Flavocytochrome P450 BM3 mutant A264E undergoes substrate-dependent formation of a novel heme iron ligand set

by the dideoxy chain termination method at the Protein and Nucleic Acid Chemistry Laboratory (PNACL) facility at the University of Leicester. A Msc I site upstream of the mutation was used in combination with Mfe I to excise a 518 bp fragment containing the A264E mutated region. The relevant fragment was resolved on a 1 % agarose gel and purified using a QIAquick gel extraction kit (Qiagen). The Mfe I/ Msc I restriction fragment was re-ligated into the backbones of pBM20 and pBM23 plasmids that had been digested with the same restriction enzymes and gel purified in the same way as the insert fragment. Correct insertion was verified by restriction enzyme digestion.


INTRODUCTION
The cytochromes P450 (P450s) constitute a superfamily of heme b-containing mono-oxygenase enzyme responsible for a huge variety of physiologically and biotechnologically important transformations (e.g. 1,2). The P450s' heme iron is ligated to the protein by a conserved cysteinate (the proximal ligand), with a water molecule usually present as the distal axial ligand (3,4). The P450s are found throughout nature and typically catalyse the reductive scission of dioxygen bound to the heme iron, frequently resulting in hydroxylation of an organic substrate.
Two successive one electron transfers to the P450 are required for oxygenation reactions. The source of the electron are reduced pyridine nucleotides (NADPH or NADH), and electron transfer is usually mediated by one or more redox partner enzymes. In hepatic P450 systems, the redox partner is the diflavin reductase NADPH-cytochrome P450 reductase (CPR), which contains FAD and FMN cofactors (1,5). In mammalian adrenal systems and many bacterial P450s, electrons are delivered via a two protein redox systems comprising an FAD-containing reductase (adrenodoxin reductase or a ferredoxin reductase) and an iron sulfur protein (ferredoxin) (6,7). However, other forms of redox systems supporting P450 catalysis are known to exist -including the direct interaction of P450 with hydrogen peroxide to facilitate fatty acid hydroxylation in the Bacillus subtilis P450 BSβ enzyme (8). In recent years, the flavocytochrome P450 BM3 enzyme from Bacillus megaterium has been studied intensively as a "model" P450 enzyme (9). P450 BM3 is a natural, soluble fusion enzyme, in which a fatty acid hydroxylase P450 (N-terminal) is joined to a CPR (C-terminal), creating an efficient electron system capable of hydroxylating a wide range of fatty acids at rates of several thousand events per minute (10,11). In certain P450s (including P450 BM3), binding of substrate facilitates the first electron transfer from the redox partner, by increasing the reduction potential of the heme iron through removal of the distal water ligand (12,13). Dioxygen then binds to the ferrous P450 heme iron. Further reduction (by another single electron transfer from the redox partner) may have important ramifications as regards the exploitation of P450s in such areas as biotransformations and toxicological applications.
Flavocytochrome P450 BM3 has been recognised as perhaps the most important P450 enzyme with respect to its capacity to perform biotechnologically-exploitable chemical transformations (9). The wild-type enzyme has been shown to catalyse regio-and stereo-selective hydroxylation and epoxygenation of long chain saturated and unsaturated fatty acids of varying chain length (11,21), and mutants generated both rationally and by forced evolution have exhibited novel properties such as hydroxylation at the ω-position on fatty acids (rather than the preferred ω-1 to ω-3 positions), and oxygenation of polycyclics, substituted fatty acids, indole, alkanes and short chain fatty acids and alcohols (e.g. 21,22,27,28). However, P450 BM3 exhibits the same structural instability observed in other P450 isoforms as regards the tendency to undergo inactivation at the heme site through P420 formation (16). In this respect, generation of a more stable P450 derivative through covalent attachment of the heme macrocycle is an attractive proposition. The fact that the heme domain of P450 BM3 is strongly related to fatty acid hydroxylases of the CYP4 family suggested that covalent heme linkage might also be feasible in this isoform, and the fact that it has not been observed for the wild-type BM3 could be explained by the presence of an alanine rather than a glutamate residue in the respective position of the conserved I helical region in this P450 (Figure 1) (29).

