Structural and Spectroscopic Characterization of P450 BM3 Mutants with Unprecedented P450 Heme Iron Ligand Sets

Two novel P450 heme iron ligand sets were generated by directed mutagenesis of the flavocytochrome P450 BM3 heme domain. The A264H and A264K variants produce Cys-Fe-His and Cys-Fe-Lys axial ligand sets, which were validated structurally and characterized by spectroscopic analysis. EPR and magnetic circular dichroism (MCD) provided fingerprints defining these P450 ligand sets. Near IR MCD spectra identified ferric low spin charge-transfer bands diagnostic of the novel ligands. For the A264K mutant, this is the first report of a Cys-Fe-Lys near-IR MCD band. Crystal structure determination showed that substrate-free A264H and A264K proteins crystallize in distinct conformations, as observed previously in substrate-free and fatty acid-bound wild-type P450 forms, respectively. This, in turn, likely reflects the positioning of the I α helix section of the protein that is required for optimal configuration of the ligands to the heme iron. One of the monomers in the asymmetric unit of the A264H crystals was in a novel conformation with a more open substrate access route to the active site. The same species was isolated for the wildtype heme domain and represents a novel conformational state of BM3 (termed SF2). The “locking” of these distinct conformations is evident from the fact that the endogenous ligands cannot be displaced by substrate or exogenous ligands. The consequent reduction of heme domain conformational heterogeneity will be important in attempts to determine atomic structure of the full-length, multidomain flavocytochrome, and thus to understand in atomic detail interactions between its heme and reductase domains.

The cytochromes P450 (P450s) 4 are a superfamily of heme b-containing monooxygenase enzymes found in organisms from all domains of life (e.g. Refs. 1 and 2). They catalyze the oxygenation (often hydroxylation) of a wide range of molecules in nature, exploiting a transient ferryl-oxo heme iron intermediate to facilitate addition of oxygen to the substrate (1). They are intensively studied due to their integral roles in, e.g., mammalian drug metabolism, steroid and sterol synthesis, polyketide antibiotic manufacture, and prokaryotic breakdown of recalcitrant pollutants (e.g. Refs. 3 and 4). The Bacillus megaterium P450 BM3 enzyme is one of the most intensively studied members of the superfamily. It is a rare example of a prokaryotic P450 that obtains electrons for reductive activation of dioxygen from a eukaryotic-type redox partner. The fatty acid hydroxylase P450 receives electrons from NADPH via a FADand FMN-containing diflavin reductase (NADPH-cytochrome P450 reductase) that is also fused to the P450 in a single polypeptide chain (5)(6)(7). The fusion arrangement facilitates efficient electron transfer between the redox cofactors in P450 BM3 and affords the enzyme the highest oxygenase rate of any P450 reported to date (Ͼ15,000 min Ϫ1 with arachidonic acid) (8). Substrate binding controls the reduction potential of the heme iron and accelerates FMN-to-heme electron transfer in the presence of fatty acids to ensure coupling of NADPH oxidation to fatty acid hydroxylation (9). The enzyme was shown recently to be functional as a fatty acid hydroxylase in the dimeric form, as demonstrated previously for eukaryotic nitricoxide synthase enzymes (10,11).
The P450s are the best characterized of the thiolate-ligated cytochromes, with a loosely associated water molecule generally located in the sixth (distal) position on the heme iron (1). This position is vacated in catalysis to allow binding of dioxygen to ferrous iron. Nitric-oxide synthase enzymes have the same heme ligand set and also activate oxygen, and chloroperoxidase (an enzyme involved in antibiotic synthesis in the fungus Caldariomyces fumago) is a further example of an enzyme with a cysteine-coordinated heme iron (12,13). However, relatively few other cysteinate-containing ligand sets are observed natu-* This work was supported in part by the UK Research Councils (Biotechnology and Biological Sciences and Research Council and Engineering and Physical Sciences Research Council). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. The atomic coordinates and structure factors ( rally. Cys-His coordination was seen in the Rhodovulum sulfidophilum SoxAX enzyme (heme 2) and in cystathionine ␤-synthase (14,15). Cys-Pro (terminal amine) coordination was also seen in the CO-sensing CooA (16). Thiolate-ligated hemoproteins have strong spectroscopic signals recognizable by, e.g., UV-visible absorption and EPR spectroscopy (17). However, identifying a ligand trans to thiolate, in absence of structural detail, can be difficult, and was the subject of detailed studies by Dawson and co-workers (18,19). Magnetic circular dichroism (MCD) in the near-IR region of ferric hemoproteins is also a sensitive technique for identifying heme ligands, using the position of charge-transfer features of high (CT 1 ) and low spin (CT LS ) heme to assign ligation (20).
