Oxygen Activation and Electron Transfer in Flavocytochrome P450 BM3

Article abstract of DOI:10.1021/ja035731o

In flavocytochrome P450 BM3, there is a conserved phenylalanine residue at position 393 (Phe393), close to Cys400, the thiolate ligand to the heme. Substitution of Phe393 by Ala, His, Tyr, and Trp has allowed us to modulate the reduction potential of the heme, while retaining the structural integrity of the enzyme's active site. Substrate binding triggers electron transfer in P450 BM3 by inducing a shift from a low- to high-spin ferric heme and a 140 mV increase in the heme reduction potential. Kinetic analysis of the mutants indicated that the spin-state shift alone accelerates the rate of heme reduction (the rate determining step for overall catalysis) by 200-fold and that the concomitant shift in reduction potential is only responsible for a modest 2-fold rate enhancement. The second step in the P450 catalytic cycle involves binding of dioxygen to the ferrous heme. The stabilities of the oxy-ferrous complexes in the mutant enzymes were also analyzed using stopped-flow kinetics. These were found to be surprisingly stable, decaying to superoxide and ferric heme at rates of 0.01-0.5 s^-1. The stability of the oxy-ferrous complexes was greater for mutants with higher reduction potentials, which had lower catalytic turnover rates but faster heme reduction rates. The catalytic rate-determining step of these enzymes can no longer be the initial heme reduction event but is likely to be either reduction of the stabilized oxy-ferrous complex, i.e., the second flavin to heme electron transfer or a subsequent protonation event. Modulating the reduction potential of P450 BM3 appears to tune the two steps in opposite directions; the potential of the wild-type enzyme appears to be optimized to maximize the overall rate of turnover. The dependence of the visible absorption spectrum of the oxy-ferrous complex on the heme reduction potential is also discussed.

Full text of DOI:10.1021/ja035731o

Published on Web 11/12/2003
Oxygen Activation and Electron Transfer in Flavocytochrome
P450 BM3
Tobias W. B. Ost,*,† Jonathan Clark,† Christopher G. Mowat,†,‡ Caroline S. Miles,‡
Malcolm D. Walkinshaw,‡ Graeme A. Reid,‡ Stephen K. Chapman,† and
Simon Daff*,†
Contribution from the School of Chemistry, UniVersity of Edinburgh, West Mains Road,
Edinburgh, EH9 3JJ, and The Institute of Cell and Molecular Biology, UniVersity of Edinburgh,
Mayfield Road, Edinburgh, EH9 3JT
Received April 22, 2003; E-mail: Simon.Daff@ed.ac.uk
Abstract: In flavocytochrome P450 BM3, there is a conserved phenylalanine residue at position 393
(Phe393), close to Cys400, the thiolate ligand to the heme. Substitution of Phe393 by Ala, His, Tyr, and
Trp has allowed us to modulate the reduction potential of the heme, while retaining the structural integrity
of the enzyme’s active site. Substrate binding triggers electron transfer in P450 BM3 by inducing a shift
from a low- to high-spin ferric heme and a 140 mV increase in the heme reduction potential. Kinetic analysis
of the mutants indicated that the spin-state shift alone accelerates the rate of heme reduction (the rate
determining step for overall catalysis) by 200-fold and that the concomitant shift in reduction potential is
only responsible for a modest 2-fold rate enhancement. The second step in the P450 catalytic cycle involves
binding of dioxygen to the ferrous heme. The stabilities of the oxy-ferrous complexes in the mutant enzymes
were also analyzed using stopped-flow kinetics. These were found to be surprisingly stable, decaying to
superoxide and ferric heme at rates of 0.01-0.5 s-1. The stability of the oxy-ferrous complexes was greater
for mutants with higher reduction potentials, which had lower catalytic turnover rates but faster heme
reduction rates. The catalytic rate-determining step of these enzymes can no longer be the initial heme
reduction event but is likely to be either reduction of the stabilized oxy-ferrous complex, i.e., the second
flavin to heme electron transfer or a subsequent protonation event. Modulating the reduction potential of
P450 BM3 appears to tune the two steps in opposite directions; the potential of the wild-type enzyme
appears to be optimized to maximize the overall rate of turnover. The dependence of the visible absorption
spectrum of the oxy-ferrous complex on the heme reduction potential is also discussed.
Introduction
axial heme ligand can be inferred from its absolute conservation
in every member of the P450 family and has been demonstrated
Cytochromes P450 (P450s) are b-heme-containing monooxy-
genases implicated in numerous biosynthetic and metabolic
processes.1 They are able to catalyze oxygen atom insertion (eq
1) into a wide variety of substrates.
to be essential for monooxygenase activity in studies of model
complexes and mutant P450s.6-9 The cysteine is believed to
provide an electronic “push” to the heme-iron, which increases
the electron density at the iron to promote the heterolytic
cleavage of heme-bound dioxygen.10,11
RH + O2 + 2e- P4508 ROH + H2O
(1)
Although the mechanism of oxygen activation by cyto-
chromes P450 has been extensively studied,12-15 the factors
This is made possible by the ability of P450s to reductively
activate molecular oxygen at the heme center. The heme-iron
is axially ligated by a cysteine thiolate ligand; this distinguishes
P450s from other oxygen activating enzymes such as the globins
and peroxidases, which utilize histidine. The electronic influence
of the axial ligand on the heme-iron in P450s is the key to their
unique catalytic abilities.2-5 The importance of cysteine as the
(3) Poulos, T. L.; Finzel, B. C.; Gunsalus, I. C.; Wagner, G. C.; Kraut, J. J.
Biol. Chem. 1985, 260, 16122-16130.
(4) Sono, M.; Roach, M. P.; Coulter, E. D.; Dawson, J. H. Chem. ReV. 1996,
96, 2841-2887.
(5) Ogliaro, F.; de Visser, S.; Shaik, S. J. Inorg. Biochem. 2002, 91, 554-
567.
(6) Higuchi, T.; Uzu, S.; Hirobe, M. J. Am. Chem. Soc. 1990, 112, 7051-
7053.
(7) Tani, F.; Matsu-ura, M.; Nakayama, S.; Nartua, Y. Coord. Chem. ReV. 2002,
226, 219-226.
(8) Auclair, K.; Moenne-Loccoz, P.; Ortiz de Montellano, P. R. J. Am. Chem.
Soc. 2001, 123, 4877-4885.
† School of Chemistry.
‡ The Institute of Cell and Molecular Biology.
(1) Ortiz de Montellano, P. R., Ed. Cytochrome P450: Structure, Mechanism
and Biochemistry, 2nd ed.; Plenum Press: New York, 1995.
(2) Dawson, J. H.; Holm, R. H.; Trundell, J. R.; Barth, G.; Linder, R. E.;
Bunnenberg, E.; Djerassi, C.; Tang, S. C. J. Am. Chem. Soc. 1976, 98,
3707-3709.
(9) Vatsis, K.; Peng, H,-M.; Coon, M. J. J. Inorg. Biochem. 2002, 91, 542-
553.
(10) Sono, M.; Dawson, J. H. J. Biol. Chem. 1982, 257, 5496-5502.
(11) Sono, M.; Andersson, L. A.; Dawson, J. H. J. Biol. Chem. 1982, 257, 8303-
8320.
(12) Gerber, N. C.; Sligar, S. G. J. Biol. Chem. 1994, 269, 4260-4266.
9
15010
J. AM. CHEM. SOC. 2003, 125, 15010-15020
10.1021/ja035731o CCC: $25.00 © 2003 American Chemical Society

Oxygen Activation in P450 BM3
A R T I C L E S
controlling activation itself are less well understood. Dioxygen
binds to ferrous heme-iron but has negligible affinity for the
ferric form. Heme reduction occurs only after substrate binding,
which induces a change in the iron coordination geometry from
octahedral to square-based pyramidal, an event which is
accompanied by a change from a low- to high-spin state at the
ferric iron.16,17 These steps initiate the P450 catalytic cycle, and
their rigorously sequential nature helps to prevent unproductive
cycling.18 In a productive cycle, the oxy-ferrous complex (once
formed) is reduced by a second electron from the reductase.