FIGURE 1 HERE
In this study, we have generated and characterized the A264E variant of P450 BM3 (in both the full length flavocytochrome and the heme domain) and determined the effect of the mutation on the catalytic, spectroscopic, thermodynamic and structural properties of the enzyme. In contrast to the results with the mammalian P450 isoforms, the introduction of a glutamate residue does not result in turnover-dependent covalent linkage of the heme macrocycle. Instead, the glutamate becomes a 6 th axial ligand to the ferric heme iron producing a completely novel heme iron ligand set (Cys-Fe-Glu), with occupancy of the glutamate (instead of water) promoted by the binding of substrates to the P450.

EXPERIMENTAL PROCEDURES Expression and purification of wild-type and mutant P450 BM3 proteins
Expression and purification of the mutant (A264E) and wild-type full-length flavocytochrome P450 BM3 and heme domains (amino acids 1-472) was performed essentially as described in previous publications (e.g. 30,31). Expression plasmids pBM20 and pBM23 (wild-type P450 BM3 heme domain and intact flavocytochrome P450, respectively) and pHMG1 and pHMG2 (the respective heme domain and intact flavocytochrome P450 A264E clones) were expressed from E. coli strain TG1 grown in Terrific Broth plus 50 µg/ml ampicillin (typically 5 litres cells) for approximately 36 hours following inoculation from an overnight culture of the relevant transformant. Cells were collected by centrifugation, resuspended in Tris.HCl (50 mM, pH 7.2) plus 1 mM EDTA (buffer A) and broken using a French press (3 passes at 950 psi) followed by sonication of the resulting suspension on ice (Bandelin Sonopuls sonicator, 5 x 20 second pulses at 50 % full power, with adequate cooling time between pulses). Extract was exchanged by dialysis into ice-cold buffer A containing the protease inhibitors benzamidine hydrochloride and phenylmethylsulfonyl fluoride (PMSF, both inhibitors at 1 mM final concentration) prior to loading onto a DEAE column pre-equilibrated in the same buffer. Enzymes were eluted in a linear gradient of 0-500 mM KCl in buffer A. The most intensely red-coloured fractions were retained, concentrated by ultrafiltration (Centriprep 30 concentratiors, Millipore) and dialysed extensively into 25 mM potassium phosphate (buffer B, pH 6.5, containing benzamidine and PMSF). Intact flavocytochrome P450 BM3 and its A264E mutant were loaded onto a mimetic yellow column (2 cm x 15 cm) and washed extensively in buffer B prior to elution with 25 mM 2' and 3' adenosine monophosphate (mixed isomers, Sigma) containing 500 mM potassium chloride. The wild-type heme domain and A264E mutant were loaded onto a hydroxyapatite column in buffer B, and eluted in a linear gradient of buffer B to 500 mM potassium phosphate (pH 6.5, plus protease inhibitors). The most intensely red-coloured fractions were pooled, concentrated and exchanged by dialysis into buffer A at 4 ºC, prior to loading on a Q-Sepharose column and eluting as described for the DEAE resin. All flavocytochrome and heme domains were pure at this stage (as judged by SDS PAGE analysis) and were concentrated by ultrafiltration to ≥500 µM, prior to dialysis into buffer A plus 50 % (v/v) glycerol and storage at -80 ºC. A264E mutant heme domain used in crystallographic trials were exchanged instead into 10 mM Tris.HCl (pH 7.2) and used directly for crystallogenesis.

Site-directed mutagenesis of the CYP102A1 gene
The A264E mutant forms of full length flavocytochrome P450 BM3 and its heme domain were constructed by overlapping mutagenic PCR. Three PCRs (A-C) were carried out in total, using the heme domain construct plasmid pBM20 (30)  The final PCR reaction (PCR C) combined the products of PCRs A and B using the primers MfeF and BamR. The product of PCR C was A-tailed using Taq DNA ploymerase and ligated into pGEM T according to the manufacturer's instructions (Promega pGEM-T Easy Vector Systems). This yielded plasmid pGEM-A264E, which was completely sequenced by the dideoxy chain termination method at the Protein and Nucleic Acid Chemistry Laboratory (PNACL) facility at the University of Leicester. A MscI site upstream of the mutation was used in combination with MfeI to excise a 518 bp fragment containing the A264E mutated region. The relevant fragment was resolved on a 1 % agarose gel and purified using a QIAquick gel extraction kit (Qiagen). The MfeI/MscI restriction fragment was re-ligated into the backbones of pBM20 and pBM23 plasmids that had been digested with the same restriction enzymes and gel purified in the same way as the insert fragment. Correct insertion was verified by restriction enzyme digestion.

