Cytochrome P450cin (CYP176A), Isolation, Expression, and Characterization*

Cytochromes P450 are members of a superfamily of hemoproteins involved in the oxidative metabolism of various physiologic and xenobiotic compounds in eukaryotes and prokaryotes. Studies on bacterial P450s, particularly those involved in monoterpene oxidation, have provided an integral contribution to our understanding of these proteins, away from the problems encountered with eukaryotic forms. We report here a novel cytochrome P450 (P450cin, CYP176A1) purified from a strain ofCitrobacter braakii that is capable of using cineole 1 as its sole source of carbon and energy. This enzyme has been purified to homogeneity and the amino acid sequences of three tryptic peptides determined. By using this information, a PCR-based cloning strategy was developed that allowed the isolation of a 4-kb DNA fragment containing the cytochrome P450cin gene (cinA). Sequencing revealed three open reading frames that were identified on the basis of sequence homology as a cytochrome P450, an NADPH-dependent flavodoxin/ferrodoxin reductase, and a flavodoxin. This arrangement suggests that P450cin may be the first isolated P450 to use a flavodoxin as its natural redox partner. Sequencing also identified the unprecedented substitution of a highly conserved, catalytically important active site threonine with an asparagine residue. The P450 gene was subcloned and heterologously expressed in Escherichia coli at ∼2000 nmol/liter of original culture, and purification was achieved by standard protocols. Postulating the native E. coli flavodoxin/flavodoxin reductase system might mimic the natural redox partners of P450cin, it was expressed inE. coli in the presence of cineole 1. A product was formedin vivo that was tentatively identified by gas chromatography-mass spectrometry as 2-hydroxycineole 2. Examination of P450cin by UV-visible spectroscopy revealed typical spectra characteristic of P450s, a high affinity for cineole 1 (K D = 0.7 μm), and a large spin state change of the heme iron associated with binding of cineole 1. These facts support the hypothesis that cineole 1 is the natural substrate for this enzyme and that P450cin catalyzes the initial monooxygenation of cineole 1 biodegradation. This constitutes the first characterization of an enzyme involved in this pathway.

The cytochrome P450s (P450s) 1 are a superfamily of oxidative hemoproteins (1, 2) that carry out an enormous variety of oxidative transformations. These range from simple alkene epoxidation and heteroatom oxidation all the way through oxygen insertion into unactivated C-H bonds and C-C bond cleavage (3). P450s are broadly categorized as either soluble, e.g. most bacterial P450s, or as membrane bound/microsomal monooxygenases, e.g. most eukaryotic P450s. All P450s, however, contain a prosthetic heme group that is ligated to the protein backbone by a cysteinyl sulfur coordinated to the heme iron. It is this heme-thiolate moiety that is responsible for much of the chemistry carried out by these enzymes. The typical reaction catalyzed by a P450 is given by Reaction 1, The exact mechanism of the oxidation reaction is still the subject of much debate, despite an enormous amount of work (3,4). The electrons derived from the nicotinamide cofactors are generally transferred to the P450 one at a time by auxiliary redox protein(s) that are characteristic of the origin of the enzymes. Thus, bacterial P450s generally employ an FAD-dependent ferrodoxin reductase and an iron-sulfur protein to mediate this transfer, whereas microsomal P450s employ a single protein that contains both an FAD and an FMN domain.
Investigations of bacterial P450s have provided much of our current understanding of P450 mechanism and structure. This is particularly true of P450 cam isolated from Pseudomonas putida, in part because it was the first P450 available in relatively large amounts (5). However, some features of P450 cam appear to be unique to this enzyme, e.g. a potassium-binding site (6), and not typical of this enzyme class. P450 cam still remains the most studied of all P450s and one of the few bacterial enzymes of this family that has been heterologously expressed with its redox partners as a catalytically active system (7). It is a biodegradative enzyme that catalyzes the conversion of camphor to 5-exo-hydroxycamphor, which initiates a cascade of biodegradative reactions that allow P. putida to live on camphor as its sole source of carbon and energy (8). P450 cam and others involved in such pathways, e.g. P450 lin (9) and P450 terp (10), are generally characterized by high substrate specificity and high rates of turnover. As such, they make attractive starting points for attempting to design biocatalysts that can carry out synthetically useful oxidative transformations. We wished to isolate a new bacterial biodegradative P450 system that could be (i) used for comparison with P450 cam , (ii) heterologously expressed as a catalytically active system, and (iii) used as the basis for constructing libraries of P450s that would be explored for their ability to catalyze useful oxidative transformations. Such a P450 system was implicated in the bacterial utilization of 1,8-cineole (cineole, 1, Scheme 1).
