Molecular and Spectroscopic Analysis of the Cytochromecbb 3 Oxidase from Pseudomonas stutzeri *

Cytochrome cbb 3 oxidase, a member of the heme-copper oxidase superfamily, is characterized by its high affinity for oxygen while retaining the ability to pump protons. These attributes are central to its proposed role in the microaerobic metabolism of proteobacteria. We have completed the first detailed spectroscopic characterization of a cytochromecbb 3 oxidase, the enzyme purified fromPseudomonas stutzeri. A combination of UV-visible and magnetic CD spectroscopies clearly identified four low-spin hemes and the high-spin heme of the active site. This heme complement is in good agreement with our analysis of the primary sequence of theccoNOPQ operon and biochemical analysis of the complex. Near-IR magnetic CD spectroscopy revealed the unexpected presence of a low-spin bishistidine-coordinated c-type heme in the complex. This was shown to be one of two c-type hemes in the CcoP subunit by separately expressing the subunit in Escherichia coli. Separate expression of CcoP also allowed us to unambiguously assign each of the signals associated with low-spin ferric hemes present in the X-band EPR spectrum of the oxidized enzyme. This work both underpins future mechanistic studies on this distinctive class of bacterial oxidases and raises questions concerning the role of CcoP in electron delivery to the catalytic subunit.

The final step in the electron transport chain of mitochondria and aerobically respiring bacteria is the 4-electron reduction of dioxygen to water. This reaction is usually catalyzed by respiratory heme-copper oxidases, e.g. mitochondrial cytochrome c oxidase, integral membrane proteins that couple the free energy of oxygen reduction to the translocation of protons. Members of the superfamily of respiratory heme-copper oxidases are represented in all domains of life and are readily identified by a highly conserved catalytic subunit (subunit I) (1)(2)(3).
The majority of heme-copper oxidases found in eubacteria fall into one of two classes according to their immediate electron donor. This can be a quinol, as in the case of cytochrome bo 3 from Escherichia coli, or a c-type cytochrome, e.g. cytochrome c 552 , which is the electron donor for cytochrome aa 3 of Paracoccus denitrificans. Quinol oxidases lack the dinuclear copper center Cu A , which is located in the hydrophilic domain of subunit II and which serves as the immediate electron acceptor in cytochrome c oxidases (4).
In recent years, a third highly diverged group of heme-copper oxidases, the cytochrome cbb 3 oxidases, have been described in proteobacteria (5). The cbb 3 oxidases are found only in proteobacteria and have been reported not only to have a very high affinity for oxygen (k m ϳ 7 nM) (6), but also to retain the ability to conserve the energy liberated from the oxygen reduction reaction (7)(8)(9). These properties give cbb 3 oxidases an essential role in the specialized energy metabolism that allows proteobacteria to colonize microaerobic environments. For instance, a cbb 3 -type enzyme is the only energy-conserving terminal oxidase whose sequence is represented in the genomes of two important human pathogens, Helicobacter pylori (10) and Campylobacter jejuni (11). Moreover, expression of a cbb 3 oxidase is also necessary for symbiotic diazotrophs to fix dinitrogen. Here, the enzyme fulfills the dual roles of maintaining a very low oxygen tension, to protect the labile nitrogenase, and allowing aerobic respiration, to support the energetically demanding process of nitrogen fixation. Indeed, a cbb 3 oxidase was first identified in Bradyrhizobium japonicum and designated fixNOQP (ccoNOQP) because expression of this gene cluster is required to support symbiotic nitrogen fixation (12).
Subsequently, the fixNOPQ (ccoNOPQ) operon has been identified in other proteobacteria, where it is always close to a second gene cluster, fixGHIS (ccoGHIS), whose expression is required for the assembly of a functional cbb 3 oxidase (13). An analysis of the evolution of heme-copper oxidases based upon multiple sequence alignments of the catalytic subunits, represented in cbb 3 oxidases by CcoN (FixN), has recently been published (3). This suggests that the cbb 3 oxidases evolved independently of the functionally distinct, but structurally related bacterial nitric-oxide reductases (14) to fulfill a specialized role in microaerobic energy metabolism (5).
Cytochrome cbb 3 oxidases are predicted to contain four subunits, although in most preparations reported to date, there is only firm evidence for the presence of CcoNOP. Like subunit I in other heme-copper oxidases, CcoN has at least 12 transmembrane helices and contains the active site, a high-spin heme (heme b 3 ) 1 magnetically coupled to an adjacent copper ion 1 To aid comparison with other heme-copper oxidases, we have adopted the following notation for the metal centers in CcoN, which is structurally related to cytochrome c oxidase subunit I. Heme b is the low-spin bishistidine-coordinated heme in CcoN that is a homolog of heme a in cytochrome c oxidase. Heme b 3 is high spin in CcoN and equivalent to heme a 3 in cytochrome c oxidase and heme o 3 in E. coli cytochrome bo 3 -quinol oxidase. In cytochrome c oxidase and quinol oxidases, this heme is magnetically coupled to a copper ion (Cu B ) to form a dinuclear center, which is the site of oxygen binding and reduction. Cu B is probably ligated by three conserved histidine residues, which also serve as ligands to Cu B in cytochrome c oxidase. known as Cu B to form a dinuclear center (15). A second heme (heme b), which serves to transfer electrons to the active site, is also contained within subunit I. Six absolutely conserved histidine residues (in helices II, VI, VII, and X) involved in ligating both heme irons as well as Cu B are diagnostic of the entire superfamily (16). Rather unusually, CcoN contains only heme B, which lacks the hydroxyethylfarnesyl substituent of the porphyrin macrocycle, found in hemes O and A, the cofactors that characterize the dinuclear centers of most classical hemecopper oxidases.
