Characterization of the low-temperature intermediates of the reaction of fully reduced soluble cytochrome oxidase with oxygen by electron-paramagnetic-resonance and optical spectroscopy.

The reaction of fully reduced soluble bovine heart cytochrome oxidase with O2 at 173K was investigated by low-temperature optical and e.p.r. spectroscopy, and the kinetics of the reaction were analysed by non-linear optimization techniques. The only e.p.r. signals seen during the course of the reaction are those attributable to low-spin cytochrome a3+ and CuA2+. Quantitative analysis of e.p.r. signals shows that, at the end point of the reaction at 173K, nearly 100% of CuA is in the cupric state but only about 40% of cytochrome a is in the ferric low-spin state. The optical spectra recorded at this stage of the reaction show incomplete oxidation of haem and the absence of a 655 nm absorption band. The only reaction scheme that accounts for both the e.p.r. and optical data is a four-intermediate mechanism involving a branching pathway. The reaction is initiated when fully reduced cytochrome oxidase reacts with O2 to form intermediate I. This is then converted into either intermediate IIA or intermediate IIB. Of these, intermediate IIB is a stable end product at 173 K, but intermediate IIA is converted into intermediate III, which is the stable state at 173 K in this branch of the mechanism. The kinetic analysis of the e.p.r. data allows the unambiguous assignments of the valence states of cytochrome a and CuA in the intermediates. Intermediate I contains cytochrome a2+ and CuA+, intermediate IIA contains low-spin cytochroma a3+ and CuA+, intermediate IIB contains cytochrome a2+ and CuA2+, and intermediate III contains low-spin cytochrome a3+ and CuA2+. The electronic state of the O2-binding CuBa3 couple during the reoxidation of cytochrome oxidase is discussed in terms of an integrated structure containing CuB, cytochrome a3 and O2.     

Characterization of the intermediates in the reaction of mixed-valence state soluble cytochrome oxidase with oxygen at low temperatures by optical and electron-paramagnetic-resonance spectroscopy.

The reaction of soluble mixed-valence-state (a3+CuA 2+.CuB + A32+) cytochrome oxidase with O2 at low temperature was studied by optical and e.p.r. spectroscopy. The existence of three intermediates [Clore & Chance (1978) Biochem. J. 173, 799-8101] was confirmed. From the e.p.r data it is clear that cytochrome a and CuA remain in the low-spin ferric and cupric states respectively throughout the reaction. No e.p.r. signals attributable to cytochrome a3 or CuB were seen in the intermediates. The difference spectra (intermediates minus unliganded mixed-valence-state cytochrome oxidase) and absolute spectra of the three intermediates were obtained. The chemcal nature of the three intermediates is discussed in terms of their spectroscopic properties. A catalytic cycle for cytochrome oxidase is proposed.     

A new ruthenium complex to study single-electron reduction of the pulsed O(H) state of detergent-solubilized cytochrome oxidase

The first step in the catalytic cycle of cytochrome oxidase, the one-electron reduction of the fully oxidized enzyme, was investigated using a new photoactive binuclear ruthenium complex, [Ru(bipyrazine)2]2(quaterpyridine), (Ru2Z). The aim of the work was to examine differences in the redox kinetics resulting from pulsing the oxidase (i.e., fully reducing the enzyme followed by reoxidation) just prior to photoreduction. Recent reports indicate transient changes in the redox behavior of the metal centers upon pulsing. The new photoreductant has a large quantum yield, allowing the kinetics data to be acquired in a single flash. The net charge of +4 on Ru2Z allows it to bind electrostatically near CuA in subunit II of cytochrome oxidase. The photoexcited state Ru(II*) of Ru2Z is reduced to Ru(I) by the sacrificial electron donor aniline, and Ru(I) then reduces CuA with yields up to 60%. A stopped-flow-flash technique was used to form the pulsed state of cytochrome oxidase (the "OH" state) from several sources (bovine heart mitochondria, Rhodobacter sphaeroides, and Paracoccus denitrificans). Upon mixing the fully reduced anaerobic enzyme with oxygenated buffer containing Ru2Z, the oxidized OH state was formed within 5 ms. Ru2Z was then excited with a laser flash to inject one electron into CuA. Electron transfer from CuA --> heme a --> heme a3/CuB was monitored by optical spectroscopy, and the results were compared with the enzyme that had not been pulsed to the OH state. Pulsing had a significant effect in the case of the bovine oxidase, but this was not observed with the bacterial oxidases. Electron transfer from CuA to heme a occurred with a rate constant of 20,000 s-1 with the bovine cytochrome oxidase, regardless of whether the enzyme had been pulsed. However, electron transfer from heme a to the heme a3/CuB center in the pulsed form was 63% complete and occurred with biphasic kinetics with rate constants of 750 s-1 and 110 s-1 and relative amplitudes of 25% and 75%. In contrast, one-electron injection into the nonpulsed O form of the bovine oxidase was only 30% complete and occurred with monophasic kinetics with a rate constant of 90 s-1. This is the first indication of a difference between the fast form of the bovine oxidase and the pulsed OH form. No reduction of heme a3 is observed, indicating that CuB is the initial electron acceptor in the one-electron reduced pulsed bovine oxidase.     