In recent work, we generated the A264E mutant of P450 BM3 and demonstrated that it coordinated the heme iron to form a novel Glu-Fe-Cys ligand set. In the substrate-free enzyme, Glu-264 coordination to the heme iron was partial, with the side chain of Glu-264 positioned close to the side chain of the active site residue Phe-87 in a proportion of molecules (21). However, binding of fatty acid to the mutant resulted in displacement of the Glu-264 side chain from the Phe-87 vicinity and the complete formation of the new Glu-Fe-Cys ligand set (21,22). In this study, we sought to capitalize on earlier work to use Ala-264 mutants of BM3 to create further novel heme iron ligand sets for structural and mechanistic investigations. We selected A264K/H variants on the basis of natural occurrence of Lys and His coordination in several other hemoproteins. In addition, the different sizes of the Lys/His amino acid side chains posed the interesting possibility that coordination might take place in different conformers of the P450.
We present here structures and spectroscopic analysis of both novel P450 heme iron ligand sets. These data enable further development of a diagnostic spectroscopic method for thiolate-ligated hemes, whereby their characteristics can, in many cases, be defined (particularly using EPR and MCD) to facilitate recognition of such species as and when they occur elsewhere in nature. Importantly, structural analysis of these substrate-free A264K/H P450 enzymes indicates that they occupy distinct conformational states considered previously to be associated with the substrate-free (A264H) and fatty acidbound (A264K) forms of the wild-type (WT) P450 BM3. Moreover, we have obtained the structure of a novel substrate-free state for both WT and A264H in which there is reorganization of mobile (mainly helical) structural elements, revealing a more open active site cavity that may be primed for ingress of fatty acid substrates to the active site.

MATERIALS AND METHODS
Molecular Biology and Protein Production-A264K and A264H mutants were generated in both the heme domain and intact flavocytochrome P450 BM3 using the Stratagene QuikChange TM kit. Primers A264KF (CATTCTTAATTAAG-GGTCATGAAACAACAAGTGG) and A264KR (CCACTTG-TTGTTTCATGACCCTTAATTAAGAATG) were used to create the A264K mutant in both the heme domain (pBM20) and flavocytochrome plasmids (pBM25). A silent BspH1 site was created to allow verification of the mutant (indicated in bold), and the mutated codon is shown underlined. Primers A264HF (CATTCTTAATTCATGGACACGAAACAACAA-GTGG) and A264HR (CCACTTGTTGTTTCGTGTCCATG-AATTAAGAATG) were used to create the A264H mutant in plasmids pBM20 and pBM25 (6). The mutated codon is underlined. The entire genes were sequenced to ensure that the desired mutation was present and that no spurious mutations occurred. Wild-type and A264K/H heme domains, and their intact flavocytochromes, were expressed in Escherichia coli strain TG1 and purified as described previously (21).
Spectroscopy and Enzyme Analysis-UV-visible spectra were collected using a Cary UV-50 scanning spectrophotometer (Varian) with a 1-cm path length quartz cuvette. Spectra were recorded for oxidized, reduced (sodium dithionite), and substrate (fatty acid)-bound forms, as described in previous studies (e.g. Refs. 8 and 21). Steady-state kinetic analysis was carried out with WT, A264K, and A264H flavocytochromes as described previously (21), using laurate and arachidonate as substrates to monitor substrate-dependent NADPH oxidation. Reductasedependent cytochrome c reduction was also monitored for WT and mutants as described previously (8,23). Reduction of P450s and binding of carbon monoxide was performed as described previously (6). Heme concentrations for WT and A26K/H mutants were determined by the pyridine hemochromagen method (24). EPR spectra were recorded using an EPR spectrometer comprising an ER200D electromagnet and microwave bridge interfaced to a EMX control system (Bruker Spectrospin), and fitted with a liquid helium flow cryostat (ESR-9, Oxford Instruments) and a dual-mode X-band cavity (Bruker type ER4116DM). Spectra were recorded for WT (400 M), A264K (745 M), and A264H (435 M) heme domains in assay buffer at 10.8 K.