Eucalypt trees produce a mixture of nearly 50 hydrocarbons termed the "essential oil" which contains extremely high concentrations of the monoterpene cineole 1. The role of cineole 1 is somewhat unclear, although it is believed to be involved in a number of functions including defense against herbivores (11,12) and pathogens (13), attracting pollinators and fruit-dispersing animals (14), and/or as an allelopathic agent (15,16). The Australian eucalypt population collectively produces and releases an estimated 500,000 tons of cineole 1 into the environment annually (17). Studies indicate that the major part of the released cineole 1 is rapidly removed by microbial oxidation (18), with lesser quantities consumed by native wildlife (19) and bushfires.
In 1979 MacRae et al. (18) isolated a soil microorganism identified as Pseudomonas flava, which could utilize cineole 1 as it sole source of carbon and energy, and oxidized derivatives of cineole 1 were identified in the growth medium. Subsequently, Trudgill and co-workers (20) identified a Rhodococcus species also capable of using cineole 1 as its sole carbon source and similarly identified oxygenated cineole derivatives. Interestingly, it was found that the initial oxidation of cineole 1 by the two organisms produced enantiomeric forms of 2-hydroxycineole 2. A metabolic pathway for the catabolism of cineole 1 in Rhodococcus sp. C1 was proposed, based on the isolated metabolites (Scheme 1). Thus, cineole 1 is initially oxidized to 2-hydroxycineole 2 by a monooxygenase, and further transformations lead to a highly functionalized hydroxyketo acid 3 that may spontaneously form the isolated dihydrofuranone 4 or be incorporated into central metabolic pathways.
Despite the relatively large amount of information concerning the chemistry of the cineole 1 biodegradative pathway, nothing was known about the enzymes involved. Early attempts at isolating the proposed monooxygenase that catalyzes cineole 1 oxidation failed (21). Several lines of evidence suggested, however, that the monooxygenase might be a P450. The biodegradative utilization's of a number of monoterpenes, including camphor, linalool, and terpineol, have all been shown to be initiated by oxidations catalyzed by P450s (21). Additionally, CYP3A4 was recently identified as being the main P450 isoform responsible for cineole 1 metabolism in humans, with the primary product, 2-hydroxycineole 2, being the same as the one implicated in bacterial biodegradation (22). Finally, the relative scarcity of other enzymes, e.g. non-heme iron oxygenases capable of carrying out such an oxidation, also suggested a P450 would be involved.
We thus set out to isolate and characterize a monooxygenase system involved in the bacterial utilization of cineole 1. Reported here is the isolation, overexpression, and purification of a P450, P450 cin , believed to be responsible for cineole 1 oxidation. In addition, the sequences of two genes that may encode the required redox partners for P450 cin are described along with preliminary characterization of the P450 that supports its involvement in the biodegradation of cineole 1.

EXPERIMENTAL PROCEDURES
Materials-Yeast extract and Bacto-tryptone were obtained from BD PharMingen. Reagent grade 1,8-cineole was obtained from Aldrich. Sodium dithionite (sodium hydrosulfite) was purchased from Aldrich. Nylon hybridization transfer membranes (GeneScreen Plus) were obtained from PerkinElmer Life Sciences. All other chemicals used were purchased from commercial sources and were of the highest grade generally available. The restriction enzymes and polynucleotide kinase were obtained from New England Biolabs, Inc. (Beverly, MA), and Roche Molecular Biochemicals based on availability. Terminal deoxynucleotidyltransferase was obtained from Promega Corp. (Madison, WI). Escherichia coli strain, DH5␣FЈ 80dlacZ⌬M15 ⌬(lacZYA-argF) U169 deoR recA1 endA1 hsdR17(r k Ϫ , m k ϩ ) phoA supE44 Ϫ thi-1 gyrA96 relA1, was obtained as frozen competent cells from Invitrogen. Sonication was achieved using a Branson Sonicator set to 50% duty cycle.
Isolation of the Microorganism-Soil samples were collected from several sites beneath eucalypt trees within a small eucalypt forest at the University of California, San Francisco. A dirt/water slurry prepared from each sample was transferred to minimal medium (23), and bacteria were selected for the ability to use 1,8-cineole as the sole carbon source. Following several rounds of serial culturing (1:50 dilutions) at 30°C, the cell culture was spread onto minimal medium plates (1.5% agar plates) with 0.5 g of cineole in the lid, yielding a single pure strain according to the protocol of Peterson and Lu (23). The strain was propagated in 10 -15 ml of minimal medium with 20 l of 1,8-cineole at 30°C.
Tryptic Digest of P450 cin -Internal amino acid sequence for P450 cin was obtained from tryptic peptides of P450 cin . These were purified by microbore high pressure liquid chromatography and sequenced using automated Edman degradation protein sequencer. These data were used in concert with that obtained via matrix-assisted laser desorption ionization-mass spectrometry to yield the sequence of three fragments ( Table I).