Cytochrome oxidase genes from Thermus thermophilus. Nucleotide sequence and analysis of the deduced primary structure of subunit IIc of cytochrome caa3.

Cytochrome caa3, a cytochrome c oxidase from Thermus thermophilus, is a two-subunit enzyme containing the four canonical metal centers of cytochrome c oxidases (cytochromes a and a3; copper centers CuA and CuB) and an additional cytochrome c. The smaller subunit contains heme C and was termed the C-protein. We have cloned the genes encoding the subunits of the oxidase and determined the nucleotide sequence of the C-protein gene. The gene and deduced primary amino acid sequences establish that both the gene and the protein are fusions with a typical subunit II sequence and a characteristic cytochrome c sequence; we now call this subunit IIc. The protein thus appears to represent a covalent joining of substrate (cytochrome c) to its enzyme (cytochrome c oxidase). In common with other subunits II, subunit IIc contains two hydrophobic segments of amino acids near the amino terminus that probably form transmembrane helices. Variability analysis of the Thermus and other subunit II sequences suggests that the two putative transmembrane helices in subunit II may be located on the surface of the hydrophobic portion of the intact cytochrome oxidase protein complex. Also in common with other subunits II is a relatively hydrophilic intermembrane domain containing a set of conserved amino acids (2 cysteines and 2 histidines) which have previously been proposed by others to serve as ligands to the CuA center. We compared the subunit IIc sequence with that of related proteins. N2O reductase of Pseudomonas stutzeri, a multi-copper protein that appears to contain a CuA site (Scott, R.A., Zumft, W.G., Coyle, C.L., and Dooley, D.M. (1989) Proc. Natl. Acad. Sci. U.S.A. 86, 4082-4086), contains a 59-residue sequence element that is homologous to the "CuA sequence motif" found in cytochrome oxidase subunits II, including all four putative copper ligands. By contrast, subunit II of the Escherichia coli quinol oxidase, cytochrome bo, also contains a region homologous to the CuA motif, but it lacks the proposed metal binding histidine and cysteine residues; this is consistent with the apparent absence of CuA from cytochrome bo.     

Evidence for a band III analogue in the near-infrared absorption spectra of cytochrome c oxidase.

Ground state near-infrared absorption spectra of fully reduced unliganded and fully reduced CO (a2+ CuA+ a3(2+)-CO CuB+) cytochrome c oxidase were investigated. Flash-photolysis time-resolved absorption difference spectra of the mixed-valence (a3+ CuA2+ a3(2+)-CO CuB+) and the fully reduced CO complexes were also studied. A band near 785 nm (epsilon approximately 50 M-1cm-1) was observed in the fully reduced unliganded enzyme and the CO photoproducts. The time-resolved 785 nm band disappeared on the same timescale (t1/2 approximately 7 ms) as CO recombined with cytochrome a3(2+). This band, which is attributed to the unliganded five coordinate ferrous cytochrome a3(2+), has some characteristics of band III in deoxy-hemoglobin and deoxy-myoglobin. A second band was observed at approximately 710 nm (epsilon approximately 80 M-1cm-1) in the fully reduced unliganded and the fully reduced CO complexes. This band, which we assign to the low spin ferrous cytochrome a, appears to be affected by the ligation state at the cytochrome a3(2+) site.     

Zinc cytochrome c fluorescence as a probe for conformational changes in cytochrome c oxidase.

