Structure and Biochemical Properties of the Alkene Producing Cytochrome P450 OleTJE (CYP152L1) from the Jeotgalicoccus sp. 8456 Bacterium*

Background: OleTJE oxidatively decarboxylates fatty acids to produce terminal alkenes. Results: OleTJE is an efficient peroxide-dependent lipid decarboxylase, with high affinity substrate binding and the capacity to be resolubilized from precipitate in an active form. Conclusion: OleTJE has key differences in active site structure and substrate binding/mechanistic properties from related CYP152 hydroxylases. Significance: OleTJE is an efficient and robust biocatalyst with applications in biofuel production. The production of hydrocarbons in nature has been documented for only a limited set of organisms, with many of the molecular components underpinning these processes only recently identified. There is an obvious scope for application of these catalysts and engineered variants thereof in the future production of biofuels. Here we present biochemical characterization and crystal structures of a cytochrome P450 fatty acid peroxygenase: the terminal alkene forming OleTJE (CYP152L1) from Jeotgalicoccus sp. 8456. OleTJE is stabilized at high ionic strength, but aggregation and precipitation of OleTJE in low salt buffer can be turned to advantage for purification, because resolubilized OleTJE is fully active and extensively dissociated from lipids. OleTJE binds avidly to a range of long chain fatty acids, and structures of both ligand-free and arachidic acid-bound OleTJE reveal that the P450 active site is preformed for fatty acid binding. OleTJE heme iron has an unusually positive redox potential (−103 mV versus normal hydrogen electrode), which is not significantly affected by substrate binding, despite extensive conversion of the heme iron to a high spin ferric state. Terminal alkenes are produced from a range of saturated fatty acids (C12–C20), and stopped-flow spectroscopy indicates a rapid reaction between peroxide and fatty acid-bound OleTJE (167 s−1 at 200 μm H2O2). Surprisingly, the active site is highly similar in structure to the related P450BSβ, which catalyzes hydroxylation of fatty acids as opposed to decarboxylation. Our data provide new insights into structural and mechanistic properties of a robust P450 with potential industrial applications.

The cytochromes P450 (P450s 2 or CYPs) are oxidases that catalyze a vast array of oxidative reactions in nature (1). These hemoproteins are found in virtually all organisms, from bacteria and archaea to humans, and are responsible for several chemical transformations that are essential, for instance, in the microbial biosynthesis of antibiotics (e.g. erythromycin in Saccharopolyspora erythraea and vancomycin in Amycolatopsis orientalis) (2,3) and in the mammalian formation of estrogens (estrone and 17␤-estradiol) through the action of the aromatase P450 (CYP19A1) on androgen substrates (androstanedione and testosterone, respectively) (4,5). The majority of characterized P450s are monooxygenases that interact with one or more redox partners to provide them with the two electrons (typically derived from NAD(P)H) required for oxidative catalysis (6). The first electron reduces the P450 cysteine thiolatecoordinated heme iron from ferric to ferrous, enabling dioxygen binding to the ferrous iron. The second electron reduces the resulting ferric-superoxo complex to the ferric-peroxo state. Two successive protonations produce first the ferric-hydroperoxo species (compound 0) and then (following the loss of a water molecule) the ferryl-oxo compound I (7) (Fig. 1). The transient and highly reactive nature of compound I prevented its definitive characterization for many years, until Rittle and Green (8) produced compound I in a large yield following rapid mixing of CYP119 (from the thermoacidophilic crenarchaeon Sulfolobus acidocaldarius) with the oxidant m-chloroperbenzoic acid and confirmed its identity using Mössbauer, EPR, and UV-visible spectroscopy. Compound I is considered to be the major oxidizing species in the P450 catalytic cycle and to be responsible for the bulk of oxidative reactions (e.g. hydroxyla-* This work was supported by a United Kingdom Biotechnology and Biolog- tion, epoxidation, oxidative demethylation, etc.) observed throughout the P450 superfamily (9,10). The vast majority of P450s use NAD(P)H-dependent redox systems consisting of either (i) an FAD-binding reductase that shuttles electrons to the P450 via a ferredoxin (or a flavodoxin in a small number of cases) or (ii) an FAD-and FMN-binding cytochrome P450 reductase, the individual flavin-binding domains of which are evolutionarily related to NAD(P)H-binding ferredoxin oxidoreductases and flavodoxins (5,11). However, other types of P450 redox partner systems exist (e.g. P450redox partner fusion enzymes, such as the CYP116B family of P450:phthalate dioxygenase reductase fusions) (12). In addition, other P450s catalyze isomerization (e.g. mammalian thromboxane synthase, CYP5A1) and dehydration (e.g. flax allene oxide synthase, CYP74A1) reactions that do not require an external source of electrons and which are completed entirely within the P450 active site (13,14). Further, through exploration of in vitro routes to driving P450 catalysis, it is now well established that the addition of hydrogen peroxide (H 2 O 2 ) or organic peroxides (e.g. cumene hydroperoxide) to P450s can facilitate substrate oxidation by directly producing compound 0, which is then protonated to generate compound I (15) (Fig.  1). This "peroxide shunt" procedure is rarely an efficient means of driving P450s, because the peroxides oxidize heme and protein. However, a small number of P450s that have evolved to exploit the peroxide shunt are now known. Notably, the Bacillus subtilis CYP152A1 (P450 BS␤ ) and the Sphingomonas paucimobilis CYP152B1 (P450 SP␣ ) naturally use H 2 O 2 to catalyze long chain fatty acid hydroxylation and are thus referred to as peroxygenases (16,17). P450 SP␣ catalyzes nearly exclusively hydroxylation at the ␣-position, whereas P450 BS␤ catalyzes hydroxylation at ␣and ␤-positions but with the majority at the ␤-position (ϳ60:40 ratio) (16).
In recent studies, Rude et al. (18) characterized a novel enzyme from the bacterium Jeotgalicoccus sp. ATCC 8456 (OleT JE ) that is 41% identical in amino sequence to P450 BS␤ and 37% identical to P450 SP␣ . OleT JE was identified as a P450 based on this sequence similarity and designated by the authors as a CYP152 P450 family member (18). The Jeotgalicoccus ATCC 8456 host strain was shown to produce a number of C18 -C20 linear and branched chain terminal alkenes, and other Jeotgalicoccus strains were shown to generate a similar spectrum of terminal alkenes in the C18 -C21 range. A His-tagged version of OleT JE was expressed in Escherichia coli and purified using nickel-nitrilotriacetic acid column chromatography and shown to catalyze formation of n-1 alkenes through H 2 O 2 -dependent decarboxylation of C14, C16, C18, and C20 saturated fatty acids (18).
In view of the potential importance of the OleT JE enzyme as a producer of terminal alkenes for exploitation in areas such as biofuels and fine chemical production, we have undertaken a study of the biochemical and biophysical properties of the isolated OleT JE (CYP152L1) enzyme and have determined its crystal structure in complex with arachidic acid. These data reveal novel properties of this biotechnologically important P450 peroxygenase. These include (i) the high catalytic efficiency of OleT JE and its capacity to be resolubilized from a precipitated form as a fully active enzyme, (ii) the extensive development of high spin (HS) heme iron in OleT JE on binding various long chain fatty acids (distinguishing it from related bacterial peroxygenases), and (iii) its unusually positive heme iron reduction potential, which is also negligibly affected by fatty acid binding despite the substrate inducing extensive HS ferric heme iron formation.