Preparation of Whole Cell DNA-Cells were grown on minimal medium (23) with 1,8-cineole as the sole carbon source at 30°C. Whole cell DNA was prepared according to Peterson et al. (24).
Generation of an Oligonucleotide Probe to Cytochrome P450 cin -A 92-bp PCR product was amplified from whole cell DNA (Citrobacter braakii) using degenerate primers 2 (5Ј-ATAT GAATTC TGG GCN ATM CAN CAH GT-3Ј and 5Ј-TATC GGATCC CK NGC RTG NGC NAC-3Ј) designed based upon five amino acids at either end of fragment 1 (Fig.  1). PCR was performed according to the following protocol: 94°C for 3 min (94°C for 1 min, ramp to 37°C over 2 min, 72°C for 1 min) for 5 cycles; (94°C for 1 min, 55°C for 1 min, 72°C for 1 min) for 25 cycles; 72°C for 10 min; 4°C. The product was treated with BamHI and EcoRI and cloned into similarly cut pCRII (Invitrogen). The insert was sequenced on an automated sequencer using M13 forward and reverse primers, confirming the central non-degenerate 42 bp coded for the 2 The Degenerate DNA code used is as follows: R ϭ A or G; K ϭ G or T; H ϭ A, C, or T; M ϭ A or C; N ϭ A, T, C, or G.  (25) and denatured at 100°C prior to use in subsequent hybridizations.
Library Construction and Screening-DNA isolated from C. braakii was digested with a bank of restriction enzymes (KpnI, PstI, SalI, PstI/HindIII, SalI/HindIII, and XhoI) and separated by electrophoresis on an agarose gel (0.7% w/v, 0.5ϫ TBE). The DNA was transferred to a nylon membrane (GeneScreen Plus PerkinElmer Life Sciences) by capillary transfer, dried, and probed with labeled oligonucleotide (26). Autoradiography revealed bands of suitable size in PstI (4 kb) and SalI (3 kb) digests. These latter digests were repeated, separated by electrophoresis, and the bands of appropriate size recovered by gel extraction.
The recovered DNA fragments were ligated into PstI-or SalI-digested pUC19 and transformed individually into library efficiency DH5␣ cells producing PstI and SalI libraries. Approximately 4,000 -5,000 recombinant cells for each fragment were plated onto 2ϫ YT plates containing ampicillin (50 g/ml) and incubated overnight. Colonies were lifted onto nylon membranes, and their DNA was fixed by successive treatments with 0.5 M NaOH, 1.5 M NaCl (4 min), 1 M Tris-HCl, pH 7.4, 1.5 M NaCl (4 min), and 50 mM Tris-HCl, pH 8.5, 50 mM EDTA, 100 mM NaCl, 1% sodium laurylsarcosine, and 250 g/ml proteinase K (20 min). Membranes were dried, and the DNA was cross-linked via UV irradiation. The membranes were screened by hybridization using the 92-bp oligonucleotide probe and labeled with 32 P as described above. Putative positive clones were picked from the original plates, grown overnight in 2ϫ YT medium containing ampicillin (50 g/ml). The identities of the clones were confirmed by restriction analysis and Southern hybridization. A positive PstI (4 kb) clone (designated pC1) was identified. Restriction analysis of pC1 revealed the SalI (3 kb) fragment (cloned as pC2) was contained entirely within the Pst (4 kb) fragment and so its isolation was not pursued further. To gain further sequence information, probes generated by the random primer method (27) using the pC1 insert as template were used to probe an EcoRI/SphI digest of C. braakii DNA. An SphI (2.9 kb) and EcoRI/SphI (5.7 kb) fragment were isolated and cloned into pUC19 to give pC3 and pC4, respectively (Fig. 2).
Sequencing-Careful restriction analysis of clones pC1 and pC4 identified a series of small internal fragments that were subcloned and sequenced using the M13 forward and reverse primers. Gaps remaining in the sequence were filled by employing suitable synthetic primers. Sequencing data were analyzed using the BLAST algorithm (NCBI Database), and three open reading frames (cinA, cinB, and cinC) were identified which were sequenced completely in both directions.
Cloning of P450 cin -Primers were designed to either end of cinA containing suitable terminal restriction sites (EcoRI and NdeI for the N-terminal primer 5Ј-CC GATTC CATATG AC TGC GAC AGT CGC-3Ј and XbaI and EcoRI for the C-terminal primer 5Ј-CC GAATTCTAGA GGA GCT TGC TCA TTC CG-3Ј). A PCR experiment was performed with pC1 as template DNA according to the following parameters: 94°C for 3 min; 30 cycles of 94°C for 45 s, 52°C for 1 min, 72°C for 1 min 30 s. The single 1.2-kb product was digested with EcoRI and inserted into the EcoRI site of pUCPKS (28). To ensure no errors had been introduced by PCR the majority of the internal sequence was replaced according to the following procedure: the clone was cut with NcoI, treated with DNA polymerase (Klenow) in a fill-in reaction, and religated to produce a selectable P450 cin mutant. Both the mutant and pC1 were digested with PpuMI/SphI. The wild-type 1.2-kb PpuMI/SphI insert from pC1 was used to replace the mutant insert in pUCPKS. An NcoI digest was performed to confirm the wild-type fragment had indeed been inserted, and sequencing (using M13 forward and reverse primers) confirmed no PCR errors were incorporated into either terminus of cinA. An EcoRI fragment was excised from pUCPKS-cinA and cloned into pUC19. This allowed convenient excision with NdeI/HindIII for insertion into pCW, producing pCW-P450 cin .
