A role for subunit III in proton uptake into the D pathway and a possible proton exit pathway in Rhodobacter sphaeroides cytochrome c oxidase

Protons are transferred from the inner surface of cytochrome c oxidase to the active site by the D and K pathways, as well as from the D pathway to the outer surface by a largely undefined proton exit route. Alteration of the initial proton acceptor of the D pathway, D132, to alanine has previously been shown to greatly inhibit oxidase turnover and slow proton uptake into the D pathway. Here it is shown that the removal of subunit III restores a substantial rate of O(2) reduction to D132A. Presumably an alternative proton acceptor for the D pathway becomes active in the absence of subunit III and D132. Thus, in the absence of subunit III cytochrome oxidase shows greater flexibility in terms of proton entry into the D pathway. In the presence of DeltaPsi and DeltapH, turnover of the wild-type oxidase or D132A is slower in the absence of subunit III. Comparison of the turnover rates of subunit III-depleted wild-type oxidase to those of the zinc-inhibited wild-type oxidase containing subunit III, both reconstituted into vesicles, leads to the hypothesis that the absence of subunit III inhibits the ability of the normal proton exit pathway to take up protons from the outside in the presence of DeltaPsi and DeltapH. Thus, subunit III appears to affect the transfer of protons from both the inner and outer surfaces of cytochrome oxidase, perhaps accounting for the long-observed lower efficiency of proton pumping by the subunit III-depleted oxidase.     

Subunit III of cytochrome c oxidase of Rhodobacter sphaeroides is required to maintain rapid proton uptake through the D pathway at physiologic pH

The catalytic core of cytochrome c oxidase is composed of three subunits where subunits I and II contain all of the redox-active metal centers and subunit III is a seven transmembrane helix protein that binds to subunit I. The N-terminal region of subunit III is adjacent to D132 of subunit I, the initial proton acceptor of the D pathway that transfers protons from the protein surface to the buried active site approximately 30 A distant. The absence of subunit III only slightly alters the initial steady-state activity of the oxidase at pH 6.5, but activity declines sharply with increasing pH, yielding an apparent pK(a) of 7.2 for steady-state O(2) reduction. When subunit III is present, cytochrome oxidase is more active at higher pH, and the apparent pK(a) of steady-state O(2) reduction is 8.5. Single-turnover experiments show that proton uptake through the D pathway at pH 8 slows from >10000 s(-1) in the presence of subunit III to 350 s(-1) in its absence. At low pH (5.5) the D pathway of the oxidase lacking subunit III regains its capacity for rapid proton uptake. Analysis of the F --> O transition indicates that the apparent pK(a) of the D pathway in the absence of subunit III is 6.8, similar to that of steady-state O(2) reduction (7.2). The pK(a) of D132 itself may decline in the absence of subunit III since its carboxylate group will be more exposed to solvent water. Alternatively, part of a proton antenna for the D pathway may be lost upon removal of subunit III. It is proposed that one role of subunit III in the normal oxidase is to maintain rapid proton uptake through the D pathway at physiologic pH.     

Slow proton transfer through the pathways for pumped protons in cytochrome c oxidase induces suicide inactivation of the enzyme

In the absence of subunit III the aa(3)-type cytochrome c oxidase exhibits a shortened catalytic life span (total number of turnovers) due to an increased probability of undergoing irreversible inactivation during steady-state turnover. Inactivation results from structural alteration of the heme a(3)-Cu(B) active site in subunit I [Hosler (2004) Biochim. Biophys. Acta 1655, 332-339]. The absence of subunit III also dramatically slows proton uptake to the active site via the D proton pathway, as well as inhibiting the proton backflow/exit pathway that connects the active site/proton pump with the outer surface of the oxidase complex. Here we demonstrate that these phenomena are linked: slow proton delivery to the active site through these pathways induces suicide inactivation, thus shortening the catalytic life span of the enzyme. Mutations that inhibit the D pathway, but not the K pathway, increase the probability of suicide inactivation. Strong inhibition of the D pathway allows suicide inactivation to occur even in the presence of subunit III. Arachidonic acid, which stimulates proton uptake by the D pathway, retards suicide inactivation. Steady-state turnover in the presence of DeltaPsi and DeltapH, which inhibits proton uptake from the inner surface of the protein, enhances suicide inactivation. Simultaneous inhibition of proton uptake from both sides of the protein by a double mutation affecting the D pathway and the proton backflow/exit pathway greatly shortens the catalytic life span of the oxidase even in the presence of subunit III. Thus, maintenance of rapid proton transfer through the D pathway and the backflow/exit pathway is one mechanism by which subunit III normally functions to prevent suicide inactivation of cytochrome c oxidase. The experiments suggest that increased lifetimes of the heme a(3) oxoferryl intermediates as well as the anionic form of Glu286 of the D pathway cause suicide inactivation in the active site.     