The x-ray structures of a number of heme-copper oxidases reported over the past 5 years have proved useful in interpreting many years of biophysical experiments on mitochondrial cytochrome c oxidase (17)(18)(19)(20). Unfortunately, few of the structural features that characterize the active site of typical hemecopper oxidases are conserved in the primary amino acid sequence in CcoN. The most notable difference is found in transmembrane helix VI, where, in cytochrome c oxidase, posttranslational modification of a tyrosine residue covalently links it to one of the histidine ligands of Cu B to form a site capable of stabilizing a radical species during dioxygen reduction (21). In the cbb 3 -type oxidases, this tyrosine residue is absent, and there is no obvious functional replacement. Two membraneassociated c-type cytochromes form part of the cbb 3 oxidase complex. The first, CcoO, is a 23-kDa monohemeprotein that appears to be the functional homolog of NorC in the related bacterial nitric-oxide reductase. NorC probably fulfills the same role of the Cu A -containing subunit II in the classical heme-copper oxidases in that it appears to serve as the immediate electron acceptor for soluble electron donors, e.g. pseudoazurin or a cytochrome c in a bc 1 -dependent electron transfer chain (22). The second, CcoP (35 kDa), has two heme c-binding sites clearly identified in the derived amino acid sequence. The function of this diheme cytochrome, for which there is no obvious homolog, is not yet clear. The cbb 3 oxidase operon also includes a fourth gene: ccoQ. With the exception of the enzyme from B. japonicum (23), there is no evidence of the protein it encodes being present in purified complexes, and its function is uncertain.
Understanding the mechanistic basis of energy conservation by bacteria living in microaerobic environments requires a preparation of cbb 3 oxidase that is stable and can be obtained in high yield. Such a preparation has recently been described in Pseudomonas stutzeri, a facultative anaerobe with the capacity to denitrify (24). We report here the first detailed spectroscopic characterization of this enzyme preparation, which we intend to underpin future mechanistic studies. To allow unambiguous spectroscopic assignments of each of the five heme groups, it was necessary to separately express in E. coli the dihemecontaining subunit CcoP.

Recombinant DNA Methods
Initially, the ccoNOQP operon of P. stutzeri was sequenced by amplifying short internal regions of the operon using degenerate primers based on multiple sequence alignments of the ccoNOQP operons of nine other proteobacterial species. All PCRs contained genomic DNA as template, 50 pmol of each oligonucleotide primer (Genosys Biotechnologies, Inc.), and Extensor Hi-Fidelity PCR Master Mix (Abgene) in a total volume of 50 l. Reactions were initiated by 2 min at 94°C, followed by 25 cycles of denaturation (30 s at 94°C), annealing (30 s at 55°C), and extension (1 min/1 kb at 72°C). After the final cycle, completion of polymerization was ensured by an additional incubation at 72°C for 3 min. The PCR products were sequenced using an automated sequencer (ABI Prism 377) at the School of Biological Sciences, University of East Anglia (Norwich, UK).
Using this initial sequence as a template, primers were designed, and overlapping genomic DNA-derived PCR fragments were cloned into either pUC18 or pBluescript KS ϩ (Table I). DNA cloning techniques were performed according to standard protocols (25). The inserts were then sequenced by MWG-BIOTECH AG (Ebersberg, Germany) using standard m13/pUC forward and reverse primers or appropriate internal primers.
To amplify the ccoP gene, forward and reverse primers were designed that contained artificial NdeI and SacI sites, respectively, which facilitated cloning into the expression vector pET21a (Novagen) to yield a new construct, pCcoPall. DNA sequencing (MWG-BIOTECH AG) using standard T7 promoter/terminator primers or internal primers confirmed that the insert was identical in sequence to wild-type ccoP.

Bacterial Strains and Growth Conditions
A single colony was used to inoculate 50 ml of LB medium, and the flask was incubated overnight with agitation (200 rpm) at 37°C. This culture was then used to inoculate 15 liters of LB medium, which was grown in a BioFlow IV fermentor (New Brunswick Ltd.) at 30°C for 28 h with constant stirring (135 rpm.). Cells were harvested at 4°C by centrifugation at 10,000 ϫ g for 20 min and resuspended in cold 20 mM Tris-HCl (pH 7.5) before being rapidly frozen in liquid nitrogen and stored at Ϫ80°C until required.
A single colony of E. coli JM109 (DE3) cells transformed with pEC86 (26) and pCcoPall was selected on L-agar, supplemented with ampicillin (100 g/ml) and chloramphenicol (35 g/ml), and used to inoculate 50 ml of LB medium containing both antibiotics. From this culture, which was incubated overnight with shaking (200 rpm) at 37°C, 1 ml was taken to inoculate 700 ml of TYP medium (supplemented with 100 g/ml ampicillin and 35 g/ml chloramphenicol) in a 2-liter flask. Typically, six 700-ml cultures were grown for 30 h with agitation (200 rpm) at 30°C, yielding ϳ50 g of cells (wet weight). The cells were harvested at 4°C by centrifugation at 10,000 ϫ g for 20 min and resuspended in cold 20 mM sodium phosphate (pH 7.5) before

Purification of Cytochrome cbb 3 Oxidase
Previously frozen P. stutzeri cells were thawed, and the following additions were made (final concentrations given): 2 mM MgCl 2 , 1 mM 4-(2-aminoethyl)benzenesulfonyl fluoride, 1 M leupeptin, 1 M pepstatin, and 5 g/ml DNase. The cells were broken by a single passage through a French pressure cell (1200 p.s.i). Approximately 15 min after the cells were disrupted, EDTA (5 mM final concentration) was added to the cell-free extract, which was then centrifuged at 10,000 ϫ g for 20 min at 4°C to remove any unbroken cells. The resulting supernatant was centrifuged at 160,000 ϫ g for 2 h at 4°C, and the pellet (which contained the cytoplasmic membranes) was resuspended in 40 ml of 20 mM Tris-HCl and 2.5 mM EDTA (pH 7.5). To remove any peripheral membrane proteins, 100 ml of 20 mM Tris-HCl, 500 mM NaCl, 2.5 mM EDTA, and 0.08% (w/v) sodium deoxycholate (pH 7.5) was added dropwise with stirring. After stirring on ice for an additional 5 min, the washed membrane fraction was sedimented by centrifugation at 125,000 ϫ g for 90 min at 4°C. Solubilization of the integral membrane proteins and subsequent purification were essentially as described by Urbani et al. (24). Purified cytochrome cbb 3 oxidase was analyzed by SDS-PAGE and both UV-visible and EPR spectroscopies prior to storage at Ϫ80°C until needed.