On protonation, this complex decomposes to the reactive oxy-
ferryl (FeIVdO) moiety responsible for oxygen atom insertion.
Although this sequence of steps is common to all P450s, the
rate at which the steps occur is highly variable as demonstrated
by the variability in the catalytic turnover rates of the many
P450 isoforms studied.1 These are influenced by additional
factors such as interprotein recognition, electron transfer,
substrate recognition, and product release. Molecular oxygen
activation must be carefully controlled in each case to ensure
efficient substrate oxygenation without uncoupling, i.e., unpro-
ductive dissociation of reduced oxygen species from the heme
prior to reaction with substrate.18 P450s must therefore balance
their ability to reduce molecular oxygen against their ability to
stabilize reactive-oxygen intermediates. Other oxygen-binding
hemoproteins have very different properties; oxidases promote
the release of reduced oxygen species, while the globins
reversibly bind dioxygen.19-21 The question of how this balance
is achieved is of fundamental importance.
equivalents directly to the heme. The unusual structure of P450
BM3 makes it an ideal model of a mammalian P450 system
contained within a single component. It is particularly useful
for studying how electron transfer is coupled to oxygen
activation in the P450s. Our earlier studies showed how
substitution of phenylalanine 393 of P450 BM3 influences the
heme reduction potential and the stability of the oxy-ferrous
complex.23,24 Here, we extend this work to examine how the
thermodynamic properties of the heme influence the key steps
in the P450 catalytic cycle: electron transfer, and oxygen
activation.
Experimental Procedures
Escherichia coli Strains and Plasmids. The preparation of plasmids
for the overexpression of full-length and heme domain constructs of
WT P450 BM3 (pBM23 and pBM20), F3939A (pCM36 and pCM80),
F393H (pCM37 and pCM81), and the F393Y (pCM109 and pCM125)
mutant enzymes has been reported previously.23 The F393W mutant
was constructed by oligonucleotide-directed mutagenesis of the wild-
type plasmids pJM23 and pJM20, respectively, using the Kunkel
method.25 The oligonucleotide primer used in the mutagenesis procedure
is shown below (mismatches are indicated by the italicized bases)
5′CTGACCGTTTCCCCACGGTTTAAACG 3′. The resulting full-
length and heme domain constructs were named pCM111 and pCM110,
respectively. Both plasmids were sequenced on a Perkin-Elmer ABI
Prism 377 DNA sequencer to ensure no secondary mutations had
occurred. The E. coli strain TG1 [supE, hsd∆5, thi, ∆(lac-proAB), F′
[tra∆36, proAB+, lacIq, lacZ∆M15]] was used for all cloning work
and for overexpression of the full-length and heme domain proteins.
Enzyme Preparations. All enzymes were isolated and purified as
described previously.23 All pure proteins were concentrated to >500
µM by ultrafiltration and were flash frozen in liquid nitrogen prior to
storage at -80 °C. Enzymes were used within 1 month of isolation.
Spectrophotometric Analysis of Fatty Acid and Carbon Mon-
oxide Binding to P450s. UV/visible absorption spectra were recorded
over the 300-800 nm range using a Shimadzu 2101 spectrophotometer
and quartz cuvettes of 1 cm path length. Typically, the concentration
of P450 BM3 used was 1-5 µM in 1 mL of assay buffer (100 mM
MOPS pH 7.0) at 30 °C. Substrate dissociation constants were
determined for arachidonate and laurate according to established
procedures.23
While the sequence identity of P450s is quite low (typically
15-25%), particular regions do exhibit a high degree of identity.
The heme binding loop runs from the heme ligand (Cys),
through a glycine residue to a phenylalanine residue (Phe393
in P450 BM3), all of which are highly conserved.22 This
structural motif is fundamentally important to P450s since the
Fe-Cys bond is the only formal bonding interaction between
the heme and the protein. Substituting the cysteine ligand results
in one of two outcomes, (i) catalytic deactivation or (ii) failure
to incorporate heme,8,9 both of which demonstrate the necessity
for a cysteine ligand but neither of which explains why.
Steady-State Kinetics. All steady-state kinetic measurements were
performed at 15 °C in air saturated assay buffer using 1 cm path length
quartz cuvettes. Initial rates of NADPH oxidation were measured, as
described previously,23 by monitoring the decrease in absorbance of
NADPH (ꢀ340 ) 6.21 mM-1 cm-1) with time, under substrate-free or
substrate-saturating ([arachidonate] ) 100 µM) conditions. Enzyme
concentrations used were typically 2-10 µM for substrate-free assays
and 10-100 nM for arachidonate-saturated assays. The quoted rate
constants (kcat) are the average of three separate experiments.
Pre-Steady-State Kinetics. All pre-steady-state measurements were
performed at 15 °C using an Applied Photophysics stopped-flow
spectrophotometer (SX.17MV) contained within an anaerobic glovebox
(Belle Technology; [O2] < 5 ppm) using either single-wavelength or
diode-array detectors.
P450 BM3 is a single component fatty acid monooxygenase
from Bacillus megaterium.45 It is composed of a P450-heme-
containing oxygenase domain, which is connected via a short
protein linker to a diflavin reductase domain. The reductase
domain is related to mammalian microsomal P450 reductase
and binds 1 equiv each of FMN and FAD. The reductase domain
functions as an NADPH dehydrogenase, supplying electron
(13) Benson, D. E.; Suslick, K. S.; Sligar, S. G. Biochemistry 1997, 36, 5104-
5107.
(14) Schlichting, I.; Berendzen, J.; Chu, K.; Stock, A. M.; Maves, S. A.; Benson,
D. E.; Sweet, R. M.; Ringe, D.; Petsko, G. A.; Sligar, S. G. Science 2000,
287, 1615-1622.
(15) Davydov, R.; Makris, T. M.; Kofman, V.; Werst, D. E.; Sligar, S. G.;
Hoffman, B. M. J. Am. Chem. Soc. 2001, 123, 1403-1415.
(16) Sligar, S. G. Biochemistry 1976, 15, 5399-5406.
(17) Tsai, R.; Yu, C. A.; Gunsalus, I. C.; Peisach, J.; Blumberg, W.; Orme-
Johnson, W. H.; Beinert, H. Proc. Natl. Acad. Sci. U.S.A. 1970, 66, 1157-
1163.
A. Heme Reduction. Rate constants for the first flavin-to-heme
electron-transfer step (kred) were determined by monitoring the formation
of the ferrous-CO (FeII-CO) adduct of the full-length flavocytochromes
with time. One syringe contained NADPH (100 µM), and the second
(18) Loida, P. J.; Sligar, S. G. Biochemistry 1993, 32, 11530-11538.
(19) Jones, R. D.; Summerville, D. A.; Basolo, F. Chem. ReV. 1979, 79, 139-
179.
(20) Chapman, S. K.; Daff, S.; Munro, A. W. Struct. Bonding 1997, 88, 39-
(23) Ost, T. W. B.; Miles, C. S.; Munro, A. W.; Murdoch, J.; Reid, G. A.;
Chapman, S. K. Biochemistry 2001, 40, 13421-13429.
70.
(21) Dawson, J. H. Science 1988, 240, 433-439.
(24) Ost, T. W. B.; Munro, A. W.; Mowat, C. G.; Taylor, P. R.; Pesseguiero,
A.; Fulco, A. J.; Cho, A. K.; Cheesman, M. R.; Walkinshaw, M. D.;
Chapman, S. K. Biochemistry 2001, 40, 13430-13438.
