Active Site Topologies of Bacterial Cytochromes P450101 (P450

THE JOURNAL 0 1992 by The American Society for Biochemistry and Molecular Biology, h e .

OF BIOLOGICAL C H E M I S ~ ~ Y Vol. 267, , No. 8, Issue of March 15, pp. 5614-5620,1992 Printed in U.S.A.

Active Site Topologies of Bacterial Cytochromes P450101 (P450,,), P450108 (P450,,), and P450102 (P450BM-3) IN SITU REARRANGEMENT OF THEIR PHENYL-IRON COMPLEXES*

(Received for publication, September 16, 1991)

Stephen F. Tuck$, Julian A. Petersong, and Paul R. Ortiz de Montellano$ll From the $Department of Pharmaceutical Chemistry, University of California, San Francisco, California 94143-0446 and the $Department of Biochemistry, University of Texas Southwestern Medical Center, Dallns, Texas 75235-9038

The reactions of cytochromes P460101 (P460,,),

phenyldiazene result in the formation of phenyl-iron complexes with absorption maxima at 474-478 nm. Treatment of the cytochrome P460 complexes with KsFe(CN)e decreases the 474-478 nm absorbance and shifts the phenyl group from the iron to the porphyrin nitrogens. Acidification and extraction of the prosthetic group from each of the ferricyanide-treated enzymes yields a different mixture of the four possible N-phenylprotoporphyrin IX regioisomers. The ratios of the regioisomers with the phenyl ring on pyrrole rings B, A, C, and D (in order of elution from the high performance liquid chromatography column) are, respectively: cytochrome P460,,, 0:O: 1:4; P46OterP, 0:O:o:l; and P460BM-3, 2:10:2:1. The isomer ratio for recombinant cytochrome P ~ ~ O B M . ~ without the cytochrome P460 reductase domain (2:9:2:1) shows that the reductase domain does not detectably perturb the active site topology of cytochrome P46oBM.s. Potassium ions modulate the intensity of the spectrum of the phenyl-iron complex of cytochrome P460,,, but do not alter the N-phenyl isomer ratio. Computer graphics analysis of the crystal structure of the cytochrome P460,, phenyl-iron complex indicates that the active site of cytochrome P460,, is open above pyrrole ring D and, to a small extent, pyrrole ring C, in complete agreement with the observed N-phenylprotoporphyrin IX regioisomer pattern. The regioisomer ratios indicate that the active site of cytochrome P45Oterp is only open above pyrrole ring D, whereas that of cytochrome P460BM.s is open to some extent above all the pyrrole rings but particularly above pyrrole ring A. The bacterial enzymes thus have topologies distinct from each other and from those of the mammalian enzymes so far investigated, which have active sites that are open to a comparable extent above pyrrole rings A and D.

P460108 (P460:,), and P460102 ( P ~ ~ O B M . ~ ) with

* This work was supported by National Institutes of Health Grants GM25515 (to P. R. 0. M.) and GM43479 (to J. A. P). Support for the core facilities of the Liver Center was provided by Grant 5 P30 DK26743. The University of California Computer Graphics facility (R. Langridge, Director) is supported by National Institutes of Health Grant RR 1081. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “adoertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

n To whom correspondence and reprint requests should be ad- dressed.

Cytochrome P450,,’ is a well characterized, soluble bacterial cytochrome P450 enzyme that allows certain strains of Pseudomonas putida to utilize camphor as their sole carbon source (2). The enzyme has been cloned and expressed in Escherichia coli (3), and its primary amino acid sequence has been determined by both protein and DNA sequencing methods (3,4). It is the only cytochrome P450 monooxygenase for which a crystal structure is available (5). The crystal structure shows that the heme is held within cytochrome P450,., by the pincer action of two helices. The L helix lies below the heme and provides the cysteine thiolate ligand to its iron atom, while the I helix lies above the heme and covers pyrrole ring B of the prosthetic group. Binding of potassium ions to the enzyme positively modulates the binding of camphor, the spin state, and the stability of the enzyme (6-8). The crystal structure suggests that a potassium ion binds to the backbone carbonyls of tyrosine 96, glycine 93, glutamates 84 and 94, and the oxygens of ordered waters 517 and 584 (5). Poulos et al. (5) have proposed that the binding of potassium increases the affinity for camphor by decreasing the mobility of tyrosine 96, which forms a hydrogen bond to the camphor carbonyl group. Parallel changes in the free energies for the binding of potassium and camphor when tyrosine 96 is converted to a phenylalanine by site-specific mutagenesis are consistent with this proposal and suggest that the enhanced mobility of tyrosine 96 in the absence of potassium increases the availability of the active site to water (9).

The reaction of phenyldiazene with cytochrome P450,,, yields a stable complex with a red-shifted absorption maximum (10). The complex has been shown by x-ray crystallog- raphy to be a a-bonded phenyl-iron complex in which the phenyl group is bound end-on to the iron atom (10). Non- bonding interactions with the I helix, which lines one edge of the active site, tilt the phenyl group in the complex approximately 10” away from a perpendicular to the heme plane. The reactions of cytochromes P4501A1, -2B1, -2B2, -2E1, -2B4, -2B10, and -2Bll with phenyldiazene yield products which, by spectroscopic criteria, are also phenyl-iron complexes (11,12). The formation of such complexes is confirmed by the fact that they can be anaerobically extracted from the

The abbreviations used are: cytochrome P450,, (P450101), cytochrome P450h, (P450108), and cytochrome P450BM.3 (P450102) according to the nomenclature of Nebert et al. (1); cytochrome P450BM.3-NR, the recombinant enzyme expressed without the reductase domain; PPIX, protoporphyrin I X heme, iron protoporphyrin IX regardless of the iron oxidation or ligation state; N.4, NB, Nc, and No, the N-phenyl-PPIX regioisomers with the phenyl on the nitrogens of pyrrole rings A, B, C, and D, respectively; HPLC, high pressure liquid chromatography; MOPS, 3-(N-morpholino) propanesulfonic acid.