Purification of CcoP
Previously frozen E. coli cells containing heterologously expressed CcoP were thawed, and the following additions were made (final concentrations given): 2 mM MgCl 2 , 1 mM 4-(2-aminoethyl)benzenesulfonyl fluoride, 1 M leupeptin, 1 M pepstatin, and 5 g/ml DNase. The cells were broken by a single passage through a French pressure cell (1200 p.s.i.). Approximately 15 min after breaking, EDTA (5 mM final concentration) was added to the cell-free extract, which was then centrifuged at 10,000 ϫ g for 20 min at 4°C to remove any unbroken cells. The resulting supernatant was centrifuged at 31,000 ϫ g for 4 h at 4°C. The pellet (which was enriched in cytoplasmic membranes containing CcoP) was resuspended in 20 mM sodium phosphate and 50 M EDTA (pH 7.5), mixed with an equal volume of 10 M urea to remove any peripheral membrane proteins, and centrifuged at 100,000 ϫ g for 90 min at 4°C. The cytoplasmic membranes were washed with ϳ200 ml of 20 mM sodium phosphate and 50 M EDTA (pH 7.5), followed by centrifugation at 100,000ϫ g for 90 min at 4°C.
The total amount of protein in the washed membranes was estimated by the BCA method (Pierce) using bovine serum albumin (1 mg/ml) as a standard. Solubilization of the membrane proteins was accomplished by stirring a suspension of the cytoplasmic membranes for 30 min at 4°C in 20 mM sodium phosphate and 50 M EDTA (pH 7.5), to which dodecyl-␤-D-maltoside (DM) 2 had been added (2.5 g of detergent/g of total protein). The extracted membranes were sedimented at 160,000 ϫ g for 60 min, and the supernatant (which contained CcoP) was loaded onto a pre-equilibrated 75-ml DEAE fast flow column (2.6 ϫ 15 cm; Amersham Biosciences). The column was washed with 150 ml of 20 mM sodium phosphate, 50 M EDTA, and 0.02% (w/v) DM (pH 7.5) before eluting the bound CcoP with a stepwise salt gradient formed in the same buffer (140 -260 mM NaCl in 20 mM increments). At each step in the gradient, the column was washed with 2 column volumes of buffer containing the appropriate salt concentration. After analysis by SDS-PAGE and both UV-visible and EPR spectroscopies, purified CcoP was stored at Ϫ80°C until needed.

Analytical Methods
Gel Electrophoresis-Protein was analyzed on 15%-SDS-polyacrylamide gels essentially as described by Laemmli (27). Gels were stained to indicate the presence of c-type cytochromes (28) or with Coomassie Brilliant Blue R-250. The following buffer was used for sample preparation irrespective of the method of protein visualization: 6 M urea, 5% (w/v) SDS, 0.1% (w/v) glycerol, and 0.05% (w/v) bromphenol blue. Samples were incubated at 50°C for 5 min directly before loading onto the gels.
Mass Spectrometry-Masses of the individual subunits were deter-mined in protein samples (ϳ5 M) using a surface-enhanced laser desorption/ionization (SELDI) mass spectrometer (Ciphergen) with H4 hydrophobic surface chips according to the manufacturer's protocol. Calibrations using horse heart myoglobin and horseradish peroxidase were done prior to each set of experiments, and the masses obtained are considered to be accurate to within 0.2%. Oxygen Uptake Measurements-The standard reaction medium (10 ml) consisted of air-saturated 20 mM sodium phosphate, 50 M EDTA, and 0.02% DM (pH 7.5). Oxygen consumption was measured polarographically at 25°C with a Clark-type electrode (EDT Instruments, Kent, UK) using 1 mM sodium ascorbate and 0.1 mM TMPD as the artificial electron donors.
Heme Quantification-The amount of heme was determined by the pyridine hemochromogen method using an inverse matrix for simultaneous determination of concentrations of hemes A, B, and C as described by Berry and Trumpower (29).
Spectroscopy-UV-visible electronic absorption spectra were recorded on a Hitachi U3100 spectrophotometer. Because cytochrome cbb 3 oxidase has a very high affinity for oxygen, totally anaerobic conditions are essential for spectroscopic work that requires the analysis of either partly or fully reduced enzyme. These conditions were obtained by allowing protein samples (prepared in a 3.5-ml five-sided cuvette (Hellma)) to stir overnight in an anaerobic (Ͻ1 ppm O 2 ) cabinet (Belle Technology, Portesham, UK). The next morning, the magnetic follower was removed, and the cuvette was made airtight with a tightfitting rubber suba-seal covered by Parafilm. All subsequent additions were made using gas-tight syringes. ␤-D-Glucose was added to a final concentration of 10 mM, followed by glucose oxidase/catalase (2 and 25 units/ml, respectively). The contents of the cuvette were mixed by inversion and left for ϳ30 min to allow for the removal of any traces of oxygen. Reduction of cytochrome cbb 3 oxidase was achieved by the addition of small aliquots of either dithionite (ϳ50 M final concentration) or 2 mM NADH plus 0.1 M phenazine methosulfate (respective final concentrations).