(22) Nelson, D. R.; Koymans, L.; Kamataki, T.; Stegeman, J. J.; Feyereisen,
R.; Waxman, D. J.; Waterman, M. R.; Gotoh, O.; Coon, M. J.; Estabrook,
R. W.; Gunsalus, I. C.; Nerbert, D. W. Pharmacogenetics 1996, 6, 1-42.
(25) Kunkel, W. H. Proc. Natl. Acad. Sci. U.S.A. 1985, 82, 488-492.
9
J. AM. CHEM. SOC. VOL. 125, NO. 49, 2003 15011

A R T I C L E S
Ost et al.
Table 1. Thermodynamic and Kinetic Parameters Determined for Wild-Type and the F393 Series of Mutant Enzymes of Flavocytochrome
P450 BM3g
enzyme
F393A
-151 ( 4
18 ( 2
240 ( 24
10 ( 1
F393H
F393Y
WT
-289 ( 5
67 ( 7
99 ( 10
140 ( 10
F393W
Substrate-bound (SB)a
-176 ( 4
21 ( 2
205 ( 21
3 ( 1
Substrate-free (SF)b
-332 ( 6
0.28 ( 0.03
reduction potential (mV)c,d
-295 ( 6
-360 ( 5
e
turnover (kcat, s-1
)
66 ( 7
41 ( 4
29 ( 3
e
heme reduction (kred, s-1
)
94 ( 10
autoxidation (kautox, × 103 s-1
)
110 ( 10
540 ( 50
e
reduction potential (mV)c,d
-312 ( 4
0.27 ( 0.03
-418 ( 6
0.12 ( 0.01
-427 ( 4
0.09 ( 0.01
-480 ( 5
0.08 ( 0.01
e
turnover (kcat, s-1
)
heme reduction
e,f
slow phase (kreds, s-1
)
0.3 ( 0.1 (71%)
3.1 (24%)
10 ( 1
0.2 ( 0.1 (48%)
1.3 (14%)
5 ( 1
0.2 ( 0.1 (18%)
0.9 (6%)
90 ( 9
0.14 ( 0.05 (28%)
2.6 (8%)
110 ( 20
0.3 ( 0.1 (11%)
11 (15%)
440 ( 50
fast phase (kredf, s-1
)
e,f
autoxidation (kautox, × 103 s-1
)
e
a Determined in the presence of saturating concentrations (100 µM) arachidonate. b Determined in the absence of substrate. c Values for wild-type, F393A,
F393H, and F393Y reported in ref 23. d Experiments conducted at 25 °C in 100 mM KPi pH 7.0 except for F393W where the buffer was 100 mM Tris/500
mM KCl, pH 7.0. e Experiments conducted at 15 °C in 100 mM MOPS pH 7.0. f Heme reduction slow phase (kreds) and fast phase (kredf), with percentage
of total heme reduction in each phase. g Experimental conditions are described in the Experimental Procedures section.
syringe contained full-length wild-type or mutant P450 BM3 (1-5 µM).
Both syringes contained anaerobic assay buffer previously saturated
with CO by bubbling for 5 min ([CO] ) ∼200µM). In the case of the
substrate-saturated experiments, full conversion (>95%) to the FeII-
CO complex was observed; the data were monophasic and were
evaluated by fitting the absorbance change to a single-exponential
function (λobs ) 450 nm (WT, F393Y, F393W), 445 nm (F393H), and
444 nm (F393A)). In the absence of substrate, incomplete conversion
to the FeII-CO complex was observed; the data were biphasic and
were evaluated by fitting the absorbance change to a double exponential
function. The slow phase constituted the greatest proportion of the
amplitude. Analyses were performed using Microcal Origin 7 software.
Values quoted are from the averages of four separate measurements.
Rate constants determined for the fast phase were less predictable;
examples are shown in Table 1.
0.5 M KCl in an anaerobic glovebox. The following mediators were
then added: 2-hydroxy-1,4-naphthoquinone (20 µM), FMN (5 µM),
benzyl viologen (10 µM) and methyl viologen (10 µM). Spectro-
electrochemical titrations were performed at 25 ( 2 °C using an Autolab
PGSTAT10 potentiostat and a Cary 50 UV/vis spectrophotometer. The
potential of the working electrode was decreased in 30 mV steps until
the enzyme was fully reduced and increased stepwise until reoxidation
was complete. After each step, the current and UV/vis absorption
spectrum were monitored until no further change occurred. This
equilibration process typically lasted 15 min. Absorbance changes were
plotted against the potential of the working electrode and analyzed using
the Nernst equation. The Ag/AgCl reference electrode employed in the
OTTLE cell was calibrated against indigotrisulfonic acid (Em ) -99
mV vs SHE) and FMN (Em ) -220 mV vs SHE) in the same buffer
conditions. All electrode potentials were corrected accordingly by +182
( 2 mV relative to the standard hydrogen electrode.
B. Autoxidation. Rate constants (kautox) for the decay of the oxy-
ferrous complexes of the wild-type and mutant P450 BM3 heme
domains were determined by monitoring the rates of autoxidation to
ferric P450 with time. The heme domain proteins (∼10 µM) were
prereduced by addition of 40 µM sodium dithionite. Excess dithionite
was removed prior to experimentation by elution of the reduced P450s
through a pre-equilibrated 10 mL G25 gel-filtration column. Reduction
was confirmed by UV/visible spectrometry on a Varian Cary 50 Bio
spectrophotometer, also contained within the glovebox. Formation of
the oxy-ferrous complex was achieved by mixing the reduced P450
(syringe A) with air-saturated assay buffer ([O2] ≈ 200 µM (syringe
B)). Formation of the oxy-ferrous species was too fast to measure (>600
s-1), yet in each case, the decay occurred over a period of seconds and
could be conveniently measured. Autoxidation of ferrous P450 (Fe2+
- O2 f Fe3+ + O2-) was followed using a PDA detector across the
range 300-700 nm. The variation in A470 with time was monophasic
for both the substrate-free and substrate-saturated experiments and was
evaluated by fitting the data to a single exponential (Microcal Origin
7). This wavelength gave the most consistent results over all the
different conditions. For substrate-free enzyme some high to low spin
conversion can be seen in the Soret region over long time scales. Over
short time scales, some oxyferrous formation can also be observed.
Optically Transparent Thin Layer Electrochemical (OTTLE)
Potentiometry. Spectroelectrochemical analysis of the F393W mutant
heme domain was conducted in an OTTLE cell constructed from a
modified quartz EPR cell with a 0.3 mm path length, containing a Pt/
Rh (95/5) gauze working electrode (wire diameter 0.06 mm, mesh size
1024 cm-1, Engelhardt, UK), a platinum wire counter electrode and a
Ag/AgCl reference electrode (model MF2052, Bioanalytical Systems,
IN 47906, USA). Enzyme samples (0.5 mL × 100-200 µM) were
eluted through a G25 column pre-equilibrated with 0.1 M Tris pH 7.5,
Crystallization and Refinement. Crystallizations of F393A, -W,
and -Y heme domains were carried out by hanging drop vapor diffusion
at 4 °C in Linbro plates. Crystals were obtained with a well solution
comprising 100 mM sodium PIPES, pH 6.5-7.5, 40 mM MgSO4, and
18-21% PEG 8000. Hanging drops of 4 µL were prepared by adding
2 µL of 40 mg/mL protein (in 50 mM TrisHCl/1 mM EDTA, pH 7.4)
to 2 µL of well solution. Plates of up to 1 × 0.3 × 0.3 mm3 were
formed after about 1 week. Crystals were immersed in well solution
containing 22% glycerol as a cryoprotectant, prior to mounting in nylon
loops and flash-cooling in liquid nitrogen. For the F393A mutant heme
domain, a data set was collected to a 2.05 Å resolution on beamline
×11 (λ ) 0.8482 Å) at DESY Hamburg, for the F393W heme domain,
a data set was collected to a 2.0 Å resolution on beamline BM14 (λ )
1.0332 Å) at ESRF Grenoble, and for the F393Y heme domain, a data
set was collected to a 2.0 Å resolution on beamline ×13 (λ ) 0.802
Å) at DESY Hamburg. All data were collected using a Mar CCD
detector. Crystals of all three mutant forms belong to space group P21.