EXPERIMENTAL PROCEDURES
Bioinformatics-The OleT JE sequence and additional members of the CYP152 family, including all known subfamilies, were BLAST-searched against a set of all the prokaryotic P450 sequences. Members of the highest scoring CYP families from these searches were used to build a tree. Sequence alignments were computed using ClustalW and checked manually for consistent alignment of known CYP motifs. Neighbor-joining trees were generated with the Phylip package (version 3.6; distributed by the author, J. Felsenstein (University of Washington, Seattle)) using ProtDist (a program in Phylip) to compute difference matrices. Trees were drawn and colored with FigTree FIGURE 1. The cytochrome P450 catalytic cycle. The reaction sequence starts at the top, with the P450 "resting" state having a ferric (Fe 3ϩ ) 6-coordinate low spin heme iron axially coordinated by cysteine thiolate (S) and a weakly bound water molecule (H 2 O). Binding of substrate (RH) displaces the water ligand to leave a 5-coordinate high spin heme iron. This is reduced by a redox partner, and the ferrous iron then binds dioxygen to form a ferricsuperoxo complex. A further single electron reduction by the redox partner generates the ferric-peroxo intermediate, which is protonated to form the transient ferric-hydroperoxo (compound 0) species. Compound 0 is further protonated and dehydrated to form the ferryl-oxo porphyrin radical cation species compound I. The compound I is considered to be the major oxidant in P450 reactions and abstracts a hydrogen from RH to produce a substrate radical, prior to "rebounding" the hydroxyl to the substrate to form hydroxylated product (ROH) and to restore the resting state of the P450 (8,55). The double-headed arrow crossing the cycle between the substrate-bound ferric P450 and compound 0 describes the catalytic mechanism for OleT JE and related peroxygenase P450s (e.g. P450 BS␤ and P450 SP␣ ) (16 -18). Direct interaction of OleT JE with H 2 O 2 produces a reactive iron-oxo species (compound 0) that is further protonated and dehydrated to form compound I, leading to either fatty acid substrate decarboxylation (major route) or fatty acid hydroxylation (minor route), as described under "Discussion" section and in the legend to Fig. 13. version 1.3.1 (available on the World Wide Web) and labeled in Adobe Illustrator CS version 11.0.0 (Adobe Systems Inc.). Sequences producing long branches on the tree were removed, and the tree was recomputed.
Expression and Purification of OleT-The gene encoding OleT JE from Jeotgalicoccus sp. ATCC 8456 was codon-optimized (for expression in E. coli), synthesized, and cloned into the pET47b (Merck Millipore, Madison, WI) vector by GenScript. The E. coli strain C41 (DE3) (Lucigen, Middleton, WI) was used as the expression host. Cells transformed with the pET47b-OleT JE plasmid were grown at 37°C with shaking at 200 rpm in total volumes of 500 ml to 3 liters of 2YT broth containing kanamycin (30 g/ml) supplemented with 500 M ␦-aminolevulinic acid. Expression of OleT JE was induced by the addition of 100 M isopropyl 1-thio-␤-D-galactopyranoside when an optical density of 0.5 (at 600 nm) was reached, at which point the incubation temperature was lowered to 25°C, and the cells were grown for a further 16 h. Cells were harvested by centrifugation at 6000 rpm, 4°C, using a JLA-8.1000 rotor in an Avanti J-26 XP centrifuge. Pellets were resuspended in a minimal volume of ice-cold buffer A (100 mM potassium phosphate (KP i ), pH 8.0), combined, and centrifuged as before. The cell pellet was then frozen at Ϫ80°C until required.
Cells were thawed at 4°C and resuspended in 3 volumes of extraction buffer/g of cell pellet. The extraction buffer consisted of buffer A containing 1 M NaCl, 20% glycerol, with a Complete TM EDTA-free protease inhibitor mixture tablet (Roche Applied Science) per 50 ml of cell suspension, DNase I (100 g/ml, bovine pancreas, Sigma-Aldrich), and lysozyme (100 g/ml, hen egg white, Sigma-Aldrich). The cells were disrupted by two passes through a French press (Thermo Scientific, Hemel Hempstead, UK), and the homogenate was centrifuged at 20,000 rpm, 4°C, for 90 min using a JA-25.50 rotor. Alternatively, cells were lysed by sonication using a Bandelin Sonopuls sonicator set to 45% amplitude with 30 30-s pulses, at 60-s intervals, with the cell suspension kept on ice throughout. The homogenate was then centrifuged as previously. The supernatant was removed, and the pH was reset to 8.0 as necessary. The sample was then incubated overnight with 10 ml (per 100 g of cell pellet) of nickel-iminodiacetic acid (Ni-IDA) chromatographic medium (Generon, Maidenhead, UK) on a rolling table at 4°C. The mixture was then poured into a column, and the collected bed of OleT JE -bound medium was washed with 10 column volumes (CV) of 100 mM KP i (pH 8.0) containing 750 mM NaCl, 20% glycerol (buffer B), and 50 mM imidazole to remove weakly bound contaminants. The column was then washed with 2 CV of the buffer B containing 125 mM imidazole, followed by 5 CV of buffer B plus 150 mM imidazole, which eluted the bulk of the OleT JE protein. The partially purified OleT JE sample was dialyzed overnight against 15 liters of buffer A at 4°C, which caused OleT JE to precipitate. Post-dialysis, precipitated protein was isolated by centrifugation at 4000 rpm, 4°C, using an A-4-62 rotor in an Eppendorf 5810 R centrifuge. The pellet was washed gently with 50 ml of buffer A, and centrifugation was repeated. OleT JE was resuspended in 5 ml of buffer A containing 1 M NaCl and 10% glycerol, which produced OleT JE at high purity (Method 1). For OleT JE destined for crystallographic studies, HRV 3C protease (Merck Millipore) was incubated with OleT JE for ϳ16 h at 4°C (50:1 g of protein/unit of protease) to remove the N-terminal polyhistidine tag. The proteolysed protein was applied to 5 ml of pre-equilibrated nickel-Sepharose resin (GE Healthcare) to bind the cleaved His tag and the tagged HRV 3C. The cleaved OleT JE was eluted from the column by washing with 100 mM KP i (pH 8.0) plus 750 mM NaCl and 10% glycerol (buffer C).
In separate preparations (avoiding the OleT JE precipitation step; Method 2), the dialysis step post-Ni-IDA chromatography was removed, and the OleT JE eluate was instead diluted (5ϫ) in buffer C and concentrated in an Amicon ultrafiltration device. The OleT JE sample was then centrifuged to clarify the sample (16,000 rpm, 4°C, using the JA-25.50 rotor), and the supernatant was then applied again to a 5-ml Ni-IDA column. The column was washed with 5 CV of buffer C containing 50 mM imidazole and then 10 CV of buffer C plus 100 mM imidazole. The His-tagged OleT JE was then eluted with 150 mM imidazole in the same buffer. All procedures generated highly purified OleT JE protein. In both cases, the pure OleT JE protein was concentrated to Ͼ20 mg/ml using a Vivaspin centrifugal concentrator (Generon), snap-frozen in liquid nitrogen, and stored at Ϫ80°C.