SDS-PAGE analysis of A264E
SDS-PAGE was performed using a Bio-Rad mini-protean II apparatus and either 10 % or 6 % polyacrylamide gels. SDS PAGE was used to establish purity of the wild-type and mutant flavocytochromes P450 BM3 and heme domains, and also to resolve the A264E flavocytochrome P450 BM3 and heme domain prior to testing for covalent attachment between the heme and P450 protein. This was done using a heme-staining procedure to establish whether heme remained associated with the protein following denaturation in SDS (32).
The A264E heme domain was resolved on 10% SDS-PAGE gels in substrate-free and arachidonate-bound forms (~ 50 µM fatty acid), both in the absence of and following incubation with excess dithionite reductant. Intact A264E flavocytochrome BM3 was run on 6% SDS-PAGE gels in the same forms as for the heme domain, and also following incubation with excess NADPH (500 µM) in presence and absence of arachidonate, and following treatment with hydrogen peroxide (20 mM). Cytochrome c (horse heart, type IV, Sigma) was used as a positive control for covalent heme ligation in 10% gels, and Shewanella frigidimarina flavocytochrome c 3 was used in 6% gels (33). A total of 2 ng enzyme was run in each lane on the SDS PAGE gels along with protein molecular molecular weight markers (NEB broad range protein marker). Gel preparation and electrophoresis were performed as described previously (34). To determine whether heme remained bound to the enzymes following denaturing gel electrophoresis (i.e. was covalently linked to the protein backbone). Gels were stained in the dark with 70 ml 0.25 M sodium acetate (pH 5.0), to which was added 30 ml of 6.3 mM 3,3',5,5'-tetramethylbenzidine (TMBZ) in methanol. Hydrogen peroxide (30 mM final concentration) was then added and the gel incubated until light blue bands appeared, indicating the presence of heme in certain proteins.
The reaction was then stopped with the addition of isopropanol (approx. 10 % v/v). The gels were photographed and then re-stained with Coomassie blue, to ensure the presence of control and sample proteins at appropriate concentrations.

Binding of substrates and ligands to wild-type and A264E P450 BM3
Binding of fatty acids and heme-coordinating inhibitors to the wild-type and A264E heme domains was analysed by optical titrations using a Cary UV-50 Bio scanning spectrophotometer (Varian). The fatty acids arachidonate, palmitoleate, palmitate, myristate and laurate were used in the titrations. The spectra for the substrate-free wild-type and A264E enzymes (typically 5-8 µM protein) were recorded at 30 ºC in assay buffer (1 ml total volume of 20 mM MOPS, pH 7.4, plus 100 mM KCl), prior to additions of the fatty acids in aliquots of 0.1-0.5 µl (using a Hamilton syringe) up to a final volume of not more than 1 % of the total volume of the solution.
Fatty acids were prepared as concentrated stocks (3-25 mM) in ethanol (arachidonate, palmitate and myristate) or DMSO (palmitoleate) or as a saturated aqueous stock in assay buffer (~ 900 mM) for laurate (31). Spectra were recorded after each addition of substrate, and a difference spectrum computed by subtraction of the starting (substrate-free) spectrum from those generated at each point in the titration. The maximal apparent absorption change induced at each point in the titration was determined by subtraction of the minimal absorption value at the trough in each difference spectrum from the maximal value at the peak (using data at the same wavelengths in each titration). The maximal changes in absorption determined in this way were plotted versus the relevant fatty acid concentration. A similar approach was taken to determine the apparent binding constant for the inhibitor 4-phenylimidazole. Data were fitted either to a rectangular hyperbola or, in cases where fatty acids bound very tightly to the P450, to a quadratic function that accounts for the quantity of enzyme consumed in the enzyme-substrate complex at each point in the titration (equation 1).
In equation 1, "A obs " is the observed absorption change at substrate/ligand concentration "S", "Et" is the total enzyme concentration and "K d " is the dissociation constant for the enzymeligand/substrate complex. All fitting of data was done using Origin software (Microcal).
The "P450" form of wild-type and A264E mutant P450 BM3 were generated by addition of a few grains of fresh sodium dithionite to the proteins (typically 4-8 µM in assay buffer), followed by slow bubbling of the solution with carbon monoxide gas for approximately one minute. The nitric oxide adduct of the P450s was generated by release of ~ 5 small bubbles into a similarly buffered solution of ferric enzyme.