5614

Active Site Topologies of Bacterial Cytochromes P450 5615

proteins and, on exposure to oxygen under acidic conditions, rearrange to give the four possible N-phenyl-PPIX regioiso- mere (10-13). As expected from earlier work with the phenyl- iron complex extracted from myoglobin, rearrangement of the complex outside of the protein results in migration of the phenyl to the four possible pyrrole nitrogens to give equal amounts of the four N-phenyl-PPIX regioisomers (13).

Recent work has shown that the phenyl-iron complexes of the mammalian cytochrome P450 enzymes can be induced to rearrange in situ by treatment with ferricyanide (11,12). The control exerted on the rearrangement reaction by the protein structure directs migration of the phenyl group to selected pyrrole nitrogens. In the case of rat cytochromes P4501A1, -2B1, -2B2, and -2E1, the only products detected in significant amounts were the N-phenyl-PPIX regioisomers NA and ND (11). The same two regioisomers predominated with the rabbit (cytochrome P4502B4), mouse (cytochrome P4502B10), and dog (cytochrome P4502Bll) isozymes, but small amounts of the Nc isomer were also detected with these enzymes (12). These results led to the conclusion that the active sites of the mammalian enzymes were open above pyrrole rings A and D, but relatively closed above pyrrole rings B and C. Unfortu- nately, efforts to validate the method by comparing the pattern of the N-phenyl adducts obtained from the cytochrome P450,,, phenyl-iron complex with the degree of exposure of the pyrrole rings in the crystal structure have been frustrated by the apparent resistance of the cytochrome P450,, complex to rearrangement within the active site (10).

We report here a modified p,rocedure that has allowed us to validate the method with cytochrome P450,,,, and use of the procedure to demonstrate that the active site topologies of two other bacterial cytochrome P450 enzymes, cytochromes P450b,, and P45bM.3, are quite different from each other and from those of the mammalian cytochrome P450 enzymes. Cytochrome P450te, is a recently isolated, soluble bacterial enzyme that specifically oxidizes a-terpineol (14).' Cyto- chrome P45hM.3 is a highly unusual and unique cytochrome P450 enzyme because: (a) it is a soluble bacterial enzyme whose amino acid sequence more closely resembles the sequence of hepatic cytochrome P450IVA1 than the bacterial cytochrome P450 enzymes (15, 16), (b) its redox partner is a flavoprotein (cytochrome P450 reductase) rather than the iron-sulfur protein usually associated with bacterial cytochrome P450 enzymes (16,17), and (c) the cytochrome P450 and cytochrome P450 reductase components are expressed as a single polypeptide in which the cytochrome P450 domain is connected to the reductase domain by a linker peptide (16, 17).

EXPERIMENTAL PROCEDURES

Materials and Methods-Phenyldiazene carboxylate azo ester was purchased from Research Organics, Inc. (Cleveland, OH). Stock solutions of phenyldiazene, typically 2.5 pl of phenyldiazene carboxylate azo ester in 200 pl of argon-saturated 1 M aqueous sodium hydroxide solution, were prepared immediately prior to use and stored on ice. Potassium ferricyanide (99+%) was purchased from Aldrich. Stock solutions of potassium ferricyanide (400 pl, 50 phi) were also prepared immediately prior to use. Methyl viologen was obtained from British Drug House (Poole, Great Britain). All reactions were carried out in 50 mM MOPS buffer freshly prepared with deionized, double-distilled water and neutralized to pH 7.0 with saturated potassium hydroxide solution. Potassium-deficient buffer was obtained by neutralizing with sodium rather than potassium hydroxide. Gel filtration of cytochrome P450- was performed using Sephadex G-25 (20-80 pm) resin equilibrated with 50 mM MOPS (pH 7.0) buffer. HPLC was performed

'Cytochrome P450,, has been assigned as the first member of cytochrome P450 family 108 (D. Nebert, personal communication).

throughout with HPLC grade solvents. All other solvents and reagents were of the highest grade and purity available. The authentic N-phenyl-PPIX regioisomers were prepared from phenylhydrazine- treated equine myoglobin as previously described (13). Cytochrome P450 concentrations were determined from the difference spectrum between the ferrous carbonyl complex of the enzyme and the ferrous, substrate-free form of the enzyme using an extinction coefficient of 91,000 M" cm" at 450 nm (18). Spectroscopic studies were carried out on a Hewlett-Packard 8450A diode array spectrophotometer.

Enzymes-The DNA fragment containing the gene encoding cytochrome P450- was cloned from whole cell DNA prepared from P. putida (ATCC 17453) using published procedures (19). The DNA was digested with HindIII and ligated into pIBI25 which had been simi- larly digested. The ligated DNA was used to transform competent E. coli DH5a cells (20). The transformed cells were plated out on 85- mm plates containing 2 X YT media plus ampicillin (100 pg/ml). The plates were screened for the presence of the clone using a kinase- labeled oligonucleotide probe which was specific for the heme binding region of cytochrome P45OCm. Clones containing the expected 4.2- kilobase insert were selected and grown up overnight in 2 X YT media plus ampicillin. Since the insert could be in either the 5' to 3' or 3' to 5' orientation, correctly oriented clones were selected by restriction digestion with SalI (21). The clones containing the gene for cytochrome P450,, were grown up in TB media (20) containing ampicillin (100 pg/ml) at 37 "C for 24 h using procedures similar to those described for putidaredoxin reductase and putidaredoxin (19). The E. coli DH5a strain and the other enzymes and reagents used for molecular cloning were obtained from Bethesda Research Laborato- ries, Life Technologies, Inc. (Gaithersburg, Maryland). The plasmid pIBI25 was obtained from International Biotechnologies, Inc. (New Haven, CT). The recombinant cytochrome P450,. was purified from E. coli by the procedure of Gunsalus and Wagner (22). Gel filtration was used to remove the camphor from the purified enzyme. For this purpose, an aliquot of cytochrome P450,, (20 p ~ , 0.5 ml) was thawed on ice and chromatographed on a Sephadex G-25 column (12 X 1.5 cm) equilibrated with 50 mM MOPS (pH 7.0). The enzyme typically eluted in a total volume of 1 ml, with a concentration of 7 p~ determined by f117 = 115,000 M" cm" (22).