Purification of Recombinant Cytochrome P450 cin -E. coli (DH5␣) cells freshly transformed with expression vector pCW-P450 cin were cultured overnight in 2 ml of 2ϫ YT/Amp (50 g/ml ampicillin) medium. This was used to inoculate four 2.8-liter fernbach flasks containing 1 liter of TB/Amp (50 g/ml ampicillin) medium. The flasks were incubated at 37°C (ϳ4.5 h) until an A 600 of 0.6 was attained. The cells were then induced with IPTG (1 mM) and incubated at 27°C for 18 h. Cells were harvested via centrifugation (6,000 ϫ g) for 20 min at 4°C and then resuspended in 80 ml of 50 mM Tris-HCl, pH 7.4, 50 mM KCl, 1 mM EDTA, 0.1 mM phenylmethylsulfonyl fluoride, 0.5 mM DL-dithiothreitol, cineole (750 M). An indication of purity of various cytochrome P450 cin preparations was obtained by measuring the ratio of the absorbance at 415 nm to that at 280 nm. Lysis was carried out by adding lysozyme (40 mg) and stirring the suspension for 1 h at 4°C, followed by sonication on ice (6 times with 1-min intervals, 50% output). The lysate was centrifuged (15,000 ϫ g) for 1 h to remove cellular debris. The supernatant was loaded onto a DEAE-ion exchange column which was then washed with 50 mM Tris-HCl, pH 7.4, 50 mM KCl, 0.5 mM DL-dithiothreitol (Buffer A). Protein was then eluted from the column by running a KCl gradient (50 -500 mM KCl) in Buffer A. Fractions containing P450 cin were detected by its characteristic Soret absorbance at 415 nm and were combined. P450 cin eluted at ϳ180 mM KCl. Pooled fractions were concentrated in an Amicon ultrafiltration cell containing a YM-30 membrane. The concentrate was then separated on an S-300 Sephacryl gel filtration column using Buffer A. P450 cin containing fractions were pooled, concentrated to ϳ3-5ml using an Amicon ultrafiltration cell (YM-30 membrane), and then loaded onto a hydroxyapatite column equilibrated with 5 mM KH 2 PO 4 , pH 7.2. A phosphate gradient (5-300 mM KH 2 PO 4 ) was run (0.5 ml/min). P450 cin eluted at ϳ120 mM KH 2 PO 4 . Fractions with an RZ of Ͼ1.2 were combined. The cytochrome P450 content was determined from the difference spectrum between samples that had been reduced with sodium dithionite and carbon monoxide saturated versus a sodium dithionite-treated reference sample using an ⑀ 450 of 125 mM Ϫ1 cm -1 . Purified protein was snap-frozen in liquid nitrogen and stored at Ϫ80°C. The procedure produced 42 mg of P450 cin per liter of culture.
Calculation of Molar Extinction Coefficients (⑀) for P450 cin -A dilute solution of P450 cin was prepared and characterized by UV in the presence of cineole. The protoheme content of the solution was determined as the pyridine hemochromogen, which was prepared as follows. Pyridine (80 l) was added to the P450 cin solution (272 l) and gently mixed. 10% SDS (40 l) was added and the volume brought to 400 l. The solution was then divided equally between two cuvettes (reference and sample). Sodium dithionite (0.5-1 mg) was added to the sample cuvette, and the difference spectra were recorded on a dual beam spectrophotometer. The absorbance at 556.5 nm (A 556.5nm -A 570nm ) was measured, and the hemochromogen content was estimated using an ⑀ value of 33.2 mM Ϫ1 (29). The following molar extinction coefficients were calculated for P450 cin : ferric, substrate-bound, ⑀ 392 ϭ 132 mM Ϫ1 cm Ϫ1 ; ferric, substrate-free, ⑀ 415 ϭ 150 mM Ϫ1 cm-1; and ferrous, substrate present, ⑀ 411 ϭ 106 mM Ϫ1 cm-1.