Studies of effects of pH, ionic strength and temperature on optical properties of the A264E heme domain
UV-visible spectra were recorded for the substrate-free form of A264E heme domain (4 µM) in 50 mM potassium phosphate in the pH range between 5.0-9.0, at 0.5 pH unit intervals. Spectral perturbations were observed and absorption data reflecting the maximal overall change between the low-and high-pH spectra (∆A 417 minus ∆A 423 with reference to the spectrum collected at pH 7.0) were plotted against pH and fitted to a sigmoid to derive an apparent pK a value accompanying the spectral conversion.

Redox Potentiometry
All redox titrations were carried out in an anaerobic glove box (Belle technology, Portesham, England) under a nitrogen atmosphere, with oxygen levels maintained at < 5 ppm. Redox titrations were carried out for both wild type heme domain P450 BM3 and the A264E BM3 heme domain (typically 6-10 µM), both in the presence and absence of arachidonic acid (~ 70 µM, K d value is < 5 µM) by the method of Dutton, and essentially as described previously (12,35,36 included to mediate in the range between +100 to -480 mV, as described previously (12,35).
The electrode was allowed to stabilize between each addition of reductant/oxidant prior to spectral acquisition and recording of the potential.
Data were analysed by plotting the absorbance at an appropriate wavelength, corresponding to the maximal absorbance change between oxidised and reduced forms, against the potential. A single electron Nernst function was then fitted to the data to describe the transition between ferric and ferrous heme iron, and the midpoint potential calculated from this data fit. For the substrate-bound titrations, arachidonate was added from a 33 mM stock in ethanol until no further change in spectral shift was observed. Data generated from the fitting procedures with wild-type P450 BM3 heme domain were in close agreement with previous studies (37).

Steady-state kinetics
The apparent rates of fatty acid-dependent NADPH oxidation catalysed by wild-type and A264E mutant flavocytochromes P450 BM3 were determined essentially as described previously (35).
All measurements were carried out in a 1cm pathlength quartz cuvette in assay buffer at 30  Measurement of reductase domain-dependent reduction of cytochrome c by wild-type and A264E mutant flavocytochromes P450 BM3 was performed essentially as described above for fatty acid turnover. However, enzyme concentration was 7 nM, and rates were determined from the accumulation of reduced products at appropriate wavelengths: for ferricyanide at 420 nm (∆ε 420 = 1010 M -1 cm -1 ) and for cytochrome c at 550 nm (∆ε 550 = 22640 M -1 cm -1 ) (38).

Determination of oxygenated fatty acid products
Turnover experiments were carried out by incubating 0.4 µM wild-type or A264E flavocytochrome P450 BM3 with 200 µM myristic acid, palmitic acid or palmitoleic acid, and 600 µM NADPH in a final volume of 5 ml. The reaction was allowed to proceed for 4 hours at 30 ˚C with stirring, before halting the reaction by acidification to pH 2.0 with hydrochloric acid.
Fatty acids were extracted from the aqueous environment into one volume of dichloromethane.
Remaining aqueous material was removed by the addition of excess solid magnesium sulphate to the mixture. Following filtration, to remove magnesium sulfate, the dichloromethane was evaporated under a stream of nitrogen gas and lipids resuspended in a small volume of methanol.
Thereafter, 20 µl samples were analysed by electrospray mass spectrometry (70 eV ionisation) using a Micromass Quattro triple quadrupole mass spectrometer. Samples from the aqueous layer were also run to ensure complete extraction and negative controls (in which no NADPH was added to the enzyme/fatty acid mixtures) were also examined to ensure that any products resulted from NADPH-dependent enzyme activity.

Molecular biology, expression and heme-binding properties
The A264E mutations were introduced into wild-type constructs of full length flavocytochrome Subsequent staining of these gels with Coomassie blue confirmed the presence of large amounts of the relevant P450 proteins, providing further proof that the b-type heme did not remain bound to the proteins in either wild-type or A264E forms.

