Engineering cytochrome P450cam into an alkane hydroxylase

Article abstract of DOI:10.1039/b300869j

The haem monooxygenase cytochrome P450cam from Pseudomonas putida has been engineered into an alkane hydroxylase. Active site amino acid residues were substituted with residues that have bulkier and more hydrophobic side-chains. The residues F87, Y96, V247 and V396, which are further away from the haem, were targeted first for substitution in order to constrain the small alkanes n-butane and propane to bind closer to the haem. We found that just two mutations could increase the alkane oxidation activity of P450cam by two orders of magnitude. The F87W/ Y96F/V247L triple mutant was then used as a basis for introducing further substitutions, at the residues T101, L244, V395 and D297 which are closer to the haem, to improve the enzyme/alkane fit and hence the alkane hydroxylase activity. The F87W/Y96F/T101L/V247L mutant oxidised n-butane with a catalytic turnover rate of 755 nmol(nmol P450cam)^-1(min) ^-1, which is comparable to the camphor oxidation activity of the wild-type (1000 min^-1). The F87W/Y96F/T101L/L244M/V247L mutant had lower n-butane oxidation activity but the highest propane oxidation rate (176 min^-1) of the P450cam enzymes studied. All P450cam enzymes gave 2-butanol and 2-propanol as the only products. Determination of the extent of uncoupling showed that hydrogen peroxide generation was the dominant uncoupling mechanism. The data indicate that further mutations at residues higher up in the active site are required to localise the substrates close to the haem and to reduce substrate mobility. These next-generation mutants will have higher activity, and may be able to catalyse the oxidation of ethane and methane. The Royal Society of Chemistry 2003.

Full text of DOI:10.1039/b300869j

View Article Online / Journal Homepage / Table of Contents for this issue
Engineering cytochrome P450cam into an alkane hydroxylase
Stephen G. Bell, Erica Orton, Helen Boyd, Julie-Anne Stevenson, Austin Riddle,
Sophie Campbell and Luet-Lok Wong*
Department of Chemistry, Inorganic Chemistry Laboratory, University of Oxford,
South Parks Road, Oxford, UK OX1 3QR
Received 22nd January 2003, Accepted 26th March 2003
First published as an Advance Article on the web 9th April 2003
The haem monooxygenase cytochrome P450cam from Pseudomonas putida has been engineered into an alkane
hydroxylase. Active site amino acid residues were substituted with residues that have bulkier and more hydrophobic
side-chains. The residues F87, Y96, V247 and V396, which are further away from the haem, were targeted first for
substitution in order to constrain the small alkanes n-butane and propane to bind closer to the haem. We found that
just two mutations could increase the alkane oxidation activity of P450cam by two orders of magnitude. The F87W/
Y96F/V247L triple mutant was then used as a basis for introducing further substitutions, at the residues T101, L244,
V395 and D297 which are closer to the haem, to improve the enzyme/alkane fit and hence the alkane hydroxylase
activity. The F87W/Y96F/T101L/V247L mutant oxidised n-butane with a catalytic turnover rate of 755 nmol(nmol
P450cam)Ϫ1(min)Ϫ1, which is comparable to the camphor oxidation activity of the wild-type (1000 minϪ1). The
F87W/Y96F/T101L/L244M/V247L mutant had lower n-butane oxidation activity but the highest propane oxidation
rate (176 minϪ1) of the P450cam enzymes studied. All P450cam enzymes gave 2-butanol and 2-propanol as the only
products. Determination of the extent of uncoupling showed that hydrogen peroxide generation was the dominant
uncoupling mechanism. The data indicate that further mutations at residues higher up in the active site are required
to localise the substrates close to the haem and to reduce substrate mobility. These next-generation mutants will have
higher activity, and may be able to catalyse the oxidation of ethane and methane.
ing at elevated temperatures and pressures, while homogeneous
catalysts do not yet have the required activity, selectivity or
Introduction
The cytochrome P450 (CYP) enzymes are haem dependent
monooxygenases that primarily catalyse the oxidation of C–H
bonds to the alcohol functionality. P450 enzymes are known to
oxidise a very extensive range of endogenous and exogenous
organic compounds, ranging from medium chain alkanes such
as n-heptane and n-octane, to steroidal and polyaromatic com-
pounds, and very large molecules such as the triterpenes and
cyclosporin.1,2 These reactions are important in biological sys-
tems because they are crucial steps in biosynthesis, cellular bio-
chemistry, metabolism, pharmacology and medicine. There has
been much effort directed at understanding the relationship
between P450 structure, function and reactivity.2 P450 enzymes
follow a common mechanism,3 in which a reactive intermediate
generated during the catalytic cycle attacks a C–H bond in a
substrate bound close to the haem iron. The nature of this C–H
bond activating intermediate remains controversial, but the
consensus is for an oxyferryl porphyrin radical cation species.4
Since the formation of this ferryl intermediate does not depend
on the stabilisation of a transition state involving the substrate,
the substrate specificity of P450 enzymes depends on what sub-
strate can enter and bind within the active site. The orientation
in which the substrate is bound, then determines which group is
closest to the reactive intermediate for attack, and hence the
products formed.
catalyst lifetime.