RESULTS
Isolation of C. braakii-A microorganism capable of utilizing 1,8-cineole 1 as its sole carbon source was isolated through culture enrichment (23) of bacteria from the cineole 1-enriched soils beneath eucalypt trees. The procedure yielded a pure strain identified using the Biolog Identification System (Biolog, Inc., Hayward, CA) (30) as C. braakii. This strain grew best at room temperature on a variety of nutrient-rich media. However, use of such media resulted in loss of the ability to grow on cineole 1 as a sole carbon source, and thus the organism was generally maintained on minimal medium (23) in the presence of cineole 1. Examination of CO difference spectra of cell lysates from these cultures indicated that a P450 was present, albeit at low concentrations.
Purification of P450 cin from C. braakii-A P450 was purified from a large scale culture of C. braakii grown on minimal medium in the presence of cineole 1. The purification protocol used was based on those published for P450 cam , P450 terp , and P450 BioI ((9, 10, 31)). A refined version of this protocol is currently used for purification of recombinant P450 cin (see below).
The procedure yielded purified protein with a peak fraction possessing an RZ (A 415 /A 280 ) of ϳ1.28. The protein isolated was clearly a P450 by CO difference spectroscopy and had a molecular mass of ϳ45 kDa by SDS-PAGE analysis. Importantly, the heme Soret absorbance at 415 nm shifted to 392 nm in the presence of cineole 1. This is characteristic of the spin state change of the heme iron (low to high spin) seen upon binding of substrate to a P450 and strongly suggested that the isolated P450 cin was involved in cineole 1 metabolism.
Tryptic Digestion and Sequencing of Protein Fragments-Nterminal sequencing of P450 cin (purified from C. braakii) using automated Edman degradation failed to produce useful data, presumably due to N-terminal blocking. However, tryptic digestion of the protein followed by high pressure liquid chromatography purification of the resultant fragments gave several peptides amenable to Edman degradation. A combination of automated Edman degradation sequencing and mass spectrometry yielded the sequences of three fragments that are shown in Table I. Fragments 1 and 2 are clearly related, differing only in the identity of the first amino acid. However, as their sequences were independently obtained, we were confident of their reliability, and this sequence was chosen as the basis upon which to design oligonucleotides for use in isolation of the gene encoding P450 cin . Fragment 3 was not useful for probe design due to the preponderance of amino acids with highly degenerate codons. We chose to amplify a larger segment of the P450 gene by PCR in order to generate a better probe, rather than simply employ short degenerate oligonucleotides. Thus, degenerate primers were designed to either end of coding sequence for fragment 1 and were chosen to minimize their inherent degeneracy (Fig 1). These were used in a PCR experiment with genomic DNA from C. braakii as template. The product from this PCR was sequenced and shown to encode the peptide fragment 1, providing the key unique 92-bp probe for P450 cin .
Southern Hybridization-Probing a series of restriction enzyme digests of C. braakii genomic DNA with the 92-bp probe labeled with 32 P by reaction with [␥-32 P]ATP in the presence of polynucleotide kinase failed to yield usable results. Believing that sensitivity was the source of our problems, we chose to label the probe with [␣-32 P]dCTP in the presence of terminal deoxynucleotidyltransferase. TdT is known to transfer up to 30 nucleotides to the 3Ј end of single-stranded DNA, which we hoped would result in an order of magnitude increase in the specific activity, and thus sensitivity, of the probe. With dCTPlabeled oligonucleotide, Southern hybridization experiments led to the isolation of a 4-kb PstI fragment that was cloned into pUC19 to give pC1. Initial results revealed that this PstI fragment did not contain a stop codon for the final gene in the putative operon (see below). Thus, random priming using pC1 insert was employed to isolate two further genomic fragments that were cloned into pUC19 as pC3 and pC4, such that the entire operon containing P450 cin could be sequenced (Fig 2).
Operon Analysis-Sequencing of the above DNA fragments revealed a putative operon consisting of three open reading frames with overlapping stop and start codons (Fig. 2). All ORFs were analyzed using the BLAST algorithm (32) and the GenBank TM data base. The first ORF (cinA) encoded a 405amino acid protein with a predicted molecular mass of 45 kDa. Homology analysis indicated that it was clearly a P450, being similar to a number of bacterial P450s (Table II), including P450 cam (27% identity and 46% similarity) but different enough to be classified as the sole member of a new P450 family CYP176A1 (2). All of the peptide fragments that had been characterized from the tryptic digest of the isolated P450 were encoded by this gene.
The two other ORFs (cinB and cinC) appeared to encode the expected redox partners for a catalytically functional P450 system. The second ORF (cinB) encoded a 49-kDa, 452-amino acid protein that was apparently an NADPH-dependent ferrodoxin reductase on the basis of sequence homology (Table II). The final gene (cinC) encoded a protein (16 kDa and 155 amino acids) that was most similar to the FMN-containing domain of cytochrome P450 reductase (Table II). The participation of a dedicated flavodoxin as a P450 redox partner has not been described previously, but E. coli flavodoxin has been reported to provide an unnatural redox partner for a variety of heterologously expressed cytochromes (33).