UV-visible Spectroscopy-Analysis of the UV-visible spectroscopic properties of OleT JE was done on a Cary 60 UV-visible spectrophotometer (Varian UK). Spectra were recorded using ϳ4 -10 M OleT JE in 100 mM KP i (pH 8.0) plus 750 mM NaCl (buffer D). Reduction of OleT JE was achieved by the addition of sodium dithionite to enzyme in buffer D made anaerobic by extensive bubbling with oxygen-free nitrogen. The ferrous-CO complex of OleT JE was formed by slow bubbling of gas into anaerobically reduced enzyme until no further absorbance change occurred. The NO complex was formed by the addition of 5-8 bubbles of NO into a sample of ferric OleT JE in anaerobic buffer.
Fatty Acid and Inhibitor Binding Titrations with OleT JE -Spectral binding titrations of OleT JE with saturated fatty acids (C12, C14, C16, C18, and C20) were performed at 25°C in buffer D. Fatty acids were from Sigma-Aldrich. Substrates (typically 0.25 mg/ml) were dissolved in 70% (v/v) EtOH (for C18 and C20) or 70% methanol (with the sodium salts of C12, C14, and C16 fatty acids) and 30% (v/v) Triton X-100 (Sigma-Aldrich). A parallel set of binding titrations was also performed using fatty acids (1 mg/ml) dissolved in 100% EtOH or methanol without Triton X-100. Prior to titrations, OleT JE samples (ϳ50 l at Ͼ20 mg/ml) in buffer C were passed through a Lipidx column of dimensions 5 ϫ 1 cm (PerkinElmer Life Sciences) in order to remove any residual lipid retained during purification of the protein from E. coli. OleT JE recovered from the column was in an extensively low spin (LS) ferric state and was used directly for titration at a final P450 concentration in the range from 5 to 10 M. Titrations were performed by stepwise additions of aliquots (0.1-1 l) of the fatty acids to the OleT JE sample (substrate additions to Ͻ1% of total volume). Spectra (800 -300 nm) were recorded for the ligand-free OleT JE and following each addition of substrate using a Cary 60 UV-visible spectrophotometer. Difference spectra at each stage in the titration were computed by subtracting the spectrum of ligandfree OleT JE from each successive fatty acid-bound spectrum collected during the titration. A pair of wavelengths were identified that defined the absorbance maximum (A peak ) and minimum (A trough ) in the difference spectra from each titration set. The overall absorbance change (A max ) at each substrate concentration point was calculated as A peak minus A trough , and A max was plotted versus [substrate]. These data were fitted using either a hyperbolic (Michaelis-Menten) function, the Morrison equation for tight binding ligands, or the Hill function (where sigmoidal behavior was observed) in order to determine dissociation constants (K d values), as described previously (19,20). Titrations and data fitting for OleT JE with dithiothreitol (DTT), imidazole, and cyanide (sodium salt) inhibitors were done in the same way as for the fatty acids, with ligands dissolved in buffer D.
Stopped-flow Analysis of Substrate Turnover-Stopped-flow absorption measurements were made using an Applied Photophysics SX18 MR stopped-flow spectrophotometer (Leatherhead, UK). Stopped-flow spectral accumulation was done using a photodiode array (PDA) detector on the same instrument. Fatty acid substrate-bound OleT JE was mixed versus different concentrations of H 2 O 2 in 100 mM KP i (pH 8.0) containing 750 mM NaCl at 25°C. OleT JE (9.2 M) was converted to an extensively HS heme iron form by mixing with arachidic acid (12 M) from a concentrated stock prepared in 80% EtOH, 20% Triton X-100. Reactions were initiated by mixing the arachidic acid bound OleT JE (4.6 M final concentration) with H 2 O 2 (3.29 -200 M final concentration). Stopped-flow traces at single wavelengths reporting on the conversion of HS OleT JE heme iron toward LS were collected over periods of up to 30 s to follow depletion of HS (390 nm) and formation of LS heme iron (418 nm). Data were analyzed and fitted using a single exponential function with the Pro-Data SX software suite (Applied Photophysics). The observed reaction rate constants (k obs values) were plotted versus the relevant H 2 O 2 concentrations, and the resultant data plot was fitted using a linear function to obtain the second order rate constant reporting on H 2 O 2 -dependent decarboxylation of substrate and the consequent heme iron spin state conversion. Entire spectral acquisition (750 -280 nm) was also done using the PDA detector for the same set of stopped-flow reactions analyzed in single wavelength mode.
Redox Potentiometry-To determine the midpoint potential for the OleT JE Fe 3ϩ /Fe 2ϩ couple, redox titrations were performed at 25°C in an anaerobic glove box (Belle Technology, Weymouth, UK) under a nitrogen atmosphere with O 2 levels maintained at less than 2 ppm. All solutions were deoxygenated by sparging with nitrogen gas. For substrate-free OleT JE , the titration was done using 9.3 M OleT JE in 100 mM KP i (pH 7.0) plus 10% glycerol. For substrate-bound OleT JE , the titration was done under the same conditions, following the addition of arachidic acid (from a 32 mM stock in 80% EtOH, 20% Triton X-100) until no further conversion of the heme iron to the HS heme state was observed (ϳ12 M arachidic acid). Mediators were added to expedite electronic equilibration in the system (2 M phenazine methosulfate, 7 M 2-hydroxy-1,4-naphthoquinone, 0.3 M methyl viologen, and 1 M benzyl viologen to mediate in the range from ϩ100 to Ϫ480 mV versus normal hydrogen electrode (NHE)) and data fitting (using the Nernst equation), and analysis was done as described previously (21)(22)(23).
EPR Analysis of OleT JE -Continuous wave X-band electron paramagnetic resonance EPR spectra of OleT JE were obtained at 10 K using a Bruker ELEXSYS E500 EPR spectrometer equipped with an ER 4122SHQ Super High Q cavity. Temperature control was effected using an Oxford Instruments ESR900 cryostat connected to an ITC 503 temperature controller. Microwave power was 0.5 milliwatt, modulation frequency was 100 kHz, and the modulation amplitude was 5 G. EPR spectra were collected for OleT JE (305 M) in the substrate-free form and for OleT JE (205 M) bound to arachidic (C20:0) acid (at a saturating concentration).
Crystallography For preparation of substrate-bound OleT JE , P450 samples were concentrated by ultrafiltration, and a stock solution of arachidic acid (32 mM) dissolved in 100% EtOH was added to a final concentration of 235 M. The concentration of EtOH did not exceed 1% of the total volume. The mother liquor was supplemented with 10% PEG 200, where an additional cryoprotectant was required, and crystals were flash-cooled in liquid nitrogen prior to data collection. Data were collected at Diamond synchrotron beamline IO4-1 and reduced and scaled using XDS (24). Structures were solved by molecular replacement with the previously solved P450 BS ␤ crystal structure (PDB 2ZQJ) using PHASER (25). Structures were refined using Refmac5 (25) and Coot (26). Final refinement statistics are given in Table 1.