The influence of subunit III of cytochrome c oxidase on the D pathway, the proton exit pathway and mechanism-based inactivation in subunit I

Although subunit III of cytochrome c oxidase is part of the catalytic core of the enzyme, its function has remained enigmatic. Comparison of the wild-type oxidase and forms lacking subunit III shows that the presence of subunit III maintains rapid proton uptake into the D pathway at the pH of the bacterial cytoplasm or mitochondrial matrix, apparently by contributing to the protein environment of D132, the initial proton acceptor of the D pathway. Subunit III also appears to contribute to the conformation of the normal proton exit pathway, allowing this pathway to take up protons from the outer surface of the oxidase in the presence of DeltaPsi and DeltapH. Subunit III prevents turnover-induced inactivation of the oxidase (suicide inactivation) and the subsequent loss of Cu(B) from the active site. This function of subunit III appears partly related to its ability to maintain rapid proton flow to the active site, thereby shortening the lifetime of reactive O(2) reduction intermediates. Analysis of proton pumping by subunit III-depleted oxidase forms leads to the proposal that the trapping of two protons in the D pathway, one on E286 and one on D132, is required for efficient proton pumping.     

Stopped-flow studies of cytochrome oxidase reconstituted into liposomes: proton pumping and control of activity

The transient kinetics of proton pumping and the electron transfer properties of cytochrome oxidase inserted into small unilamellar vesicles have been investigated by stopped-flow spectrophotometry. In the presence of valinomycin, proton pumping and cytochrome c oxidation by cytochrome oxidase are synchronous up to rate constants of approximately 9 sec-1. Moreover, the enzyme depleted of subunit III ("three-less oxidase") was also shown to pump protons, although with a significantly smaller stoichiometry. Thus, subunit III is not the only (or even the main) proton channel, although it may be involved in the regulation of activity. The kinetics of cytochrome c oxidation by COV in the absence and in the presence of ionophores have been investigated. Analysis of the time course of the process in the transient and steady state phases indicates that the onset of control by the electrochemical gradient follows the transfer of four electrons, i.e., one complete turnover of the oxidase. Two possible alternative interpretations for the control of the turnover phase are presented and discussed.     

Proton translocation by a native and subunit III-depleted cytochrome c oxidase reconstituted into phospholipid vesicles. Use of fluorescein-phosphatidylethanolamine as an intravesicular pH indicator

The existence of a proton pump associated with bovine cytochrome c oxidase (EC 1.9.3.1) has over the last few years been a matter of considerable dispute. In an attempt to resolve some of the problems with the measuring system we have synthesized fluorescein-phosphatidylethanolamine which when reconstituted with cytochrome c oxidase into phospholipid vesicles provided a reliable indicator of the intravesicular pH. It was observed that cytochrome c oxidase catalyzed the abstraction of almost 2 protons from the intravesicular medium/molecule of ferrocytochrome c oxidized. In parallel experiments whereby the extravesicular pH was measured with an electrode it was found that the enzyme appeared to be responsible for the appearance of almost 1.0 proton/molecule of ferrocytochrome c oxidized. Taken together these data unequivocally demonstrate that cytochrome c oxidase behaves as a proton pump. Furthermore, the other proton which was abstracted is believed to be used for the process of the reduction of oxygen. Similar experiments were performed with a cytochrome c oxidase preparation which was devoid of subunit III. Under these circumstances the enzyme appeared to be unable to translocate protons across the vesicular membrane but was competent to abstract protons from the intravesicular medium for the reduction of oxygen.     