Numerous microorganisms are known to oxidise alkanes as
part of their metabolic or catabolic pathways, and indeed some
can grow on the chemically inert alkanes as the sole carbon and
energy source. Many alkane metabolising yeast strains use P450
enzymes to activate alkanes such as n-octane; the most well
characterised is the CYP52 enzyme from Candida maltosa.12
Alkane hydroxylase is a non-haem iron enzyme from Pseudo-
monas oleovorans that oxidises linear alkanes to the terminal
alcohols.13–15 The most spectacular are the methane-oxidising
bacteria such as Methylococcus capsulatus (Bath)16–19 and
Methylosinus trichosporium Ob3b,20–22 which grow on methane
by using methane monooxygenases (MMO) to activate the
highly inert methane C–H bond. MMO can also oxidise longer
chain alkanes, although further oxidation of the initial alcohol
products is a potential problem. The P450 enzymes have also
been investigated for alkane oxidation. We have described the
engineering of P450cam for the oxidation of the short chain
alkanes butane, pentane and hexane.23–25 Forced evolution of
P450BM-3 has generated a mutant with 13 mutations, none of
which was within the active site, that oxidised n-octane with the
very high NADPH turnover rate of 8000 minϪ1, and n-butane
(1830 minϪ1) and propane (850 minϪ1) are also oxidised at fast
rates.26
The unique C–H bond activation activity of cytochrome
P450 enzymes have no equivalents in classical synthetic meth-
odologies, and so these enzymes have many potential appli-
cations in biotransformation.5–7 Research efforts in this area
have focussed on the soluble bacterial enzymes, in particular the
structurally characterised P450cam from Pseudomonas putida
We,27–31 and others,32–39 have shown that P450cam can be
engineered to oxidise many different classes of compounds that
are structurally unrelated to the natural substrate, camphor. We
report here the results our recent efforts to engineer P450cam
into an alkane hydroxylase, with particular emphasis on the
oxidation of the smaller, gaseous alkanes. We have previously
shown that increasing the hydrophobicity of the active site with
the Y96A and Y96F mutations promoted the oxidation of
alkanes,23 and that decreasing the active site volume favoured
the oxidation of less sterically demanding alkanes.24 For
instance the Y96F/V247L mutant oxidised hexane at a faster
rate than 3-methylpentane while the Y96A/V247L mutant
showed the opposite activity trend. Our approach was therefore
to decrease the active site volume by targeting residues in the
P450cam substrate-binding pocket for substitutions with amino
3,8
and P450BM-3 from Bacillus megaterium.9–11 The selective
catalytic oxidation of alkanes to alcohols without further oxid-
ation to the aldehydes or ketones, is a major target of catalysis
research. An efficient process for the oxidation of methane to
methanol, ethane to ethanol, and of the higher homologues,
especially to the terminal alcohols, will have very significant
impact on industrial processes by providing new feedstocks for
the synthesis of complex molecules. All current industrial
alkane oxidation processes utilise metal oxide catalysts operat-
T h i s j o u r n a l i s © T h e R o y a l S o c i e t y o f C h e m i s t r y 2 0 0 3
D a l t o n T r a n s . , 2 0 0 3 , 2 1 3 3 – 2 1 4 0
2133

View Article Online
acids with bulkier side-chains, in order to promote the binding
and oxidation of the small alkane molecules.
6.22 mMϪ1 cmϪ1. The maximum concentration of ethanol in an
incubation reaction was <1% v/v, and control experiments
showed that ethanol on its own at up to 10% v/v did not induce
any increase in the rate of NADH consumption.
Materials and methods
Incubations with the gaseous alkanes n-butane, propane,
ethane and methane were carried out in 4-mL capacity cuvettes
equipped with a screw cap and Teflon septum seal. The final
volume of incubations was 2.5 mL. The concentration of the
reaction components was the same as those for reactions with
n-pentane. In a typical reaction the volume of buffer required to
take the final volume to 2.5 mL was ca. 1.6 mL. Half of this
required volume of buffer was saturated with the gaseous
alkane substrate, while the other half was a buffer solution sat-
urated with oxygen. The cuvette was immediately sealed with
the Teflon-backed screw cap and the reaction was initiated by
the addition of NADH via a hypodermic syringe needle pushed
through the Teflon septum.
For many mutant/n-butane, and some mutant/propane com-
binations it was necessary to reduce the concentration of
NADH to initiate the reaction because the coupling efficiencies
were so high that the alkane substrate was fully consumed. In
these cases only 200 µM of NADH was added so as to maintain
the substrate in excess and to keep the P450cam enzyme satur-
ated with substrate. Interestingly even when the n-butane or
propane were exhausted in incubations, there was no further
oxidation of the 2-butanol or 2-propanol to the ketones.
General
Enzymes for the molecular biology work were from New
England Biolabs, UK. Buffer components were from Anachem,
UK. The alkane substrates, the alcohol products, and general
reagents were from Sigma/Aldrich or Lancaster, UK. NADH
was from Roche Diagnostics, UK. UV/Vis spectra were
recorded on a Varian CARY 1E double-beam spectrophoto-
meter equipped with a Peltier cell temperature controller
(
0.1 ЊC). Gas chromatography analyses were performed on a
Fisons Instruments 8000 series gas chromatograph equipped
with a flame ionisation detector (FID). The n-pentane oxid-
ation products were analysed on a DB-1 fused silica column
(0.25 mm i.d. × 30 m) or a β-DEX chiral phase column
(0.25 mm i.d. × 30 m), while a SPB-1 column (0.5 mm i.d. × 60
m) together with a special injector liner was used for the lighter
alcohols: the butanols, the propanols, ethanol and methanol.
Enzymes and molecular biology
General DNA and microbiological manipulations were carried
out by standard methods.40 The P450cam enzyme and the
physiological electron transfer co-factor proteins putidaredoxin
and putidaredoxin reductase, were expressed in Escherichia coli
and purified as previously described.41–43 The proteins were
readily purified by anion exchange chromatography and were
stored at Ϫ20 ЊC in buffers containing 50% v/v glycerol as
cryoprotectant. The glycerol was removed from the proteins
immediately prior to activity assays by gel filtration on a 5-mL
bed volume PD-10 column (Amersham Biosciences, UK), elut-
ing with 50 mM Tris, pH 7.4. Site-directed mutagenesis was
carried out by a polymerase chain reaction (PCR)-based
method using the QuikChange kit from Stratagene using oligo-
nucleotides designed following the manufacturer’s guidelines.
For example the oligonucleotide 5Ј-TACGACTTCATTCC-
CCTCTCGATGGATCCGCC-3Ј, and its reverse complement,
Assays for the alkane oxidation products
After all the NADH had been consumed in an incubation,
3-methyl-2-pentanol was added as an ethanol stock to the
n-pentane oxidation reactions to act as an internal standard
(final concentration 300 µM) for the solid phase extraction and
GC analysis steps. Organics in a 1-mL aliquot of the incubation
mixtures were adsorbed onto a Varian Bond-Elut C18 column
(1 mL matrix volume) which had been equilibrated by washing
with 1 mL methanol followed by 1 mL 50 mM Tris, pH 7.4. The
column was then washed with 1 mL 50 mM Tris, pH 7.4, and
dried under vacuum for 5 min. Bound organics were eluted with
300 µL chloroform and then analysed by GC on the DB-1
column. The injector temperature was 120 ЊC and the FID
was held at 250 ЊC. The temperature of the DB-1 column was
maintained at 40 ЊC. Under these conditions 2-pentanol and
3-pentanol were eluted as one peak at 3.89 min. Further analy-
sis with a β-DEX chiral phase column (temperature profile:
40 ЊC for 1 min then 3 ЊC per min to 100 ЊC and then main-
tained at 100 ЊC for 8 min) showed that all the P450cam
were used to introduce the Thr101
Leu (T101L) mutation
(the codon for residue 101 is underlined). Oligonucleotides
for site-directed mutagenesis were from MWG-Biotech. New
mutants were identified and then fully sequenced by automated
DNA sequencing on an ABI 377XL Prism DNA sequencer
by the DNA sequencing facility at the Department of Bio-
chemistry, University of Oxford. All P450cam enzymes
enzymes gave virtually statistical mixtures (2-pentanol
:
described in this work also contained the Cys334
Ala
3-pentanol ≈ 2 : 1) of these compounds, with small amounts
(<2% total) of 2-pentanone and 3-pentanone also detected.