EPR spectra were recorded using an ER-200D X-band spectrometer (Spectrospin, Bruker) equipped with a liquid helium flow cryostat (ESR-9, Oxford Instruments) and interfaced to an ESP1600 computer. Spectra at 9.66 GHz were collected using 2.0-milliwatt (mW) microwave power with a modulation frequency and amplitude of 100 kHz and 1 millitesla, respectively.
Room temperature magnetic circular dichroism (RT-MCD) spectra were recorded on either a Jasco J-500D (240 -1000 nm) or a Jasco J-730 (800 -2000 nm) circular dichrograph. An Oxford Instruments superconducting solenoid with a 25-mm room temperature bore was used to generate magnetic fields of up to 6 teslas. MCD spectral intensities depend linearly on the magnetic field at room temperature and are expressed as ⌬⑀ per unit magnetic field (M Ϫ1 cm Ϫ1 T Ϫ1 ) (31). All samples prepared for magneto-optical spectroscopy were dissolved in deuterated 50 mM HEPES and 50 M EDTA (pH* 7.4), where pH* is the apparent pH of the deuterated buffer. Spectra were exported as ASCII files and replotted in Axum Version 6.0 (MathSoft Inc.).

DNA Sequence of the P. stutzeri ccoNOPQ Operon
The first requirement of this study was to obtain an authentic DNA sequence of the ccoNOQP operon from P. stutzeri. The derived amino acid sequence was expected to inform as to the number of c-type hemes in the complex and their potential second axial ligands and to confirm the presence of the three conserved heme ligands in the catalytic subunit (CcoN). To this end, a series of degenerate primers based on multiple sequence alignments were designed that allowed the amplification and sequencing of ϳ85% of the ccoNOQP operon, but neither the 5Јor 3Ј-regions.
The sequence of the 3Ј-end of the ccoNOQP operon was obtained as follows. Because ccoG, whose expression is required for the assembly of a functional cytochrome cbb 3 , is located immediately downstream of the ccoNOQP operon in Rhodobacter capsulatus, Rhodobacter sphaeroides, P. denitrificans, Pseudomonas aeruginosa, B. japonicum, and Sinorhizobium meliloti, it was possible to design a degenerate primer in the 5Ј-region of ccoG with a high degree of confidence. The PCR product that we obtained clearly contained both the 3Ј-end of ccoP and some 300 bp of ccoG.
The sequence of the 5Ј-end of the ccoNOQP operon was obtained by taking account of the similar organization of the upstream region in both P. stutzeri and P. aeruginosa ( Fig. 1) (30,31). The aerotaxis receptor gene (aer) is located upstream of ccoNOQP in P. aeruginosa. A degenerate primer based on a multiple sequence alignment of four aer sequences was designed, which allowed amplification of both the 5Ј-end of ccoN and a 100-bp region immediately upstream. The size of the PCR product and the sequence data ( Fig. 1) suggest that unlike P. aeruginosa, the ccoNOQP operon in P. stutzeri is not tandemly repeated.
In total, a continuous region of 3579 bp was sequenced (data not shown). This region spans four open reading frames in close proximity (ccoN, ccoO, ccoQ, and ccoP) that together account for 3155 bp of sequence. The amino termini of CcoN, CcoO, and CcoP obtained by direct sequencing of the polypeptides (Ref. 24 and data not shown) are in good agreement with the derived sequence from the corresponding genes. An anaerobox (TTGAT-N 4 -GTCAA), a consensus sequence that recognizes members of the FNR family of transcriptional regulators, is located Ϫ88 and Ϫ102 bp upstream of the ccoN start codon. This is consistent with the observation that FnrA, a genuine FNR-type regulator, is required for the expression of cytochrome cbb 3 oxidase in P. stutzeri (32). The start codon of ccoG lies 137 bp downstream of ccoP. The sequenced region contains an additional 303 bp that encodes the N terminus of CcoG, which has significant homology to CcoG of P. aeruginosa. However, unlike some Rhizobia and P. denitrificans, there is no anaerobox in the ccoG promoter region of P. stutzeri (33).
Multiple sequence alignments indicated the six conserved histidine residues in CcoN responsible for ligating: heme b (His 60 and His 348 ), heme b 3 (His 218 ), and Cu B (His 258 , His 259 , and His 346 ). There are one and two conserved CXXCH motifs in CcoO and CcoP, respectively, that diagnose the covalent binding of c-type hemes (Fig. 2). It was originally assumed that all three c-type hemes would have a methionine as a second axial ligand to the heme iron (12). However, the emergence of more ccoNOQP sequences from diverse proteobacteria has made it apparent that both CcoO and CcoP also contain a single conserved histidine residue that in principle could also fulfill this role (Fig. 2).

Purification and Characterization of Cytochrome cbb 3 Oxidase
To obtain purified cytochrome cbb 3 oxidase from P. stutzeri for spectroscopic analysis, we essentially followed the method recently reported by Urbani et al. (24).
SDS-PAGE-The properties of the cytochrome cbb 3 oxidase preparation used in this study and that reported by Urbani et al. (24) are broadly similar. Analysis of the final purification product on a 15% SDS-polyacrylamide gel followed by staining with Coomassie Blue revealed bands with apparent molecular masses of 42, 32, and 23 kDa. These correspond to CcoN, CcoP, and CcoO, respectively. The additional band around 90 kDa reported by Urbani et al. (24) was not present in our final purification product when a sample buffer containing urea was used to prevent protein aggregation. As is the case with most other previously reported cytochrome cbb 3 oxidase preparations, we found no evidence of CcoQ being part of the enzyme complex after purification. The discrepancy between the apparent molecular mass of CcoN determined by SDS-PAGE and that predicted from its DNA sequence is probably caused by the irregular (e.g. faster) migration of the very hydrophobic CcoN subunit on the gels. Analysis of the holoenzyme on SDS-polyacrylamide gels that were subsequently stained for 3,3Ј,5,5Јtetramethylbenzidine-mediated heme peroxidase activity (28) revealed two strong bands with apparent molecular masses of 32 and 23 kDa. These correspond to the c-type heme containing subunits CcoP and CcoO.