Zinc cytochrome c forms tight 1:1 complexes with a variety of derivatives of cytochrome c oxidase. On complex-formation the fluorescence of zinc cytochrome c is diminished. Titrations of zinc cytochrome c with cytochrome c oxidase, followed through the fluorescence emission of the former, have yielded both binding constants (K approximately 7 x 10(6) M-1 for the fully oxidized and 2 x 10(7) M-1 for the fully reduced enzyme) and distance information. Comparison of steady-state measurements obtained by absorbance and fluorescence spectroscopy in the presence and in the absence of cyanide show that it is the reduction of cytochrome a and/or CuA that triggers a conformational change: this increases the zinc cytochrome c to acceptor (most probably cytochrome a itself) distance by some 0.5 nm. Ligand binding to the fully oxidized or fully reduced enzyme leaves the extent of fluorescence quenching unchanged, whereas binding of cyanide to the half-reduced enzyme (a2+CuA+CuB2+-CN(-)-a3(3+)) enhances fluorescence emission relative to that for the fully reduced enzyme, implying further relative movement of donor and acceptor.     

Extended X-ray absorption fine structure of copper in CuA-depleted, p-(hydroxymercuri)benzoate-modified, and native cytochrome c oxidase

Cytochrome c oxidase contains four redox-active metal centers: two heme irons, cytochromes a and a3, and two copper ions, CuA and CuB. Due to the paucity of spectroscopic signatures for both copper sites in cytochrome c oxidase, the ligands and structures for these sites have remained ambiguous. The specific depletion of CuA from the p-(hydroxymercuri)benzoate- (pHMB-) modified cytochrome c oxidase recently reported [Gelles, J., & Chan, S. I. (1985) Biochemistry 24, 3963-3972] is herein described. Characterization of this enzyme shows that the structures of the remaining metal centers are essentially unperturbed by the CuA modification and depletion (P. M. Li, J. Gelles, and S. I. Chan, unpublished results). Copper extended X-ray absorption fine structure (EXAFS) measurements on the CuA-depleted cytochrome c oxidase reveal coordination of three (N, O) ligands and one (S, Cl) ligand at the CuB site. Comparison of EXAFS results obtained for the CuA-depleted, pHMB-modified, and "unmodified control" enzymes has allowed the deconvolution of the EXAFS in terms of the inner coordination spheres for CuA as well as CuB. On the basis of these data, it is found that the structure for the CuA site is consistent with two (N, O) ligands and two S ligands.     

Interactions of sulphide and other ligands with cytochrome c oxidase. An electron-paramagnetic-resonance study.

The effect of sulphide on resting oxidized cytochrome c oxidase was studied by both e.p.r. and optical-absorption spectroscopy. Excess sulphide causes some reduction of cytochrome a, CuA and CuB, and the formation of the cytochrome a3-SH complex after about 1 min. After several hours in the presence of excess sulphide only the e.p.r. signals due to low-spin ferricytochrome a3-SH persist, giving a partially reduced species. Re-oxidation of this partially reduced sulphide-bound enzyme by ferricyanide makes all of the metal centres except CuB detectable by e.p.r. We conclude that sulphide has reduced and binds to CuB as well as to ferricytochrome a3. Sulphide binding to cuprous CuB may raise its mid-point potential and make re-oxidation difficult. Addition of reductant (ascorbate + NNN'N'-tetramethyl-p-phenylenediamine) and sulphide together to the oxidized resting enzyme produces a species in which cytochrome a and CuA are nearly completely reduced and cytochrome a3 is e.p.r.-detectable as approx. 80% of one haem in the low-spin sulphide-bound complex. The g = 12 signal of this partially reduced derivative is almost unchanged in magnitude relative to that of the resting enzyme; this suggests that the g = 12 signal may arise from less than 20% of the enzyme and that it may be relatively unreactive to both ligation and reduction. Such a reactivity pattern of the g = 12 form of the oxidase is also demonstrated with the ligands F- and NO, which are thought to bind to cytochrome a3 and CuB respectively.     

