3
laurate hydroxylation in the I401E mutant. The coupling of NADPH oxidation to lauric acid
oxidation was ≥40% for all mutant enzymes analysed in this study.

The perturbed potentials of the I401E/L86E heme domains resulted in stabilization of the ferrous-
oxy form (compared to WT), enabling the species to be readily characterized optically (Figure 5)
and for samples to be frozen for EPR analysis. However, EPR showed no signal attributable to a
ferric superoxo form, even in the I401E mutant in which lifetime of the oxy complex is
substantially increased. It is considered that electronic redistribution of this oxy complex occurs
in the P450 catalytic cycle to favour a ferric superoxo species, thus withdrawing electron density
from the heme iron to favour a second electron transfer that leads to a transient ferric peroxo
intermediate. The ferrous-oxy complexes of P450cam and human aromatase are also EPR silent,
although Mossbauer and resonance Raman analysis of the P450cam complex provide compelling
Biochemical Journal Immediate Publication. Published on 24 Feb 2010 as manuscript BJ20091603
THIS IS NOT THE VERSION OF RECORD - see doi:10.1042/BJ20091603
Accepted Manuscript
Licenced copy. Copying is not permitted, except with prior permission and as allowed by law.
© 2010 The Authors Journal compilation © 2010 Portland Press Limited

Page 15

Structure and catalysis in P450 BM3 mutants

15
data for ferric heme iron and oxygen in the superoxide form [54,55]. Thus, we conclude that the
WT and mutant P450s BM3 analysed here are likely in the ferric superoxo form, with their ferric
heme iron and bound superoxide antiferromagnetically coupled to produce an EPR silent species.

Conclusions
The I401E, L86E and F261E glutamate replacement mutants near the heme 5-methyl group in
P450 BM3 induced perturbations to substrate/ligand affinity (including improved Kd values for
laurate and arachidonate in I401E/L86E mutants) and kinetics of substrate oxidation.
Conformational flexibility of mutant heme domains was apparent both from evidence of some
coordination of heme iron by Glu264/Glu86 in low temperature EPR studies of the relevant
mutants, and from the distinctive conformational states in which the three mutants were
crystallized. The mutations induced heme potential changes in all mutants, with considerable
positive shifts in the Fe(III)/Fe(II) potential of the L86E/I401E mutants resulting in greater
stability of the ferrous-oxy species, and an ~2-fold increase in FMN-to-heme electron transfer
rate in the substrate-bound I401E flavocyochrome. All mutants retained lauric acid hydroxylase
activity, with altered regioselectivity in the F261E (ω-1 oxidation favoured) and I401E (ω-3
oxidation favoured) mutants. Although there was no evidence of covalent linkage of heme 5-
methyl to the mutant proteins, similar strategies of introducing charged residues in the heme
environment may prove useful for stabilization of heme oxy complexes, and for alteration of
substrate selectivity and regioselectivity of substrate oxidation in BM3 and other P450s.

Acknowledgements
The authors acknowledge the financial support of the UK Biotechnology and Biological Sciences
Research Council (BBSRC, grants BB/F00252/1 and BB/F00883X1) for this research. We are
also grateful to Prof. W. Ewen Smith (University of Strathclyde, UK) for access to resonance
Raman facilities.


References
1) Denisov, I.G., Makris, T.M., Sligar, S.G. and Schlichting, I. (2005) Structure and chemistry of
cytochrome P450. Chem. Rev. 105, 2253-2277.
2) Guengerich, F.P., Wu, Z.L. and Bartleson, C.J. (2005). Function of human cytochrome P450s:
characterization of the orphans. Biochem. Biophys. Res. Commun. 338, 465-469.
3) Cupp-Vickery, J.R., Han, O., Hutchinson, C.R. and Poulos, T.L. (1996). Substrate-assisted
catalysis in cytochrome P450eryF. Nat. Struct. Biol. 3, 632-637.
4) Lawson, R.J., Leys, D., Sutcliffe, M.J., Kemp, C.A., Cheesman, M.R., Smith, S.J., Clarkson,
J., Smith, W.E., Haq, I., Perkins, J.B. and Munro, A.W. (2004). Thermodynamic and biophysical
characerterization of cytochrome P450 BioI from Bacillus subtilis. Biochemistry 43, 12410-
12426.
5) Poulos, T.L. (2003). The past and present of P450cam structural biology. Biochem. Biophys.
Res. Commun. 312, 35-39.
6) Munro, A.W., Leys, D.G., McLean, K.J., Marshall, K.R., Ost, T.W., Daff, S., Miles, C.S.,
Chapman, S.K., Lysek, D.A., Moser, C.C., Page, C.C. and Dutton, P.L. (2002). P450 BM3: The
very model of a modern flavocytochrome. Trends Biochem. Sci. 27, 250-257.
7) Perera, R., Sono, M., Sigman, J.A., Pfister, T.D., Lu, Y. and Dawson, J.H. (2003). Neutral
thiol as a proximal ligand to ferrous heme iron: implications for heme proteins that lose cysteine
thiolate ligation on reduction. Proc. Natl. Acad. Sci. USA 100, 3641-3646.
8) Munro, A.W., Girvan, H.M. and McLean, K.J. (2007). Cytochrome P450-redox partner fusion
enzymes. Biochim. Biophys. Acta 1770, 345-359.
Biochemical Journal Immediate Publication. Published on 24 Feb 2010 as manuscript BJ20091603
THIS IS NOT THE VERSION OF RECORD - see doi:10.1042/BJ20091603
Accepted Manuscript
Licenced copy. Copying is not permitted, except with prior permission and as allowed by law.
© 2010 The Authors Journal compilation © 2010 Portland Press Limited