Analysis of Products Formed by OleT JE in Reactions with H 2 O 2 and Fatty
Acids-OleT JE reactions with long chain saturated fatty acids (C12-C20) were set up as follows. 5-ml reactions were done in buffer D, with 250 M dodecanoic acid (sodium salt), palmitic acid or arachidic acid, 500 M hydrogen peroxide, and 0.6 M OleT JE . The final reaction mixtures were incubated for periods up to 30 min at room temperature. 1 ml of the reaction mixture was then extracted (at different reaction times) with an equal volume of HPLC grade heptane, and the sample was centrifuged at 14,000 rpm for 20 min. The top layer was then analyzed by GC/MS. Analysis was done using a Thermo Fisher DSQ II GC/MS instrument with a 30 m ϫ 0.25 mm ϫ 0.25-m ZB5MS GC column (Phenomenex). Injection was cold on-column. The oven program was set so that an initial temperature of 50°C was ramped at 10°C/min to 300°C post-injection. Electronic ionization was used, and ions in the range of 40 -640 m/z were scanned at two scans/s.

RESULTS
Classification of OleT JE as CYP152L1-There are currently 21,039 named cytochrome P450 sequences (see the Cytochrome P450 Homepage). Approximately 6% are bacterial (1254 sequences), and an additional 48 are from archaea. Initial BLAST searches with OleT JE showed that it was less than 40% identical to most known CYP152 sequences and barely over the 40% recommended cut-off for CYP family membership to two CYP152 sequences (having 41% identity with CYP152A1 from B. subtilis and 40% with CYP152A2 from Clostridium acetobutylicum). The location of the OleT JE sequence in a phylogenetic tree (as CYP152L1) strongly argues for inclusion in the distinct CYP152 clade. The same logic applies to the renamed CYP152M1 from Enterococcus faecium, which has a long branch in the tree. This sequence was previously named CYP241A1, but that nomenclature has been changed based on its inclusion within the CYP152 clade. A second sequence, CYP152L2 from Staphylococcus massiliensis S46, is 64% identical to CYP152L1 (Fig. 2).
Expression and Purification of OleT JE -The OleT JE gene was codon-optimized for expression in E. coli, and preliminary studies revealed that the enzyme was expressed well in a number of E. coli strains. The C41 (DE3) strain (Lucigen) was selected for protein production with the gene cloned into pET47b via the BamHI and EcoRI restriction sites with a 6-His N-terminal tag and transcribed using the T7-lac RNA polymerase/promoter system. Expression cell extracts were red in color, indicative of the production of a heme protein. However, our initial studies revealed that the OleT JE protein precipitated on dialysis following elution from an Ni-IDA protein in the first chromatographic purification step. Previous studies by Rude et al. (18) used high salt (NaCl) concentration in several purification buffers, and in view of this and the halophilic nature of the host bacterium (Jeotgalicoccus sp. ATCC 8456), we considered that the protein might be stabilized in solution at high ionic strength. This proved to be the case, and it was found that the precipitation of OleT JE could be used to advantage, because resolubilization of the centrifuged protein pellet in buffer A containing 1 M NaCl and 10% glycerol produced an OleT JE sample with a P450-like heme spectrum (A max at ϳ418 nm). SDS-PAGE at this stage also indicated the protein to be extensively purified (purification Method 1). Specifically for crystalliza-tion, the OleT JE His tag was removed by incubation with HRV 3C protease, and the mixture was loaded onto a nickel-Sepharose column. Washing the column in buffer C (100 mM KP i (pH 8.0) plus 750 mM NaCl and 10% glycerol) resulted in elution of a highly purified tag-free OleT JE protein (Fig. 3) and the retention of the cleaved His tag and the tagged protease on the column.
Having identified the issues with the propensity of OleT JE to aggregate at low ionic strength, an alternative strategy was developed to avoid its precipitation, by eluting OleT JE from Ni-IDA in the high salt buffer C, centrifuging the sample, and then reapplying to Ni-IDA resin equilibrated in buffer C. By washing the column with increasing concentrations of imidazole in buffer C, His-tagged OleT JE was eluted at 150 mM imidazole in a highly pure form (purification Method 2). A typical yield of purified OleT JE was ϳ20 mg/liter of E. coli cell culture using either Method 1 or Method 2 for protein purification.
UV-visible Absorption Properties of OleT JE -Rude et al. (18) inferred the cytochrome P450 nature of OleT JE from amino acid sequence similarities to peroxygenase members of the CYP152 family of P450s and demonstrated in vitro that cell extracts of Jeogalicoccus sp. ATCC 8456 could decarboxylate the saturated fatty acids arachidic acid (C20) and stearic acid (C18) to their respective n-1 terminal alkenes (1-nonadecene and 1-heptadecene, respectively). A His-tagged OleT JE isolated from E. coli was also shown to catalyze stearic acid decarboxylation in an H 2 O 2 -dependent reaction (18). However, UV-visible absorption features typical of a P450 enzyme were not presented in this earlier study. Fig. 4 shows characteristic absorption spectra for pure OleT JE in its oxidized (ferric, Fe 3ϩ ) and sodium dithionite-reduced (ferrous, Fe 2ϩ ) forms and for the ferrous-carbon monoxide (Fe 2ϩ -CO) and ferric-nitric oxide (Fe 3ϩ -NO) species. The resting (ferric) form of OleT JE shows a heme spectrum typical of a P450 enzyme with its ferric heme iron in an LS state. The major absorption feature (the Soret band) is at 418 nm, with the smaller ␣and ␤-bands in the visible region at ϳ566 and 535 nm, respectively. These values are similar to those of other LS bacterial P450s (e.g. the Bacillus megaterium P450 BM3 (CYP102A1) heme domain with maxima at 418, 534, and 568 nm and the Mycobacterium tuberculosis CYP121A1 at 416.5, 538, and 568 nm) (27,28). The two methods of preparing OleT JE (i.e. with or without a protein precipitation step) produced identical oxidized OleT JE spectra. Any residual imida- zole ligand from nickel column chromatography (in both cases) was extensively depleted by ultrafiltration used to concentrate the proteins and thus did not produce any imidazole-ligated OleT JE heme iron. Reduction of OleT JE with sodium dithionite produced a ferrous hemoprotein with the Soret band diminished in intensity and shifted to 414 nm. In the visible (heme Q-band) region, a single, slightly asymmetric feature is seen at ϳ540 nm. The blue shift of the Soret spectrum on reduction indicates substantial retention of cysteine thiolate proximal coordination in the OleT JE ferrous state, and the spectral maxima are similar to those features seen in, for example, the well characterized Pseudomonas putida camphor hydroxylase P450cam (CYP101A1, 411 and 540 nm) and in the explosive degrading P450 XplA from Rhodococcus rhodochrous strain 11Y (CYP177A1, 408 and 542 nm) (23,29). The addition of carbon monoxide to anaerobically reduced OleT JE produced a characteristic P450 heme spectrum with the Soret band red-shifted to 449 nm and a Q-band feature at 551 nm. A small shoulder on the Soret feature at ϳ423 nm probably indicates a minor proportion (ϳ5%) of the P420 (probably cysteine thiol-coordinated) form of the OleT JE Fe 2ϩ -CO complex. The NO-bound ferric OleT JE spectrum is also typical of other P450-NO adducts, with an asymmetric Soret feature (ϳ427 nm) and distinctive, enhanced intensity ␣and ␤-bands at ϳ573 and 540 nm (30). Using the method of Berry and Trumpower (31), an extinction coefficient of ⑀ 418 ϭ 91.5 mM Ϫ1 cm Ϫ1 was established for the LS ferric form of OleT JE (31,32).