The mechanism by which oxygen and cytochrome c increase the rate of electron transfer from cytochrome a to cytochrome a3 of cytochrome c oxidase

When cytochrome c oxidase is isolated from mitochondria, the purified enzyme requires both cytochrome c and O2 to achieve its maximum rate of internal electron transfer from cytochrome a to cytochrome a3. When reductants other than cytochrome c are used, the rate of internal electron transfer is very slow. In this paper we offer an explanation for the slow reduction of cytochrome a3 when reductants other than cytochrome c are used and for the apparent allosteric effects of cytochrome c and O2. Our model is based on the conventional understanding of cytochrome oxidase mechanism (i.e. electron transfer from cytochrome a/CuA to cytochrome a3/CuB), but assumes a relatively rapid two-electron transfer between cytochrome a/CuA and cytochrome a3/CuB and a thermodynamic equilibrium in the "resting" enzyme (the enzyme as isolated) which favors reduced cytochrome a and oxidized cytochrome a3. Using the kinetic constants that are known for this reaction, we find that the activating effects of O2 and cytochrome c on the rate of electron transfer from cytochrome a to cytochrome a3 conform to the predictions of the model and so provide no evidence of any allosteric effects or control of cytochrome c oxidase by O2 or cytochrome c.     

The mixed valence state of the oxidase binuclear centre: how Thermus thermophilus cytochrome ba3 differs from classical aa3 in the aerobic steady state and when inhibited by cyanide

In the aerobic steady state of the classical eukaryotic cytochrome c oxidase, three aa(3) redox metal centres (cytochrome a, CuA and CuB) are partially reduced while the fourth, cytochrome a(3), remains almost fully oxidized. Turnover depends primarily upon the rate of cytochrome a(3) reduction. When prokaryotic cytochrome c-552 oxidase (ba(3)) of Thermus thermophilus turns over, three different metal centres (cytochromes b, a(3) and CuA) share the steady state electrons; it is the fourth, CuB, that apparently remains almost fully oxidized until anaerobiosis. Cytochrome a(3) stays partially reduced during turnover and a possible P/F state may also be populated. Cyanide traps the aerobic ba(3) CuB centre in the a(3)(2+)CNCuB(2+) state; the corresponding eukaryotic cyanide trapped state is a(3)(3+)CNCuB(+). Both states become the fully reduced a(3)(2+)CNCuB(+) upon anaerobiosis. The different reactivities of the aa(3) and ba(3) binuclear centres may be correlated with the very different proximal histidine N-Fe distances in the two enzymes (3.3 A for ba(3) compared to 1.9 A for aa(3)) which may in turn relate to the functioning of thermophilic Thermus cytochrome ba(3) in vivo at a very elevated temperature. But the differences may also just exemplify how evolution can find surprisingly different solutions to the common problem of electron transfer to oxygen. Some of these alternatives were potentially enshrined in a model of the oxidase reaction already adopted by Gerry Babcock in the early 1990s.     

Single electron reduction of 'slow' and 'fast' cytochrome-c oxidase.

Evidence is presented that single electron reduction is sufficient for rapid electron transfer (k greater than 20 s-1 at pH 8.0 in 0.43 M potassium EDTA) between haem a/CuA and the binuclear centre in 'fast' oxidase, whereas in 'slow' oxidase intramolecular electron transfer is slow even when both CuA and haem a are reduced (k congruent to 0.01 s-1). However, while a single electron can equilibrate rapidly between CuA, haem a and CuB in 'fast' oxidase, it seems that equilibration with haem a3 is relatively slow (k congruent to 2 s-1). Electron transfer between cytochrome c and CuA/haem a is similar for both types of enzyme (k congruent to 2.4 x 10(5) M-1.s-1).     

Some properties of Nitrosomonas europaea cytochrome c oxidase (aa3-type) which lacks CuA

From Nitrosomonas europaea which had been cultivated in a medium deficient in copper, cytochrome c oxidase (aa3-type) which did not have CuA was purified. The oxidase did not show the 830-nm peak and its ESR spectrum differed greatly from that of the normal enzyme, which has two copper atoms, CuA and CuB, per molecule. However, the oxidase which did not have CuA showed almost the same cytochrome c oxidizing activity as the normal oxidase.     

Routes of electron transfer in beef heart cytochrome c oxidase: is there a unique pathway used by all reductants?