MCD spectra were recorded using JASCO J/810 and J/730 dichrographs in the near UV-visible and near-IR regions, respectively, using an Oxford Instrument superconducting solenoid with a 25-mm ambient bore to generate a magnetic field of 6 Tesla. A 0.1-cm path length cuvette was used to record near-IR spectra with sample concentrations the same as those used for EPR spectral collection. UV spectra were recorded for WT (30 M), A264K (64 M), and A264H (60 M) heme domains with 50 mM HEPES in deuterium oxide (pH*, 7.0) as buffer (where pH* is the apparent pH measured in D 2 O using a standard glass pH electrode).
Crystallography and Data Collection-The WT, A264K, and A264H heme domains were crystallized by the sitting drop method at 4°C. Drops were prepared by addition of 2 l of the mother liquor to 2 l of a 15 mg/ml protein solution in 10 mM Tris.HCl (pH 7.5). Wild-type and A264K crystals were obtained in 100 mM cacodylate (pH 6.0) containing 160 mM MgCl 2 and 16% polyethylene glycol 3350. Crystals of A264H were obtained in 100 mM cacodylate (pH 6.0) containing 130 mM MgCl 2 and 18% polyethylene glycol 3350. Crystals were flash-cooled in liquid nitrogen after brief soaking in mother liquor supplemented with 10% polyethylene glycol 200. Data were collected on a single crystal at the European Synchrotron Radiation Facility, Grenoble ID14.2 (for WT and A264K) and, in case of A264H, for a single crystal at Deutsches Elektronen Synchrotron (DESY) Hamburg, beamline X11. Data were reduced and scaled using DENZO and SCALEPACK (25). The WT and A264H structures were solved using the molecular replacement program AMoRe (26) with the WT structure as search model (PDB code 1BU7). The A264K structure was solved using the available A264E structure (PDB code 1SMI) as starting model. Atomic coordinates and B-factors were refined using the maximum likelihood based Refmac5 (27). Data collection statistics and final refinement parameters can be found in Table 1.

Expression and Purification of A264K/H Variants
The A264K and A264H mutants of P450 BM3 were constructed in both the isolated heme domain (residues 1-472) and in the full-length flavocytochrome, as described under "Materials and Methods." All constructs were expressed in E. coli to similar levels as the WT proteins and were purified in similar yield to WT BM3 and its heme domain, using standard methods (21). All proteins had full incorporation of heme cofactor, and the mutant flavocytochromes were replete with FAD and FMN.

Spectroscopic and Catalytic Properties of A264H and A264K Variants
UV-visible Spectroscopy-UV-visible absorption spectra of the A264K and A264H heme domain mutants were recorded for the oxidized, ligand-free enzymes and compared directly with spectra for the WT P450 BM3 heme domain. The Soret maximum for WT heme domain was at 418 nm, whereas that for the mutants was shifted to 424 nm (A264K) and 427 nm (A264H). The shorter wavelength (␦) band was of greater intensity in the mutants and had a more distinct peak shifted to ϳ362 nm from ϳ360 nm in the WT heme domain (Fig. 1). Spectral changes were also observed in the visible region (reduced intensity of the ␣ band and a red shift in the ␤ band in both mutants) ( Fig. 1, inset). The ␣/␤ bands for A264K(H) were at 571/542 nm (576/544 nm) compared with 569/535 nm for WT BM3. Heme band shifts were essentially identical for the A264K/H flavocytochromes (once changes in spectrum due to flavin contribution were accounted for), indicating that the reductase domain does not impact significantly on heme coordination in these enzymes. The shift of the Soret band to longer wavelengths (a type II P450 spectral shift) is consistent with the replacement of the distal water molecule (the sixth ligand to the heme iron) with a stronger field ligand. In light of preceding studies with the A264E enzyme, it appeared clear that the side chains of His-264 and Lys-264 coordinated to the heme iron in the A264H/K mutants, giving rise to distinctive perturbations of the optical spectrum (21). In the A264E mutant, Glu-264 partially occupies the distal coordination position on the heme iron in the substrate-free form, but complete occupancy was induced on binding of fatty acid substrates, as verified spectroscopically and by determination of atomic structures for substrate-free and palmitoleate-bound A264E heme domain (21,22). By contrast, addition of various fatty acid substrates (arachidonic acid, lauric acid, and palmitoleic acid) induced no significant spectral change in the A264K/H enzymes, suggesting that coordination of heme iron by the respective side chains was complete in the absence of fatty acids. In addition, negligible spectral perturbation was induced on addition of imidazole or substituted azoles.