Analysis of Substrate and Inhibitor
Binding to OleT JE -The binding of substrates to P450s is often associated with alteration of the spin state of their ferric heme iron, usually through displacing its weakly bound sixth ligand water molecule and inducing a shift toward the HS form (e.g. see Refs. 33 and 34). For OleT JE , we investigated the binding of a series of saturated fatty acids (C12-C20) and found that in all cases, the lipids induced a LS to HS transition, with the Soret band shifting from 418 nm toward 394 nm. Table 2 shows fatty acid binding K d data for OleT JE (purified using Method 1) and using fatty acid stocks dissolved in alcohol or in alcohol containing 30% (v/v) Triton X-100 (see "Experimental Procedures"). The K d values and the extent of HS heme iron developed were improved in all cases in the presence of the detergent, although Triton X-100 alone induces no spin state change (e.g. 0.67 Ϯ 0.03 M versus 6.20 Ϯ 0.26 M for palmitic acid). The extent of spin state change induced varied according to chain length, with the longer chain fatty acids (C18:0 and C20:0) inducing a more complete conversion to the HS ferric state than observed for the C12:0 to C16:0 fatty acids). For a titration using an arachidic acid (C20:0) stock, including Triton X-100, the HS conversion was almost complete (estimated at Ն95%), as shown in Fig. 5A. In contrast, lauric acid (C12:0) produced ϳ52% HS at saturation (Fig. 5B). For studies with non-precipitated OleT JE (prepared using Method 2) in the presence of Triton X-100, tight binding of fatty acids was again observed (e.g. K d values of 1.54 Ϯ 0.19 M for arachidic acid and 12.7 Ϯ 0.3 M for lauric acid) ( Table 2). However, the K d values increased by approximately an order of magnitude for all fatty acids tested compared with those for OleT JE prepared by Method 1 (Table 2). Thus, contrary to what may have been expected, the resolubilized OleT JE shows higher affinity than the non-precipitated form for the panel of fatty acid substrates tested.
Binding of cyanide and imidazole to OleT JE produced typical type II P450 heme absorption shifts to longer wavelength. Soret shifts to 433 nm (K d Ͼ 10 mM) and 424 nm (K d ϭ 193 Ϯ 11 M) were observed for cyanide and imidazole, respectively. The binding of DTT to OleT JE was also analyzed in view of the report from Rude et al. (18), which indicated that DTT could support OleT JE fatty acid decarboxylase activity by producing H 2 O 2 under aerobic conditions in the presence of the P450 heme iron (18,35). However, in previous studies, we showed that DTT coordinated the heme iron in the explosive degrading XplA P450 (23). DTT is known to bind P450 heme iron, and ligation is feasible in both DTT thiol and thiolate forms (36,37). Fig. 5C shows data from a spectral titration of OleT JE with DTT in buffer D. The DTT-bound form has three distinct absorption features in the Soret region, with peaks at 372 and 423 nm, and a strong absorbance shoulder at ϳ460 nm. The central band is the most intense. The 423 nm peak arises from distal ligation of DTT thiol to OleT JE heme iron, whereas the outer peaks result from a split (hyperporphyrin) Soret spectrum in which DTT thiolate ligates the iron (36,37). Comparable spectral maxima are at 374, 423.5, and 453.5 nm for XplA (23). In XplA, the intensities of the three absorbance bands are quite similar, but in OleT JE , the outer bands are much weaker than the 423 nm feature, suggesting that DTT favors heme ligation in the thiol state under the conditions used. Fig. 5C (inset) shows fitting of  DTT-induced heme absorption change for OleT JE , leading to a K d of 159 Ϯ 7 M. In the study by Rude et al. (18), DTT at 200 M was used to support OleT JE catalysis. However, our data indicate that substantial inhibition of OleT JE probably occurs under such conditions.

Determination of the Heme Iron Redox Potentials of Substrate-free and Substrate-bound OleT JE -Fatty acid binding to
OleT JE induces substantial shifts in heme iron spin state equilibrium toward HS (e.g. see Fig. 5A), and such shifts in spin state equilibrium are often associated with the heme iron developing a more positive potential and becoming easier to reduce (e.g. see Refs. 22 and 33). Spectroelectrochemical titrations were done for both substrate-free and arachidic acid-bound forms of OleT JE to determine the midpoint potentials for the heme iron Fe 3ϩ /Fe 2ϩ couples (versus NHE). Despite the extensive HS heme content in the arachidic acid-bound OleT JE , its heme potential (Ϫ105 Ϯ 6 mV) is not significantly different from that of the substrate-free form (Ϫ103 Ϯ 6 mV) (Fig. 6). In both cases, the heme iron potentials are quite positive compared with many bacterial P450s, which rely on NAD(P)H-dependent electron transfer from protein redox partner systems. Examples include the camphor binding-induced shift in heme iron potential from Ϫ300 to Ϫ170 mV (versus NHE) in P450cam (enabling electron transfer from the ferredoxin partner at Ϫ240 mV) (33,38) and the arachidonic acid-induced shift in potential from Ϫ429 to Ϫ289 mV (versus NHE) in P450 BM3 (22). However, unlike the aforementioned P450s, OleT JE is evolutionarily adapted to interact directly with H 2 O 2 in order to form reactive iron-oxo species (initially the ferric-hydroperoxo compound 0, which is probably transformed to the ferryl-oxo compound I), and its positive potential is probably a consequence of the environment of the heme and its cysteine thiolate ligand. The fact that the OleT JE heme potential is effectively unchanged in the HS substrate-bound form may be a consequence of the proximity of a negatively charged substrate carboxylate group to the heme iron in the arachidic acid-bound form.
Another notable feature in the spectra for the reduced forms of substrate-free and arachidic acid-bound OleT JE is that neither form a unique spectral species that could be assigned to a cysteine thiolate-coordinated ferrous P450 heme iron. As shown in Fig. 6, the UV-visible spectrum for OleT JE immediately following reduction has its Soret feature at 414 nm, with a small shoulder at ϳ423 nm, indicative of a mixture of Cys thio-late-coordinated (major species) and thiol-coordinated (minor species) forms. In the redox titration for substrate-free OleT JE (Fig. 6A), the Soret peak for the reduced P450 is split into two components, with a peak at 406 nm and a shoulder at ϳ425 nm. The former probably represents thiolate-coordinated ferrous OleT JE , and the latter the thiol-coordinated form (39). A similar phenomenon is seen for the arachidic acid-bound OleT JE (Fig.  6B), although in this case, the main peak is at 420 nm with a shoulder at ϳ400 nm, suggesting a higher proportion of the thiol-coordinated ferrous form in the substrate-bound OleT JE . For both substrate-free and arachidic acid-bound OleT JE redox titrations, it is evident that there is a single set of isosbestic points throughout the titrations, indicating that the equilibrium between thiol-and thiolate-coordinated ferrous forms remains constant as the concentration of ferrous OleT JE accumulates. The Soret isosbestic point is at 408 nm for the arachidic acid-bound form and at 410 nm for substrate-free OleT JE . Thus, under the same redox titration conditions, arachidic acid substrate binding seems to push the ferrous heme cysteine thiolate/thiol equilibrium slightly further toward the thiol-coordinated state.