The concentration of the pentanol products in incubation
mixtures was determined by calibrating the concentration
response of the FID to 2-pentanol. Mixtures containing differ-
ent concentrations of 2-pentanol and all the components of a
normal incubation, except NADH and substrate, were
extracted and analysed as for normal incubations. Control
analyses showed that, as expected, the FID response to the iso-
meric 2-pentanol and 3-pentanol were identical, within experi-
mental error. The plot of the ratio of the 2-pentanol peak area
to the internal standard, against the 2-pentanol concentration,
gave a straight line passing through the origin. The absolute
quantity of pentanol products formed by enzymatic turnover
in an incubation mixture was then readily determined. The
coupling efficiency was the percentage of NADH consumed
that led to product formation. Control experiments showed
that the solid phase extraction procedure gave highly repro-
ducible results, with <5% variation across a large number of
analyses.
(C334A) base mutation which prevents protein dimerization via
disulfide bond formation.44 For convenience, the C334A base
mutant is referred to as ‘wild-type’, and the Y96F/C334A
double mutant as Y96F, etc. The Y96F, F87W/Y96F, Y96F/
V247L and F87W/Y96F/V247L mutants had been reported
previously.30
NADH turnover rate determinations
All incubations were carried out at 30 ЊC. Incubation mixtures
contained 50 mM Tris, pH 7.4, 200 mM KCl, 1 µM P450cam,
10 µM putidaredoxin, 1 µM putidaredoxin reductase, and 30 µg
mLϪ1 bovine liver catalase,23 except where the extent of un-
coupling to form hydrogen peroxide was to be determined. The
mixtures were oxygenated and then equilibrated at 30 ЊC for
2 min. For n-pentane oxidation reactions (final incubation vol-
ume 1.7 mL), the substrate was added as a 10 mM stock in
ethanol to a nominal final concentration of 500 µM. NADH
was added as a 20 mg mLϪ1 stock solution in 50 mM Tris,
pH 7.4, to ca. 400 µM (final A340 = 2.5) and the absorbance at
340 nm was monitored. The rate of NADH consumption was
For the lighter alcohols formed by the oxidation of gaseous
alkanes, the solid phase extraction method was less reliable and
calculated from the slope of the time-course plot using ε340
=
D a l t o n T r a n s . , 2 0 0 3 , 2 1 3 3 – 2 1 4 0
2134

View Article Online
Fig. 1 The active site structure of wild-type cytochrome P450cam with bound camphor. Both figures are views from the side, with depth cue, to
show the Tier-like arrangement of active site residues and the location of the residues. Side-chains of residues in the separate Tiers are coloured
differently – Tier 1: green, Tier 2: cyan, Tier 3: purple. The backbone nitrogen is coloured blue, and the backbone carbonyl oxygen is omitted for
clarity. The camphor carbonyl group forms a hydrogen bond with the phenolic side-chain of Tyr-96. There are complementary van der Waals
contacts between camphor and the side-chains of Phe-87, Thr-185, Leu-244, Val-247 and Val-295. The figures were generated with the program
WebLab Viewer from the entry 2cpp from the Protein Data Bank.
in fact not suitable for ethanol and methanol. Extraction with
organic solvents was not satisfactory because of the poor parti-
tion of methanol, ethanol and the propanols into the organic
phase. In addition even contamination of the solvents by <0.1%
v/v of the lighter alcohols became an issue. Furthermore the
DB-1 column could not resolve 2-propanol from most of the
common organic solvents. We therefore used a SPB-1 column
(0.5 mm i.d. × 60 m) for the analysis of these alcohols. The
SPB-1 column also had the added advantage of being tolerant
of water and thus allowed the direct injection of incubation
mixtures for GC analysis. However, this required a special
injector liner because the normal straight glass tube liners did
not give clean and consistent vaporisation of water which has
the tendency to shatter into microscopic droplets upon initial
contact with the hot surface. This “beading” of water normally
results in broad and poorly resolved peaks. With the special
liner we were able to obtain well-defined peaks consistently.
For the analysis of aqueous reaction samples, 90 µL of the
reaction mixture was added to a 10-µL aliquot of a 200 µM
solution of 1-pentanol in water; the 1-pentanol acting as the
internal standard. A 2-µL aliquot of such a mixture was
injected directly onto the column. The oven temperature was
held at 40 ЊC for 4 min and then increased at 5 ЊC minϪ1 to
100 ЊC, and held at 100 ЊC for 2 min. The retention times
were: methanol 2.8 min, ethanol 3.5 min, 2-propanol 4.8 min,
2-butanol 8.5 min, and 1-pentanol (internal standard) 16.0 min.
All the P450cam enzymes studied in this work oxidised
n-butane to 2-butanol only, and 2-propanol was the only
product from propane.
such as n-pentane and then to seek additional mutations
that will promote the oxidation of n-butane and propane.
Since the alkane substrates are smaller than camphor, active site
substitutions were made with amino acids such as leucine,
methionine, phenylalanine and tryptophan, which have larger
side-chains. The expectation was that, as additional mutations
were introduced, the volume of the P450cam active site will be
reduced, and the substrate specificity will be increasingly biased
towards the smaller, gaseous alkanes.
The structure of wild-type P450cam with bound camphor is
shown in Fig. 1.45 Sligar and coworkers have described this
structure as being made up of three Tiers.3,34 In Tier 1 are the
residues T101, L244 and V295, which are close to the haem.
The side-chains of F87, Y96, V247, I395 and V396 in Tier 2
almost form a ring above Tier 1, while the side-chain of T185
and, to a lesser extent, M184, cover the top of the active site.