UV-visible spectroscopy
Despite the apparent lack of covalent heme ligation, close examination of the electronic spectrum of the oxidized forms of the A264E enzymes indicated that the heme signals were slightly shifted with respect to the wild-type forms. The Soret band maximum for both oxidized, substrate-free A264E flavocytochrome and heme domains was shifted to longer wavelength by approximately 1-2 nm with respect to the wild-type forms (from ~ 418 nm to 420.5 nm), and there were similar small perturbations in the α/β band region (from ~ 568 nm to 571 nm for the α band; from 534 nm to 538 nm for the β band). In the dithionite-reduced form, the spectral properties of the ferrous forms of the substrate-free wild-type and A264E mutants were virtually indistinguishable, with the Soret band shifted to 411 nm and the α/β bands apparently fused with a maximum at 546 nm (Figure 2A). The spectral properties of the ferrous-carbon monoxy complexes were also determined, showing a Soret shift to ~ 449 nm for the A264E heme domain cf ~ 448 nm for the wild-type enzyme (30,40). To analyse further the spectral properties of the A264E mutant, a nitrosyl complex was generated by bubbling the A264E heme domain with NO gas (5 small bubbles). A Soret band shift to 435 nm was observed (cf 434 nm for wild-type BM3 heme domain). As with wild-type enzyme, there was a marked increase in intensity of the alpha and beta bands, with their maxima shifted to 575 nm and 543.5 nm, respectively. The binding of a tight-binding azole inhibitor (4-phenylimidazole) was also performed. In this case, the spectral features of the complex were virtually indistinguishable from those of wild-type P450 BM3, with the Soret band shifted to 425.5 nm, and with changes in alpha/beta band intensity and shifts to 575/544 nm ( Figure 2B).

FIGURE 2 HERE
To examine spectral effects on the binding of fatty acid substrates, optical binding titrations were done for the A264E heme domain using a number of fatty acids (lauric acid, myristic acid, palmitic acid, palmitoleic acid and arachidonic acid) known to bind tightly to the wild-type enzyme. Surprisingly, spectral changes observed on addition of fatty acids to A264E were distinct from those observed previously for wild-type P450 BM3 and various mutant forms (e.g. the most effective amongst those tested. This aspect of the mutant's behaviour is discussed in more detail in the Discussion section. Thus, rather than undergoing a substrate-dependent optical transition typical of increased high-spin heme iron content, the A264E mutant shows instead a type II transition usually observed on ligation of inhibitors to the heme iron (e.g. imidazoles, see Figure 2B). This type of optical transition likely indicates reinforcement of the low-spin form of the cytochrome. Data for the optical titration with arachidonic acid are shown in Figure 3.

FIGURE 3 HERE
Notwithstanding the unusual spectral changes observed, the apparent binding constants (

TABLE 1 HERE
An observation from studies of the pH-dependence of the UV-visible electronic spectrum of the A264E mutants was that the absorption maximum of the Soret band was very sensitive to pH changes in the range between 5-9.5 (in which there was not any significant destruction of the heme). Spectra for the A264E heme domain were recorded in potassium phosphate at several pH values across this range. At pH 7.5, the Soret maximum was located at 420.5 nm, as seen in the buffer used for fatty acid-binding titrations and kinetic studies (e.g. see Figure 2). As the pH was However, rates of fatty acid-dependent NADPH oxidation were considerably slower than those for the wild-type P450 BM3 ( Table 1). In the absence of fatty acids, both wild-type and A264E flavocytochrome P450 BM3 oxidise NADPH at a slow rate (~ 5 min -1 ). In the presence of either lauric acid or myristic acid, there was considerable stimulation of NADPH oxidase activity, despite the fact that addition of these fatty acids did not induce any considerable changes in the optical spectrum of the A264E mutant. Enzyme activity in the presence of palmitic acid, palmitoleic acid or arachidonic acid was even higher, albeit rather less than that observed with wild-type P450 BM3. (Table 1). Thus, it appears that the flavocytochrome P450 BM3 A264E mutant retains considerable levels of activity with long chain fatty acids, despite the fact that these fatty acids induce either negligible change towards high spin, or inhibitor-like optical change to the heme spectrum.

Fatty acid oxygenation
In view of the apparently conflicting data indicating high levels of fatty acid-dependent NADPH oxidation despite type II optical shifts induced by the same fatty acids, we undertook studies to establish whether NADPH oxidation was linked to oxygenation of myristic acid, palmitic acid and palmitoleic acid. Turnover studies were performed as described in the Experimental section, and products were examined by mass spectrometry. Products were evident from turnover of each of these substrates. In the case of myristic acid, there was similar levels of production of monooxygenated product for both wild-type and A264E flavocytochrome P450 BM3. With palmitic acid, mass spectrometry showed the presence of both mono-and di-oxygenated products for both wild-type and A264E enzymes, consistent with previous data and indicative that a primary hydroxylated product can act as a substrate for a second round of oxidation (41,42). With the mono-unsaturated fatty acid palmitoleic acid (cis-9-hexadecenoic acid) as the substrate, both mono-and di-oxygenated products were observed for both wild-type and A264E enzymes, again with rather lower amounts of products with the mutant enzyme. In previous studies, Fulco and co-workers demonstrated that both hydroxylation (close to the ω-terminus) and epoxidation (across the C 9 -C 10 double bond) of palmitoleic acid are catalysed by wild-type P450 BM3 (43,44). Under the same experimental conditions, the amounts of products generated from palmitic acid and palmitoleic acid by the A264E mutant were ~ 30-40 % lower than those produced by wild-type P450 BM3, consistent with the differences in steady-state kinetics shown in Table 1.
In parallel studies of peroxide production during fatty acid turnover, there was no significant difference between the wild-type and A264E enzymes, suggesting that both enzymes couple NADPH oxidation to fatty acid oxygenation tightly, but that the A264E mutant is a much slower hydroxylase than the wild-type P450 BM3 (45). Thus, despite the unusual spectral conversions produced on binding long chain fatty acids, the A264E flavocytochrome retains the capacity to oxygenate fatty acids.