Subcloning and Expression of P450 cin -A PCR-based approach was used to clone the cinA gene into the expression vector pCWori (34). Expression of the encoded protein after induction with IPTG was monitored by SDS-PAGE and CO difference spectroscopy of cell lysates. It was found that although polypeptide production occurred at temperatures from 25 to 37°C, P450 holoprotein was only observed when the growth temperature was below 28°C. CO difference spectroscopy indicated that P450 was produced at concentrations of ϳ2000 nmol/liter of original culture. Characterization of the purified protein by N-terminal sequencing and electrospray mass spectrometry gave similar results. Both techniques clearly identified the protein as P450 cin and indicated that some degree of N-terminal processing of the protein had occurred. The degree of processing varied between different preparations, with some having the major species present derived  1. PCR primer design based on tryptic peptide fragment 1. a, forward PCR primer; b, fragment 1; c, reverse PCR primer. The restriction sites engineered into the 5Ј ends of the primers are indicated. Amino acid residues in parentheses are portions of the peptides not encoded by the oligonucleotide primers. Degenerate nucleotide codes are shown in boldface. F1PCR-F and F1PCR-R are 72-and 128-fold degenerate, respectively. from loss of the N-terminal methionine, whereas others had the predominant protein derived from loss of the first six amino acids of the predicted sequence.
In Vivo Oxidation of Cineole by P450 cin -When P450 cin was expressed in E. coli in the presence of cineole 1, a metabolite was produced with mass spectral characteristics consistent with that of 2-hydroxycineole 2. This compound was not produced when E. coli not expressing P450 cin was grown in the presence of cineole 1. This result strongly supports the hypothesis that P450 cin catalyzes the first monooxygenation in the biodegradation of cineole 1.
Spectral Characteristics of Cytochrome P450 cin -The protein was first examined by UV-visible spectroscopy (Fig. 3) that revealed the typical characteristics of a bacterial P450, a heme Soret at 415 nm in the absence of substrate which shifted to 392 nm upon the addition of cineole 1. This spectral change is correlated with a change in the heme iron from low to high spin and is associated with the binding of substrate to bacterial P450s and as such was the first evidence that cineole 1 was indeed the natural substrate of this enzyme. Reduction of the protein with dithionite and treatment with CO resulted in the formation of a CO complex with the typical 449 nm absorbance of a P450.
1,8-Cineole Binding by Cytochrome P450 cin -The addition of 1,8-cineole 1 to the substrate-free form of the enzyme results in a decrease in absorbance at 415 and 568 nm and an increase in absorbance at 392 and 643 nm, indicating the conversion of low spin hemoprotein to the high spin form (35). Difference spectroscopy (Fig. 4) was used to calculate a dissociation constant for the enzyme-substrate complex (K D ) by adding cineole 1 to a solution of P450 cin and monitoring the absorbance at 415 and 392 nm. The difference (measured as A 392 -A 415 ) was plotted against the cineole 1 concentration to produce a substratebinding curve (Fig. 4)

DISCUSSION
Prior to the beginning of this work, much was known about the chemistry of cineole 1 biodegradation. Two organisms had been independently isolated previously that could live on cineole 1 as their sole source of carbon and energy (18,20). From an investigation of the oxidized cineole metabolites isolated from growth media, a consensus biodegradative pathway was proposed (Scheme 1), albeit proceeding via intermediates with different absolute stereochemistry in each of the two organisms. Nothing was known about the enzymes involved in the proposed pathway, despite some early investigations (21). Our results support the hypothesis that a P450 is responsible for the initial step in the biodegradation of cineole 1.
We isolated from the soil of a eucalypt forest a strain of C. braakii that could utilize cineole 1 for growth and energy. C. braakii is a Gram-negative motile rod that is a common environmental organism and that has recently been reclassified from Citrobacter freundii (36). P. flava and Rhodococcus, in which cineole 1 utilization had been found previously, are also common Gram-negative organisms. Given the quantity of cineole 1 that is believed to be released into the environment by eucalypts, we were not surprised to find that a third genus of bacteria has developed the ability to grow on this compound. The organism grew poorly on minimal medium in our hands when cultured on a large scale, and although it grew well on nutrient rich-media, such growth resulted in loss of its ability to utilize cineole 1. This latter fact perhaps suggests that the genes responsible for cineole 1 metabolism are plasmid-borne as has been shown for the P450s involved in camphor and terpineol utilization (10,37). The difficulty in growing the organism clearly meant that isolation of the gene encoding the cineole monooxygenase would be essential for its characterization. Initial approaches to isolating the P450 gene focused on its expected homology with other P450s, because this seemed to offer the most direct route to P450 cin . Homology-based methodology in bacterial P450 systems is particularly difficult, however, because most of these proteins reside in sparsely populated families with only limited regional homology to one another. However, with this in mind, a degenerate oligonucleotide was constructed based upon the conserved region that includes the cysteine that ligates the heme iron and used to probe genomic DNA from C. braakii. A second protocol used this oligonucleotide in a genomic PCR with another degenerate primer based upon a conserved region seen in the redoxin reductase partners of P450 cam and P450 terp . The latter experiment was based on the hypothesis that the P450 cin gene would be followed directly by those of its redox partners and that the first of these would encode a redoxin reductase as seen in the P450 cam and P450 terp operons. Both approaches were unsuccessful, due in part to the limited sensitivity of the degenerate oligonucleotides. In addition, although the architecture of the P450 cin operon was as expected, the proteins encoded constitute an entirely novel redox system for P450s with no homology to those found with other bacterial P450s.