Cytochromes P ~ ~ O B M . ~ (23) and P ~ ~ % M . ~ - N R were expressed in E. coli and purified to h~mogeneity.~ Cytochrome P450t, was purified from the wild-type Pseudomonas sp which had been grown on a- terpineol as the sole source of carbon and energy (14).' The purification procedure was very similar to that for a purification of cytochrome P450,- (24). The samples were stored in 50-p1 aliquots containing approximately 20 nmol of enzyme in 10 mM potassium phosphate buffer. Each aliquot was thawed only once prior to use to avoid freeze-thaw enzyme loss. Typically, the concentrated enzyme was thawed on ice and diluted to a concentration of 10 ~ L M with 50 mM MOPS (pH 7.0) buffer.

Formation of the Iron-Phenyl Complex-To a solution of the de- sired cytochrome P450 (5 nmol, 500 pl) at 25 "C was added l pl (70 nmol) of phenyldiazene stock solution (0.14 mM final concentration). In experiments with cytochrome P45hM.3, a new chromophore appeared at 474 nm within 10 min of treatment with phenyldiazene, with concomitant loss of Soret absorbance at 416 nm. The 474 nm absorbance appeared as a shoulder on the declining absorbance of the 416 nm peak. After 10 min, no further changes were detected at either 474 or 416 nm. For incubations with cytochromes P ~ ~ O B ~ . ~ - N R , P450,, and P45OCm, complex formation was indicated by a maximum at 478 nm and was complete within 10, 5, and 5 min, respectively,

Actiue Site-directed Iron-Phenyl to Nitrogen-Phenyl Shift-To the solution of the cytochrome P450 iron-phenyl complex was added 1-3 pl (50-150 nmol) of potassium ferricyanide stock solution. For cytochromes P45oBM.3, P~~OBM.,-NR, and P450,,, this resulted in an immediate loss of the iron-phenyl complex chromophore. After 5 min, the solution was added to 5 ml of a freshly prepared solution of 5:95 (v/v) 18 M sulfuric acidacetonitrile. This was allowed to sit for 1 h at 4 "C before the organic phase was removed in uacuo. The residue was redissolved in 2 ml of 0.9 M aqueous sulfuric acid, and the porphyrin adducts were extracted three times with a total of 3 ml of dichloromethane. The organic residue obtained by concentrating the extracts in uacuo was dissolved in 150 pl of HPLC solvent system A and was analyzed as described below. For cytochrome P450,,, addi-

T. Oster, S. S. Boddupalli, and J. A. Peterson, unpublished results. J. A. Peterson, unpublished data.

5616 Active Site Topologies of Bacterial Cytochromes P450 tion of up to 6 p1 of potassium ferricyanide stock solution significantly decreased the intensity of the 478 nm chromophore but did not completely eliminate it. One approach to circumvent this problem was to incubate the complex with ferricyanide for 5 min in a closed vial and then to blow argon over the solution surface for 1 h to produce an anaerobic environment. The modified hemes were then isolated by adding 20 pl of argon-saturated 2 M aqueous hydrochloric acid and extracting the mixture anaerobically with three 250-pl aliquots of argon-saturated butanone containing 0.025% w/v butyl- ated hydroxytoluene as an antioxidant. The combined butanone extracts were then allowed to sit in air for 30 min to promote decomposition of the residual phenyl-iron complexes rather than the iron-to-nitrogen shift of the phenyl group. This procedure avoided non-protein-directed shift of the residual iron-phenyl complex. To the butanone extract was then added 750 p1 of 18 M sulfuric acidacetonitrile (5:95) and 1 ml of 0.9 M aqueous sulfuric acid. The solution was concentrated in vacuo, and 1.5 ml of 0.9 M aqueous sulfuric acid was added. Alternatively, the phenyl-iron complex could be oxidized completely in situ by including methyl viologen in the reaction. To the solution of the cytochrome P450,, phenyl-iron complex was added 5 pl of a 1 mM stock solution of methyl viologen, giving a 1:l molar ratio of the dye to the enzyme complex. Subsequent addition of 3 p1 of ferricyanide stock solution resulted in complete loss of the 478 nm absorbance within 20 min. The modifiedporphyrins were extracted and analyzed by HPLC as described below for the other cytochrome P450 enzymes.

The Effect of Potassium on Iron-Phenyl Complex Formation and Active Site Topology of Cytochromes P450,., and P450,,-The de- sired cytochrome P450 (10 nmol) was prepared in 50 mM MOPS (pH 7.0) buffer neutralized with sodium hydroxide (1 ml), and the solution was split into two equal aliquots. To one aliquot was added 40 pl of 1.25 M aqueous potassium chloride solution, giving a final potassium concentration of 100 mM. Both aliquots were individually treated with phenyldiazene and potassium ferricyanide as described above. The potassium concentration introduced with the ferricyanide (0.1 mM) was 10-100 times lower than the Kd for potassium and is negligible?