Spectroscopic analysis
Optical  (Figure 4). Thus, the direction of movement of the Soret band (to shorter wavelength) mimics the optical transition seen in the electronic absorption spectrum. However, the near UV-visible CD spectrum of the substrate-free A264E heme domain has its Soret band at ~422 nm. On addition of arachidonic acid, this feature sharpens and shifts to 426 nm, the same wavelength as seen for the Soret maximum in the optical spectrum of arachidonate-bound A264E heme domain (Figure 4). Thus, near UV-visible CD spectroscopy shows much more marked differences in the spectra of wildtype and A264E heme enzymes than does optical absorption spectroscopy. A large spectral difference is seen between the near UV-visible CD spectra substrate-free forms of wild-type and

EPR:
EPR spectroscopy of the arachidonate-bound and substrate-free forms of the wild-type and A264E heme domains were recorded, and are shown in Figure 5A. The EPR spectrum for wildtype P450 BM3 is as previously reported (46) and typical for low-spin ferric P450s which all give rise to spectra with g z in the range 2.40-2.45 (47)(48)(49). On binding of substrate, the ferric spin equilibrium is perturbed and the iron becomes a mixture of low-and high-spin. The latter appears in the EPR with features at g = 8.18, 3.44, 1.66 which originate in the lowest (m s = "±½") Kramers doublet of the S = 5/2 ferric ion. In the "low-field limit" where the axial zero-field-splitting parameter is greater than the Zeeman splittings (D >> gβH) these values correspond to a rhombicity of E/D = 0.11. In ferric hemoproteins, such substantial rhombicities are found only with thiolate ligation. The EPR spectra of the substrate-free and arachidonatebound A264E mutant of P450 BM3 both suggest the presence of several low-spin ferric species, all with g-values which indicate that thiolate coordination is maintained. A264E appears to contain two distinct forms with g z ~ 2.56 and 2.43 respectively. Each feature shows structure indicative of further minor heterogeneities. There is no example of cysteinate proximal to a neutral oxygen ligand giving a g z -value greater than 2.45. The g z ~ 2.56 must therefore indicate ligation different from that of the wild-type and is strong evidence for coordination of the distal glutamate. In support of this, the formate derivatives of P450 cam and chloroperoxidase give g z at 2.55 and 2.59 respectively (50,51). The feature at g z ~ 2.43 suggests that a proportion of the sample retains a distal water ligand although it appears that the distal mutation causes some heterogeneity in this sub-population. In contrast to the wild-type P450 BM3 enzyme, when substrate is bound to the A264E mutant, there is no significant switch to a high-spin form.
Instead, changes in the EPR spectrum for arachidonate-bound A264E indicate differences in the distribution of the low-spin species, with a diminution of the contribution from a Cys-aqua ligated form and a simultaneous increase in the proportion of Cys-Glu coordinated species, indicating that substrate-binding promotes the ligation of Glu 264 to the heme iron.

FIGURE 5 HERE
A comparison of the EPR spectra for 4-phenylimidazole-bound forms of wild-type and A264E mutant heme domains is shown in Figure 5B. There is some heterogeneity in the wild-type complex, with the major triplet of g-values at 2.57, 2.26 and 1.86/1.85. The g z signal is broadened, suggesting that there may be a split population of two conformers with a slightly different g z component, and paired with the two subtly different g x components. Minor signals at 2.44 and 1.92 may reflect a small proportion of non-ligated enzyme. In the azole-bound A264E complex, there appears to be one predominant species with g-values at 2.58, 2.26 and 1.86. The homogeneity of this spectrum is in part due to the apparently complete ligation of the azole to the heme iron in this sample (removing residual aqua-ligated components seen in the wild-type spectrum), consistent with the tighter K d value determined for binding of 4-phenylimidazole to A264E (c.f. wild-type) from spectral titrations (see Table 1). However, it appears to be the case that a single conformational form of azole-bound heme is present in A264E, whereas there may be two distinct species in the wild-type P450.