Crystals of the F393A heme domain were found to have the following
cell dimensions: a ) 58.694 Å, b ) 152.911 Å, c ) 61.197 Å, and â
) 94.640°. Those of the F393W heme domain had the following
dimensions: a ) 58.911 Å, b ) 153.538 Å, c ) 61.430 Å, and â )
94.424°. Those of the F393Y heme domain had the following
dimensions: a ) 59.608 Å, b ) 152.983 Å, c ) 61.774 Å, and â )
94.539°. Data processing was carried out using the HKL package.2626
The wild-type flavocytochrome P450 BM3 heme domain structure
(2HPD),2727 stripped of water, was used as the initial model. Electron
(26) Otwinowski, Z.; Minor, W. Methods Enzymol. 1997, 276, 307-326
(27) Ravichandran, K. G.; Boddupalli, S. S.; Hasemann, C. A.; Peterson, J. A.;
Deisenhofer, J. Science 1993, 261, 731-736.
9
15012 J. AM. CHEM. SOC. VOL. 125, NO. 49, 2003

Oxygen Activation in P450 BM3
A R T I C L E S
density fitting was carried out using the program TURBO-FRODO2828
(in Silicon Graphics Geometry Partners Directory 86, Silicon Graphics,
Mountain View, CA). Structure refinement was carried out using
Refmac.2929 The atomic coordinates have been deposited in the Protein
Data Bank (1P0V:F393A; 1P0W:F393W; 1P0X:F393Y).
Results
Characterization of the F393W Mutant. The ferrous
F393W mutant (in both full-length and heme domain) formed
a typical P450 complex with CO, with a Soret absorption
maximum at 449 nm. Titration of the mutant enzyme with
substrate (either arachidonate or laurate) induced a shift in the
absorption spectrum consistent with the low- to high-spin
transition of ferric heme iron. The substrate dissociation
constants were determined to be 578 ( 35 µM for laurate and
2.0 ( 0.4 µM for arachidonate, both of which are similar to
values determined for the wild-type enzyme. The most obvious
change caused by the mutation is a large decrease in the heme
reduction potential in the presence and absence of substrate.
These values are -480 ( 5 mV and -360 ( 5 mV for the
substrate-free and substrate-saturated F393W heme domain,
respectively (Table 1), compared to values of -427 mV and
-289 mV for the wild-type enzyme, respectively. These
potentials are substantially more negative than the midpoint
potential of NADPH (-320 mV) and probably also of the FMN
cofactor of P450 BM3,48 making electron transfer to the heme
in the F393W mutant enzyme thermodynamically unfavorable
even in the presence of substrate. In the wild-type enzyme, heme
reduction in the presence of NADPH is favored only after
substrate binding. This effect, together with the spin-state
change, acts as an electron-transfer control mechanism. The
other mutants included in this study have been characterized
previously:23,24 the F393A and F393H mutants both have higher
reduction potentials than the wild-type enzyme and blue-shifted
Soret absorption bands for their ferrous heme-CO complexes.
The F393Y mutant, on the other hand, is similar to the wild-
type enzyme in these respects. All the mutants have similar
substrate binding affinities and associated spin-state shifts. This
is the first indication that some key properties of the enzyme
active site are unaffected by the mutation despite the large shifts
either way in heme reduction potential.
Steady-State Kinetics. The kcat values for wild-type P450
BM3 and the F393A, -H, -Y, and -W mutants are shown in
Table 1. The background rate constant for NADPH oxidation
in the absence of substrate is small (kcat < 0.3 s-1) for all the
mutants, and in all cases, the kcat increases by 2 orders of
magnitude (18-67 s-1) in the presence of substrate. This is
consistent with the established idea that substrate binding gates
electron transfer to the heme, such that reducing equivalents
are only consumed when substrate is present. The F393W
mutant was found to couple NADPH consumption to substrate
monooxygenation with greater than 90 % efficiency. The other
mutants and the wild-type enzyme couple similarly well23 (i.e.
>80%) except for the F393A mutant, which is slightly less
efficient (56% coupling). The substrate saturated kcat varies
across the series of enzymes studied (WT ≈ F393Y > F393W
. F393H ≈ F393A) by approximately 3-fold, but there is no
Figure 1. Heme reduction of flavocytochrome P450 BM3. Panels A and
B show the time courses for formation of the P450 complex from the
substrate-saturated (A) and substrate-free (B) forms of the wild-type (WT)
and mutant enzymes. Each trace is labeled with the one letter code
corresponding to the particular F393X substitution and is shown fitted to a
single (panel A) or double (panel B) exponential function. The abscissa in
panel B has been rescaled to show the proportion of ferrous CO complex
formed during the second (slow) exponential phase (kreds). For wavelengths,
see Experimental Procedures. Panel C shows the absorption spectra of
substrate-free ferric (solid line), substrate-bound ferric (broken line), and
the ferrous-CO complex (dotted line) of the wild-type enzyme.
linear correlation with reduction potential. The wild-type enzyme
has the largest kcat (67 s-1), and this is similar to the F393Y
mutant (which has a similar reduction potential). The F393W
mutant turns over at half this rate and has a more negative
reduction potential. The F393H and F393A mutants on the other
hand are slower still but have more positive reduction potentials.
This complex trend suggests that the reduction potential of the
wild-type enzyme is optimized to give the fastest overall
catalytic rates.
Heme Reduction. In a series of stopped-flow experiments,
rate constants were determined for the formation of ferrous CO
complex on reaction of the holoenzymes with NADPH (Figure
1). These data (Table 1) approximate to the rate of first electron
transfer to the heme. The ferrous form generated by electron
transfer to the heme, is trapped by CO binding.49 Binding of
CO by reduced enzyme is rapid, occurring at rates in excess of
1000 s-1 in CO-saturated buffer in the wild-type and mutant
enzymes. The overall process has many individual steps, most
of which occur quickly, nevertheless, slight lag phases can be
observed in some of the traces in Figure 1. A contributing factor
may be reduction of the flavins, which is also a multiphasic
process observed at 450 nm as a decrease in absorbance.49 In
the absence of substrate the heme reduction traces are biphasic,
with a small fast phase being followed by a larger slow phase.
The fast phase showed no obvious trend and varies erratically
(examples only are presented in Table 1). Larger fast phases
are obtained with enzyme containing a larger proportion of high-
spin heme this has been found to result from contamination with
detergents during preparation, which bind irreversibly to the
enzyme. The slow phase was consistent for all the mutants, with
very little change across the series. However, its amplitude
appears to increase with reduction potential. It appears that the
electron-transfer process is an equilibrium, the position of which
depends on the reduction potential of the CO-bound heme and
(28) Roussel, A.; Cambillau, C. Silicon Graphics Geometry Partners Dictionary
86; Silicon Graphics: Mountain View, CA.
(29) Murshudov, G. N.; Vagin, A. A.; Dodson, E. J. Acta Crystallogr. 1997,
D53, 240-255.
9
J. AM. CHEM. SOC. VOL. 125, NO. 49, 2003 15013

A R T I C L E S
Ost et al.
Scheme 1
that of the FMN. The rate constants observed are then the sum
of the forward and reverse rate constants. With this taken into
account, in the absence of substrate, heme reduction appears to
be slow enough to limit steady-state turnover for all the enzymes
studied (kreds < 0.4 s-1). For the F393W mutant, the rate
SF
constant (slow phase) determined for heme reduction is 5-fold
faster than kcat, but only 11% heme reduction occurs in the slow
phase, suggesting that the reverse process is favored by the
extremely negative heme reduction potential.