HPLC Analysis of N-Phenylprotoporphyrin IX Adducts-100 pl of the porphyrin sample dissolved in HPLC solvent system A (see below) was analyzed by HPLC on a Hewlett-Packard HP 1090 diode array system fitted with an Alltech 5-pm Partisil ODS 3 column. The column was eluted with an 8020 mixture of solvent systems A and B, where system A is 64:l methanol:water:acetic acid, and system B is 10:1 methano1:acetic acid. Elution was isocratic for 30 min followed by a 1-min gradient to 100% B and 5-min isocratic elution at 100% B. The diode array detector was set to monitor the eluent at 416 nm, with a 4 nm bandwidth and with respect to the absorbance at 600 nm. N-Phenyl-PPIX adducts were identified by comparison of their retention times and spectra, and co-elution, with the authentic N- phenyl-PPIX isomers. Quantitation of the peak areas was performed using Hewlett-Packard integration software. Adduct regioisomers could be detected if they corresponded to 2.5% or more of the heme in a normal incubation with 2 nmol of enzyme. A zero is indicated in the product ratios if the isomer is not detected and therefore accounts for less than 2.5% of the original heme.

RESULTS

Cytochrome P450ca,-Reaction of cytochrome P450,,, with phenyldiazene, as previously reported (lo), results within 2- 4 min in the formation of a phenyl-iron complex with an absorption maximum at 478 nm (Fig. 1). Formation of the new absorption is accompanied by a concomitant decrease in the Soret absorption at 416 nm (10). The absorption maximum of the complex was previously reported as 474 nm, but the actual maximum appears to be at 478 nm. The maximum is shifted from 478 to 474 nm due to superposition of the absorbance at 478 nm on the declining edge of a background absorbance. Addition of ferricyanide to the complex repro- ducibly decreases, but does not completely eliminate, the peak at 478 nm even if the amount of ferricyanide or the time of the incubation with the oxidizing agent is increased. However,

S. F. Tuck, J. A. Peterson, and P. R. Ortiz de Montellano, unpublished data.

0.0 1 400 500 600

WAVELENGTH (nm)

FIG. 1. Absorption spectrum changes observed in the reaction of cytochrome P450,, with phenyldiazene. Spectrum of cytochrome P450,, before (-”) and after (- --) reaction with phenyldiazene in K+-containing medium and after further reaction with ferricyanide and methyl viologen (. . . .).

51 4

W

g - P 3 8 9

2 -

10 20 30 TIME (rnin)

FIG. 2. HPLC of the N-phenyl-PPIX adducts. HPLC of the N-phenyl-PPIX adducts obtained from phenyldiazene-treated cytochrome P450,, after oxidation of the intact protein complex with ferricyanide and methyl viologen followed by aerobic extraction (a), ferricyanide alone followed by anaerobic extraction of the prosthetic heme group (b), and ferricyanide oxidation of a reaction in K’- deficient medium followed by anaerobic extraction of the prosthetic heme group (c). HPLC conditions are given under “Experimental Procedures.” The arrows point to the N-phenylprotoporphyrin IX isomers, and the letters indicate the pyrrole ring substituted in each.

inclusion of methyl viologen in the oxidation mixture as a mediator resulted in complete loss of the 478 nm peak. Methyl viologen was found some time ago to facilitate electron transfer from dithionite to cytochrome P450,, (25). The oxidized enzyme complex was worked up by adding it aerobically at 4 “C to 5% (v/v) sulfuric acid in acetonitrile, allowing the mixture to stand for 1 h, concentrating it under vacuum, and extracting the prosthetic group into dichloromethane. HPLC analysis of the prosthetic group extracted from the enzyme oxidized by ferricyanide alone shows that it consists of a 1:1:2:4 mixture of the N-phenyl-PPIX regioisomers Ng, NA, Nc, and ND, respectively (not shown). Analogous analysis of the enzyme fully oxidized by ferricyanide and methyl viologen yielded essentially a 0:0:1:4 ratio of the Ng, NA, Nc, and ND regioisomers (Fig. 2). No more than traces of NB and NA are detected. The identities of the four products are confirmed by


their co-elution and spectroscopic identity with the authentic isomers obtained from phenylhydrazine-treated myoglobin (13).

The ratio of the regioisomers obtained by oxidation of the complex with ferricyanide alone is unreliable because partial retention of the 478 nm absorbance suggests that a part of the complex is resistant to its oxidative action. Previous work has demonstrated that phenyl-iron heme complexes free of the protein matrix rearrange under aerobic, acidic conditions to an essentially equal mixture of the four possible regioisomers of N-phenyl-PPIX (10, 11, 13). It is therefore likely that the regiospecificity of the active site phenyl group migration is partially obscured by rearrangement of the residual phenyl- iron complex to an equal mixture of the four regioisomers outside of the protein. Indeed, the regioisomer ratio obtained when the reaction is flushed with argon before acidifying and anaerobically extracting the prosthetic group, conditions which decompose residual phenyl-iron complexes to products other than N-phenyl-PPIX, produces an NB:NA:Nc:ND regioisomer ratio (0:0:1:4) identical with that obtained from the protein oxidized with ferricyanide and methyl viologen (Fig. 2). It is therefore clear that only the Nc and ND isomers are formed in significant amounts by in situ rearrangement of the complex.

The effect of potassium ions on formation and rearrangement of the phenyl-iron complex was examined to determine if the process is sensitive to the structural changes mediated by binding of this cation. Incubation of cytochrome P450,, with phenyldiazene in a medium containing sodium rather than potassium cations results in the appearance of a relatively broad and attenuated absorption at 474 nm (Fig. 3). In contrast, a normal spectrum is observed in a parallel incubation supplemented with 100 mM KC1 (Fig. 3). Treat- ment of the complex formed in the absence of potassium ions with ferricyanide results in partial loss of the 474 nm band. When the ferricyanide-treated protein is worked up anaerobically as already described to suppress the formation of N- phenyl products outside of the active site, the NB:NA:Nc:ND regioisomers were obtained in a 001:4 ratio (Fig. 2). This is exactly the same ratio as is obtained in the presence of potassium ions. The regiochemistry of the phenyl-nitrogen shift is therefore not sensitive to the differences in the mobility of Tyr-96 mediated by the binding of potassium.