In parallel studies of steady-state kinetics of NADPHdependent cytochrome c reduction and fatty acid oxygenation by the A264K/H flavocytochromes, there was no significant stimulation of NADPH oxidation induced by addition of fatty acids (and no hydroxylated products isolated), but levels of cytochrome c reductase activity were essentially identical to those of WT BM3. In addition, neither A264K/H was able to form any significant amount of ferrous-CO complex at either 450 nm or 420 nm on reduction with dithionite and bubbling with carbon monoxide. This suggests that the distal ligand remains firmly bound on reduction and that neither dioxygen (under turnover conditions) nor CO can displace the Lys/His-264 ligands. However, spectral changes observed on reduction suggest that there is some heterogeneity in the reduced species formed, with putative changes in both proximal and distal heme coordinations. The complex nature of these species is currently under investigation, but it is clear that the reduction potential of both mutants is more negative than the wild-type P450, due to incomplete reduction by dithionite.
EPR Spectroscopy-EPR spectra of both A264H/K ferric BM3 heme domains contain a single set of rhombic features and show shifts in g-values from the WT P450 (A264K: g z ϭ 2.47, g y ϭ 2.26, and g x ϭ 1.91; A264H: 2.50, 2.26, and 1.89; and WT: 2.42, 2.26, and 1.92) (Fig. 2). The homogeneity of the EPR spectra of the A264K/H mutants contrasts with the heterogeneity observed in the EPR spectrum of substrate-free A264E heme domain, in which signals from both Cys-Fe-H 2 O and Cys-Fe-Glu components were present (21). Thus, these data are consistent with the absence of UV-visible perturbations on fatty acid substrate binding to the oxidized A264K/H heme domains, and indicative that heme iron coordination by the Lys/His-264 ligands is complete for these mutants in both substrate-free and fatty acid-bound forms.
The g-values determined indicate a change in heme environment that is consistent with coordination of nitrogenous side chains at the distal position, as was inferred from UV-visible studies. Imidazole complexes of P450 BM3 (CYP102A1) and P450 cam (CYP101A1) have EPR spectra with similar g-values    (29).
MCD-Near IR (NIR) and UV-visible room temperature MCD spectra of A264K and A264H are shown in Fig. 3. In the NIR region, the position of a charge-transfer (CT) band can be diagnostic for the nature of heme iron ligand sets (20). The NIR MCD spectra are characteristic of ferric low spin heme iron in both heme domain mutants (Fig. 3A). Thiolate coordination is indicated by band position and characteristic low intensities by comparison with other low spin hemes (for example histidineligated heme proteins). For A264H, the NIR low spin chargetransfer (CT LS ) peak is at 1155 nm, shifted by Ͼ70 nm with respect to that for the WT enzyme. This peak suggests heme thiolate and nitrogenous ligation and is comparable to published examples (e.g. myoglobin plus HS Ϫ , 1200 nm; WT BM3 plus imidazole (at 4.2 K), 1180 nm; SoxAX heme 2, 1150 nm; ferric CooA (Pro/Cys Ϫ ), 1190 nm) (14, 29 -32). For A264K, the NIR CT LS band is at 1100 nm, shifted 20 nm from the WT value. This shift represents coordination of the K264 side chain to the thiolate-ligated heme iron. This, to our knowledge, is the first report of a cysteine/lysine MCD CT LS band (although a cysteine/amine ligation was observed for CooA (16)).