Loida and Sligar reported that the V247A (Tier 2) and T101M
(Tier 1) mutations reduced the ethylbenzene oxidation activity
of P450cam.34 It was suggested that this arose because the
V247A mutation created space in the upper part of the active
site to allow the substrate to bind away from the haem, while
the T101M mutation restricted the space available close to the
haem and thus forced the substrate to bind further away. Amino
acid substitutions at the Tier 3 residue T185 had relatively
minor effects on the oxidation of a series of alkylbenzene com-
pounds.34 We reasoned that bulky substitutions at T185 may
not be necessary if it proved possible to reduce the space in the
upper parts of the P450cam active site by bulky substitutions in
Tier 2. This approach has been successfully applied to the pro-
tein engineering of P450cam for the reductive dehalogenation
of chlorinated ethanes,3 and the oxidation of alkanes,23–25 and
polychlorinated benzenes.29,30 After a good combination of
mutations in Tier 2 was found, we would introduce additional
bulky substitutions in Tier 1 to improve further the alkane
oxidation activity.
Uncoupling via hydrogen peroxide formation
The extent of uncoupling of NADH consumption to hydrogen
peroxide generation was determined for n-butane and propane
oxidation by a standard assay using phenol, 4-aminoantipyrine
and horseradish peroxidase. The concentration of the quinone-
imine product formed was determined from the absorbance at
n-Pentane oxidation activity
500 nm by using ε500 = 6850 MϪ1 cmϪ1
.
The n-pentane oxidation activity of the first series of mutants
designed to establish a good combination of mutations in the
upper part of the P450cam active site are given in Table 1.
Bulky amino acid substitutions were carried out at the Tier 2
sites F87, Y96, V247 and V396. In addition the F98W mutation
was also introduced. The side-chain of F98 does not form
part of the active site but it contacts the side-chain of Y96,
Results and discussion
Engineering P450cam for alkane oxidation
We adopted the approach of searching for a P450cam mutant
with high activity for the oxidation of a longer chain alkane
D a l t o n T r a n s . , 2 0 0 3 , 2 1 3 3 – 2 1 4 0
2135

View Article Online
Table 1 The pentane oxidation activity of wild-type and mutants of P450cam. The NADH turnover and the product formation rates are given in
nmol per nmol of P450cam per min. The data are means of at least three experiments, with all data for each parameter being within 15% of the mean.
n.d.: no product detected
P450cam enzyme
NADH oxidation rate
Product formation rate
Coupling (%)
Wild-Type
Y96F
Y96F/V247A
Y96F/V247L
6.4
405
152
510
32
0.07
106
31
227
0.4
1.1
29.5
22
44.5
1.3
Y96F/F98W
F87W/Y96F
471
367
510
64
26
76
192
26.5
237
n.d.
1.0
42.5
7.2
46.5
n.d.
3.8
F87W/Y96F/F98W
F87W/Y96F/V247L
F87W/Y96W/V247L
F87W/Y96F/F98W/V247L
F87W/Y96F/V247L/V396L
1.7
2.3
and substitutions here may influence the alkane oxidation
activity.
intermediate does not undergo rapid attack on the substrate but
is instead reduced by the electron transfer co-factor protein
back to the ferric resting state. The overall catalytic cycle would
then result in the four-electron reduction of dioxygen to two
molecules of water, i.e. oxidase activity (Fig. 2). Interconversion
between different substrate binding orientations could also lead
to the transient approach of water molecules into the active
site.49,50 The presence of these water molecules could lead to
protonation of the ferric hydroperoxy intermediate at the iron-
bound oxygen, resulting in loss of hydrogen peroxide. Extra
water molecules in the oxygen binding groove in the kink in the
distal I helix, around the T252 side-chain,45,51 could also inter-
fere with the delivery of protons to the hydroperoxy oxygen
atom which is not bound to the haem, again leading to peroxide
dissociation. The most stable location of substrate binding may
be very close to the haem iron, resulting in steric interaction
between the substrate and the bound dioxygen or hydroperoxy
ligand. It should be noted that if the bulky substitutions
reduced the active site volume by too much, an unnatural
substrate such as n-pentane might be forced to bind so close to
the haem iron that such steric interactions would give rise
to uncoupling. Finally, there may be steric hindrance between
the substrate and the hydroperoxy ligand during the inter-
conversion between different substrate binding modes. Both
these steric effects could weaken the iron–hydroperoxide
interaction and lead to the loss of hydrogen peroxide.49
The Y96F mutation greatly enhanced the coupling efficiency
of n-pentane oxidation (29.5%) compared to the wild-type
enzyme (1.1%). The major uncoupling pathway for both
enzymes was hydrogen peroxide formation, but this pathway
was much reduced in the mutant. Since n-pentane is smaller
than camphor and also very conformationally flexible, the
Y96F mutation on its own is not expected to prevent com-
pletely the entrance of water molecules to the active site or to
confer a unique n-pentane binding orientation. Nevertheless
the increased active site hydrophobicity in the mutant should
facilitate the expulsion of active site water molecules and thus
retard the peroxide uncoupling pathway.
All the P450cam enzymes studied gave virtually statistical
(2 : 1) mixtures of 2-pentanol and 3-pentanol. Some mutants
also gave small (<2% in total) amounts of the correspond-
ing pentanones, but 1-pentanol or pentanal were not observed.
The radical rebound mechanism generally accepted for P450
catalysis suggests that n-pentane is sufficiently mobile within
the active site of the P450cam enzymes such that the more
reactive secondary C–H bonds are preferentially attacked over
the primary C–H bonds at the terminal carbons.
The activity data (Table 1) clearly show that even a small
number of active site mutations can greatly increase the alkane
oxidation activity of P450cam. For comparison the camphor
oxidation activity of wild-type P450cam under identical condi-
tions is 1000 minϪ1. The largest increase in activity over the
wild-type was imparted by the Y96F mutation first described by
Atkins and Sligar.46 The increase was a result of both enhanced
NADH oxidation rate and coupling efficiency (the product
yield based on NADH consumed). Under the assay conditions
the rate-limiting step for the NADH oxidation activity is the
first electron transfer from reduced putidaredoxin to the ferric,
substrate-bound P450cam. According to the Marcus Theory of
electron transfer,47 this reaction is faster for a higher thermo-
dynamic driving force and lower reorganisation energy. Binding
of a substrate sufficiently close to the haem could displace
the haem iron sixth ligand, water, leading to an increase in the
haem reduction potential and a decrease in the reorganisation
energy of electron transfer because the haem in both the ferric
and ferrous forms would be five-coordinate.48 However, these
criteria do not restrict the motion of the bound substrate—as
long as it is sufficiently close to the haem to cause dissociation
of the water ligand, fast electron transfer should ensue. We have
suggested previously that the Y96F mutation promotes the
binding and oxidation of hydrophobic organic molecules by
increasing the active site hydrophobicity and facilitating the
displacement of water molecules from the active site.23 Thus
n-pentane is bound more strongly and the haem ligand
water displaced more readily in the Y96F mutant, leading to
faster electron transfer and higher NADH oxidation activity
compared to the wild-type enzyme.