Stopped-flow Analysis of OleT JE Turnover Kinetics-In order to determine the kinetics of H 2 O 2 -dependent fatty acid oxidation, we exploited the fact that turnover of bound substrate is accompanied by a reconversion of OleT JE heme iron spin state from HS to LS as the substrate is decarboxylated. The two states of the P450 have considerably different heme spectra, and thus we used stopped-flow absorbance spectroscopy to measure the rate constants for LS OleT JE values) for the binding of fatty acids to OleT JE were determined by optical titration, as described under "Fatty Acid and Inhibitor Binding Titrations with OleT JE ." The K d values stated are for titrations done using OleT JE prepared using Method 1 and using fatty acid stocks made up in alcohol containing 30% (v/v) Triton X-100 (A) or in alcohol without the addition of the detergent (B). The final set of K d data (C) are for OleT JE prepared using Method 2 and using fatty acid stocks in alcohol with 30% (v/v) Triton X-100. For protein prepared using Method 1, the K d values are lower, and the extent of spectral shift toward the ferric HS state is improved in the presence of Triton X-100, which in itself does not induce any significant HS heme development. In the case of OleT JE prepared using Method 2, the fatty acid K d values are consistently higher than those for enzyme produced using Method 1 from comparable titrations with Triton X-100 in the fatty acid stocks. The percentage of HS heme is estimated with reference to the nearly complete conversion to HS of the OleT JE sample titrated with arachidic acid, including 30% Triton X-100 (estimated at 95% HS at apparent saturation with arachidic acid).  Fig. 7B (inset) shows the accompanying stopped-flow data for this reaction at 417 nm and 7.58 M H 2 O 2 , with data fitted accurately using a single exponential function to give a k obs of 12.50 Ϯ 1.16 s Ϫ1 . EPR Analysis of OleT JE -The continuous wave X-band EPR spectrum of substrate-free OleT JE (prepared using Method 1) displays features attributable to the S ϭ 1 ⁄ 2 LS ferric heme with a thiolate-proximal ligand to the iron and a distal ligating water molecule (Fig. 8A). Several such LS forms with rhombic anisotropy are evident from the multiplicity of lines observed, and the resolvable contributions at g z show g values ranging from those typical for LS ferric P450s (2.43 and 2.48) (e.g. see Refs. 39 -42) to those associated with chloroperoxidases and the fatty acid hydroxylase P450 SP␣ (CYP152B1) (2.55, 2.61, and possibly 2.70) (17,43). Overall, the EPR spectrum suggests a large, water-filled site with multiple coordination geometries and hydrogen bonding partners available to the distal water ligand. The addition of substrate, arachidic acid, produces a very different EPR spectrum dominated by two S ϭ 5 ⁄ 2 rhombic HS ferric thiolateligated heme signals having five-coordinate iron. The g values are 7.76, 3.84, and 1.75 for one signal and 7.76, 3.67, and 1.71 for the second (the signal at g ϭ 4.26 arises from nonspecifically bound non-heme iron) (Fig. 8B). The differences between these two forms reflect small differences in the ligand field geometry and as such are likely to be a result of distortions of the heme group and the thiolate ligand rather than any change in the ligation of the heme iron. The observation of high spin heme is in contrast to P450 SP␣ , which shows no spin state change on substrate binding (17) and where x-ray crystallography has shown that the heme retains the water sixth ligand when substrate is bound. Approximately 15% of the protein (as determined by relative integration of the low spin forms, accounting for differences in concentration and subtraction of baselines to account for underlying high spin species) is converted to a new LS species with g values of g z ϭ 2.46, g y ϭ 2.25, and g x ϭ 1.89, which is not present in the substrate free enzyme. It is likely that this minor LS species is in equilibrium with the HS form.

Fatty acid
OleT JE -catalyzed Substrate Turnover-OleT JE turnover assays were done using H 2 O 2 and with a range of saturated fatty acids (C12-C20), as described under "Experimental Procedures." As reported by Rude et al. (18), products were identified and characterized as terminal alkenes. Fig. 9 shows formation of 1-alkene products from arachidic acid (forming 1-nonadecene) and from lauric acid (forming 1-undecene). Rude et al. (18)  reported (from in vivo and/or in vitro analyses) data consistent with the oxidative decarboxylation of fatty acids in the chain length range 14 -22, forming the n-1 terminal alkenes (18). Here we demonstrate that the chain length selectivity of OleT JE extends to the C12:0 saturated fatty acid lauric acid.
Crystal Structure of OleT JE -The structure of the OleT JE P450 was determined to a resolution of 2.3 Å for the substratefree form (PDB 4L54) and to 2.5 Å for the arachidic acid fatty acid-bound form (PDB 4L40). Despite the fact that both forms were obtained from different crystallization conditions, the crystal packing is identical, with little difference in structure noted between the C20-bound and ligand-free OleT JE P450s (root mean square deviation of 0.125 for 410 C ␣ atoms). The overall OleT JE P450 structure closely resembles the related peroxygenase P450 BS␤ with a root mean square deviation of 0.99 Å for 379 C ␣ atoms (Fig. 10). Areas where significant deviations occur are located at the FG-loop and the adjacent C-terminal loop regions. Both of these regions line the fatty acid binding pocket, more specifically the area involved in binding the fatty acid tail. In OleT JE , residue Leu-177 from the FG-loop closes the narrow access channel to the solvent that is present in both P450 BS␤ and the related P450 SP␣ . The OleT JE fatty acid binding pocket is elongated in comparison with P450 BS␤ , due to three point mutations in the N-terminal ␤-sheet region (from FIGURE 6. Determination of the OleT JE heme iron reduction potential in its substrate-free and arachidic acid-bound forms. A, data from a spectroelectrochemical redox titration of ligand-free OleT JE (8.1 M). The spectrum for the oxidized enzyme (solid line) shows the Soret maximum at 419 nm, whereas that for the fully dithionite-reduced P450 (dashed line) has its Soret maximum at 406 nm and shows a single feature in the Q-band region at ϳ560 nm. Intermediate spectra are shown in dotted lines. Arrows, direction of absorption changes observed during the reductive part of the titration. Inset, plot of absorbance at the Soret peak (417 nm) versus the applied potential corrected for the NHE. Data are fitted using the Nernst equation to give a midpoint potential of E 0 Ј ϭ Ϫ103 Ϯ 6 mV. B, redox titration for arachidic acid-bound OleT JE (8.1 M; thick solid line). The oxidized substrate-bound species has its Soret maximum at 395 nm, and the fully reduced form (thick dotted line) has a maximum at ϳ420 nm. Intermediate spectra in the titration are shown in dotted lines. Arrows again indicate absorption changes observed during the reductive part of the titration. The inset shows a plot of absorbance at the substrate-bound Soret peak (395 nm) versus the applied potential corrected for the NHE, with data fitted using the Nernst equation to yield E 0 Ј ϭ Ϫ105 Ϯ 6 mV. P450 BS␤ Ile-25, Leu-41, and Leu-315 to the corresponding OleT JE Thr-24, Ala-40, and Ala-317) (Fig. 11). This provides additional space to accommodate fatty acids with a chain length up to C20.