MCD spectra for the A264K/H heme domains were recorded in the near-UV-visible region and are shown in Fig. 3B, alongside the WT heme domain spectrum. The A264K and A264H mutants show clear shifts in band positions and relative intensity with respect to WT. The general spectral pattern for each mutant is again indicative of low spin ferric heme, consistent with UV-visible and EPR spectroscopy. The CT 2 band has shifted from 575 nm in WT heme domain to 583 nm and 587 nm, respectively, for the A264K and A264H mutants. In studies of ligand complexes of P450 cam (CYP101A1), the CT 2 band for unligated P450 was at 575 nm, for the 1-octylamine complex at 580 nm, and for the imidazole complex at 585 nm, similar to our values for the BM3 WT and A264K/H mutants (18). The general spectral patterns of both A264K/H mutants reveal the Soret derivative crossover at 425 nm, compared with 419 nm for WT BM3. The 425 nm crossover is indicative of nitrogen coordination to the heme iron. This is similar to the value (426 nm) for the imidazole complex of CYP101A1, with its octylamine complex at 421 nm, and that for unligated P450 at 416 nm (14,18). The feature is of greater intensity in A264K/H than in WT, with a peak-to-trough intensity of ϳ90 mM Ϫ1 cm Ϫ1 . Elsewhere in the visible MCD spectra, peaks for the WT BM3 (CYP101A1), A264K (CYP101A1-octylamine), and A264H (CYP101A1-imidazole) are located at 521 nm (518), 526 nm (526), and 533 nm (530), respectively (18). Thus, UV-visible MCD spectral features of CYP101A1 imidazole/amine complexes are consistent with distal coordination of BM3 heme iron by Lys-264/His-264 side chains. In general, the shifts in the MCD spectra of the mutants with respect to those of the WT P450 are greater in the His-264 mutant than the Lys-264 mutant, but show the same overall trend.

Crystal Structures of BM3 A264K/H Mutants
Spectroscopic analysis thus indicated that the A264H/K mutants adopted unprecedented cytochrome P450 heme iron ligand sets in solution state at both ambient and cryogenic tem-  peratures. Crystal structures were determined to ascertain structural properties of the mutants and to establish the heme iron coordination features.
Crystal structures were determined for the A264K/H heme domains (to 2.4 Å/1.9 Å; see Table 1 for data collection statistics and final refinement parameters) confirming that Lys/His-264 side chains coordinate the P450 heme iron (Fig. 4, A and B). In preceding studies of the WT BM3 heme domain, two distinct conformational states were seen, dependent on fatty acid substrate binding. Differences between substrate-free (SF conformation) and substrate-bound (SB conformation) structures are evident by I helix deformation (more kinked in SB conformation) and motion of F/G and B/C helices, leading to active site closure and substrate burial (33,34). For A264K, iron coordination (as shown in our previous studies with the A264E heme domain (22)) occurs with an equilibrium shift to the SB conformation. This is a consequence of configuration of the Lys-264 side chain required for optimal binding to the iron. The I helix is forced outward with respect to the iron, triggering the SF to SB conformational switch. In the A264E mutant the SB conformation is adopted independent of heme ligation by the glutamate side chain (21,22). By contrast, the A264H heme domain crystallizes in the SF conformation. We postulate this occurs because the SF conformation, rather than the SB conformation, satisfies the steric constraints of the imidazole side chain of His-264. Fig. 4C demonstrates structural differences in these new P450 species. The angle between planes of the heme and imidazole ring of His-264 is almost exactly 90°, and the imidazole is orientated at ϳ45°to the pyrrole nitrogens, minimizing unfavorable interactions. Iron coordination by His-264 is unlikely to occur in the SB conformation, because correct orientation of the imidazole moiety with respect to the heme plane would be dramatically impaired following I helix relocation. Thus, His-264 is likely displaced from the heme iron in the case that the SB conformation is adopted by this mutant. As discussed above, the A264H/K flavocytochromes were inactive, as fatty acid oxygenases and fatty acid substrates did not induce spectral shifts (unlike WT BM3). Thus, A264K/H enzymes are "locked" in their hexacoordinated states, with oxygen/substrate unable to displace nitrogenous ligands. Although the A264K/H mutants crystallize in distinct space groups and with different cell packing, both have two molecules in the asymmetric unit cell. In A264K, conformations of both monomers are essen-   (35). However, A264H molecule A, although clearly in a substrate-free conformation, adopts a markedly different organization of the mobile regions (mainly A, BЈ/C, and F/G helices) that has not been observed before. As detailed below, we have determined the structure of the WT BM3 heme domain in the same conformational state, revealing it is not a consequence of the A264H mutation.