Cytochrome c oxidase oxidizes several hydrogen donors, including TMPD (N,N,N',N'-tetramethyl-p-phenyl-enediamine) and DMPT (2-amino-6,7-dimethyl-5,6,7,8-tetrahydropterine), in the absence of the physiological substrate cytochrome c. Maximal enzyme turnovers with TMPD and DMPT alone are rather less than with cytochrome c, but much greater than previously reported if extrapolated to high reductant levels and (or) to 100% reduction of cytochrome a in the steady state. The presence of cytochrome c is, therefore, not necessary for substantial intramolecular electron transfer to occur in the oxidase. A direct bimolecular reduction of cytochrome a by TMPD is sufficient to account for the turnover of the enzyme. CuA may not be an essential component of the TMPD oxidase pathway. DMPT oxidation seems to occur more rapidly than the DMPT--cytochrome a reduction rate and may therefore imply mediation of CuA. Both "resting" and "pulsed" oxidases contain rapid-turnover and slow-turnover species, as determined by aerobic steady-state reduction of cytochrome a by TMPD. Only the "rapid" fraction (approximately 70% of the total with resting and approximately 85% of the total with pulsed) is involved in turnover. We conclude that electron transfer to the a3CuB binuclear centre can occur either from cytochrome a or CuA, depending upon the redox state of the binuclear centre. Under steady-state conditions, cytochrome a and CuA may not always be in rapid equilibrium. Rapid enzyme turnover by either natural or artificial substrates may require reduction of both and two pathways of electron transfer to the a3CuB centre.     

Determination of the optical properties of CuA(II) in bovine cytochrome c oxidase using magnetic circular dichroism as an optical detector of paramagnetic resonance

This paper describes the use of a novel magnetic circular dichroism-microwave double resonance (MCD-ODMR) experiment to study the optical properties of the EPR detectable copper center, CuA2+, in bovine cytochrome c oxidase. By irradiating the sample with a monochromatic microwave beam of appropriate frequency it is possible to quench partially the MCD intensity of the features due to CuA2+. In this way the MCD bands from this center have been identified even in the presence of overlapping optical transitions from the haem centers of this enzyme. The resulting spectrum compares well with that reported some years ago from this laboratory and obtained by measuring MCD magnetisation characteristics. In addition the shapes of the MCD-ODMR lines obtained in a plot of MCD intensity against magnetic field strength have been analyzed to yield the relative polarizations of the optical transitions of CuA2+ which contribute to the MCD spectrum. All of the bands observed between 450 and 850 nm are predominantly polarized in the xy plane perpendicular to the direction of the g-tensor component of CuA2+ at g parallel = 2.18. This suggests that all of the CuA2+ ligands that contribute to the optical charge-transfer transitions in the visible region lie approximately in the basal plane. Possible structures for CuA2+ can now be suggested.     

Infrared evidence of azide binding to iron, copper, and non-metal sites in heart cytochrome c oxidase.

Interactions of azide ion with bovine heart cytochrome c oxidase (CcO) at five redox levels (IV) to (0), obtained by zero to four electron reduction of fully oxidized enzyme CcO(IV), were monitored by infrared and visible/Soret spectra. Partially reduced CcO gave three azide asymmetric stretch band at 2040, 2016, and 2004 cm-1 for CcO(III)N3 and two at 2040 and 2016 cm-1 for CcO(II)N3 and CcO(I)N3. Resting CcO(IV) reacts with N3- to give one band at 2041 cm-1 assigned to CuB2+N3 and another at 2051 cm-1 to N3- that is associated with protein but is not bound to a metal ion. At high azide concentrations the weak association of many azide molecules with non-metal protein sites was observed at all redox levels. These findings provide direct evidence for 1) N3- binding to CuB as well as Fea3 in partially reduced enzyme, but no binding to Fea3 in fully oxidized enzyme and no binding to either metal in fully reduced enzyme; 2) a long range effect of the oxidation state of Fea or CuA on ligand binding at heme a3, but not at CuB; and 3) an insensitivity of either Fea3 or CuB ligand site to changes in ligand or oxidation state at the other site. The observed independence of the Fea3 and CuB sites provides further support for Fea3(3)+ OOH, rather than Fea3(3)+ OOCuB2+, as an intermediate in the reduction of O2 to water by the oxidase.     