MCD
The MCD spectra for wild-type P450 BM3 heme domain are consistent with those we have reported earlier (46). The room temperature near UV-visible MCD spectra of substrate-free wildtype and A264E P450 BM3, and the arachidonate-bound A264E enzyme (Figure 6A) each show a pattern of bands typical for low-spin ferric hemes with a thiolate ligand. The unusually low MCD intensity in both the Soret band (400-420 nm) and the α/β region (500-600 nm) is also characteristic of such species (52)(53)(54)(55)(56)(57)(58)(59)(60)(61). A small additional negative feature, at ~655 nm, is part of a derivative-shaped charge-transfer band and arises from the presence of a low level (≤ 15%) of the high-spin form (see 62). The spectrum of substrate-bound wild-type P450 BM3 is very different and shows that the low-spin heme is now the minority species (~25%). The CT band near 655 nm has increased in intensity and other high-spin bands are evident at 360-405 nm, and as a shoulder at ~555 nm. Low-spin ferric hemes also give rise to a porphyrin-to-ferric chargetransfer transition at longer wavelengths. This appears as a positive signed band in the MCD and has been located in the room temperature NIR MCD for wild-type P450 BM3 minus substrate, and for the A264E mutant in both the absence and presence of substrate at ~1080 nm ( Figure   6B) as was previously reported for WT BM3-P450 at low-temperature (46). Close inspection of these three spectra reveals qualitative and quantitative differences. The peak position for the transition shifts from ~ 1075 nm for the wild-type substrate-free enzyme, through to ~1080 nm for the substrate-free A264E mutant, and to ~1085 nm for the arachidonate-bound A264E mutant. Differences in both the breadth and intensity of the CT band are discernible ( Figure 6B).
The exact energy of this CT transition is generally diagnostic of the two heme axial ligands (63,64), but the influence of the second ligand is somewhat reduced in the presence of thiolate, as

FIGURE 6 HERE
Addition of arachidonic acid substrate to wild-type P450 BM3 results in a marked change in the NIR MCD spectrum. Consistent with the switch to predominantly high-spin heme that was observed at UV-visible wavelengths, the low-spin CT band near 1100 nm is significantly diminished. The derivative-shaped MCD band centred at ~900 nm is the CT band characteristic of high-spin ferric heme and is extremely similar to that reported for substrate-bound cytochrome P450 cam (66).

Potentiometric analysis
Previous studies have shown that fatty acid binding to wild-type P450 BM3 is accompanied by loss of the aqua ligand to the heme iron and a shift in the heme iron spin-state equilibrium towards the high-spin form (21,30). In P450 BM3 (as in P450 cam), this is accompanied by a change in the heme iron reduction potential of ~ 130-140 mV (from -427 mV to -289 mV for P450 BM3) (12,37).