More protein-specific information was thus required, and this was obtained by isolating native P450 cin from C. braakii. Protein purification proved remarkably straightforward, with essentially homogenous protein obtained in three steps using a protocol based upon that published for P450 cam and P450 BioI . Protein sequence, obtained via trypsinolysis and peptide sequencing, was used to design degenerate PCR primers to either end of a large peptide fragment (Fig. 1). A 92-bp oligonucleotide probe for P450 cin was then generated via genomic PCR. This oligonucleotide was labeled with 32 P by reaction with [␥-32 P]ATP in the presence of polynucleotide kinase, but probing a series of restriction enzyme digests of C. braakii genomic DNA failed to yield usable results. Believing that sensitivity was the source of our problems, we chose to label the probe with [␣-32 P]dCTP in the presence of terminal deoxynucleotidyltransferase (25). TdT is known to transfer up to 30 nucleotides to the 3Ј end of single-stranded DNA, which results in an order of magnitude increase in the specific activity and thus sensitivity of the probe. Utilization of [␣-32 P]dCTP-labeled oligonucleotide resulted in the eventual isolation of DNA fragments containing the P450 cin gene.
Sequencing of the isolated DNA revealed a putative P450 cin operon that contained three open reading frames with overlapping start and stop codons (Fig. 2). The first of these genes clearly encoded a P450, as seen by its strong similarity to a number of other bacterial P450s (Table II). The encoded protein was classified as CYP176A1 and trivially denoted as P450 cin . Although most of the conserved residues found in P450s were also seen in P450 cin , there was one striking exception (Fig. 5). Threonine 252 in P450 cam is located on the I helix, a conserved structural feature of P450s, and is believed to hydrogen-bond to a water molecule during the oxygen activation process and thus direct delivery of a proton to the distal oxygen in the irondioxygen complex (38,39). This specific delivery is believed to result in cleavage of the O-O bond to yield the active ferryl species; delivery of a proton to the proximal oxygen is thought to result in the dissociation of hydrogen peroxide (so-called uncoupling). Given the essential role of this residue in catalysis it is not surprising that a threonine or serine is seen in this position in essentially all known P450s. An exception occurs in P450eryF (CYP107A1), in which it is replaced by an alanine; in this enzyme a hydroxyl group located on the substrate is believed responsible for the positioning of the required water molecule. In P450 cin , however, sequence alignments suggest that the threonine has been replaced with an asparagine. Clearly, the amide functionality of the asparagine could serve to hydrogen-bond to a water molecule and thus direct proton delivery and cleavage of the dioxygen bond. However, this would be the first time that a hydroxyl group has not filled this role, and further experimentation will be required to validate this hypothesis.
The two further genes discovered during the sequencing of the putative operon appear to encode potential redox partners for P450 cin . Although this arrangement is identical to that seen in the P450 cam and P450 terp operons, the identity of these redox partners is quite different. In general there are two types of P450, as classified by their auxiliary electron transport partners. In bacteria and mitochondria (class I P450s) there are two proteins that mediate the transfer of electrons from the pyridine nucleotide to the P450; one is an FAD-containing ferrodoxin reductase, and the other is an iron-sulfur protein, either 2Fe-2S or 3Fe-4S. In eukaryotes (class II P450s), there is generally a single flavoprotein, cytochrome P450 reductase that contains both an FAD and an FMN cofactor, which mediates P450 reduction. Other less common arrangements also exist for electron transfer. P450 nor (40) accepts electrons directly from NADH, whereas P450 BM-3 (41), a bacterial P450, incorporates both a class II-like reductase and a P450 domain into a single polypeptide. Class III P450s, e.g. CYP74A, do not require exogenous redox partners because they act upon peroxide-containing substrates (42)(43)(44). Analysis of the second gene in the P450 cin operon revealed that it had strong homology with adrenoredoxin reductase, the ferrodoxin reductase that functions in the adrenal mitochondrial system (Table II). However, the third gene of the operon appeared to be a flavodoxin and not an iron-sulfur-containing protein, because a BLAST search indicated that it was most homologous to the FMN domain of CPR (Table II). This suggests that P450 cin utilizes a novel, natural redox system for electron transfer, which consists of an FADdependent flavodoxin reductase and an FMN-containing flavodoxin. This system is intermediate between the classical class I bacterial P450 that utilizes two proteins, but which includes one with an Fe-S cluster, and the class II P450 that only employs a single protein but utilizes FAD and FMN cofactors. Although there is no precedent for this type of system occurring naturally, it has been reported that flavodoxins can act as "unnatural" redox partners for P450s. Waterman and co-workers (33,45) have shown that the flavodoxin reductase and flavodoxin systems from both E. coli and Anabaena will serve as functional redox partners for a variety of (heterologous) mitochondrial P450s. It should be noted that there is as yet no firm experimental evidence that links P450 cin with the two proteins as its redox partners. However, as mentioned, the arrangement of the genes within the operon is identical to that seen with both P450 cam and P450 terp and their redox partners.