Uncoupling of reducing equivalents from the electron
donor (NADH or NADPH) away from product formation is a
major feature of P450 catalysis, especially in the oxidation of
unnatural substrates. For these compounds, it is possible that
the NADH consumption activity is very high but the rate of
product formation is slow (see Table 1) because most the
reducing equivalents are channelled away to form hydrogen
peroxide and water.27 The extent of uncoupling is therefore a
stringent measure of the match between the substrate and the
enzyme active site than the NADH oxidation rate.34 A number
of uncoupling mechanisms can be envisaged. If the substrate is
mobile, a certain population of the enzyme could have a sub-
strate bound too far away from the haem, such that the ferryl
Addition of the bulky substitutions F87W or V247L, to the
Y96F mutation, doubled the n-pentane oxidation activity,
mainly by increasing the coupling efficiency, whilst the V247A
mutation reduced the activity by 60%. These results are consist-
ent with the proposal that bulky substitutions in Tier 2 should
increase the activity and coupling of the oxidation of a small
substrate by constraining it to bind close to the haem, while
mutations that generate space in the upper part of the pocket
should reduce the activity.34 When the Y96F, F87W and V247L
mutations were combined, the Y96F/F87W/V247L mutant did
not show significantly higher n-pentane oxidation activity than
the double mutants, suggesting that there may be competition
between the effects of the F87W and V247L mutations.
Some mutations had unexpected effects. The Y96W, F98W
and V396L mutations reduced the activity by as much as two
orders of magnitude. It is not clear why the V396L mutation
D a l t o n T r a n s . , 2 0 0 3 , 2 1 3 3 – 2 1 4 0
2136

View Article Online
Table 2 The n-butane and propane oxidation activity of wild-type and mutants of P450cam. The rates of NADH turnover and the formation of
product (2-butanol and 2-propanol) are given in nmol per nmol of P450cam per min. The data are means of at least experiments, with all data for
each parameter being within 15% of the mean
n-Butane
Propane
NADH
oxidation rate
Product
formation rate
Coupling
(%)
NADH
oxidation rate
Product
formation rate
Coupling
(%)
P450cam enzyme
Wild-Type
Y96F
F87W/Y96F
Y96F/V247L
F87W/Y96F/V247L
F87W/Y96F/T101L
Y96F/T101L/V247L
F87W/Y96F/T101L/V247L
F87W/Y96F/T101M/V247L
F87W/Y96F/V247L/V295I
F87W/Y96F/V247L/V396L
F87W/Y96F/T101L/V247L/D297M
F87W/Y96F/T101L/V247L/L244M
8
100
200
250
263
546
423
795
373
363
52
0.4
42
132
188
89
459
186
755
205
99
12.5
243
520
0.5
42
66
75
34
85
44
95
55
9
18
60
110
82
198
152
346
164
124
7.5
214
266
0.5
2.2
12
43
14
51
20
111
66
3.5
1.5
56
0.5
12
20
39
17
26
13
32
40
3.0
18
26
66
27.5
24
63
385
578
90
176
Fig. 2 (A) The catalytic mechanism of cytochrome P450cam showing the three potential uncoupling pathways of auto-oxidation of the ferrous-oxy
form (ii), hydrogen peroxide generation and oxidase reactivity. The auto-oxidation pathway is much slower than the other steps and can be ignored in
most cases. (B) Three mechanisms of the peroxide uncoupling pathway. In (i) the accessibility of the haem-bound oxygen to water molecules leads to
protonation and loss of hydrogen peroxide; this reaction should require the participation of a general acid on the side-chain of a residue. The
schematic in (ii) shows the steric hindrance between the hydroperoxy ligand and a substrate that is bound too close to the haem iron which can
weaken the FeIII–OOH interaction and cause dissociation of the hydroperoxide anion. In (iii) the possibility of multiple substrate-binding orien-
tations is shown with two R–H substrate molecules. The arrow represents a trajectory of interconversion between different binding orientations
which brings the substrate close to the haem-bound hydroperoxy ligand. Steric interactions could again weaken ligand binding, resulting in the loss
of the hydroperoxide ion.
reduced the n-pentane oxidation activity in the F87W/Y96F/
The oxidation of n-butane and propane
V247L/V396L mutant, although the presence of four such
bulky substitutions in Tier 2 might have unpredictable effects
such as causing the structure to change and the actives site to
open up compared to the wild-type. The detrimental effects of
the Y96W mutation on P450cam substrate oxidation activity
has been reported.39 The F98 side-chain is not part of the active
site but rather the interior of the protein, although it contacts
the side-chain at the 96 position. It is possible that the F98W
mutation caused structural changes that reduced the activity. In
addition, the Y96 side-chain is proposed to be part of the sub-
strate access channel,45,52 and so both the Y96W and F98W
mutations could also affect substrate entry.
Before adding mutations to the F87W/Y96F/V247L triple
mutant, we examined the n-butane and propane oxidation
activity of all the mutants tested for n-pentane oxidation. The
results for some mutants are given in Table 2. All P450cam
mutants we examined in the present work oxidised n-butane to
2-butanol and propane to 2-propanol, with no evidence for the
formation of the terminal alcohols. Hence, as with n-pentane,
the bound alkanes molecules appeared to be sufficiently mobile
within the active site and the ferryl intermediate oxidised the
more reactive secondary C–H bonds. Unlike for n-pentane,
however, there was no further oxidation of the secondary
alcohol products to the ketones, even after the n-butane and
propane substrates were exhausted in the reactions.
From the results of the initial screening of mutations, the
F87W/Y96F/V247L mutant was chosen as the starting mutant
for introducing substitutions at Tier 1 residues to promote the
oxidation of the smaller, gaseous alkanes.