Q-band ENDOR (electron nuclear double resonance) of the high-affinity ubisemiquinone center in cytochrome bo3 from Escherichia coli

Electron nuclear double resonance (ENDOR) was performed on the protein-bound, stabilized, high-affinity ubisemiquinone radical, QH*-, of bo3 quinol oxidase to determine its electronic spin distribution and to probe its interaction with its surroundings. Until this present work, such ENDOR studies of protein-stabilized ubisemiquinone centers have only been done on photosynthetic reaction centers whose function is to reduce a ubiquinol pool. In contrast, QH*- serves to oxidize a ubiquinol pool in the course of electron transfer from the ubiquinol pool to the oxygen-consuming center of terminal bo3 oxidase. As documented by large hyperfine couplings (>10 MHz) to nonexchangeable protons on the QH*- ubisemiquinone ring, we provide evidence for an electronic distribution on QH*- that is different from that of the semiquinones of reaction centers. Since the ubisemiquinone itself is physically nearly identical in both QH*- and the bacterial photosynthetic reaction centers, this electronic difference is evidently a function of the local protein environment. Interaction of QH*- with this local protein environment was explicitly shown by exchangeable deuteron ENDOR that implied hydrogen bonding to the quinone and by weak proton hyperfine couplings to the local protein matrix.     

The photoreactivity of the copper-NO complexes in cytochrome c oxidase and in other copper-containing proteins

The complexes of NO with CuB of cytochrome c oxidase in which cytochrome a3 may or may not be ligated to cyanide or fluoride are photodissociable. NO does not appear to react with CuB in complexes of cytochrome c oxidase in which sulphide or mercaptans are ligated to the haem iron of cytochrome a3. A comparison is made between the photoreactivity of the complexes of NO with cytochrome c oxidase and those with ceruloplasmin, ascorbate oxidase, and haemocyanin. It is shown that the photoreactivity of CuB 2+.NO in cytochrome c oxidase is not unique for this enzyme, but may also be observed in the complexes of NO with type-1 copper-containing enzymes. This would suggest that the ligation of CuB in cytochrome c oxidase shows some similarity to type-1 copper in blue oxidases.     

Structural models of the redox centres in cytochrome oxidase.

Evolutionary conservation, predicted membrane topography of the subunits, and known chemical and physical properties of the catalytic metals in cytochrome oxidase provided the basis for plausible structural models of the enzyme's redox centres. Subunit II probably binds one of the copper ions (CuA) whilst subunit I is likely to bind the two haems (a and a3) and the other redox-active copper (CuB). Two cysteine and two histidine residues of subunit II are the likely ligands of CuA, forming a centre that may be structurally similar to that in azurin. The two haems may be sandwiched between two transmembranous segments of subunit I, one of which also provides a histidine ligand to CuB. A third segment may provide two more histidine ligands to the latter. The model was constructed with a 4 A Fe-Cu distance in the binuclear haem a3-CuB centre, and a 14 A distance between the haem irons. The subunit I model involves only three transmembranous helices which bind three catalytic metal groups. The fit of this model to several known physicochemical properties of the redox centres is analysed.     

Electron spin relaxation of CuA and cytochrome a in cytochrome c oxidase. Comparison to heme, copper, and sulfur radical complexes

The method of continuous saturation has been used to measure the electron spin relaxation parameter T1T2 at temperatures between 10 and 50 K for a variety of S = 1/2 species including: CuA and cytochrome a of cytochrome c oxidase, the type 1 copper in several blue copper proteins, the type 2 copper in laccase, inorganic Cu(II) complexes, sulfur radicals, and low spin heme proteins. The temperature dependence and the magnitude of T1T2 for all of the species examined are accounted for by assuming that the Van Vleck Raman process dominates the electron spin-lattice relaxation. Over the entire temperature range examined, the relaxation of the type 1 coppers in six to seven times faster than that of type 2 copper, inorganic copper, and sulfur radicals, in spite of the similar g-anisotropies of these species. This result may indicate that the coupling of the phonon bath to the spin center is more effective in type 1 coppers than in the other complexes studied. The relaxation of CuA of cytochrome oxidase exhibits an unusual temperature dependence relative to the other copper complexes studied, suggesting that the protein environment of this center is different from that of the other copper centers studied and/or that CuA is influenced by a magnetic dipolar interaction with another, faster-relaxing paramagnetic site in the enzyme. A comparison of the saturation characteristics of the CuA EPR signal in native and partially reduced CO complexes of the enzyme also suggests the existence of such an interaction. The implications of these results with respect to the disposition of the metal centers in cytochrome oxidase are discussed.