Mass Spectrometry-Analysis of the enzyme complex by SELDI mass spectrometry revealed three components with masses of 23,414, 34,907, and 52,851 Da. These values are averages from three separate protein preparations and are in good agreement with the molecular masses predicted from the gene sequence (CcoO, 23,463 Da; CcoP, 34,984 Da; and CcoN, 53,197 Da) after appropriate adjustment for the presence of the cofactors. Again, there was no evidence of CcoQ in the purified enzyme complex. Given the low molecular mass of CcoQ (6717 Da) that was derived from the gene sequence and its predicted hydrophobicity, it was considered likely that it would be resolved by SELDI mass spectrometry if it were present.
Oxygen Uptake Measurements-Oxygen uptake was measured with 0.1 mM TMPD as the immediate electron donor. No oxygen consumption was observed without the addition of enzyme. By varying the pH while maintaining the ionic strength using 0.1 M MES, 0.1 M Tricine, 0.2 M ethanolamine, 50 M EDTA, and 0.02% (w/v) DM as a buffer, the optimal pH for TMPD-mediated oxidase activity of cytochrome cbb 3 was found to be pH 7.5. Under conditions optimized for pH and ionic strength (20 mM sodium phosphate, 50 M EDTA, and 0.02% (w/v) DM (pH 7.5)), a turnover of 250 electrons/s was recorded. Activity was completely abolished by the addition of 100 M KCN to the assay medium.
Heme Determination-The molar extinction coefficient at 411 nm (the Soret maximum of the oxidized enzyme) was determined from the heme content measured by the pyridine hemochromogen method (29). Each independent measurement was done in parallel with myoglobin (heme B) and horse heart cytochrome c (heme C) standards to take account of the small, but variable degree of incomplete reduction of heme B. This analysis confirmed that the complex contained five hemes, with hemes B and C being present at a molar ratio of 2:3. Calculation of ⑀ 411 using the inverse matrix method described by Berry and Trumpower (29) yielded a value of 5.85 ϫ 10 5 M Ϫ1 cm Ϫ1 (Ϯ5 ϫ 10 3 ; n ϭ 10).
UV-visible Spectroscopy-The electronic absorption spectrum of fully oxidized cytochrome cbb 3 oxidase as isolated has a Soret maximum at 411 nm and two broad, but well defined FIG. 1. Strategy for sequencing the ccoNOQP operon of P. stutzeri. A, organization of the ccoNOQP operon in the closely related P. aeruginosa. In this species, ccoNOPQ is present as a tandem repeat. The aeg gene, which encodes a protein involved in aerotaxis signal transduction, is located upstream of the first copy. The ccoGHIS operon is located downstream. B, organization around the single copy of the ccoNOPQ operon in P. stutzeri. Note the presence of aer and ccoGHIS relative to the single copy. C, enlargement of the P. stutzeri ccoNOQP operon, with the regions cloned and sequenced underlined.
features in the visible region between 530 and 550 nm (Fig. 3). In all preparations that that we have studied so far, there is an additional weaker feature centered at 645 nm. This was assigned to one of a pair of ligand-to-metal charge-transfer bands of the high-spin ferric heme b 3 (see below and Refs. 34 and 35). The intensity of this feature varied somewhat between preparations, probably due to some degree of heterogeneity in the dinuclear center, which is a phenomenon well described in other heme-copper oxidases (36). Upon complete reduction of the anaerobic enzyme with excess dithionite, the Soret peak shifted to 417 nm with a distinct shoulder at 420 nm. Also two features intensified in the visible region at 551 and 521 nm with distinct shoulders at 559 and 528 nm, respectively. The peaks at 417, 521, and 551 nm are characteristic of ferrous c-type hemes, whereas the shoulders at 420, 528, and 559 nm are indicative of ferrous b-type hemes.
RT-MCD Spectroscopy-The UV-visible MCD spectrum informs as to the number and spin state of the ferric hemes present in an enzyme, whereas the NIR-MCD spectrum diagnoses the axial ligands of low-spin ferric hemes. The UV-visible region of the RT-MCD spectrum of fully oxidized cytochrome cbb 3 oxidase as isolated is dominated by a derivative-shaped feature in the Soret region of the spectrum ( crossover ϭ 411 nm) (Fig. 4). The form and intensity of this feature are consistent with the presence of four low-spin ferric hemes, as is the intensity of the trough at 569 nm. We suppose these features arise from the three c-type hemes in CcoO and CcoP and the magnetically isolated low-spin ferric heme b in CcoN.
The relatively weak contribution of a single high-spin ferric heme, arising from the active-site heme b 3 of CcoN, to the UV-visible MCD spectrum would be difficult to see against a background of four low-spin ferric hemes. However, we have previously described a feature in the MCD spectrum of other heme-copper oxidases that corresponds to the ligand-to-metal charge-transfer band seen in the UV-visible electronic absorption spectrum. This feature, known as CT 2 , is one of a pair of ligand-to-metal charge-transfer bands that indicate the presence of high-spin ferric b-or o-type hemes at the active sites of cytochrome bo 3 (34, 35) and bacterial nitric-oxide reductase (37). The energy and intensity of CT 2 depend on the nature of the distal (exogenous) ligand bound to the heme (35,37). A feature corresponding to CT 2 is clearly resolved in the RT-MCD spectrum of cytochrome cbb 3 oxidase (Fig. 4A) and has a minimum at ϳ635 nm that we take to indicate the presence of water as the sixth ligand.