Atomic Structure Reveals a Novel Conformational State of the BM3 Heme Domain
In parallel studies to those on the A264K/H mutants, we identified similar crystallization conditions for the WT SF heme domain that generated crystals in the same space group as obtained for the A264H mutant. The WT crystals diffracted to atomic resolution, and the structure revealed identical conformations to those observed for monomers in the A264H heme domain crystal structure. Thus, the novel substrate-free con-formation observed in monomer A is not confined to the A264H heme mutant. Instead, the structures reflect conformational freedom available to P450 BM3. This new conformation is hereafter termed SF2, and is shown overlaid with the preceding SB and SF structures (Fig.  5A) (34,35). The altered positions of the individual ␣-helices are markedly dissimilar to the SB conformation seen for substrate-free A264E and A264K mutant structures. The rearrangement from the SF1 to the SF2 conformation leads to a widening of the active site access channel (Fig. 5, B-D). Access to the active site is further enhanced by the high mobility of the F/G loop, which is disordered in the crystal structures, leading to further opening of the active site cavity that likely facilitates ingress of fatty acid substrate to the active site.
Despite marked overall differences between monomers 1 and 2 in the WT crystal structure (which display conformations SF1 and SF2, respectively), there are no significant differences in the immediate vicinity of the heme binding pocket and active site of the enzyme between the two states. The major structural rearrangements accompanying SF1 to SF2 transition occur at peripheral segments of the protein. However, there are some interesting differences observed between both WT SF1 and SF2 structures and the previously determined SF1 crystal structure (1BU7) with respect to heme conformation and coordination (35). In the final electron density maps for both monomers, there is evidence for a minor occupancy of a second heme conformation, in which the heme plane is flipped 180°. A similar observation was made for the atomic resolution structure of Mycobacterium tuberculosis CYP121, where the relative occupancy of conformers was determined ϳ70:30 (36). In contrast to CYP121, relative occupancy of the minor heme conformer in BM3 is at the detection threshold, likely accounting for Ͻ20% of the population.
Although the iron-sulfur (Fe-Cys) bond distance refines to a value similar to those previously reported (2.35 Å), the geometry of the water iron ligation is markedly different. Electron density clearly reveals the sixth ligand water molecule to be only ϳ50% occupied with an oxygen-iron distance of 2.20 Å. Strong electron density is also seen for a partially occupied second water molecule in a position close to, and mutually exclusive with, the sixth aqua ligand (Fig. 6A). The resulting picture reveals a water molecule that occupies either a position directly ligating to the sixth position on the heme iron, or else resides in a distinct position nearby, with relative occupancies of ϳ50:50. This is again similar to what was seen for the CYP121 structure, where additional density of the water molecule was interpreted as partially bound dioxygen. However, the distance between both atoms in the BM3 heme domain density is significantly larger than the interatomic distance observed for diatomic oxygen. Thus, an alternative model for CYP121 (as presented here for BM3 in light of this high resolution structural data) is dual occupancy of the sixth ligand water molecule at ratios of ϳ80:20 (ligated:unbound) for CYP121 and ϳ50:50 for P450 BM3. For both BM3 and CYP121, the non-ligating water is in close proximity to the heme plane, stabilized only by a few hydrogen bonds with neighboring residues, and presumably occupying the position of the non-ligating oxygen atom when dioxygen is bound. It is uncertain if the observed on/off water equilibrium in P450 crystals truly reflects solution behavior of the oxidized P450s, or is a consequence of x-ray photoreduction andconcomitantligationchanges.However,noclearx-raydosedependent effects on the electron density have been observed.
Slightly further from the heme iron, a small change in conformation and hydrogen bonding pattern of the I-helix is observed on comparing the present SF1/SF2 WT structures with previously determined SF1 structures (33,34). Regardless of overall conformation, in both monomers the backbone oxygen of Ile-263 is shifted outward from the ␣-helix and hydrogen bonds to a water molecule rather than the Glu-267 amide nitrogen, as was previously observed (Fig. 6B) (34). Although this is an additional deformation from the ideal ␣-helical geometry (in addition to lack of direct hydrogen bonding between the main chains of Ala-264 and Thr-268), it does not resemble the deformation seen upon rearrangement to the SB conformation.