Surface proton donors for the D-pathway of cytochrome c oxidase in the absence of subunit III

The major proton-transfer pathway into the buried active site of cytochrome c oxidase (CcO) is the D-pathway that begins with the subunit I residue Asp-132 on the inner protein surface (the cytoplasmic surface of the aa3-type CcO of Rhodobacter sphaeroides). Asp-132 is surrounded by residues from both subunits I and III. In the absence of subunit III, CcO retains activity, but the functional characteristics of the D-pathway are significantly altered such that the transfer of protons from Asp-132 into the pathway becomes the rate-limiting step. Determination of the pH-dependence of the rate constant for D-pathway proton uptake during the single-turnover of CcO indicates that the pKa of Asp-132 in the absence of subunit III is approximately 7. The removal of subunit III also allows for alternative surface proton donor/acceptors other than Asp-132. With Asp-132 altered to alanine, the rate constant for D-pathway proton uptake is very slow (5 s(-1)) in the presence of subunit III. Once subunit III is removed, the proton uptake rate constant increases 80-fold, to 400 s(-1). The pKa associated with this uptake is >10, and the initial proton donor/acceptor in D132A III (-) is proposed to be a water of the D-pathway rather than an amino acid residue. Arachidonic acid (Aa), which stimulates the activity of several D-pathway mutant CcOs, appears to become the initial proton donor/acceptor in the absence of subunit III, whether or not Asp-132 is altered. Aa shifts the pKa of the initial proton donor to 7.6 for both wild-type (WT) III (-) and D132A III (-). The results indicate that subunit III creates a barrier that helps prevent protons from donors other than Asp-132 from directly accessing the internal waters of the D-pathway, while the subunit also provides an environment that increases the rate at which Asp-132 transfers protons into the D-pathway.     

Characterization of electron-transfer and proton-translocation activities in bovine heart mitochondrial cytochrome c oxidase deficient in subunit III

The electron-transfer and proton-translocation activities of cytochrome c oxidase deficient in subunit III (Mr 29 884) prepared by native gel electrophoresis [Ludwig, B., Downer, N. W., & Capaldi, R. A. (1979) Biochemistry 18, 1401-1407] have been investigated. This preparation has been depleted of 82-87% of its subunit III content as quantitated by Coomassie Brilliant Blue staining intensity on sodium dodecyl sulfate-polyacrylamide gel electrophoresis and [14C]dicyclohexylcarbodiimide labeling. The maximum rate of electron transfer of the subunit III deficient enzyme at pH 6.5 is 383 s-1, 78% of control enzyme. Neither the high-affinity site (Km = 10(-8) M) nor the low-affinity site (Km = 10(-6) M) of the cytochrome c kinetic interaction with cytochrome c oxidase is affected by the removal of subunit III. Subunit III deficient cytochrome c oxidase retains the ability to bind cytochrome c in both the high- and low-affinity sites as determined in direct thermodynamic binding experiments. Liposomes containing this preparation exhibit a respiratory control ratio [Hinkle, P. C., Kim, J. J., & Racker, E. (1972) J. Biol. Chem. 247, 1338-1341] of 3.9, while liposomes containing control enzyme exhibit a ratio of 4.3, suggesting that they have a similar proton permeability. Vectorial proton translocation initiated by the addition of ferrocytochrome c in liposomes containing subunit III deficient enzyme is decreased by 64% compared to those containing control enzyme. When the proton-translocated to electron-transferred ratio is measured in these phospholipid vesicles at constant enzyme turnover, removal of subunit III from the enzyme decreases the ratio from 0.52 to 0.21, a 60% decrease.(ABSTRACT TRUNCATED AT 250 WORDS)     

Subunit III-depleted cytochrome c oxidase provides insight into the process of proton uptake by proteins

We review studies of subunit III-depleted cytochrome c oxidase (CcO III (-)) that elucidate the structural basis of steady-state proton uptake from solvent into an internal proton transfer pathway. The removal of subunit III from R. sphaeroides CcO makes proton uptake into the D pathway a rate-determining step, such that measurements of the pH dependence of steady-state O(2) consumption can be used to compare the rate and functional pK(a) of proton uptake by D pathways containing different initial proton acceptors. The removal of subunit III also promotes spontaneous suicide inactivation by CcO, greatly shortening its catalytic lifespan. Because the probability of suicide inactivation is controlled by the rate at which the D pathway delivers protons to the active site, measurements of catalytic lifespan provide a second method to compare the relative efficacy of proton uptake by engineered CcO III (-) forms. These simple experimental systems have been used to explore general questions of proton uptake by proteins, such as the functional value of an initial proton acceptor, whether an initial acceptor must be surface-exposed, which side chains will function as initial proton acceptors and whether multiple acceptors can speed proton uptake.     