Thus, the UV-visible MCD spectroscopic analysis of the oxidized enzyme complex is consistent with the presence of five hemes, four low-spin and one high-spin, which is in good agreement both with our analysis of the derived amino acid sequence of the ccoNOPQ operon and the analysis of the heme content of the purified enzyme. To obtain information about the axial ligands of the four low-spin hemes, we recorded the NIR-MCD spectrum of oxidized cytochrome cbb 3 oxidase at room temperature. The spectrum contains two prominent, but overlapping features with peaks at 1580 and 1790 nm (Fig. 4B), characteristic of the porphyrin-to-ferric charge-transfer transitions of low-spin ferric hemes. The energies of such transitions report the heme axial ligands (38,39). In the case of oxidized cytochrome cbb 3 , the maxima at 1580 and 1790 nm arise from low-spin ferric hemes that have bishistidine (His/His) and His/ Met axial ligands, respectively. This confirms the presence of both His/His-coordinated hemes (heme b) and His/Met (c-type hemes) as predicted by our analysis of the derived amino acid sequence. However, the relative intensities of the two features indicate the presence of His/His-and His/Met-coordinated lowspin ferric hemes at a ratio of 2:2, and not 1:3 as anticipated. Therefore, of the three low-spin c-type hemes present in CcoO and CcoP, rather unexpectedly, one must have His/His axial ligation.
X-band EPR Spectroscopy-Having established both the number of hemes in the cytochrome cbb 3 complex and axial ligands of the four low-spin hemes, we wished to apply this information to assign the signals in the X-band EPR spectrum of fully oxidized cytochrome cbb 3 oxidase (Fig. 5). In fact, this spectrum is rather complex and clearly contains signals that arise from both high-and low-spin ferric hemes. By analogy to other heme-copper oxidases, the high-spin ferric heme in the dinuclear center (heme b 3 ) is expected to be EPR-silent due to its being weakly coupled to the nearby Cu B (II) ion (40). However, a signal at g ϭ 6.02 must arise from a small amount of uncoupled high-spin ferric heme b 3 . This can be explained in terms of a small proportion of the preparation being damaged or lacking Cu B , something that we have previously observed for cytochrome bo 3 from E. coli (41). This view is consistent with our observation that the X-band EPR spectrum of one preparation of cytochrome cbb 3 contained a much larger high-spin ferric heme signal (data not shown). Oxygen uptake assays done on the same sample revealed it to have one-third of the catalytic activity of the other samples. These results suggest a failure to incorporate Cu B into the anomalous sample, which would account for both the magnetically isolated high-spin heme and the low activity. Signals at g ϭ 4.3 and 2.06 represent minor levels of adventitious Fe(III) and Cu(II), respectively.
The complexity of the EPR signals present in the X-band spectrum of the cytochrome cbb 3 oxidase complex made assigning any given signal to a particular heme very difficult. In particular, we wished to identify the EPR signature of the His/His-ligated c-type hemes that we had identified in our NIR-MCD experiments. To address this issue, CcoP was expressed separately in E. coli with the aim of recording the X-band EPR spectra under similar conditions. The features at g ϭ 2.99, 2.24, and 1.53 are immediately recognizable as the spectrum of a bishistidine-liganded heme, in which the ligand planes are orientated approximately parallel. This is typical for the isolated low-spin heme in the major subunit of all heme-copper oxidases examined and for the equivalent heme in nitric-oxide reductase. This was therefore assigned to the low-spin heme of CcoN. There are clearly at least two additional features overlapping with the g ϭ 2.99 peak (indicated by arrows in Fig. 5), which are undoubtedly the g z signals of two of the other low-spin hemes identified in the MCD. The derivative at g ϭ 3.30 and the additional detail on the g ϭ 2.24 signal were again assigned as the g y features of these other hemes. Because EPR spectra are dependent not only on the nature of the axial ligands, but also on their orientation, it is not straightforward to correlate these species with those identified in the MCD spectra. The small isolated peak at g ϭ 3.47 is the g z feature of the second limiting type of low-spin ferric heme EPR, a "large g max " spectrum for which the g y and g x features are broad and often not detected (42). Again, these features are not unique to any one pair of axial ligands, and there are a number of examples in the literature of such spectra arising from both His/His and His/Met coordination (43)(44)(45).
The addition of reductant (NADH/phenazine methosulfate) to the sample led to the loss of all signals, with the exception of the high-spin ferric heme feature at g ϭ 6.02 and the rhombic low-spin spectrum with g z ϭ 2.99, although the latter was reduced in intensity (data not shown). These were, however, totally abolished by the addition of sodium dithionite.

Expression and Purification of CcoP
E. coli JM109 (DE3) cells doubly transformed with pCcoPall and pEC86 were grown in batch culture, and the extent of cell growth was monitored. Periodically, cells were harvested and analyzed on SDS-polyacrylamide gels that were stained for the presence of c-type hemes. These experiments indicated that the best time to harvest the cells to obtain maximum yields of CcoP was after 30 h, by which time the cells were in mid/late exponential phase (data not shown). Note that heme incorporation was significantly lower before 24 h and that cells harvested after 36 h were entering stationary phase and seemed to contain significant amounts of cytochrome bd oxidase, which made purification of CcoP difficult (see below). Cells harvested at 30 h yielded 5 mg of pure CcoP/liter of culture.
Subcellular fractionation of the harvested cells was done according to the method of Thöny-Meyer et al. (46) and examined for the presence of CcoP using SDS-PAGE followed by heme staining. Recombinant CcoP was clearly visible as a 32-kDa protein that is associated with the inner membrane fraction. A second faint band, which ran at ϳ17 kDa, could sometimes be seen in the heme-stained gel and arises from CcmE, the protein in the cytochrome c maturation pathway responsible for heme insertion (47). SELDI experiments done on two different enzyme preparations of recombinant CcoP both gave a single peak with a mass of 34,943 Da (preparation 1) and 35,057 Da (preparation 2). These are in good agreement with the molecular masses predicted from the derived amino acid sequence and the mass spectroscopic analysis of the holoenzyme (see above).