The activity trends for the smaller alkanes paralleled those
for n-pentane. The activity was increased as the F87W, Y96F
D a l t o n T r a n s . , 2 0 0 3 , 2 1 3 3 – 2 1 4 0
2137

View Article Online
and V247L mutations were added. The V396L mutation was
detrimental to n-butane and propane oxidation by P450cam
(Table 2), as was the F98W mutation (data not shown). The
competition between the effects of the F87W and V247L
mutations was again apparent, which raised the possibility
that there might be changes in the structure of the enzyme
when all three mutations are combined. However, the crystal
structure of this triple mutant with 1,3,5-trichlorobenzene or
(ϩ)-α-pinene bound within the active site showed that the struc-
ture of both the peptide backbone and the side-chains were
virtually superimposable on that of the wild-type.53,54 Since
these two substrates are more rigid and sterically demanding
than the linear alkanes, it is most unlikely that the alkanes
would induce structural changes. A more plausible explanation
is that the F87 and V247 residues are located across the Tier 2
circle of residues (Fig. 1), such that increasing the side-chain
volume at one position could push the substrate towards the
other side-chain, improving the enzyme–substrate fit. However,
when the side-chain volume is increased at both positions, the
substrate is forced somewhere else in Tier 2, without significant
improvement in activity.
The Tier 1 residues T101, L244, V295 and D297 are close to
the haem.34 We began further mutagenesis at the 295 position.
It has been reported that most mutations at V295, except the
V295I mutation, destabilise the P450cam protein fold.55 When
the V295I mutation was added to the F87W/Y96F/V247L triple
mutant, the rate of NADH oxidation was increased slightly for
both n-butane and propane, but crucially the coupling effi-
ciency was lowered, especially for propane (Table 2). It is likely
that the larger isoleucine side-chain at the 295 position forced
the smaller propane molecule to bind away from the haem,
hence the coupling was greatly reduced. In contrast, for the
larger n-butane, the F87W/Y96F/V247L combination of
mutations was more effective in holding the substrate close to
the haem, thus the coupling was largely maintained.
We further investigated the possibility that the effects of the
F87W and V247L mutations may be in competition with each
other, by introducing the Tier 1 substitution T101L to the
F87W/Y96F, Y96F/V247L and F87W/Y96F/V247L mutants.
The data in Table 2 clearly show that the T101L mutation
significantly increased the n-butane and propane oxidation rate
when combined with the F87W/Y96F double mutant, but
interestingly it decreased the activity when added to the Y96F/
V247L mutant. Therefore, the location of the mutation added
to the starting Y96F mutation is important. The T101L
mutation most probably forced n-butane and propane to bind
away from the haem in the Y96F/T101L/V247L mutant, most
likely to the vicinity of the F87 side-chain which is situated
directly above the 101 side-chain (Fig. 1). However, this option
is not available in the F87W/Y96F/T101L mutant.
F87W/Y96F/T101L/V247L mutant was 111 minϪ1, with 32%
coupling (Table 2). The 220-fold rate enhancement for propane
oxidation over the wild-type was less than for n-butane, in all
probability reflecting the need for other mutations in order to
achieve the best fit for the smaller propane molecule.
The dramatic effect of the T101L mutation prompted us to
introduce the T101M mutation. We found that it promoted
both the NADH oxidation rate and coupling efficiency
for alkane oxidation, but it was less effective than the T101L
mutation (Table 2). However, it is interesting to note that the
coupling for propane oxidation was in fact higher than that
with the T101L mutation. Hence, analogous to the V295I
mutation, the T101M mutation could prove to be effective once
the NADH oxidation rate has been increased by suitable
combination of mutations at other active site residues.
Finally we investigated the effect of adding bulky substi-
tutions at the Tier 1 residues L244 and D297, to the F87W/
Y96F/T101L/V247L mutant. The D297 side-chain does not
contact camphor but forms a hydrogen bond to one haem pro-
pionate group (Fig. 1). The D297M mutation was introduced to
probe the effect of a hydrophobic and bulky substitution at a
residue close to the haem. Interestingly this mutation did not
affect the stability of the protein fold even though the carb-
oxylate side-chain of D297 forms a specific interaction with the
haem prosthetic group. However, the D297M mutation reduced
the n-butane and propane oxidation activity of the parent
quadruple mutant (Table 2). In hindsight this is perhaps not so
surprising because the D297 residue is quite far away from the
haem iron, and even a long side-chain such as that in Met may
not be sufficient to influence substrate binding and oxidation by
the haem.
The side-chain of the L244 residue is already fairly bulky, and
so mutations to phenylalanine and tryptophan were considered.
However, the aromatic side-chains of these residues might not
protrude into the active site sufficiently to promote the binding
of propane. Therefore we used the L244M mutation to intro-
duce a longer, linear side-chain. This mutation slightly reduced
the n-butane oxidation of the F87W/Y96F/T101L/V247L
mutant, but the coupling was maintained at a very high level
(Table 2). Most significantly, this mutation doubled the coup-
ling for propane oxidation, from 32 to 66%, so that the propane
oxidation activity was increased to 176 minϪ1, the highest of all
the P450cam mutants studied in this work, even though the
NADH oxidation activity was lowered than the parent quad-
ruple mutant. Again we see the beneficial or potentially bene-
ficial effects of bulky substitutions in Tier 1. This result also
reinforces the need, shown by the T101M and V295I mutations,
to optimise the Tier 2 mutations further so that the NADH
oxidation activity is maintained while the coupling is improved
by Tier 1 mutations.
When all four mutations were combined to form the F87W/
Y96F/T101L/V247L mutant, these factors and the apparently
detrimental effects of the F87W/V247L combination of
mutations disappeared, and the activity was the highest. These
data highlight the often subtle differences between the effects of
individual mutations, and especially when they are combined.