Spectroscopic and functional properties of subunit III-depleted cytochrome oxidase

Beef heart cytochrome c oxidase has been depleted of subunit III by treatment with chymotrypsin. The removal of subunit III has been evaluated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and gel fluorography of preparations of the oxidase labeled with [14C]dicyclohexylcarbodiimide prior to proteolysis. Removal of subunit III resulted in a perturbation of the visible spectrum of reduced cytochrome oxidase. Subunit III-depleted oxidase is spectroscopically very similar to the oxidase from Paracoccus denitrificans. When reconstituted into liposomes, the depleted enzyme still pumped protons in response to a pulse of reduced cytochrome c. The H+/e- stoichiometry averaged 0.5. Redox-linked proton translocation could be observed only when respiratory control ratios were higher than 3 and the reductant pulse was of a magnitude that allowed for no more than 5 turnovers of the oxidase.     

What is the essential proton-translocating molecular machinery in cytochrome oxidase?

Pulses of O2 added to anaerobic mitochondria in the presence of antimycin, but in the absence of exogenous reductants, led to H+ translocation until the amount of oxidizing equivalents exceeded the number of endogenous reducing equivalents capable of rapid reduction of cytochrome oxidase. This demonstrates that either the heme of cytochrome alpha or that CuA is the redox center, the function of which is coupled to proton translocation in cytochrome oxidase. Chemical labeling of subunit III of cytochrome oxidase by dicyclocarbodiimide (DCCD), or removal of this subunit by treatment of the enzyme at high pH, results in loss of proton translocation by the isolated and membrane-reconstituted enzyme. Possible roles of subunit III in proton translocation are discussed.     

An IgA monoclonal antibody directed against type III antigen on group B streptococci acts as an opsonin

We have investigated the mechanisms by which a murine IgA mAb directed against the type III Ag (IgA anti-III mAb) of group B streptococci (GBS) protects neonatal rats from lethal infection with these organisms. Purified IgA anti-III mAb enhanced phagocytosis of type III GBS by rat peritoneal macrophages in vitro by fourfold compared with phagocytosis of buffer-treated GBS. In the absence of antibody, neonatal rat serum did not promote phagocytosis, but addition of neonatal rat serum to GBS opsonized with IgA anti-III led to a sevenfold increase in phagocytosis. Heat inactivation of C destroyed the ability of neonatal rat serum to enhance phagocytosis in the presence of IgA. C3 deposition was observed when GBS coated with IgA anti-III mAb were incubated in untreated neonatal rat serum or in serum treated with Mg/EGTA. This latter observation suggested that C3 deposition occurred through activation of the alternative pathway. The control IgA mAb MOPC 315 did not enhance GBS ingestion or C3 deposition on GBS. Depletion of C in vivo by using cobra venom factor abolished the protective effect of IgA anti-III mAb in the neonatal rat model. These data suggest that the ability of this IgA to activate C further enhances its opsonic activity and may be essential for its protective effect in vivo.     

Altering conserved lipid binding sites in cytochrome c oxidase of Rhodobacter sphaeroides perturbs the interaction between subunits I and III and promotes suicide inactivation of the enzyme

Subunit III of the three-subunit catalytic core of cytochrome c oxidase (CcO) contains no metal centers, but it does bind two lipids, within a deep cleft, in binding sites conserved from bacteria to humans. Subunit III binds to subunit I, where it prevents the spontaneous suicide inactivation of CcO by decreasing the probability of side reactions at the heme-Cu O2 reduction site in subunit I. Subunit III prevents suicide inactivation by (1) maintaining adequate rates of proton delivery to the heme-Cu active site and (2) stabilizing the structure of the active site during turnover [Mills and Hosler (2005) Biochemistry 44, 4656]. Here, we first show that mutating several individual residues of the conserved lipid binding sites in subunit III disturbs the subunit I-III interface. Then, two lipid binding site mutants were constructed with an affinity tag on subunit III such that the mutant CcOs could be isolated with 100% subunit III. R226A eliminates an ion pair to the phosphate of the outermost lipid of the cleft, while W59A-F86A disrupts interactions with the fatty acid tails of both lipids. Once these mutant CcOs are placed into soybean phospholipid vesicles, where extensive exchange of bacterial for soybean lipids takes place, it is shown that altering the lipid binding sites mimics a major loss of subunit III, even though subunit III is completely retained, in that suicide inactivation becomes much more probable. The rate of proton delivery to the active site remains rapid, ruling out slow proton uptake as the primary reason for increased suicide inactivation upon alteration of the lipid binding sites. We conclude that altering the lipid binding sites of subunit III may promote side reactions leading to suicide inactivation by allowing greater movement to occur in and around the O2 reduction site of subunit I during the catalytic cycle.     