Figure 2. Formation and decay of the oxy-ferrous species of the P450
BM3 F393A-heme domain at 15 °C under substrate-free (panel A) and
substrate-saturating conditions (panel B). In both cases, the oxy-ferrous
species is shown by the broken line, the solid lines correspond to ferric
low-spin (panel A) and high-spin P450 (panel B). Panel C shows the
exponential decay (kautox) of substrate-free (upper trace) and substrate-
saturated (lower trace) oxy-ferrous complexes, monitored at 470 nm. Both
are shown fitted to single-exponential decay functions.
In the presence of substrate, the rate constants for heme
reduction increase by 2 orders of magnitude (kred is 29 s-1 or
more) for all the enzymes studied. However, there is now a
clear correlation with the heme reduction potential. At more
negative reduction potentials (i.e., for the F393W and F393Y
mutants and the wild-type enzyme) the rate of heme reduction
slows and is similar to the rate of catalytic turnover, indicating
that FMN to heme electron transfer is rate determining. At more
positive reduction potentials, the rate of heme reduction is
faster: kred for the F393A mutant in the presence of substrate
is 10-fold greater than for the F393W mutant. However, the
rate of catalytic turnover is slower, such that kred is now 10-
fold greater than kcat. This indicates that a change in the rate-
determining step has occurred for the F393A and F393H
mutants.
Oxy-Ferrous Decay. In the P450 catalytic cycle, heme
reduction is followed by the binding of dioxygen, to form an
oxy-ferrous/superoxy-ferric complex (Scheme 1). This is the
starting point for the multistep oxygen activation process. Using
rapid-scanning UV/vis stopped-flow spectrophotometry we were
able to obtain spectra of the oxy-ferrous complexes of all the
mutants and the wild-type enzyme and monitor their rates of
decay to ferric heme, probably via the dissociation of superoxide.
Figures 2-4 show the spectra of the wild-type enzyme and the
F393A and F393W mutants in the presence and absence of
substrate, and their respective time-courses for decay. The Soret
peak positions were at 423 and 425 nm for the substrate-free
and substrate-bound oxy-ferrous complexes respectively, regard-
less of mutation, but differences were observed in the long-
wavelength region. These are shown in Figure 5, panel C. The
similarities of the Soret bands of the mutant enzymes spectra
indicate that they are all of oxy-ferrous complexes. The
differences observed in the R-â region are not, therefore, due
to contamination by oxidation products, but indicate that the
heme electronic structure is perturbed by the mutations.
Although decay of the oxy-ferrous complex is a nonproduc-
tive process so far as catalytic turnover is concerned, its rate is
indicative of complex stability. The oxy-ferrous complexes of
all the F393X enzymes decayed to the ferric high-spin form in
the presence of substrate and primarily to low-spin ferric heme
in the absence of substrate, as expected. However, in the latter
reaction, a proportion of the enzyme remained high-spin
(<20%), possibly due to oxidative damage of the active site.
Rate constants for the autoxidation of the ferrous P450s in the
absence (kautoxSF) and presence of substrate (kautoxSB) are shown
in Table 1. Two major observations can be made. First, the
autoxidation rate is essentially independent of the presence of
substrate. This indicates that substrate has little effect on the
stability of the oxy-ferrous complex in this P450 and is unlikely
to be involved directly in catalytic oxygen activation. The slight
increase in stability may be a reflection of the increased
hydrophobicity of the active site when substrate is bound.
Second, the oxy-ferrous decay rate constants correlate with
the heme reduction potentials. The F393H oxy-ferrous complex
SB
was the most long-lived (kautox ) 3.0 × 10-3 s-1), with a
40-fold slower autoxidation rate than the oxy-ferrous complex
of the wild-type enzyme (kautoxSB ) 0.14 s-1). The F393W oxy-
ferrous complex, on the other hand, decayed 4 times faster
(kautox ) 0.54 s-1) than that of the wild-type enzyme. The
SB
oxy-ferrous complex is therefore destabilized in mutants with
lower reduction potentials and Vice Versa. The formation of
superoxide on decay of the oxy-ferrous complex is likely to
contribute to uncoupled turnover of the enzymes. The viability
of this can be ascertained by comparing the rate constants for
oxy-ferrous decay to the steady-state turnover rates. In the
presence of substrate, catalytic turnover is 2 orders of magnitude
faster than autoxidation for all the mutants, indicating that
uncoupling via this route is not viable. Coupling percentages
9
15014 J. AM. CHEM. SOC. VOL. 125, NO. 49, 2003

Oxygen Activation in P450 BM3
A R T I C L E S
Figure 4. Formation and decay of the oxy-ferrous species of the P450
BM3 F393W-heme domain at 15 °C under substrate-free (panel A) and
substrate-saturating conditions (panel B). In both cases, the oxy-ferrous
species is shown by the broken line, the solid lines correspond to ferric
low-spin (panel A) and high-spin P450 (panel B). Panel C shows the
exponential decay (kautox) of substrate-free (upper trace) and substrate-
saturated (lower trace) oxy-ferrous complexes, monitored at 470 nm. Both
are shown fitted to single-exponential decay functions.
Figure 3. Formation and decay of the oxy-ferrous species of the wild-
type P450 BM3 heme domain at 15 °C under substrate-free (panel A) and
substrate-saturating conditions (panel B). In both cases, the oxy-ferrous
species is shown by the broken line, the solid lines correspond to ferric
low-spin (panel A) and high-spin P450 (panel B). Panel C shows the
exponential decay (kautox) of substrate-free (upper trace) and substrate-
saturated (lower trace) oxy-ferrous complexes, monitored at 470 nm. Both
are shown fitted to single-exponential decay functions.
were found to be greater than 90% for all mutants except the
F393A mutant (56%), which must uncouple after the second
electron transfer event. In the absence of substrate, the rate of
background turnover appears to be limited by the rate of heme
reduction, however, subsequent electron transfer events are
likely to be faster than autoxidation avoiding the generation of
superoxide. For the F393A and F393H mutants substrate-free
turnover is > 30-fold faster than autoxidation, indicating that
this is the case.
Crystal Structures. Data sets to a resolution of 2.0 Å (2.05
Å for F393A) were used to refine the structures of the F393A,
F393W, and F393Y heme domains to final R-factors of 16.87%
(Rfree ) 23.73%), 15.93% (Rfree ) 22.13%), and 17.17% (Rfree
) 23.70%) respectively (Table 2). For the F393A-heme domain,
the final model consists of two protein molecules (molecule A
comprising residues Lys3-Leu188 and Asn201-Leu455; mol-
ecule B comprising Lys3-Leu188 and Tyr198-Leu455) each
containing 1 heme and a total of 1408 water molecules. For
the F393W-heme domain and F393Y-heme domain, each of
the final models consists of two protein molecules (molecule
A comprising residues Lys3-Leu188 and Asn201-Leu455;
molecule B comprising Lys3-Arg190 and Pro196-Leu455) each
containing 1 heme and a total of 1420 and 1238 water molecules,
respectively. For all molecules, the electron density for the first
two N-terminal residues and all (or most of) the loop region
containing residues Gln189-Glu200 was uninterpretable, so these
have been omitted from the model. The RMSD fit of all
backbone atoms for each of the F393X-heme domain structures
and the wild-type heme domain is 0.3 Å indicating no major
differences between the four structures. Due to the presence of
two molecules in the asymmetric unit for each of these models,
Figure 5. Comparison of the visible absorption spectra of P450 BM3 with
oxyhemoglobin. Panel A shows the characteristic visible absorption spectra
of the ferric high-spin (FeIII (HS)), ferric low-spin (FeIII (LS)) and
carbonmonoxy-ferrous (FeII-CO (LS)) forms of the wild-type P450 BM3
heme domain. Panel B shows the visible absorption spectrum of oxy-
hemoglobin (FeII-O2). Panel C shows the visible absorption spectra of the
oxy-ferrous complexes (FeII-O2) of F393A, wild-type, and F393W P450
BM3 heme domains.
the RMSD values stated are the average over both molecules.