The only contaminant present in the membrane fraction that contributed to the UV-visible spectrum was cytochrome bd oxidase. Cytochrome bd is readily identified in the reduced spectrum by peaks at 560 and 635 nm. Initially, the presence of cytochrome bd in the membranes was problematic because it eluted from the anion-exchange column at ϳ195 mM NaCl, just before CcoP, which eluted at ϳ220 mM NaCl. Consequently, further purification of CcoP sometimes required rechromatography on a second anion-exchange column with the same shallow stepwise gradient as the first. CcoP obtained by this method, although often partially reduced, was pure as judged by SDS-PAGE analysis (Fig. 6).
Analysis of the heme content of CcoP using the pyridine hemochromogen method confirmed the presence of two c-type hemes and established a molar extinction coefficient for oxidized CcoP: ⑀ 408 ϭ 2.7 ϫ 10 5 M Ϫ1 cm Ϫ1 . Analysis of the samples used for EPR spectroscopy by the pyridine hemochromogen method was used to confirm that no spectroscopic contaminants (in particular, cytochrome bd), were present.

Spectroscopic Analysis of Recombinant CcoP
UV-visible Spectroscopy-The electronic absorption spectrum of the fully oxidized CcoP subunit contains a Soret maximum at 408 nm and two broad, but resolved peaks in the visible region (Fig. 7). There was no feature beyond 551 nm that would indicate the presence of a high-spin ferric heme. Partial reduction of CcoP with ascorbate resulted in a shift in the Soret peak to 414 nm and a decrease in its intensity. It also led to a change in the visible region with the appearance of highly resolved bands at 521 and 551 nm, whose intensity indicated that one of the two c-type cytochromes had become reduced. Subsequent complete reduction with excess dithionite caused the Soret peak to intensify and to move to 416 nm. It also caused an increase in the intensity of the features in the visible region at 521 and 551 nm, consistent with the reduction of a second c-type cytochrome.
X-band EPR Spectroscopy-The X-band EPR spectrum of oxidized CcoP contains signals arising from two low-spin hemes (Fig. 8). The features at g ϭ 2.98, 2.25, and 1.53 are due to a low-spin ferric heme with rhombic symmetry. However, a broader signal is clearly resolved as a shoulder on the left of g z ϭ 2.98. When the microwave power was increased from 2 to 20 mW, the g z feature could be clearly resolved into two signals (g ϭ 3.19 and 2.98) corresponding to two different hemes. The ability to separate them in this manner is a consequence of different spin-relaxation states of the two hemes. The feature at g ϭ 2.47 arises from a heme with His/OH Ϫ ligation. When the sample was incompletely reduced with ascorbate, g z ϭ 2.98 remained. When the sample was then fully reduced with excess dithionite, this feature disappeared (data not shown).
NIR-MCD Spectroscopy-Our EPR measurements indicated that of the two low-spin c-type hemes in CcoP, only one was readily reduced by NADH in the presence of the mediator phenazine methosulfate (reduction potential (E m ) ϳ ϩ80 mV) (Fig. 8). This implies considerable spacing of the reduction potential of the two hemes. Typically, His/Met-ligated c-type hemes have a midpoint reduction potential of over ϩ150 mV and are readily reduced by ascorbate. This suggested that the second heme must have a reduction potential of Ͻ50 mV, a value that would be commensurate with His/His ligation. Confirmation of this came from the NIR-MCD spectrum of recombinant CcoP (data not shown). This reveals the presence of both His/Met-and His/His-coordinated low-spin hemes. Consequently, we can be certain that a His/His-ligated c-type heme is associated with CcoP. DISCUSSION Since they were first recognized as a novel and distinctive class of heme-copper oxidases in B. japonicum, cytochrome cbb 3 oxidases have been purified from several species of proteobac-teria, including P. denitrificans (7), R. sphaeroides (48), R. capsulatus (49) B. japonicum (6), and P. stutzeri (24). Rather poor yields and a tendency for the complex to dissociate have made  detailed biochemical characterization difficult, and a structure is not yet available, although crystallization conditions have been reported (24). However, there is evidence that cytochrome cbb 3 oxidases are characterized by an unusually high affinity for oxygen (k m ϳ 7 nM) (6) combined with an ability to pump protons (9). The present study, using the purified enzyme from P. stutzeri, represents the first detailed spectroscopic characterization of this little understood class of heme-copper oxidases and provides some interesting insights into the organization of the enzyme complex as well as raises some questions concerning the role of CcoP.
To ensure that the enzyme from P. stutzeri really represents a prototypic cytochrome cbb 3 oxidase, we sequenced the gene cluster (ccoNOPQ) that encodes the four structural genes together with the immediate flanking regions of the genome. The derived amino acid sequence of CcoNOP is ϳ80% identical to that of P. aeruginosa enzyme, an observation that suggests that the P. stutzeri enzyme is indeed a typical cytochrome cbb 3 oxidase. The hydrodynamic properties of the purified P. stutzeri cytochrome cbb 3 oxidase, prepared in identical fashion to the enzyme used for the spectroscopic studies described here, have recently been reported (24). Analytical centrifugation combined with mass spectroscopy showed that each complex contained CcoN, CcoP, and CcoO at a ratio of 1:1:1. Biochemical analysis of the enzyme purified for this work is consistent with these findings. Moreover, comparison of the experimentally determined N termini (24) and molecular masses (this study) of CcoN, CcoO, and CcoP is in good agreement with the derived amino acid sequences reported here. These derived amino acid sequences also indicated the presence of three c-type hemes; in common with other cytochrome cbb 3 oxidases, the enzyme isolated from P. stutzeri contains two membrane-anchored subunits that contain c-type cytochromes. These are the monoheme CcoO (23 kDa) and the diheme CcoP (35 kDa), one or both of which may serve to transfer electrons derived from the bc 1 complex to the catalytic subunit. Despite the presence of a single heme c-binding site, CcoO exhibits minimal similarity to known c-type cytochromes (50), with the notable exception of NorC, a membrane-anchored cytochrome c subunit of nitricoxide reductase (51). Equilibrium redox titrations (49) have identified two electrochemically distinct c-type cytochromes with E m ϭ ϩ265 and ϩ320 mV and E m ϭ ϩ385 mV for heme b. Further titrations of a membrane fraction derived from a mutant (M7) that incorporates only CcoO into the enzyme complex allowed the assignment of the ϩ320-mV reduction potential to the c-type heme in this subunit. Mutagenesis studies on the assembly and function of the individual subunits are also consistent with these assignments (23). These results led to the proposal that electrons are passed from the electron donor to heme b via CcoP and CcoO in this order, although the requirement for such a degenerate electron transfer chain is far from clear.