Strikingly, the rate of n-butane oxidation by the F87W/Y96F/
T101L/V247L mutant was 755 nmol per nmol of P450cam
mutant per min, and with 95% coupling efficiency. This activity
is 1900 times faster than that of the wild-type, and is compar-
able to the camphor oxidation activity of wild-type P450cam
(1000 minϪ1, >95% coupling). Hence the combined effects of
the four mutations in Tiers 1 and 2 appeared to have generated
an active site with near perfect fit for n-butane, thus shutting
down the different uncoupling pathways. We note that the
n-pentane oxidation activity of this quadruple mutant was an
order of magnitude lower than that of the parent F87W/Y96F/
V247L mutant (data not shown), suggesting that our strategy
of reducing the active site volume to promote the oxidation of
smaller alkanes may be valid. The propane oxidation rate of the
The extent of uncoupling by peroxide formation
In order to provide further insight into the oxidation of
n-butane and propane by P450cam mutants we determined the
extent of uncoupling to form hydrogen peroxide for a number
of mutants by a standard calorimetric assay. The data are
shown in histogram form in Fig. 3. The most evident trend is
that the peroxide uncoupling pathway dominates propane oxid-
ation while n-butane oxidation shows much more uncoupling
via the oxidase pathway. The Y96F/T101L/V247L mutant
showed some oxidase uncoupling, but this was completely elim-
inated by introducing the F87W mutation to give the F87W/
Y96F/T101L/V247L mutant. This observation strongly sug-
gests that the role of the F87W mutation is to close down the
top of the active site so that n-butane and propane are bind
close to the ferryl intermediate, thus reducing uncoupling
via the oxidase pathway.
The different effect of the T101L and T101M mutations
is interesting. The T101M mutation reduced the coupling of
D a l t o n T r a n s . , 2 0 0 3 , 2 1 3 3 – 2 1 4 0
2138

View Article Online
oxidation rates as high as ca. 50 minϪ1 under atmospheric
pressure. The reactions are clearly slow and almost totally
uncoupled, indicating that there is still some way to go before
these smallest of alkanes can be bound sufficiently close to the
haem to lead to fast NADH oxidation with good coupling to
product formation. Reactions were also carried out at 10 atm
partial pressures of both the alkane and oxygen. We were not
able to monitor the NADH oxidation rates in these reactions,
but no products were detected.
Summary and conclusions
We have shown conclusively that cytochrome P450cam can be
engineered to an alkane hydroxylase, with activities that can
approach that for the oxidation of the natural substrate, cam-
phor. The enzyme was successfully engineered for the oxidation
of n-butane and propane by the approach of improving the
enzyme/substrate fit via reduction of the active site volume. We
first introduced substitutions with amino acids with bulky side-
chains at residues located higher up in the active site, and then
used similar substitutions at residues close to the haem. The
results also suggest a number of other additional mutations
that could further increase the activity. However, some of the
mutations at certain sites had unpredictable consequences.
Moreover, even the most active mutants for propane oxidation
did not oxidise ethane or methane, and further protein engin-
eering is required for these, the smallest alkanes. Since all the
alkanes are oxidised at the secondary carbon positions, these
substrates remain mobile, at least conformationally, within the
active site. Therefore it appears that it is relatively straight-
forward to engineer a P450 enzyme to oxidise an unnatural
substrate with high activity and coupling, but manipulating the
binding orientation and hence product selectivity is a more
difficult problem, especially for flexible molecules.
Acknowledgements
Fig. 3 The channelling of reducing equivalents from NADH to the
formation of product and to the uncoupling pathways that generate
hydrogen peroxide and water (oxidase pathway). (A) n-butane
oxidation, and (B) propane oxidation. The P450cam mutants are 1:
Y96F/T101L/V247L, 2: F87W/Y96F/T101L.V247L, 3: F87W/Y96F/
T101M/V247L, 4: F87W/Y96F/T101L/L244M/V247L, and 5: F87W/
Y96F/T101L/V247L/D297M.
This work was supported by the Biotechnology and Biological
Sciences Research Council and the Engineering and Physical
Sciences Research Council (EPSRC), UK, under grant B10666.
J.-A. S. was supported by a studentship from the EPSRC. We
also thank the referees for helpful comments and suggestions.
n-butane oxidation by promoting the peroxide and oxidase
pathway equally, while it decreased the peroxide uncoupling for
propane oxidation without increasing the oxidase pathway.
Hence this mutation appears to push n-butane away from
the haem, most likely due to steric hindrance, to reduce the
efficiency of substrate attack by the ferryl intermediate. With
propane, however, it appears to bring the substrate closer to
the haem, presumably by increasing the van der Waals inter-
actions between the side-chain and substrate. This reduces
access of water to the haem iron, etc., and uncoupling via the
peroxide pathway is reduced. The D297M mutation signifi-
cantly increased the uncoupling via the oxidase pathway. The
reasons for this are not clear. The L244M mutation reduced the
peroxide uncoupling for propane oxidation from 73 to 25%
while increasing the oxidase pathway to 9%. The reduction of
peroxide uncoupling is consistent with a reduction in active
site volume and hence water access to the haem iron. The slight
increase in oxidase uncoupling suggests a longer average dis-
tance of the substrate from the ferryl intermediate and supports
the need for further engineering near the top of the active site to
constrain propane to bind closer to the haem.
References
1 F. P. Guengerich, J. Biol. Chem., 1991, 266, 10019.
2 P. R. Ortiz de Montellano, in Cytochrome P450: Structure,
Mechanism and Biochemistry, Plenum, New York, 1995.
3 E. J. Mueller, P. J. Loida and S. G. Sligar, in Cytochrome P450:
Structure, Mechanism and Biochemistry, ed. P. R. Ortiz de
Montellano, Plenum, New York, 1995.
4 J. T. Groves and Y.-Z. Han, in Cytochrome P-450: Structure,
Mechanism and Biochemistry, ed. P. R. Ortiz de Montellano,
Plenum, New York, 1995.
5 D. G. Kellner, S. A. Maves and S. G. Sligar, Curr. Opin. Biotechnol.,
1997, 8, 274.
6 L. L. Wong, Curr. Opin. Chem. Biol., 1998, 2, 263.
7 C. S. Miles, T. W. Ost, M. A. Noble, A. W. Munro and
S. K. Chapman, Biochim. Biophys. Acta, 2000, 1543, 383.
8 I. C. Gunsalus and G. C. Wagner, Methods Enzymol., 1978, 52, 166.
9 L. O. Narhi and A. J. Fulco, J. Biol. Chem., 1986, 261, 7160.
10 Y. Miura and A. J. Fulco, Biochim. Biophys. Acta, 1975, 388, 305.
11 S. S. Boddupalli, B. C. Pramanik, C. A. Slaughter, R. W. Estabrook
and J. A. Peterson, Arch. Biochem. Biophys., 1992, 292, 20.