Cytochrome P450,,-The Soret band of cytochrome P450te, is converted within 2-4 min to a new absorption maximum at 478 nm in the reaction of the enzyme with

0.0 I 400 500 600

WAVELENGTH (nm)

FIG. 3. Absorption spectrum of cytochrome P450,, after reaction with phenyldiazene in K+-rich (--) and K+-deficient (Na+-rich) (- - -) buffer.

WAC\\ t + 1

10 20 30 TIME (min)

FIG. 4. HPLC analysis of the N-phenyl-PPIX isomers obtained from phenyldiazene-treated cytochrome P450,, after oxidation of the protein complex with ferricyanide. HPLC conditions are given under “Experimental Procedures.” The peaks at 15-16 min, by spectroscopic and co-elution criteria, are not NE. The arrows point to the N-phenylprotoporphyrin IX isomers, and the letters indicate the pyrrole ring substituted in each.

phenyldiazene (not shown). The phenyl-iron complex sig- nalled by this absorption, in contrast to that of cytochrome P450,.,, is not very stable and decays to unknown products within several minutes of its formation. Ferricyanide must therefore be added to the complex formed with this isozyme as soon as the absorbance at 478 nm ceases to increase. The ferricyanide causes immediate and complete loss of the 478 nm absorbance, showing that the complex of cytochrome P450t,,, in contrast to that of cytochrome P450,.,, is very sensitive to the oxidative action of ferricyanide. In agreement with this inference, HPLC analysis of the prosthetic group extracted aerobically from the ferricyanide-treated enzyme yields exclusively the ND regioisomer of N-phenyl-PPIX (Fig. 4). The absence of other regioisomers clearly shows that the phenyl-iron complex, as expected from the spectroscopic changes, is no longer present when the protein is transferred aerobically to the acid solution. Identical results are obtained if the reaction is repeated in buffer containing sodium rather than potassium salts (not shown), indicating that potassium does not detectably alter the active site topology.

Cytochrome P450BM.3-Incubation of cytochrome P45oBM.3 with phenyldiazene gives rise to the characteristic phenyl- iron heme complex absorption maximum at 474 nm (not shown). Treatment of the protein complex with ferricyanide causes complete disappearance of the 474 nm band, indicating complete elimination of the phenyl-iron complex. The solution therefore need not be made anaerobic prior to acidification and prosthetic group extraction. HPLC analysis of the heme fraction shows that in situ rearrangement of the phenyl- iron complex yields the NB:NA:Nc:ND regioisomers in a 2:10:2:1 ratio (Fig. 5). The phenyl group thus preferentially migrates to pyrrole ring A and, to a lesser extent, to pyrrole rings B and C.6

The intriguing possibility that the regiospecificity of the in situ iron-nitrogen phenyl shift might be sensitive to the binding of cytochrome P450 reductase has been explored using cytochrome P ~ ~ O B M . ~ - N R , a mutant of cytochrome P45OBM+ from which the cytochrome P450 reductase domain has been deleted? The cytochrome P450 domain of cytochrome

Workup of the cytochrome P45oBM.3 reaction with acetic rather than sulfuric acid, as done in earlier work (10-12), yields NB:NA:Nc:ND in a 1:5:3:1 ratio (B. A. Swanson, unpublished results). NA is still the dominant isomer but Nc is a somewhat more important secondary product. Pyrrole ring C may be less rigidly covered in the complex than pyrrole rings B and D.

5618 Active Site Topologies of Bacterial Cytochromes P450

r 1

15 20 30 TIME (min)

FIG. 5. HPLC of the N-phenyl-PPIX regioisomers. HPLC of the N-phenyl-PPIX regioisomers obtained from phenyldiazene- treated cytochrome P 4 5 0 ~ ~ 3 after oxidation of the intact phenyl-iron protein complex with ferricyanide and extraction by the HC1 procedure (a) and by analogous treatment of cytochrome P45&~.3-NF4 the recombinant protein without the flavoprotein reductase domain (21). The HPLC conditions are given under "Experimental Procedures." The arrows point to the N-phenylprotoporphyrin IX isomers, and the letters indicate the pyrrole ring substituted in each.

P45bM-3 was obtained by expressing a suitably truncated gene in E. coli. The spectroscopic properties of the truncated recombinant protein suggest that its active site domain is unperturbed. The reaction of cytochrome p 4 5 b ~ - ~ - N R with phenyldiazene and subsequent treatment with ferricyanide proceeded exactly as with the intact protein containing both the cytochrome P450 and cytochrome P450 reductase domains (not shown). HPLC analysis shows that NB:NA:NC:ND are produced in a 2:92:1 ratio (Fig. 5). This is essentially the same as the ratio obtained with the untruncated protein.

DISCUSSION

The reaction of cytochrome P450 enzymes with phenyldiazene has been shown to yield the corresponding a-bonded phenyl-iron complexes (10-12). A definitive x-ray crystal structure of the phenyl-iron complex has been obtained only for cytochrome P450,,, (lo), but convincing evidence for the formation of similar complexes with other cytochrome P450 enzymes is provided by the demonstration that the prosthetic group of spectroscopically similar complexes can be extracted and converted by oxidizing agents to a roughly equal mixture of the N-phenyl-PPIX regioisomers (11, 12). Formation of the four regioisomers of N-phenyl-PPIX in very different amounts when the phenyl-iron shift occurs within rather than outside the active site presumably reflects the steric environment of the active site and has therefore been used to derive schematic topological maps of the active sites. It has not been possible previously to test the assumption that the regioisomer pattern provides a true representation of the steric environment and can therefore be used to reconstruct the active site topology. This test can be carried out only with cytochrome P450,,, for which a crystal structure is available, but earlier efforts to promote the iron-nitrogen migration of the phenyl group within its active site were unsuccessful (10). As reported here, reaction with ferricyanide alone does, in fact, promote the iron-nitrogen shift, but a fraction of the cytochrome P450,, phenyl-iron complexes is refractory to oxidation under these conditions. The persistence of the phenyl-iron complex spectrum, albeit diminished in intensity, explains the earlier erroneous conclusion that the shift did not occur (10). Resistance of a fraction of the cytochrome P45OC,, complexes to ferricyanide is anomalous because this reagent causes the