DISCUSSION
The capacity of the eukaryotic family 4 cytochromes P450 to link their heme macrocycle covalently to the protein backbone has been one of the most significant discoveries in P450 research in recent years (e.g. 23,24). From a biotechnological perspective, the ability to covalently link the porphyrin to the P450 protein matrix is attractive for at least two reasons.
Firstly, the presence of the glutamate and the glutamate ligation process was shown to enhance catalytic activity in rabbit CYP4B1 (26), and enhancement of catalytic rate is clearly a desirable feature to endow on an enzyme. Secondly (and the major reason), the capacity of P450s to undergo conversion to the inactive "P420" form (in which native cysteinate heme ligation is lost) is well recognized, and heme can even be dissociated completely from the P450 enzyme under moderately denaturing conditions (e.g. 67). Thus, the ability to covalently tether heme should promote longevity of P450 activity, particularly since the P450/P420 conversion has been shown to be reversible in selected P450s, including P450 BM3 (e.g. 68). From a perspective of exploitation of P450 enzymes, the bacterial enzymes P450 cam and P450 BM3 are the most intensively studied. Rational mutagenesis of P450 cam has produced variants of the camphor hydroxylase that are able to oxygenate molecules such as butane and propane (69). Rational mutagenesis of the fatty acid hydroxylase P450 BM3 has produced variants that catalyse oxygenation of fatty acids at different positions from the wild-type form, and in which substrate selectivity has been converted towards short chain fatty acids and polycyclic aromatic hydrocarbons such as phenanthrene (21,70). Forced evolution of P450 BM3 has also produced an efficient alkane hydroxylase enzyme (22). It is clear that P450 BM3 has great biotechnological potential for production of functionalised hydrocarbons. In view of this potential and the close relationship between P450 BM3 and the eukaryotic family 4 fatty acid hydroxylases, we mutagenised P450 BM3 to introduce the glutamate conserved in the eukaryotic However, the spectra for the A264E mutant are considerably different from those for the wildtype heme domain. In the absence of substrate the EPR spectrum of A264E is typical for a lowspin heme, but shows substantial heterogeneity in the signal that we consider result from an ensemble of forms present. The observation of g-values identical to those of the substrate-free wild-type heme domain indicates that a significant proportion of the mutant persists in a "nativelike" water-bound form, although small shoulders on these features suggest some minor heterogeneity (Fig. 5A). The other three g z components clustered around g = 2.56 are assigned Phe87 is known to be a conformationally flexible amino acid, which is important in controlling substrate selectivity and for interaction with the ω-terminal carbon of fatty acid substrates of P450 BM3, preventing hydroxylation at this position (72,73). Potentially, interactions between Phe87 and Glu264 in its ligated form provide further explanations for the heterogeneity in the substrate-free EPR spectrum. In the arachidonate-bound form of A264E the heterogeneity is reduced, due partly to the loss of much of the aqua-ligated form as the fatty acid "drives" on the glutamate ligand. It is likely that this occurs due to the substrate displacing unligated glutamate from its other favoured conformation, in which it interacts with the phenyl group of Phe87.
However, the apparent degree of heterogeneity in the EPR spectrum is further reduced due to alterations in the proportions of the different glutamate-ligated forms of A264E. Potentially, this also reflects the re-positioning of Phe87 following binding of substrate, and the position(s) However, binding of long chain fatty acids forces the equilibrium heavily in favour of the novel Cys-Glu ligand set, in similar fashion as binding of these substrates induces aqua ligand displacement and formation of a 5-coordinate high-spin heme iron species in the wild-type P450 BM3. An intriguing aspect evident from this study is that the A264E flavocytochrome P450 BM3 enzyme retains considerable fatty acid oxygenase activity, at least towards those fatty acids that are efficient in inducing the switch to Cys-Glu coordination. This indicates that the Glu ligand must be displaced (at least in a proportion of the enzyme) following the first reduction step in the catalytic cycle. This should allow oxygen to bind, reductive scission of dioxygen to occur following the second electron reduction of the iron, and for oxygenation of the tightly bound fatty acid substrate. In reporting their structure of the palmitoleate-bound form of the wild-type P450 BM3 heme domain, Li and Poulos noted that the distance between the ω-carbon of the bound substrate and the heme iron was too great for oxidative attack of the substrate, and that further structural change resulting in re-positioning of the substrate closer to the heme was required following reduction of the ferric iron (72). NMR studies of the ferric and ferrous P450 indicated that there should be a 6 Å movement of substrate subsequent to heme iron reduction (78). Thus, it appears clear that the Cys-Glu coordination is broken following heme iron reduction, enabling binding of oxygen and for catalysis to ensue, albeit with lower affinity than in the wild-type enzyme. Potentiometric analysis (Fig. 7)  likely scenario is that the Cys-aqua species is reduced preferentially (i.e. has a more positive reduction potential than the Cys-Glu species) and that the equilibrium between the Cys-aqua and Cys-Glu ferric forms is drawn towards the former as the reductive titration progresses. The altered heme iron reduction potential of the A264E mutant is discussed further in the accompanying paper, in light of major structural changes observed for the mutant.
In conclusion, the A264E variant of P450 BM3 produces an unprecedented heme iron ligand set

Figure 6
Magnetic Circular Dichroism studies of the A264E heme domain MCD spectra in the near-UV visible region (Fig. 6A) and the NIR region (Fig. 6B) Fig. 6A, since in this case the spectral change on reduction occurs over a more compressed wavelength range (from 419.5 nm to 410 nm). Inset shows a data fit (as in Fig. 6A), producing a heme iron midpoint reduction potential of -316 ± 3 mV.