It is interesting to note that although the biodegradative P450s such as P450 cam , P450 terp , and P450 cin occur in an operon closely associated with their redox partners, this is not the case for many other bacterial P450s uncovered by genome sequencing projects. For example, in Mycobacterium tuberculosis of the ϳ20 P450s identified, at least 14 have no obvious operon-associated redox partners. Perhaps this implies that biodegradative P450s are necessarily organized into operons with their electron transfer proteins to ensure coordinated expression and efficient functioning. Other bacterial P450s, which may not be required to operate as efficiently as biodegradative ones that supply the cell carbon and energy, may employ general redox partners such as cellular flavodoxins or ferrodoxins.
P450 cin was heterologously expressed in E. coli at a level of ϳ2000 nmol/liter of culture. Purification, including a hydroxyapatite column (46) as a final polishing step, gave essentially a homogeneous protein (ϳ1000 nmol/liter) after three steps. Its identity was confirmed by N-terminal sequencing and electrospray mass spectrometry, which also indicated that some Nterminal processing (truncation of up to six amino acids) had occurred in a batch-dependent manner. However, spectral analysis of the protein (see below) revealed no effect of this inhomogeneity of the protein. Although the catalytic activity remains to be explored, it is not expected that this variation will significantly alter its efficiency, given that the N terminus is not implicated in either the mechanism of oxygen activation or the interaction with redox partners in any other bacterial P450 (47).
We postulated that as P450 cin was believed to have a flavodoxin as a redox partner, it might utilize the endogenous E. coli flavodoxin/flavodoxin reductase system. Waterman and co-workers (33,45) had reported that this system was respon- sible for the observed catalytic activity in a variety of heterologously expressed P450s. Thus, we expressed P450 cin in the presence of cineole 1, and careful gas chromatography-mass spectrometry analysis of the organic extracts of the growth medium revealed the presence of a metabolite tentatively identified as 2-hydroxycineole 2. This compound was not produced when cineole 1 was added to the growth medium of E. coli not expressing P450 cin . While not conclusive, this result clearly demonstrates that P450 cin is capable of cineole 1 hydroxylation and strongly suggests that it is involved in its biodegradation.
UV-visible analysis of the heterologously expressed P450 cin revealed the typical characteristics of a bacterial P450 (Fig. 3). Significantly, a spin state change of the heme iron from low spin to high spin was observed in the presence of cineole 1, typical of a bacterial P450 in the presence of its substrate. This and the tight binding of cineole 1 to P450 cin (K D ϭ 0.7 M) strongly support our hypothesis that the natural substrate for this enzyme is cineole 1, because it mirrors exactly the behavior of other terpene-degrading P450s with their substrates, e.g. P450 cam (5), P450 terp (10), and P450 lin (9). These observations combined with our in vivo turnover results also provide the first real evidence that a P450 is involved in the biodegradation of cineole 1. It will be of interest to determine the way in which this strong, specific binding of cineole 1 is achieved because it presents little, if any, opportunity for H-bonding. Such interactions are known to play an important role in the specific binding of other terpene substrates with their biodegradative P450 (48). The oxygen in cineole 1 is present as a ditertiary ether and, as such, is unlikely to participate in any H-bonding. It is hoped that crystallization trials currently underway may shed light on the specificity of this enzyme-substrate interaction.
Conclusion-A new bacterial P450 system has been isolated and the P450 cloned and overexpressed. The high level of expression, as well as the ease of purification, encourage us in the belief that this P450 may serve as a useful model for P450s in general and bacterial P450s in particular. It contains a number of unique features, such as an unusual asparagine substitution for a catalytically essential threonine, which should help to increase our fundamental understanding of P450 mechanism. In addition, it appears that P450 cin is the first known P450 to utilize a flavodoxin as its natural redox partner. Experiments designed to explore further the catalytic activity and the interactions of P450 cin with both its substrate and redox partners are currently in progress.