complete disappearance of the chromophore of the complex with all the other cytochrome P450 enzymes so far studied. However, the phenyl-iron complex of cytochrome P45OCa, is completely oxidized by ferricyanide if methyl viologen is in- cluded in the reaction as a mediator or effector (Fig. 1). The use of methyl viologen was suggested by the earlier finding that the sluggish reduction of cytochrome P45Ocn, by dithionite is greatly accelerated by the inclusion of methyl viologen (25).

Oxidation of the cytochrome P450- phenyl-iron complex by ferricyanide and methyl viologen, or by ferricyanide alone if the products are worked up anaerobically, produces Nc and ND in a 1:4 ratio (Fig. 2). This result suggests that pyrrole rings A and B are masked in the active site of the enzyme, and that the region above pyrrole ring D is substantially more open than that above pyrrole ring C. This precisely mimics the active site topology deduced from the crystal structure of cytochrome P45OCm. Computer graphics analysis of the region of the heme that is solvent-accessible in the crystal structure of the phenyl-iron complex from which the phenyl group has been removed by computer processing (Fig. 6) shows that pyrrole ring D is completely available to solvent, pyrrole ring C is partially available, and pyrrole rings A and B are completely masked by residues from a portion of the I helix and are therefore unavailable. The general topology inferred from the crystal structure of the phenyl-iron complex is the same as that deduced from the crystal structure of the camphor- bound enzyme (5). The phenyl shift data provide evidence that the active site conformation is approximately the same for cytochrome P450- in solution as in the crystalline state. Most importantly, however, the observation that the N- phenyl isomer pattern faithfully reports the steric environment of the cytochrome P450,,, active site provides strong support for the validity of the regioisomer pattern as a probe of the active site topology.

Binding of potassium to cytochrome P45OC,,, decreases the mobility of tyrosine 96 and increases the rigidity of the active site (6-9). It was initially thought that the resistance of a fraction of the cytochrome P45Ocn, phenyl-iron complex to treatment with ferricyanide might be due to the presence of two enzyme populations, one potassium-bound and the other

FIG. 6. Solvent-accessible surface of the heme group (red) and the residues in the active site (white) determined from the crystal structure of the cytochrome P4SO- phenyl-iron complex (10) after deletion of the phenyl coordinates. The solvent-accessible surfaces were calculated by the method of COMOUY (26). Comparison of the two surfaces shows that the solvent-accessible regions of the heme not masked by active site residues are over pyrrole ring D and, to a smaller extent, pyrrole ring C.


potassium-free. Potassium binding does affect the phenyl- iron complex because reaction of cytochrome P450,, with phenyldiazene in a potassium-deficient medium yields a weaker, less well defined spectrum for the complex (Fig. 3). However, oxidation of the complex with ferricyanide gives the same N-phenyl-PPIX isomer mixture that is obtained when the reaction is carried out in potassium-rich medium (Fig. 2). Potassium binding thus increases complex formation and enhances the spectroscopic definition of the complex, presumably because it decreases the deformability and water acces- sibility of the active site, but it does not alter the active site topology to the extent that it is detectable by the phenyl shift method. The degree of potassium saturation of the enzyme also does not appear to be responsible for the resistance of the phenyl-iron complex to the ferricyanide-mediated iron- nitrogen shift in the absence of methyl viologen because resistance was observed in the potassium-poor and potassium- rich buffers.

Cytochrome P450,,, a second bacterial enzyme with sub- stantial (25%) sequence identity with cytochrome P450,., (14)‘ was examined to determine the degree of topological similarity between the active sites of two closely related bacterial enzymes. Cytochrome P450te, readily forms a phenyl-iron complex, but the complex is unstable, unlike the complexes of cytochrome P450,,, and all other cytochrome P450 enzymes studied to date. It is necessary to add ferricyanide immediately after the complex is formed to successfully promote the iron-nitrogen shift. The sequence of amino acids in cytochrome P450,, immediately adjacent to the conserved Gly-X-X-Thr of the putative I helix (GHDTTSSSSGGA) is significantly more polar than that of any other known cytochrome P450.4 It is possible that the resulting higher polarity of the active site contributes to the instability of the cytochrome P450te, phenyl-iron complex.

Only the No isomer is formed by in situ rearrangement of the cytochrome P450,, phenyl-iron complex (Fig. 4). The conclusion that the active site is primarily open above pyrrole ring D complements the inference that the same region is open in cytochrome P450,,,, although the open region in cytochrome P450,,, but not cytochrome P450,, extends to a small extent over pyrrole ring C . A model of the active site of cytochrome P450,, is given in Fig. 7. Potassium appears not to play the same role in cytochrome P450te, that it does in cytochrome P450,, because the spectrum of the phenyl-iron complex and the N-phenyl-PPIX isomer generated by ferricyanide treatment are the same whether potassium is present or absent from the medium. In addition, Tyr-96 ofcytochrome P450,.,, which mediates the effect of potassium ions (5,9), is

replaced by a cysteine residue in the optimal alignment of cytochromes P450,., and P450teT6 This is consistent with the finding that potassium cations do not alter the observable catalytic or spectroscopic properties of the enzyme:

Cytochrome P450BM.3, in contrast to cytochrome P450,,, is a very different bacterial enzyme from cytochrome P45OC,,. The most notable differences are the fact that the redox partner for cytochrome P450BM.3 is a flavoprotein analogous to the mammalian cytochrome P450 reductase, and the fact that the hemoprotein and flavoprotein are part of the same polypeptide (15-17). In addition, cytochrome P45hM.3 oxidizes fatty acids rather than small monoterpenes (27, 28). It is therefore quite interesting, but perhaps not surprising, that ferricyanide treatment of the cytochrome P 4 5 0 ~ ~ . 3 phenyl- iron complex primarily yields NA and little ND, the dominant or exclusive regioisomer obtained with the other two bacterial enzymes (Fig. 5). The regioisomer pattern is quite different- not only from those of cytochromes P450,, and P450tem, but also from those of the mammalian isozymes investigated to date, all of which yield comparable amounts of NA and ND and, in some cases, minor amounts of Nc (Table I) (11, 12). A schematic model of the cytochrome P450BM-3 active site is given in Fig. 7.