DISCUSSION
The work presented here provides novel insights into structure and conformation of the Bacillus megaterium fatty acid hydroxylase P450 BM3 and identifies novel heme iron ligand sets in the A264K/H mutants. In our preceding studies of the A264E mutant of BM3, we demonstrated that the Glu-264 side chain coordinated the heme iron in the distal position in a substrate-dependent manner (i.e. coordination was complete in the fatty acid-bound form) (21). Here we show that the distal heme coordination also occurs in the A264H and A264K variants but that the structural and catalytic properties of these mutants are markedly different from those of the A264E mutant. Heme iron ligation is complete in substrate-free A264K/H heme domain enzymes and remains so on binding fatty acids, with Soret bands red-shifted to 424/427 nm, respectively. Similar types of optical shifts are observed on inhibitory coordination of P450 heme iron by amine and imidazole/ azole inhibitors (e.g. Refs. 18 and 37).
The reductase domain (in the flavocytochrome mutants) does not perturb the spectral properties, and these enzymes are not functional fatty acid hydroxylases (while retaining WT-level reductase domain-dependent activity). Thus, the properties of these enzymes contrast with the A264E enzyme, in which NADPH-dependent fatty acid hydroxylation was catalyzed, albeit at substantially reduced levels with respect to WT (21). In A264E, heme iron reduction and/or turnover-dependent conformational alterations enable displacement of the Glu-264 ligand in favor of dioxygen and enable catalysis. In the A264K/H mutants this does not occur to any significant extent, as evident from both the absence of fatty acid hydroxylase activity and the inability of the enzymes to form Fe 2ϩ ⅐CO complexes.
The stability of the distal ligands in the ferrous A264K/H enzyme forms is manifest also in the fact that a carbon monoxy complex cannot be formed to any significant extent. This prevents estimation of the A264K/H P450 concentration by the method of Omura and Sato (38). We instead compared protein concentrations of WT and A264K/H mutants using the pyridine hemochromagen assay, and this indicated that the extinction coefficients of the mutant heme domains were not considerably altered from that for WT (⑀ 424 ϭ 97.5 mM Ϫ1 cm Ϫ1 for A264K, compared with ⑀ 427 ϭ 98 mM Ϫ1 cm Ϫ1 for A264H and ⑀ 418 ϭ 95 mM Ϫ1 cm Ϫ1 for WT BM3 heme domain) (39).
Spectroscopic analysis of the A264K/H enzymes provides fingerprints for these novel P450 heme iron ligand sets. Although the BM3 A264H His-Fe-Cys ligand set has been observed previously in heme 2 of SoxAX, cystathionine ␤-synthase, and in a M60C variant of cytochrome c, this is the first example of this ligand set in a b-type hemoprotein (14,15,28). Moreover, the BM3 A264K Lys-Fe-Cys ligand set is the first example of this species in any hemoprotein. Ours is the first report for the NIR CT LS MCD band for this type of ligand set. In the CO-sensing transcriptional activator CooA, the axial ligands are provided by cysteine and an N-terminal proline residue; thus this protein is also coordinated by a (terminal) amine. In CooA a Cys-to-His ligand switch follows heme iron reduction, and this precedes replacement of the proline by CO as a distal ligand (40). In BM3 A264K reduction does not appear to lead to ligand displacement, because no CO adduct is formed. Comparative studies of the near UV-visible MCD features of the A264K/H mutants indicate similar spectral features to those obtained previously for the imidazole (cf. A264H) and octylamine (cf. A264K) complexes of CYP101A1 (18).