In addition, the RMSD fit between both molecules (A and B)
9
J. AM. CHEM. SOC. VOL. 125, NO. 49, 2003 15015

A R T I C L E S
Ost et al.
Table 2. Data Collection and Refinement Statistics
structural difference associated with the change in the side chain
caused by the substitution of phenylalanine 393, the heme
binding region of all of these mutant forms remains relatively
unperturbed by the particular substitution. The exception to this
is the F393A mutant where a conformational change in the
orientation of a nearby surface glutamine residue (Gln403)
occurs. Comparison of the WT and the other mutant structures
shows that the amide side chain of Gln403 points away from
the heme, out toward the protein:solvent interface. However,
in the case of the F393A-heme domain mutant structure, this
glutamine has flipped ∼ 45° toward the heme, with the amide
head occupying the void left by the substitution of the bulky
phenyl side chain of F393 by alanine (Figure 7a). The amide
group of Gln403 is the same distance (3.6 Å) from the heme
ligand as the histidine side chain is in the F393H mutant (3.7
Å). Both may form hydrogen-bonding interactions with Cys400.
This would explain why the redox properties of the two mutant
enzymes are so similar; they are the result of a common
interaction. A hydrogen bond to the axially ligating cysteine
would be expected to reduce its potency as a strong σ and π
donor, leading to the increase in heme reduction potential
observed. The imidazole group of the F393H mutant may also
mediate the reduction potential by electrostatically interacting
F393A
F393W
F393Y
resolution (Å)
20.0-2.05
361 447
66 956
98.7
8.6
8.6
20.0-2.0
498 626
72 855
99.6
13.2
6.5
17.0-2.0
837 703
67 437
91.0
total no. of reflections
no. of unique reflections
completeness (%)
(I)/[σ(I)]
10.0
Rmergea (%)
8.5
R
merge in outer shell (%)
24.5
20.9
18.3
(2.12-2.05 Å) (2.12-2.0 Å) (2.07-2.0 Å)
Rcrystb (%)
16.87
23.73
15.93
22.13
17.17
23.70
Rfreeb (%)
RMSD from ideal values:
bond lengths (Å)
bond angles (deg)
ramachandran analysis:
most favored (%)
0.013
2.8
0.012
2.4
0.013
2.6
90.9
90.7
8.8
91.4
8.1
additionally allowed (%) 8.6
a Rmerge ) ΣhΣi|I(h) - II(h)|/ΣhΣiII(h), where II(h) and I(h) are the ith
b
and mean measurement of reflection h, respectively. Rcryst ) Σh|Fo - Fc|/
Σh Fo, where Fo and Fc are the observed and calculated structure factor
amplitudes of reflection h, respectively. Rfree is the test reflection data set,
5% selected randomly for cross validation during crystallographic refine-
ment.
is ∼0.2 Å for each of the mutant enzyme structures. Electron
density for the heme binding region of each of the F393X-heme
domain structures is shown in Figure 6. Apart from the expected
Figure 6. Stereoviews of the heme-binding region of P450 BM3 F393A (A), F393Y (B), and F393W (C). Electron density is shown in cyan, while the
heme and peptide are shown in atom-type colors (red ) O; blue ) N; yellow ) S; orange ) Fe). In each case, the substituted residue (X393) and the
important glutamine 403 (Gln403) are labeled. The electron density map was calculated using Fourier coefficients (2Fo - Fc), where Fo and Fc are the
observed and calculated structure factors, respectively. The contour level is 1.0σ, where σ is the RMS electron density. This diagram was generated using
BOBSCRIPT46 and RASTER 3D47.
9
15016 J. AM. CHEM. SOC. VOL. 125, NO. 49, 2003

Oxygen Activation in P450 BM3
A R T I C L E S
Figure 7. Overlaid stereoviews of the heme-binding region of P450 BM3 heme domain. Panel A shows the overlay of the F393A (atom type colors) and
F393H (magenta) structures. Panel B shows the overlay of wild-type (atom type colors) with the F393Y (red) and the F393W (green) mutant structures. This
diagram was generated using BOBSCRIPT46 and RASTER 3D47.
with the heme,23 whereas the F393A mutant is unable to do
this. The F393W-heme domain mutant structure (Figure 7b)
shows that the indole nitrogen of Trp393 is not in the correct
orientation to hydrogen-bond with Cys400, but the increased
size of the residue side chain results in a closer interaction
between it and the heme macrocycle. The F393W mutant has a
more negative reduction potential than the wild-type enzyme.
This is probably due to changes in the dielectric environment
of the heme and its axial ligand, which appear to be heavily
shielded from the aqueous solvent in this mutant enzyme.
low to high-spin FeIII) and a reduction potential increase to favor
heme reduction. Consequently, the wild-type enzyme’s turnover
rate is 500-fold higher with substrate than without. The substrate-
free F393A and -H mutants have similar reduction potentials
to the substrate-bound wild-type enzyme; therefore, by compar-
ing the heme reduction rates of the wild-type enzyme and the
F393A or -H mutants (Table 1) and assuming that the mutations
only affect the electronic properties of the heme, we are able to
calculate the individual contributions of the spin-state shift and
the reduction potential shift to the overall acceleration effect.
These factors have been the subject of intense scrutiny,16,31 but
to date, the importance of the relative contribution of each effect
remains experimentally unverified. Despite its similar reduction
potential, the substrate-free F393A mutant has a 200-fold slower
turnover rate than the substrate-bound wild-type enzyme.
Conversely, the substrate-bound F393W mutant, which has a
70 mV lower reduction potential, is only 50% slower. It is clear,
therefore, that the spin-state shift is the dominant activation
mechanism and is 100-fold more important than the reduction
potential shift. The rate constants for heme reduction add further
insight; these are low for all the substrate-free enzymes, which
have low-spin heme and increase by at least 70-fold on substrate
binding. For the substrate-bound enzymes, the rate constants
for heme reduction increase linearly with reduction potential
(Figure 8) but only by a factor of 8 over 200 mV. Clearly this
is not enough to cause a 500-fold increase in kcat on substrate
binding. The results are consistent with nonadiabatic electron
transfer theory,32 which predicts a Gaussian dependence of
electron-transfer rate on driving force (∆G). The linear relation-
ship obtained is almost certainly an artifact of the limited
reduction potential range studied; the plot would be expected
Discussion
Substitutions of Phe393 of P450 BM3 have been shown to
modulate the reduction potential of the heme over a 200 mV
range, while retaining the structural integrity of the enzyme’s
active site. The fact that the active site of the enzyme is
structurally unaffected by the mutations enables a quantitative
examination of the effect of modulating heme reduction potential
on individual steps in the P450 catalytic cycle. Consequently,
we are able to probe the kinetics and thermodynamics of heme
reduction and oxygen activation in a structurally consistent
environment. The electronic properties of the flavins in the
reductase domain of the enzyme have been shown to be
unperturbed, both kinetically and thermodynamically, by the
presence of the heme domain.48 It is unlikely, therefore, that
the mutations would have a significant influence on these
cofactors.
P450 BM3 is a supremely efficient monooxygenase system,
able to hydroxylate unactivated alkyl chains at rates of >100
s-1 (at 25 °C) with >95% coupling between substrates.23,30
Coupling is aided by a mechanism, common in P450s, in which
substrate binding induces a simultaneous spin-state shift (from
(31) Honeychurch, M. J.; Hill, A. O.; Wong, L.-L. FEBS Lett. 1999, 451, 351-
353.