A caveat exists to the conclusion that the active site topology of cytochrome P450BM.3 is quite different from those of cytochromes P450,,,, P450,, and the mammalian enzymes so far studied. The hemoprotein is fully complexed in cytochrome P450BM.3 with the reductase because both are part of the same protein (16, 17), whereas all the other topologies were determined in the absence of the appropriate redox partners. The possibility therefore exists that the differences in active site topologies are to some extent caused by complexation with the electron transfer partner. To examine this question, the reaction of phenyldiazene with a recombinant cytochrome P450BM.3 protein from which the reductase domain had been genetically deleted was examined. The truncated protein forms the phenyl-iron complex normally and gives, after ferricyanide treatment, virtually the same N- phenyl-PPIX regioisomer distribution as is obtained with intact cytochrome P450BM.3 (Fig. 5 and Table I). It is therefore clear that complexation of the reductase and hemoprotein domains does not detectably alter the topology of the active site.

TABLE I Ratio of the N-phenyl-PPIX regioisomers formed by migration of the phenyl group from the iron to the heme pyrrole nitrogens within the

intact protein complexes P450 isozyme Reference

. . . . . . . . . . . . . Me .................... .......................................... 9 1

. . .......................... 4 Y’Y \,Me

r

. . . . . . . . COzH ’ ’ ‘COzH ‘ 1

. , ..g / \ * * . . . . . . . .

Me \‘ D Me

. . . . . . .

. . . . . . . . . . CO,H CO&

A B FIG. 7. Models of the active sites of cytochrome P450t,,, (A)

and cytochrome P4S0eM-s (B) based on the N-phenyl-PPIX regioisomer patterns obtained with the two hemoproteins. The absolute orientation of the heme is assumed in the models to be that found in cytochrome P450,,,.

0:2:01 0:2:03 0:3:02 0:1:02 0:3:2:3 0:3:1:3 0:4:1:2 1:5:3:1 2:102:1 2:9:2:1 001:4 00O:l

11 llb llb 11 12 12 12 This work This work This work This work This work

A value of zero indicates the isomer was not detectable. Isomers can be detected if they correspond to 2.5% or more of the original heme in a normal incubation with 2 nmol of enzyme.

* The values cited here were obtained by the procedure described under “Experimental Procedures.” They differ slightly from the reported values (P4502B1, 0:1:01.7; P4502B2, 0:l:Ol). The earlier values were obtained by a procedure that gives low product yields that are difficult to quantify and are therefore less accurate (11).

5620 Active Site Topologies of Bacterial Cytochromes P450

The structure of cytochrome P450,, has been used to predict the tertiary structures of other forms of the enzyme for which crystallographic information is not available (29- 33). Whether these predictions are valid or not, sequence assignments of over 150 different eucaryotic and procaryotic cytochromes P450 led to the conclusion that certain distin- guishing features are present in all of these enzymes. Promi- nent among these is the presence of an a-helical sequence (I helix) of amino acids centered around a highly conserved threonine (Thr-252 in cytochrome P450,,,) (5, 30). In cytochrome P450,.,, this helix is part of the substrate binding pocket and is thought to participate in oxygen orientation and possibly oxygen activation (5,3435). Most of the regions of the heme ring which are blocked from reaction with the activated phenyl group are covered by amino acid residues from this helix (Fig. 5). The absence of NB from the regioisomer mixture obtained from the cytochrome P450,, complex is expected because pyrrole ring B is completely covered in the enzyme by the I helix. The absence of NB from the regioisomer mixtures obtained with all the isozymes studied to date except cytochrome P450BM.3 is consistent with the inference from sequence correlations that the I helix is a highly conserved domain (29-32). The formation of NB as a minor (40%) product of the phenyl shift reaction with cytochrome P450BM.3, however, suggests that the region above pyrrole ring B is less obstructed by active site residues in cytochrome P450BM.3 than in other isozymes despite conser- vation of the I helix domain. One strong implication of the different regiospecificities of the phenyl shift reaction with the different cytochrome P450 enzymes is that even though the I helix or a structural analogue may be present in all of them, its structural relationship with respect to the heme ring varies.

In conclusion, the correspondence between the active site topology predicted by the phenyl-iron shift and the crystal structure of cytochrome P450,- validates use of the phenyl shift as a predictive topological probe. We have used this predictive ability to study the active site geometries of two other soluble bacterial cytochromes P450: P45hM.3 and P450,,. The former is related to microsomal cytochromes P450, while the latter is similar to cytochrome P450,.,. The predicted active site geometry of cytochrome P450BM-3, with pyrrole ring A primarily accessible to solvent, is dramatically different from that of cytochromes P450,,, and P450,,, in which ring D is predominantly accessible to solvent. We conclude from these results that there are significant structural differences in the active site geometries of these two different classes of proteins.

Acknowledgments-We thank Julia A. Fruetel for a sample of cytochrome P450,,, Michelle Jacobsen for samples of cytochrome P450@,p, P450BM.3, and P450F"-NR, Dr. Barbara A. Swanson for preliminary work on the cytochrome P45oBM.3 system, and Dr. Sandra Graham-Lorence for computer graphics analysis of the solvent- ac- cessible surface in the crystal structure of the cytochrome P450,, phenyl-iron complex.