The active site of OleT JE is remarkably similar to P450 BS␤ , despite the fact that both enzymes favor distinct catalytic reactions (decarboxylation versus hydroxylation, respectively). The fatty acid carboxylate group is bound by the conserved Arg-245 and placed approximately perpendicular to the heme plane in OleT JE , with the C ␣ and C ␤ carbons closest to the heme iron (at 5.1 and 5.7 Å, respectively; Fig. 12). No other direct polar contacts are made between the carboxylate headgroup and the protein. A water molecule is seen to occupy a position between His-85 and the C20 carboxylate, in close proximity to both moieties (at distances of 3.3 and 2.7 Å, respectively). This water molecule is too distant from the heme iron (3.3 Å) to act as a direct ligand, and the heme is pentacoordinate in the substratebound OleT JE P450 structure. In comparison, the substratefree OleT JE active site contains several water molecules with ill defined density above the heme plane, probably corresponding to a range of distinct water structures in the absence of the substrate, consistent with conclusions from EPR data.
One of the few key differences between the OleT JE and P450 BS ␤ active sites is the switch from His-85 (OleT JE ) to Gln-85 (P450 BS␤ ). In OleT JE , the His-85 imidazole side chain points into the active site, directly toward the heme iron (at a distance of 5.8 Å). The imidazole moiety is sandwiched between the heme edge and Phe-79 and makes no polar contacts with other amino acids. In addition, the A and B pyrrole groups of the OleT JE P450 heme are distorted, with the effect of moving these closer to the substrate by ϳ0.8 Å when compared with P450 BS␤ . This is independent of substrate binding because the same deviation from planarity can be observed for the substrate-free OleT JE P450 heme group. The difference in heme conformation between OleT JE and P450 BS␤ cannot be attributed directly to mutations in the heme vicinity because most heme binding residues are identical, and it appears instead to be linked to small changes in the position of secondary elements containing heme binding residues.

DISCUSSION
In this paper, we present the first structural and detailed kinetic and biochemical characterization of the Jeotgalicoccus OleT JE enzyme, which catalyzes the production of terminal alkenes from long chain fatty acids, an industrially relevant reaction. The enzyme is one of a growing number of P450 enzymes that has evolved to use hydrogen peroxide (rather than NAD(P)H-dependent redox partners) in order to form reactive iron-oxo species for substrate oxidation (5,16). Its closest bacterial relatives are the S. paucimobilis P450 SP␣ and the B. subtilis P450 BS␤ , which catalyze predominantly the ␣and ␤-hydroxylation of long chain fatty acids (44,45). OleT JE was classified as CYP152L1, in the same family as the peroxygenases P450 SP␣ (CYP152B1) and P450 BS␤ (CYP152A1) and the Clostridium acetobutylicum CYP152A2, a further fatty acid ␣-hydroxylase (46). OleT JE is assigned as the first member of a new CYP152 subfamily (CYP152L1).
Crystal structure data are confirmatory of the typical P450 fold in OleT JE and also of a close evolutionary relationship with the other structurally resolved peroxygenases P450 SP␣ and P450 BS␤ . OleT JE does not undergo any major structural change on binding the C20:0 lipid arachidic acid, which nearly completely occupies the entire P450 active site cavity. The OleT JE fatty acid binding mode is very similar to that seen in P450 SP␣ and P450 BS␤ . The major differences seen in OleT JE relate to the active site His-85 residue (which is replaced by a glutamine residue in both the hydroxylases P450 SP␣ and P450 BS␤ ) and the distortion of its heme cofactor (as compared with these hydroxylases). Despite extensive structural identity between OleT JE , P450 SP␣ , and P450 BS␤ , and the similarities in their spec-ificity for long chain fatty acids, there is a major difference in reaction mechanism and product formation. OleT JE catalyzes mainly oxidative decarboxylation to form terminal alkenes from fatty acids, whereas P450 SP␣ and P450 BS␤ favor hydroxylation at the ␣and ␤-carbons, respectively (16 -18). Rude et al. (18) reported that the Q85H mutant of P450 BS␤ catalyzes decarboxylation of palmitic acid (to 1-pentadecene) ϳ1.5-fold faster than the WT enzyme and also enhances palmitic acid ␤-hydroxylation at the expense of ␣-hydroxylation, thus confirming the importance of the residue at this position. The initial reaction with H 2 O 2 in each of these P450s should form the reactive iron-oxo species compound 0 (the ferric hydroperoxo intermediate in the P450 catalytic cycle) using the peroxide "shunt" mechanism (7). However, a fundamental difference in the OleT JE reaction mechanism leads to formation of the terminal alkene as the major product (rather than hydroxylated fatty acids).
To explain this phenomenon, we are drawn to the role of the active site histidine (His-85) in OleT JE . It has been postulated that the high pK a of compound II in P450s generally ensures that compound I in these enzymes effectively abstracts a hydrogen atom from the substrate (as opposed to a single electron). However, in OleT JE , His-85 could function to donate a proton to compound I, concomitant with its reduction to compound II by abstraction of an electron from the fatty acid carboxylate moiety. The formation of the carboxylate radical would lead to homolytic scission of the substrate C-C ␣ bond, following hydrogen atom abstraction from the C ␣ position by compound II. In the hydroxylases, such a reaction is unlikely to occur in the absence of a proton donor to form protonated compound II, and these enzymes are thus likely to use compound I to abstract hydrogen from either the C ␣ or the C ␤ position, ultimately leading to monohydroxylation of substrate (Fig. 13) (47). FIGURE 9. Oxidative decarboxylation of arachidic acid and lauric acid by OleT JE . A, total ion count from GC separation of the C19 terminal alkene 1-nonadecene in the reaction of OleT JE with arachidic acid (C20:0) (top), with mass spectrometric analysis of the major peak at 16.21 min confirming its identity (bottom; inset highlighting the region of the 1-nonadecene mass ion with m/z ϭ 266). B, total ion count from GC separation of the C11 terminal alkene 1-undecene following reaction of OleT JE with lauric acid (C12:0) (top), with mass spectrometric analysis of the major peak at 6.63 min confirming its identity (bottom; inset highlighting the region of the 1-undecene mass ion with m/z ϭ 154).   In early studies of OleT JE , we noted the propensity of the enzyme to aggregate and precipitate in typical (low salt) buffer conditions. However, we were able to turn this to our advantage through demonstrating that precipitation of the OleT JE P450 (induced through dialysis into low salt buffer) could readily be reversed by resolubilization (in high salt buffer) in a form that retained thiolate-coordinated heme and the ability to bind fatty acids. Moreover, OleT JE purified using this route (Method 1) had a higher proportion of LS ferric heme iron (probably due to displacement of lipid retained in its active site during isolation from E. coli) than did the enzyme isolated without the precipitation step (Method 2). OleT JE prepared using Method 1 was readily crystallized to produce the structural data shown in Figs. 10 -12 for substrate-free and arachidic acid-bound forms.