(30) Narhi, L. O.; Fulco, A. J. J. Biol. Chem. 1986, 261, 7160-7169.
(32) Marcus, R. A.; Sutin, N. Biochem. Biophys. Acta 1985, 811, 265-322.
9
J. AM. CHEM. SOC. VOL. 125, NO. 49, 2003 15017

A R T I C L E S
Ost et al.
spectra of the oxy-ferrous and oxyferryl complexes of P450cam
under cryogenic conditions.
Generation of the oxy-ferrous P450 complex33 is the initial
oxygen activation step, but this species is not directly involved
in substrate hydroxylation.34 In the absence of additional
reducing equivalents, it decays spontaneously to ferric P450 and
superoxide (autoxidation). While this is an unproductive reaction
so far as catalysis is concerned, the rate of autoxidation is a
quantitative measure of the stability of the oxy-ferrous species.
As shown in Figure 8, the rate of decay increases exponentially
as the reduction potential of the heme decreases for the mutants
studied, which suggests a direct relationship between the heme
reduction potential and the activation energy for dissociation
of superoxide. This is clearly reasonable for a simple bond
scission event. Scheme 1 illustrates the relationship between
the heme reduction potential and the rate of autoxidation. The
reduction potential of the dioxygen/superoxide couple is constant
at approximately -160 mV; therefore, changes in the reduction
potential of the heme will be manifested solely in changes to
the equilibrium dissociation constants for dioxygen (Kox) and
superoxide (Ksox) from oxy-ferrous heme. The rate constants
for association are second order and diffusion controlled and
are therefore unlikely to show much dependence on reduction
potential. The dissociation rate constant for dioxygen (k-ox),
on the other hand, is likely to be affected by changes in the
FeII-dioxygen bond energy
The oxy-ferrous complex is shown in Scheme 1 along with
its alternative resonance form, the superoxy-ferric complex;
however, the true nature of the complex lies somewhere between
these two extremes. The oxy-ferrous/superoxy-ferric content of
this complex is determined by the relative positions of the Fe
3d-orbital energy levels and the oxygen π* orbital after bonding.
This will also determine the strength of bonding and, therefore,
the rate of autoxidation. A simplified MO diagram is shown in
Figure 9.19 According to this model, the oxy-ferrous/superoxy-
ferric character of the complex is determined by the energy gap
Figure 8. Dependence of kinetic parameters on reduction potential. Panel
A shows the plot of kcat versus reduction potential (b); panel B shows the
plot of kred (left axis, y, fitted to a straight line) and kautox (right axis, O,
fitted to an exponential decay) versus reduction potential. The arrow denotes
the position of wild-type P450 BM3.
to adopt Gaussian characteristics out with this range. Certainly,
it is difficult to believe that ket would drop to zero at -375 mV
as suggested by the ordinate intercept. Nevertheless, the clear
trend observed with reduction potential does suggest that the
electron-transfer rate is not being heavily perturbed by structural
changes at the heme-flavin interface, which would be likely
to introduce random scatter. The mutated residue does lie within
the proposed area of interdomain contact50 and may be involved
in a pathway for electron transfer, but the structural changes
induced by the mutations appear to have little effect on the rate
of electron transfer; i.e., there is only an 8-fold variation across
the series, which correlates with the shifts in the heme reduction
potential.
Electron transfer in the substrate free enzyme is complicated
by the fact that the heme is low-spin, six-coordinate ferric prior
to reduction but becomes high-spin ferrous afterwards. The
driving force for electron transfer in this case depends on the
reduction potential of the heme in a fixed low-spin six-
coordinate environment, which is likely to be very negative.
Electron transfer may also be initiated by spontaneous spin-
state equilibration. In this case, the proportion of heme in the
high-spin state (populated thermally) determines the rate of
electron transfer. Both mechanisms appear to be extremely
inefficient for all the enzymes studied, until a large-scale spin-
state shift is induced by substrate binding.
Various steps in the P450 catalytic cycle have been postulated
to be rate determining. These include the following: transfer
of the first electron to the heme (this is certainly the step
activated by substrate binding); transfer of the second electron;
oxygen activation (possibly involving protonation of a peroxo-
ferric species); and product release.1-4 It is also likely that the
rate-determining step will vary between different P450 enzymes,
given the variety of redox partners and substrates utilized by
these systems. Nevertheless, all P450s share the same mecha-
nism of oxygen activation. All the intermediate heme-oxy
complexes on the catalytic pathway are unstable, making their
characterization difficult. Great progress has recently been made
in this area by Sligar, Petsko, and co-workers,14 who have used
time-dependent X-ray crystallography to obtain structures and
(∆) between the iron dπ (dxz, dyz) and the π*O molecular orbitals.
2
If ∆ is large, the low-spin complex with oxy-ferrous character
will be favored. Decreasing ∆ results in the population of the
π*O orbital and superoxy-ferric character. In our series of
2
mutants, ∆ will be dependent on the energy of the Fe dxz,dyz
orbitals prior to oxygen binding, which will determine both their
position relative to the π*O orbital and the magnitude of their
2
bonding interaction with it. The energy of the Fe dxz,dyz orbitals
relates directly to the reduction potential of the heme and is
likely to be dependent on the π-donor strength of the thiolate
axial ligand. Additional H-bonding interactions (i.e., in the
F393A and F393H mutants) are likely to weaken the π-donor
ability of the thiolate and increase the reduction potential of
the heme. Electrostatic interactions between the imidizolium ion
that may have been formed in the F393H mutant and the heme
would also lower the energy of the Fe 3d-orbitals and increase
the reduction potential of the heme. The π-bonding interaction
between the dxz and π*O orbitals will shift electron density from
2
the iron to the oxygen and will cause the two orbitals to move
further apart in energy. However, if the π-bonding interaction
is weak, then an increase in the energy of the d orbitals caused
(33) Peterson, J. A.; Ishimura, Y.; Griffen, B. W. Arch. Biochem. Biophys. 1972,
149, 197-208.
(34) Lipscomb, J. D.; Sligar, S. G.; Namtvedt, M. J.; Gunsalus, I. C. J. Biol.
Chem. 1976, 251, 1116-1124.
9
15018 J. AM. CHEM. SOC. VOL. 125, NO. 49, 2003

Oxygen Activation in P450 BM3
A R T I C L E S
Figure 9. Qualitative molecular orbital diagram for P450 oxy-ferrous complexes (adapted from ref 19). This scheme illustrates how possible changes in the
π-donor character of the axial ligand could cause changes in ∆. The two situations shown are where ∆ > pairing energy (FeII-O2) and where ∆ < pairing
energy (FeIII-O2-).
by a mutation (e.g., F393W) will take the nonbonding dyz orbital
show that the oxyferrous complexes of the mutants are different
and have altered electronic properties.
closer in energy to the π*O orbital giving the electrons a greater
2
tendency to unpair.