The allosteric ATP-inhibition of cytochrome c oxidase activity is reversibly switched on by cAMP-dependent phosphorylation

In previous studies the allosteric inhibition of cytochrome c oxidase at high intramitochondrial ATP/ADP-ratios via binding of the nucleotides to the matrix domain of subunit IV was demonstrated. Here we show that the allosteric ATP-inhibition of the isolated bovine heart enzyme is switched on by cAMP-dependent phosphorylation with protein kinase A of subunits II (and/or III) and Vb, and switched off by subsequent incubation with protein phosphatase 1. It is suggested that after cAMP-dependent phosphorylation of cytochrome c oxidase mitochondrial respiration is controlled by the ATP/ADP-ratio keeping the proton motive force Deltap low, and the efficiency of energy transduction high. After Ca(2+)-induced dephosphorylation this control is lost, accompanied by increase of Deltap, slip of proton pumping (decreased H(+)/e(-) stoichiometry), and increase of the rate of respiration and ATP-synthesis at a decreased efficiency of energy transduction.     

Disposition of polar and nonpolar residues on outer surfaces of transmembrane helical segments of proteins involved in proton translocation

"Helical wheel" projections of transmembrane helical segments of membrane proteins involved in proton translocation were constructed. The particular proteins studied were the uncF protein subunit of the Escherichia coli proton-ATPase, the uncE protein subunit of the E. coli proton-ATPase, and cytochrome oxidase subunit III. Clear demarcation of polar and nonpolar regions on surfaces of transmembrane helical segments was seen in the uncF protein and in uncE protein helical segment two, but not in uncE protein helical segment one. The transmembrane segment of cytochrome oxidase subunit III which includes the dicyclohexylcarbodiimide (DCCD)-reactive residue was very similar to E. coli uncE protein helical segment two. The DCCD-reactive residue in both was clearly located on a nonpolar surface.     

G204D, a mutation that blocks the proton-conducting D-channel of the aa3-type cytochrome c oxidase from Rhodobacter sphaeroides

Cytochrome c oxidase uses the free energy of oxygen reduction to establish a transmembrane proton gradient. The proton-conducting D-channel in this enzyme is the major input pathway for protons which go to the binuclear center for water formation ("chemical protons") and likely the only input pathway for protons that get translocated across the lipid membrane ("pumped protons"). The D-channel starts at an acidic residue near the protein surface (D132, Rhodobacter sphaeroides numbering) and leads to another acidic residue near the binuclear center. Recent studies have shown that mutants that introduce an additional acidic residue in the channel (N139D) have the remarkable effect of accelerating steady-state oxidase activity but completely eliminating proton pumping. In this work, an aspartic acid was introduced at the position of glycine 204, G204D, which is also within the D-channel, and the effects were examined. In contrast to N139D, the G204D mutation results in a dramatic decrease of the steady-state oxygen reductase activity (<2% of wild type) [Aagaard, A., and Brzezinski, P. (2001) FEBS Lett. 494, 157-160]. The residual activity is not coupled to the proton pump, and furthermore, in reconstituted vesicles the mutant enzyme exhibits a reverse respiration control ratio; i.e., the mutant oxidase activity is stimulated rather than inhibited when working against a protonmotive force. Hence, the mutant behaves very much like the D132N, which blocks proton uptake through the D-channel. Single-turnover experiments show that the rate-limiting step in the reaction of O2 with the fully reduced G204D mutant is the F --> O transition, similar to the D132N mutant. The block of the D-channel in the D132N mutant can be partly bypassed by biochemically removing subunit III from the enzyme, indicating that removal of the subunit reveals an alternate entrance for protons to the channel. However, this is not observed with the G204D mutant. This suggests that the cryptic entrance to the D-channel that is revealed by the removal of subunit III is between the levels of G204 and D132.     

Influence of N-ethoxycarbonyl-2-ethoxy-1,2-dihydroquinoline modification on proton translocation and membrane potential of reconstituted cytochrome-c oxidase support "proton slippage".