The binding of fatty acids displaces the axial water ligand from the heme iron of OleT JE , leading to a shift of the ferric heme iron toward the HS state and to accompanying large changes in the heme absorption spectrum. This provides the basis for determination of fatty acid K d values by optical titration. Many of the fatty acids tested for binding to OleT JE have very limited solubility in water. For this reason, we investigated their suspension in detergent solutions, and found that although there was no evidence for binding of Triton X-100 to OleT JE , the K d values for fatty acids suspended in 30% Triton X-100 were markedly improved over those suspended in alcohol alone for OleT JE prepared by Method 1. By comparison, the K d values for Triton X-100-suspended lipids with OleT JE prepared by Method 2 were ϳ5-20-fold greater, possibly due to the P450 prepared without the precipitation step retaining contaminant short lipids that hinder the binding of the C12 to C20 fatty acids tested ( Table 2). The robust nature of OleT JE and its stability to low salt precipitation and resolubilization in high salt is unusual in the P450 superfamily but perhaps not surprising given the halotolerant nature of the Jeotgalicoccus bacterial genus. The cysteine thiolate ligand to the OleT JE heme iron is clearly retained following its resolubilization, as evidenced by retention of catalytic activity and the production of a Fe(II)CO (P450) complex with Soret maximum at 449 nm (Fig. 4). The stability of OleT JE to precipitation and resolubilization thus provides a convenient route to its purification and also bodes well for its application in synthesis of terminal alkenes.
Stopped-flow kinetic analysis of OleT JE indicated that the extensively HS fatty acid-bound form is rapidly reconverted back to the LS ferric form upon mixing with H 2 O 2 , with a second order rate constant of 0.80 Ϯ 0.02 M Ϫ1 s Ϫ1 with respect to H 2 O 2 concentration. Previous studies of P450 SP␣ reported a specific activity of 838 min Ϫ1 for the ␣-hydroxylation of myristic acid using 200 M H 2 O 2 , whereas P450 BS␤ catalyzed the ␣/␤-hydroxylation of myristic acid with a specific activity of 209 min Ϫ1 at 100 M H 2 O 2 (16,44,48). The OleT JE rate constants determined here are substantially faster than those for P450 SP␣ and P450 BS␤ , although they report only on the catalytic steps of substrate decarboxylation, alkene displacement from the active site, and restoration of the water ligand to the OleT JE heme iron. However, in turnover studies with different fatty acids, considerable amounts of alkene products were observed, consistent with efficient decarboxylase activity of OleT JE . Another notable difference between these peroxygenases lies in their optical response to binding fatty acid substrates. For P450 SP␣ , there is negligible heme iron spin state and associated Soret absorption change upon binding fatty acids, whereas with P450 BS␤ , some HS shift does occur, but to a much smaller extent than is observed for OleT JE with its best substrates (49). Because binding of H 2 O 2 requires displacement of the sixth ligand water on the P450 heme iron, this process may be much more efficient for OleT JE compared with P450 SP␣ and P450 BS␤ . Whereas Rude et al. (18) suggested the use of DTT as a peroxide-generating agent to drive catalysis by OleT JE , our data here indicate that DTT coordinates the ferric heme iron to inhibit the enzyme with a K d of 159 M and thus is unlikely to be as effective as H 2 O 2 in driving OleT JE activity (Fig. 5C).
In its evolution toward peroxygenase activity, OleT JE has undergone mutations that reinforce its divergence from the typical class I and II P450 enzymes that interact with NAD(P)H- FIGURE 13. Proposed mechanism for OleT JE . The cytochrome P450 OleT JE catalyzes oxidative decarboxylation of long chain fatty acids as its major reaction, whereas the highly related P450 BS␤ produces predominantly ␤-hydroxylated fatty acids from the same substrates (16,17). OleT JE His85 (replaced by Gln-85 in P450 BS␤ ) is proposed to act as a proton donor to the ferryl-oxo porphyrin radical cation (compound I) intermediate in this P450, concomitant with its reduction to compound II by an electron abstracted from the fatty acid carboxylate. Homolytic scission of the C-C ␣ bond, concomitant with hydrogen abstraction from C ␣ to compound II, leads to production of the terminal alkene and CO 2 . For the hydroxylase P450 BS␤ , this reaction cannot occur in the absence of an appropriate proton donor to compound II, with the catalytic outcome instead being a typical P450 hydroxylation at the C ␤ position. dependent redox partner enzymes and reductively activate oxygen bound to the heme iron to facilitate substrate oxidation (50). Most notably, it is evident that the acid-alcohol amino acid pair in the P450 I helix that is common to oxidase P450s (e.g. Asp-251 and Thr-252 in P450cam or Glu-267 and Thr-268 in P450 BM3) is not conserved in OleT JE or in either P450 BS␤ or P450 SP␣ (44,45,51,52). Because protonation reactions on heme iron-bound dioxygen do not feature in the catalytic cycle of the peroxygenases, this motif is dispensed with in these enzymes. Instead, the acid-alcohol pair is replaced in OleT JE by Arg-245 (which binds the substrate carboxylate) and Pro-246 in a highly conserved I-helix region of the CYP152 family enzymes (7). A conserved phenylalanine residue (Phe-393 in P450 BM3; Phe-350 in P450cam) in the P450 heme binding region is also absent from the bacterial peroxygenases. Phe-393 interacts with the heme thiolate bond in CYP102A1 and was shown to be important in maximizing P450 catalytic efficiency through regulating heme iron potential and thus heme iron reduction rate and ferrous-oxy complex stability in studies of various Phe-393 variants (53,54). In OleT JE and the other bacterial peroxygenases, the absence of this extensively conserved phenylalanine and amino acid insertions into this "heme binding loop" consensus region indicate that heme thermodynamic properties are regulated differently (45). Our potentiometric data show that OleT JE has a very positive heme iron Fe(III)/Fe(II) redox potential (Ϫ103 mV versus NHE), which is not significantly altered in the HS form when bound to arachidic acid (Ϫ105 mV) (Fig. 6). Thus, the absence of key residues in OleT JE , BS ␤ , and SP ␣ is a clear indicator of their distinct evolutionary pathway, such that protonation of iron-oxo intermediates or substrate-dependent regulation of electron transfer from redox partners has been bypassed through direct use of H 2 O 2 . Similarly, the introduction of the conserved arginine-proline pair (Arg-245 and Pro-246 in OleT JE ) in place of the acid/alcohol pair in each of the these structurally characterized peroxygenase P450s is clearly an evolutionary adaptation that facilitates positioning of fatty acid substrate to enable its oxidative decarboxylation or ␣/␤-hydroxylation. However, a more detailed understanding of general structural determinants that might define a peroxygenase P450 "class" may require further structural/mechanistic studies of more divergent enzymes.
In conclusion, we have determined novel structural and biochemical properties of an unusual P450 peroxygenase enzyme (CYP152L1, OleT JE ) that catalyzes oxidative decarboxylation of fatty acids to produce terminal alkenes. OleT JE is a catalytically efficient and structurally robust P450 with great potential in the synthesis of a variety of alkenes as "drop-in" biofuels (i.e. biofuels that may be used interchangeably with conventional fuels) or chemical reagents. Ongoing research is directed at engineering altered fatty acid chain length selectivity into OleT JE in order to enable production of short chain, volatile alkenes.