12 T. Zimmer, T. Iida, W. H. Schunck, Y. Yoshida, A. Ohta and
M. Takagi, Biochem. Biophys. Res. Commun., 1998, 251, 244.
13 A. Bosetti, J. B. Van Beilen, H. Preusting, R. G. Lageveen and
B. Witholt, Enzyme Microb. Technol., 1992, 14, 702.
14 J. B. van Beilen, J. Kingma and B. Witholt, Enzyme Microb.
Technol., 1994, 16, 904.
Ethane and methane
15 I. E. Staijen, J. B. Van Beilen and B. Witholt, Eur. J. Biochem., 2000,
267, 1957.
16 J. Colby, D. I. Stirling and H. Dalton, Biochem. J., 1977, 165, 395.
17 H. Dalton, Adv. Appl. Microbiol., 1980, 26, 71.
None of the P450cam mutants showed any significant activity
for the oxidation of ethane and methane even though the most
active mutants for n-butane and propane oxidation had NADH
D a l t o n T r a n s . , 2 0 0 3 , 2 1 3 3 – 2 1 4 0
2139

View Article Online
18 J. Green and H. Dalton, J. Biol. Chem., 1989, 264, 17698.
19 S. J. Pilkington and H. Dalton, Methods Enzymol., 1990, 188, 181.
20 G. M. Tonge, D. E. Harrison and I. J. Higgins, Biochem. J., 1977,
161, 333.
21 J. D. Lipscomb, Annu. Rev. Microbiol., 1994, 48, 371.
22 B. G. Fox, W. A. Froland, J. E. Dege and J. D. Lipscomb, J. Biol.
Chem., 1989, 264, 10023.
23 J.-A. Stevenson, A. C. G. Westlake, C. Whittock and L.-L. Wong,
J. Am. Chem. Soc., 1996, 118, 12846.
24 J.-A. Stevenson, J. K. Bearpark and L.-L. Wong, New J. Chem.,
1998, 551.
25 S. G. Bell, J.-A. Stevenson, H. D. Boyd, S. Campbell, A. D. Riddle,
E. L. Orton and L.-L. Wong, Chem. Commun., 2002, 490.
26 A. Glieder, E. T. Farinas and F. H. Arnold, Nat. Biotechnol., 2002,
20, 1135.
27 C. F. Harford-Cross, A. B. Carmichael, F. K. Allan, P. A. England,
D. A. Rouch and L.-L. Wong, Prot. Eng., 2000, 13, 121.
28 D. P. Nickerson, C. F. Harford-Cross, S. R. Fulcher and L. L. Wong,
FEBS Lett., 1997, 405, 153.
37 M. P. Mayhew, A. E. Roitberg, Y. Tewari, M. J. Holden,
D. J. Vanderah and V. L. Vilker, New J. Chem., 2002, 26, 35.
38 K. J. French, M. D. Strickler, D. A. Rock, G. A. Bennett,
J. L. Wahlstrom, B. M. Goldstein and J. P. Jones, Biochemistry, 2001,
40, 9532.
39 K. J. French, D. A. Rock, J. I. Manchester, B. M. Goldstein and
J. P. Jones, Arch. Biochem. Biophys., 2002, 398, 188.
40 J. Sambrook, E. F. Fritsch and T. Maniatis, Molecular Cloning:
A Laboratory Manual, Cold Spring Harbor Laboratory Press,
New York, 1989.
41 A. C. Westlake, C. F. Harford-Cross, J. Donovan and L. L. Wong,
Eur. J. Biochem., 1999, 265, 929.
42 J. A. Peterson, M. C. Lorence and B. Amarneh, J. Biol. Chem., 1990,
265, 6066.
43 T. Yasukochi, O. Okada, T. Hara, Y. Sagara, K. Sekimizu and
T. Horiuchi, Biochim. Biophys. Acta, 1994, 1204, 84.
44 D. P. Nickerson and L.-L. Wong, Prot. Eng., 1997, 10, 1357.
45 T. L. Poulos, B. C. Finzel and A. J. Howard, J. Mol. Biol., 1987, 195,
687.
29 J. P. Jones, E. J. O’Hare and L. L. Wong, Chem. Commun., 2000,
247.
46 W. M. Atkins and S. G. Sligar, J. Biol. Chem., 1988, 263, 18842.
47 R. A. Marcus and N. Sutin, Biochim. Biophys. Acta, 1985, 811, 265.
48 M. J. Honeychurch, H. A. O. Hill and L. L. Wong, FEBS Lett.,
1999, 451, 351.
30 J. P. Jones, E. J. O’Hare and L. L. Wong, Eur. J. Biochem., 2001, 268,
1460.
31 S. G. Bell, R. J. Sowden and L.-L. Wong, Chem. Commun., 2001,
635.
49 R. Raag and T. L. Poulos, Biochemistry, 1991, 30, 2674.
50 R. Raag and T. L. Poulos, Biochemistry, 1989, 28, 917.
51 R. Raag and T. L. Poulos, Biochemistry, 1989, 28, 7586.
52 T. L. Poulos, B. C. Finzel and A. J. Howard, Biochemistry, 1986, 25,
5314.
53 X. Chen, A. Christopher, J. P. Jones, S. G. Bell, Q. Guo, F. Xu,
Z. Rao and L. L. Wong, J. Biol. Chem., 2002, 277, 37519.
54 S. G. Bell, X. Chen, R. J. Sowden, F. Xu, J. N. Williams, L. L. Wong
and Z. Rao, J. Am. Chem. Soc., 2003, 125, 705.
32 J. A. Fruetel, J. R. Collins, D. L. Camper, G. H. Loew and P. R. Ortiz
de Montellano, J. Am. Chem. Soc., 1992, 114, 6987.
33 D. Filipovic, M. D. Paulsen, P. J. Loida, S. G. Sligar and
R. L. Ornstein, Biochem. Biophys. Res. Commun., 1992, 189, 488.
34 P. J. Loida and S. G. Sligar, Biochemistry, 1993, 32, 11530.
35 O. Sibbesen, Z. Zhang and P. R. Ortiz de Montellano,
Arch. Biochem. Biophys., 1998, 353, 285.
36 D. A. Grayson, Y. B. Tewari, M. P. Mayhew, V. L. Vilker and
R. N. Goldberg, Arch. Biochem. Biophys., 1996, 332, 239.
55 S. G. Sligar, D. Filipovic and P. S. Stayton, Methods Enzymol., 1991,
206, 31.
D a l t o n T r a n s . , 2 0 0 3 , 2 1 3 3 – 2 1 4 0
2140