The cytochrome cbb 3 oxidase operon includes a fourth gene, ccoQ, which is predicted to encode a small membrane-bound polypeptide. The presence of CcoQ in a purified complex has so far been demonstrated immunologically only in B. japonicum (52), and its function remains unclear. In-frame deletion mutants of ccoQ constructed in B. japonicum (23) and R. sphaeroides (53) have no apparent effect upon the assembly or activity of cytochrome cbb 3 oxidase, although the amount of cytochrome c appears to be somewhat decreased in the ⌬ccoQ mutant of B. japonicum (23). There is some recent evidence to suggest that, in R. sphaeroides, CcoQ serves as a "transponder" in an as yet undefined signal transduction pathway that controls the expression of photosynthesis-related genes in response to the flux of electrons through cytochrome cbb 3 oxidase.
It has been suggested that this specific role for CcoQ may perhaps be related to the presence of two histidine residues that are conserved in R. sphaeroides and R. capsulatus, but that are not present in non-photosynthetic species (53,54).
The combination of UV-visible and RT-MCD spectroscopies confirmed the presence of five different hemes in the oxidized enzyme. Of these, four are low-spin species, and the fifth is the high-spin heme b 3 in the active site. Because there is no evidence of a significant amount of either high-spin heme b 3 or Cu B in the X-band EPR spectrum of the oxidized enzyme, we conclude that these two species must be magnetically coupled to form an EPR-silent species. Even at higher microwave powers and lower temperatures (4 K), we could find no evidence of the kind of broad EPR features that diagnose weak coupling between heme o 3 and Cu B in the dinuclear center of E. coli cytochrome bo 3 (40,55).
An additional conclusion from our spectroscopic analysis is that all three c-type hemes must be low-spin six-coordinate species. This in itself was not surprising, except that the NIR-MCD spectroscopic data indicated that one of these hemes must have His/His coordination. Isolation of heterologously expressed CcoP confirmed that the His/His-ligated c-type heme is associated with this subunit rather than CcoO, which, as noted earlier, is a distant relative of NorC. Further evidence for the His/His-ligated heme comes from the X-band EPR and UV-visible electronic absorption spectra of CcoP after partial reduction with ascorbate (E m ϳ ϩ50 mV). In general, His/Hisligated c-type hemes have reduction potentials that are ϳ200 mV lower than those of His/Met-ligated species due to the relative stabilization of the ferrous state (50).
Thus far, there are only two published EPR spectra of cytochrome cbb 3 oxidases in the literature, of preparations of the enzyme from R. capsulatus (49) and R. sphaeroides (48), both of which were recorded to seek evidence of a Cu A site. Although rather dilute, both samples generated spectra clearly demonstrating the absence of a dinuclear Cu A center and also indicating a coupled active site along with the presence of low-spin ferric heme. The X-band EPR spectrum of oxidized cytochrome cbb 3 oxidase in this report is rather more complex, with signals arising from at least four low-spin ferric hemes. To interpret this spectrum, we must consider also the MCD data and the X-band EPR spectrum of CcoP. Together, these data provide three important pieces of information. First, the large g max signal in the EPR spectrum of the complex, which resembles a similar signal in the structurally related NorC, almost certainly arises from the single His/Met-coordinated c-type heme in CcoO. Second, the intense well resolved signal that dominates the complex rhombic trio (g z ϭ 2.99, g y ϭ 2.23, and g x ϭ 1.50) is absent from the EPR spectrum of CcoP and was consequently assigned to heme b of CcoN. Finally, the two other contributors to this rhombic trio (g z ϭ 3.19 and 2.98) that appear as shoulders on the main heme b signal are both present in the EPR spectrum of CcoP. The g z ϭ 3.19 signal disappeared upon reduction with ascorbate and was consequently assigned to the His/Met-ligated species, leaving the g z ϭ 2.98 component, which was assigned to the His/His-ligated c-type heme.
It is well established that oxidases, which terminate aerobic respiratory chains, accept relatively high potential electrons from soluble electron donors or directly from quinols. For example, Cu A , which serves as the immediate electron acceptor for c-type cytochromes in aa 3 -type oxidases, has a reduction potential of approximately ϩ240 mV (56). In bacterial nitricoxide reductase, a close relative of the heme-copper oxidases, the role of electron acceptor is fulfilled by NorC, a membranebound protein containing a single c-type heme (EЈ 0 ϭ ϩ310 mV) (43). Consequently, the question arises as to why cytochrome cbb 3 oxidases, which cannot use quinol as an electron donor, apparently require three redox centers to transfer electrons to the catalytic subunit, when in other heme-copper oxidases, one suffices. Moreover, the presence of a c-type heme in CcoP with a reduction potential far lower than that of proposed electron donors raises evident questions concerning function, not least because it has been established that CcoNO subcomplexes from both P. denitrificans (7) and B. japonicum (23) are catalytically competent. The possible role(s) of the CcoP subunit in the cytochrome cbb 3 oxidase complex and/or in the microaerobic response of proteobacteria are the subject of ongoing investigation in our laboratory.