Bovine heart cytochrome-c oxidase was reconstituted in liposomes and modified with N-ethoxycarbonyl-2-ethoxy-1,2-dihydroquinoline (EEDQ). EEDQ reacted mainly with subunits II and III and to a lower extent with subunit I, as shown by difference labeling with [14C]dicyclohexylcarbodiimide. EEDQ treatment of cytochrome-c oxidase vesicles influenced ferrocytochrome c-induced proton pumping by reducing maximally the H+/e- stoichiometry from 0.84 (control) to 0.24, but had only small effects on respiration, respiratory control ratio, and proton conductivity of the proteoliposomes. By titrating the reaction rate of the control and the modified cytochrome-c oxidase vesicles versus the membrane potential, as measured with a Ph3MeP+ electrode, saturation curves are obtained, which in both cases approach 225 mV. The ratios of electron transport rates of the two proton pumps at various membrane potentials decrease between 160 and 225 mV from about 2.2 to 1, indicating that the nonlinear flow/force relationship of these proton pumps is at least partly due to "slippage" of proton pumping.     

Alternative initial proton acceptors for the D pathway of Rhodobacter sphaeroides cytochrome c oxidase

To characterize protein structures that control proton uptake, we assayed forms of cytochrome c oxidase (CcO) containing a carboxyl or a thiol group in line with the initial, internal waters of the D pathway for proton transfer in the presence and absence of subunit III. Subunit III provides approximately half of the protein surrounding the entry region of the D pathway. The N139D/D132N mutant contains a carboxyl group 6 ? within the D pathway and lacks the normal, surface-exposed proton acceptor, Asp-132. With subunit III, the steady-state activity of this mutant is slow, but once subunit III is removed, its activity is the same as that of wild-type CcO lacking subunit III (～1800 H+/s). Thus, a carboxyl group～25% within the pathway enhances proton uptake even though the carboxyl has no direct contact with bulk solvent. Protons from solvent apparently move to internal Asp-139 through a short file of waters, normally blocked by subunit III. Cys-139 also supports rapid steady-state proton uptake, demonstrating that an anion other than a carboxyl can attract and transfer protons into the D pathway. When both Asp-132 and Asp/Cys-139 are present, the removal of subunit III increases CcO activity to rates greater than that of normal CcO because of simultaneous proton uptake by two initial acceptors. The results show how the environment of the initial proton acceptor for the D pathway in these CcO forms dictates the pH range of CcO activity, with implications for the function of Asp-132, the normal proton acceptor.     

Transient kinetics of subunit-III-depleted cytochrome c oxidase.

Cytochrome c oxidase from ox heart was depleted of subunit III and its transient kinetic properties studied by stopped-flow and flash photolysis. It was found that the overall mechanism of electron transfer is very similar for subunit-III-depleted and native oxidase, although significant differences in some kinetic parameters have been detected. These include the second-order rate constant for cytochrome c oxidation and the rate-limiting step of the overall process. Moreover, at low cytochrome c/oxidase ratios (where the number of reducing equivalents is insufficient), the rate of reoxidation of cytochrome a was found to be very slow, even in air, and in fact for the subunit-III-depleted enzyme is even slower than for the native oxidase. The stability of reduced cytochrome a excludes the likelihood that removal of subunit III leads to a new O2-binding site, and the result may be relevant to the lowered vectorial H+/e- stoichiometry. The subunit-III-depleted oxidase can be pulsed under appropriate conditions and its combination with CO is unchanged, as shown by kinetic experiments and difference spectroscopy.     

Identification of the structural subunits required for formation of the metal centers in subunit I of cytochrome c oxidase of Rhodobacter sphaeroides

Genetic manipulation of the aa(3)-type cytochrome c oxidase of Rhodobacter sphaeroides was used to determine the minimal structural subunit associations required for the assembly of the heme A and copper centers of subunit I. In the absence of the genes for subunits II and III, expression of the gene for subunit I in Rb. sphaeroides allowed purification of a form of free subunit I (subunit I(a)()) that contained a single heme A. No copper was present in this protein, indicating that the heme a(3)-Cu(B) active site was not assembled. In cells expressing the genes for subunits I and II, but not subunit III, two oxidase forms were synthesized that were copurified by histidine affinity chromatography and separated by anion-exchange chromatography. One form was a highly active subunit I-II oxidase containing a full complement of structurally normal metal centers. This shows that association of subunit II with subunit I is required for stable formation of the active site in subunit I. In contrast, subunit III is not required for the formation of any of the metal centers or for the production of an oxidase with wild-type activity. The second product of the cells lacking subunit III was a large amount of a free form of subunit I that appeared identical to subunit I(a)(). Since significant amounts of subunit I(a)() were also isolated from wild-type cells, it is likely that subunit I(a)() will be present in any preparation of the aa(3)-type oxidase isolated via an affinity tag on subunit I.