In our series of mutants the active-site structure is retained,
making it unlikely that interactions between bound dioxygen
and active site residues could have been altered. Other oxygen-
binding hemoproteins utilize active-site residues to stabilize
bound dioxygen. Both the globins and peroxidases form
stabilized oxy-ferrous heme complexes in which histidine
residues (and also an Arg for peroxidase) interact with the bound
dioxygen.37,38 The active sites of P450s are often hydrophobic,
with proton delivery occurring via an active-site Thr residue
(Thr268 in P450 BM3) working in tandem with an Asp.39 This
environment is likely to disfavor occupation of the oxygen π
orbitals, either through a spin-state change or via bonding
interactions. The overall result is an activated P450 oxy-ferrous
complex in which its stability with respect to superoxide
dissociation (which is decreased by a lower heme reduction
potential) is balanced against its rate of reduction. Measuring
the rate of electron transfer to the oxyferrous complex directly
is beyond the scope of this study, but either this step or a
protonation event leading to oxygen activation is a likely rate-
determining step for the F393A and -H mutants. There is a
complex relationship between kcat and the heme reduction
potential in the P450 BM3 mutants (Table 1). At low potentials,
electron transfer to the heme is clearly rate determining, whereas,
at high potentials, a subsequent catalytic step becomes important
(Figure 8). The increased stability of the oxyferrous complexes
observed is an indication that the second electron transfer event
Some insight into the nature of the oxy-ferrous complex can
be gained from the long-wavelength visible spectra shown in
Figure 5. Panel A shows the spectra of high-spin ferric, low-
spin ferric, and carbonmonoxy-ferrous forms of wild-type P450
BM3. Compared with these are the spectra of oxyhemoglobin
(panel B) and the oxy-ferrous complexes of the F393A, WT,
and F393W mutants of P450 BM3 (panel C). The bond
vibrational energy of bound dioxygen in oxyhemoglobin
determined using FTIR spectroscopy (υO-O ) 1107 cm-1
)
indicates that the oxy-ferrous form of oxyhemoglobin has much
of its electron density delocalized across the dioxygen bond.35,36
Panel C compares the spectra of the oxy-ferrous complexes of
the P450 BM3 enzymes. The spectrum of the F393A mutant
closely resembles the carbonmonoxy-ferrous form of the
enzyme, with fused R and â bands. That of the F393W mutant
resembles the ferric low-spin spectrum and that of oxyhemo-
globin, whereas the spectrum of the wild-type oxy-ferrous
complex is intermediate. This trend is likely to be caused by
changes in the energies of the Fe 3d orbitals in the oxy-ferrous
complex and may reflect a shift in the balance of oxy-ferrous/
superoxy-ferric character or a shift of electron density toward
the oxygen, as a result of changes in the π-bonding properties
of the heme axial ligand. It is impossible to determine the
electronic configuration of the complex from the visible
absorption spectra alone. However, the changes observed do
(37) Phillips, S. E.; Schoenborn, B. P. Nature 1981, 292, 81-82.
(38) Miller, M. A.; Shaw, A.; Kraut, J. Nat. Struct. Biol. 1994, 1, 524-531.
(39) Yeom, H,; Sligar, S. G.; Li, H.; Poulos, T. L.; Fulco, A. J. Biochemistry
1995, 34, 14733-14740.
(35) Momenteau, M.; Reed, C. A. Chem. ReV. 1994, 94, 659-698.
(36) Wittenberg, J. B.; Wittenberg, B. A.; Peisach, J.; Blumberg, W. E. Proc.
Natl. Acad. Sci. U.S.A. 1970, 67, 1846-1853.
9
J. AM. CHEM. SOC. VOL. 125, NO. 49, 2003 15019

A R T I C L E S
Ost et al.
may have been slowed. There are two mechanisms by which a
second electron can be accepted by the heme system: (1) An
electron can be transferred directly into the dioxygen π* orbital.
This orbital is, however, further from the electron donor (FMN)
and is high in energy. (2) If the dioxygen π* orbital is already
populated to some extent, the electron can transfer into the Fe
dyz orbital, which will now be half-filled. This would be more
favorable. Thus, the electronic nature of the oxyferrous complex
holds the key to the subsequent steps of catalysis and warrants
further investigation.
Recent studies on P450cam39,40 and nitric oxide synthase41
(NOS) have shown that removal of hydrogen bonds to the
thiolate heme ligand decreases the reduction potential of the
iron. Introducing such hydrogen bonds clearly has the opposite
effect, as demonstrated in this study23 and that of Ueyama et
al.42 It is likely that the presence/absence of a hydrogen bond
affects the iron reduction potential by modulating the π- and
σ-donor characteristics of the thiolate ligand. It is also interesting
to compare the oxy-ferrous complexes of P450cam and P450
BM3. The visible spectrum of the former has fused R and â
bands, and the complex is relatively stable (kautox ) 0.001 s-1).
These properties are remarkably like those of the F393A and
F393H mutants of P450 BM3 and may indicate that their 3d
orbitals are of similar energy. It is also worth noting that the
camphor-bound reduction potential of P450cam is also similar
to those of the mutants (-175 mV). In P450cam, electron
transfer is rate limiting, primarily due to the requirement for
formation of the productive P450/putidaredoxin complex.43,44
For P450cam, a stable oxy-ferrous complex may be required
to compensate for any delay in the delivery of the second
electron from putidaredoxin. For P450 BM3, the integral
reductase domain ensures the rapid delivery of two successive
electron equivalents49 and the stability of the oxy-ferrous
complex is poised to maximize the rates of both.
In summary, we have shown that decreases in the heme
reduction potential lead to heme reduction being rate limiting,
whereas increases in the heme reduction potential lead to
increases in the stability of the oxy-ferrous complex. The latter
may result in the second FMN to heme electron transfer being
rate limiting. The 3d orbital energies of the heme-iron in wild-
type P450 BM3 are optimized such that both the first electron
transfer and the subsequent oxygen activation steps occur at
similar rates, maximizing the overall rate of catalytic turnover.
We have also shown that the first electron transfer is gated
almost entirely by the substrate-induced spin-state shift and that
the contribution of the simultaneous reduction potential shift is
negligible by comparison.
Acknowledgment. The research was performed with support
from the Edinburgh Protein Interaction Centre (EPIC), the
BBRSC (studentship to TWBO & JC and Postdoctoral funding
to CSM), and The Royal Society (Fellowship to S.D.). We thank
ESRF Grenoble and DESY Hamburg for the use of synchrotron
facilities. Synchrotron Infrastructure Action of the Improving
Human Potential Program to the EMBL Hamburg Outstation,
Contract Number HPRI-CT-1999-00017.
JA035731O
(40) Yoshioka, S.; Takahashi, S.; Ishimori, K.; Morishima, I. J. Inorg. Biochem.
2000, 81, 141-151.
(45) Nahri, L. O.; Fulco, A. J. J. Biol. Chem. 1987, 262, 6683-6690.
(46) Esnouf, R. M. J. Mol. Graphics 1997, 15, 132-136.
(47) Merritt, E. A.; Bacon, D. J. Methods Enzymol. 1997, 277, 505-524.
(48) Daff, S. N.; Chapman, S. K.; Turner, K. L.; Holt, R. A.; Govindaraj, S.;
Poulos, T. L.; Munro, A. W. Biochemistry 1997, 36, 13816-13823.
(49) Sevrioukova, I.; Shaffer, C.; Ballou, D. P.; Peterson, J. A. Biochemistry
1996, 35, 7058-7068.
(41) Adak, S.; Crooks, C.; Wang, Q.; Crane, B. R.; Tainer, J. A.; Getzoff, E.
D.; Stuehr, D. J. J. Biol. Chem. 1999, 274, 26907-26911.
(42) Ueyama, N.; Nishikawa, N.; Yamada, Y.; Okamura, T.; Oka, S.; Sakurai,
H.; Nakamura, A. Inorg. Chem. 1998, 37, 2415-2421.
(43) Hui Bon Hoa, G.; Begard, E.; Debye, P.; Gunsalus, I. C. Biochemistry
1978, 17, 2835-2839.
(44) Hintz, M. J.; Mock, D. M.; Peterson, L. L.; Tuttle, K.; Peterson, J. A. J.
Biol. Chem. 1982, 257, 14324-14332.
(50) Sevrioukova, I. F.; Li, H. Y.; Zhang, H.; Peterson, J. A.; Poulos, T. L.
Proc. Natl. Acad. Sci. 1999, 96, 1863-1868.
9
15020 J. AM. CHEM. SOC. VOL. 125, NO. 49, 2003