The atomic structures of the A264K/H heme domains revealed interesting alterations in protein conformation and led also to our identifying a new conformational form of P450 BM3. As with the A264E heme domain, A264K adopts a SB conformation as a result of structural constraints compatible with the Lys-Fe(III) bond (22,41). In any case, our recent studies suggest that the structural change between SF and SB forms reflects natural conformational freedom occurring in this P450 regardless of substrate binding and that the SB conformation may be a conformation that binds substrate tighter (i.e. has lower K d for fatty acids), rather than being a form induced by substrate binding per se (21,22). By contrast, the A264H heme domain adopts an SF conformation akin to that in the WT P450 (33). Further, the monomers in the asymmetric unit cell display distinct SF conformations, with SF1 being near-identical to that previously observed for the substrate-free WT P450, and SF2 being a novel form with altered positions of key secondary structural elements and more open access to the P450 active site. As confirmation that this new form was not unique to the A264H mutant, we isolated crystals of WT BM3 heme domain in the same space group and solved the atomic structure to show that the same SF2 conformation was adopted in one of the two monomers in these crystals. At this stage, it is unclear whether this novel conformation is a result of crystal packing forces distinct from previous crystal forms, or whether it represents a bona fide solution conformation. However, the SF2 structure reveals a more open entrance to the substrate binding site as a result of alterations in position of the F/G loop and other mobile regions. Therefore, this particular conformational state of the P450 appears primed for substrate entry and argues in favor of multiple quasi-stable conformers that may exist in solution. The high resolution (1.2 Å) structure revealed conformational heterogeneity of the heme macrocycle (bound in two conformations related by a 180°rotation about the axis of symmetry across the CH ␣ -Fe-CH ␦ atoms of the molecule) in both the SF1 and SF2 forms, as was observed previously for CYP121 (36). In eukaryotic CYP4 family P450s (fatty acid -hydroxylases with evolutionary links to BM3), a natural, turnover-dependent cross-linking of methyl group(s) to a conserved glutamate residue was observed (42). Stabilization of heme binding afforded by covalent linkage of heme to protein has drawn interest with respect to inducing this process in relevant point mutants of biotechnologically important P450s, such as BM3 and P450 cam (CYP101A1) (21,30). However, heterogeneity of heme orientation (seen in mammalian CYPs 4A1 and 4A3) may occur commonly in recombinant P450s (42). Dual positions of the heme macrocycle will place reactive methyl group(s) in dif-ferent environments, making complete cross-linking unlikely without further engineering to enable linkage of the heme bound in the second conformation, or may block linkage in the second conformation altogether.
The new high resolution WT structure also reveals partial occupancy of the distal position by a water molecule, with a second water molecule bound adjacent and in a mutually exclusive position to the coordinating water. This possibly reflects an equilibrium whereby the distal water is more mobile (i.e. can readily dissociate from the iron in absence of substrate), and the novel position of the second water molecule is distinct from those previously reported in the N-palmitoylglycine-bound heme domain structure (34).
Preceding attempts to crystallize the full-length P450 BM3 have been unsuccessful. This is a consequence of its complex, multidomain nature, its dimeric state, and conformational heterogeneity in both reductase and heme domains (e.g. Refs. 10, 22, 33, and 41). Here, we have produced heme domains "locked" into SB (A264K) or SF (A264H) conformations, thus reducing the conformational heterogeneity associated with the heme domain. Recently, we have also solved the structure of the BM3 FAD/NADP(H)-binding domain, achieved by engineering out a solvent-exposed cysteine (Cys-773) that promotes nonspecific disulfide bridge formation hindering crystallization efforts. 5 Thus, we have created mutant forms of P450 BM3 domains in which structural heterogeneity is minimized, and generation of similar mutant versions for the full-length protein could lead to successful structure determination of this multidomain enzyme.
In conclusion, we report generation and structural/spectroscopic characterization of two novel BM3 heme ligation state mutants. The Cys-Fe-His ligand set is seen naturally in only two non-P450s (14,15). The Cys-Fe-Lys ligand set is unprecedented. Spectroscopic studies provide fingerprints for these species. MCD studies provide important reference spectra (including a distinctive NIR CT band) that characterize these species and that could be important in identifying similarly ligated hemoproteins elsewhere in nature. Structural constraints dictate that A264H and A264K occupy different conformations, similar to those seen in crystals of substrate-free and palmitoleate-bound forms of WT heme domain (33,41). These locked forms will be useful in studies to structurally resolve the flavocytochrome, which in turn will be important for rationalizing molecular interactions between P450s and NADPH-cytochrome P450 reductase enzymes.