REFERENCES 1. Nebert, D. W., Nelson, D. R., Coon, M. J., Estabrook, R. W.,

Feyereisen, R., Fujii-Kuriyama, Y., Gonzalez, F. J., Guengerich, F. P., Gunsalus, I. C., Johnson, E. F., Loper, J. C., Sato, R.,

Waterman, M. R., and Waxman, D. J. (1991) DNA Cell Biol.

2. Sligar, S. G., and Murray, R. I. (1986) in Cytochrome P-450: Structure, Mechanism, and Biochemistry (Ortiz de Montellano, P. R., ed) pp. 429-503, Plenum Publishing Corp., New York

3. Unger, B. P., Gunsalus, I. C., and Sligar, S. G. (1986) J. Biol.

4. Haniu, M., Armes, L. G., Yasunobu, K. T., Shastry, B. A., and Gunsalus, I. C. (1982) J. Biol. Chem. 2 5 7 , 12664-12671

5. Poulos, T. L., Finzel, B. C., and Howard, A. J. (1987) J. Mol.

6. Peterson, J. (1971) Arch. Biochem. Biophys. 144,678-693 7. Lange, R., Hui Bon Hoa, G., Debey, P., and Gunsalus, I. C. (1979)

8. Hui Bon Hoa, G., and Marden, M. (1982) Eur. J. Biochem. 124,

9. Di Primo, C., Hui Bon Hoa, G., Douzou, P., and Sligar, S. (1990) J. Biol. Chem. 265,5361-5363

10. h a g , R., Swanson, B. A., Poulos, T. L., and Ortiz de Montellano, P. R. (1990) Biochemistry 29,8119-8126

11. Swanson, B. A., Dutton, D. R., Lunetta, J. M., Yang, C. S., and Ortiz de Montellano, P. R. (1991) J. Biol. Chem. 266 , 19258- 19264

12. Swanson, B. A., Bornheim, L. M., Halpert, J. R., and Ortiz de Montellano, P. R. (1991) Arch. Biochem. Biophys. 292,42-46

13. Swanson, B. A., and Ortiz de Montellano, P. R. (1991) J. Am.

14. Lu, J.-Y., and Peterson, J. A. (1990) FASEB J. 4 , A2243 15. Fulco, A. J. (1991) Annu. Reu. Pharmacol. Toricol. 31 , 177-203 16. Ruettinger, R. T., Wen, L.-P., and Fulco, A. J. (1989) J. Biol.

17. Nahri, L. O., and Fulco, A. J. (1987) J. Biol. Chem. 262 , 6683-

18. Omura, T., and Sato, R. (1964) J. Biol. Chem. 239, 2370-2378 19. Peterson, J. A., Lorence, M. C., and Amarneh, B. (1990) J. Biol.

Chem. 265,6066-6073 20. Sambrook, J., Fritsch, E. F., and Maniatis, T. (1989) Molecular

Cloning: A Laboratory Manual, 2nd Ed, pp. 1.98-1.99, Cold

York Spring Harbor Laboratory Press, Cold Spring Harbor, New

21. Unger, B. P., Sligar, S. G., and Gunsalus, I. C. (1986) in The Bacteria: A Treatise on Structure and Function (Sokatch, J. R., ed) Vol. X, pp. 557-589, Academic Press, San Diego

22. Gunsalus, I. C., and Wagner, G. C. (1978) Methods Enzymol. 5 2 ,

23. Boddupalli, S. S., Estabrook, R. W., and Peterson, J. A. (1990) J. Biol. Chem. 266,4233-4239

24. OKeefe, D. H., Ebel, R. E., and Peterson, J. A. (1978) Methods Enzymol. 6 2 , 151-157

25. Peterson, J. A., White, R. E., Yasukochi, Y., Coomes, M. L., O'Keefe, D. H., Ebel, R. E., Masters, B. S. S., Ballou, D. P., and Coon, M. J. (1977) J. Biol. Chem. 252,4431-4434

10 , 1-14

Chem. 261,1158-1163

Biol. 195,687-700

Eur. J. Biochem. 94,491-496

311-315

Chem. SOC. 113,8146-8153

Chem. 264,10987-10995

6690

166-188

26. Connolly, M. L. (1983) Science 221 , 709-713 27. Miura, Y., and Fulco, A. J. (1975) Biochim. Biophys. Acta 388,

28. Ho, P. P., and Fulco, A. J. (1976) Biochim. Biophys. Acta 431 ,

29. Nelson, D. R., and Strobel, H. W. (1988) J. Biol. Chem. 263 ,

30. Nelson, D. R., and Strobel, H. W. (1989) Biochemistry 28,656- 660

31. Edwards, R. J., Murray, B. P., Boobis, A. R., and Davies, D. S. (1989) Biochemistry 28.3762-3770

32. Tretiakov, V. E., Degtyarenko, K. N., Uvarov, V. Y., and Ar-

33. Zvelebil, M. J. J. M., Wolf, c. R., and Sternberg, M. J. E. (1991) chakov, A. I. (1989) Arch. Biochem. Biophys. 276,429-439

Protein Eng. 4, 271-282 34. Imai, M., Shimada, H., Watanabe, Y., Matsushima-Hibiya, Y.,

Makino, R., Koga, H., Horiuchi, T., and Ishimura, Y. (1989) Proc. Natl. Acad. Sci. U. S. A.. 8 6 , 7823-7827

35. Martinis, S. A., Atkins, W. M., Stayton, P. S., and Sligar, S. G. (1989) J. Am. Chem. SOC. 111,9252-9253

305-317

249-256

6038-6050

Documents

Active Site Topologies of Bacterial Cytochromes P450101 (P450