A Common Mechanism for the Interaction of Nitric Oxide with the Oxidized Binuclear Centre and Oxygen Intermediates of Cytochromec Oxidase*

The reactions of nitric oxide (NO) with fully oxidized cytochrome c oxidase (O) and the intermediates P and F have been investigated by optical spectroscopy, using both static and kinetic methods. The reaction of NO with O leads to a rapid (∼100 s−1) electron ejection from the binuclear center to cytochrome a and CuA. The reaction with the intermediates P and F leads to the depletion of these species in slower reactions, yielding the fully oxidized enzyme. The fastest optical change, however, takes place within the dead time of the stopped-flow apparatus (∼1 ms), and corresponds to the formation of the F intermediate (580 nm) upon reaction of NO with a species that we postulate is at the peroxide oxidation level. This species can be formulated as either Fe5+ = O CuB 2+or Fe4+ = O CuB 3+, and it is spectrally distinct from the P intermediate (607 nm). All of these reactions have been rationalized through a mechanism in which NO reacts with CuB 2+, generating the nitrosonium species CuB 1+ NO+, which upon hydration yields nitrous acid and CuB 1+. This is followed by redox equilibration of CuB with Fea/CuA or Fea3 (in which Fea and Fea3 are the iron centers of cytochromes a and a 3, respectively). In agreement with this hypothesis, our results indicate that nitrite is rapidly formed within the binuclear center following the addition of NO to the three species tested (O, P, and F). This work suggests that nitrosylation at CuB 2+ instead of at Fea3 2+ is a key event in the fast inhibition of cytochrome c oxidase by NO.

The reactions of nitric oxide (NO) with fully oxidized cytochrome c oxidase (O) and the intermediates P and F have been investigated by optical spectroscopy, using both static and kinetic methods. The reaction of NO with O leads to a rapid (ϳ100 s ؊1 ) electron ejection from the binuclear center to cytochrome a and Cu A . The reaction with the intermediates P and F leads to the depletion of these species in slower reactions, yielding the fully oxidized enzyme. The fastest optical change, however, takes place within the dead time of the stopped-flow apparatus (ϳ1 ms), and corresponds to the formation of the F intermediate (580 nm) upon reaction of NO with a species that we postulate is at the peroxide oxidation level. This species can be formulated as either Fe 5؉ ‫؍‬ O Cu B 2؉ or Fe 4؉ ‫؍‬ O Cu B 3؉ , and it is spectrally distinct from the P intermediate (607 nm). All of these reactions have been rationalized through a mechanism in which NO reacts with Cu B 2؉ , generating the nitrosonium species Cu B 1؉ NO ؉ , which upon hydration yields nitrous acid and Cu B 1؉ . This is followed by redox equilibration of Cu B with Fe a /Cu A or Fe a3 (in which Fe a and Fe a3 are the iron centers of cytochromes a and a 3 , respectively). In agreement with this hypothesis, our results indicate that nitrite is rapidly formed within the binuclear center following the addition of NO to the three species tested (O, P, and F). This work suggests that nitrosylation at Cu B 2؉ instead of at Fe a3 2؉ is a key event in the fast inhibition of cytochrome c oxidase by NO.
Cytochrome c oxidase (ferrocytochrome c oxidoreductase, EC 1.9.3.1), the terminal enzyme in the mitochondrial respiratory chain, catalyzes the reduction of molecular oxygen to water (1). This process is coupled to proton translocation across the inner membrane. The enzyme contains four redox-active centers. Electron entry from cytochrome c, the natural substrate, occurs via a diatomic copper center, Cu A . After rapid equilibrium with Fe a , 1 the electron is transferred to Cu B and Fe a3 , which together con-stitute the binuclear center, where oxygen is reduced.
Although oxygen binds with low affinity (10 3 M Ϫ1 ) (2) to reduced Fe a3 , rapid electron transfer from the two reduced metals comprising the binuclear center to molecular oxygen ensures that oxygen remains bound as a peroxy species. In this way, oxygen is kinetically trapped and further reduction can take place (3,4). The successive steps leading to the formation of water have been studied by different spectroscopic techniques (see Ref. 5 for review). The spectral signatures of two of these intermediates, which exhibit bands at 607 and 580 nm in the difference spectrum with respect to the oxidized enzyme, were first reported by Wikström (6), by reversing the electron transfer reaction, and further characterized by Wikström and Morgan (7). These authors assigned the spectral signatures at 607 and 580 nm to a ferric peroxy (P) and ferryl oxo (F) species, respectively. These assignments, however, have been challenged by a number of authors (8 -10). For example, Resonance Raman results obtained by Proshlyakov et al. (10) suggest that the 607 nm band originates from an oxoferryl structure. On the other hand, the same authors have been unable to identify the putative peroxy species in their system in turnover sustained by H 2 O 2 (11). However, irrespective of the assignments, there seems to exist a general agreement that compound F (580 nm) is a ferryl oxo species and that it is one electron more reduced than compound P (607 nm) (11,12).
One of the ways to solve the problem of the identity of the P intermediate could perhaps be through the use of a suitable probe. The possibility that nitric oxide (NO), a powerful reversible inhibitor of cytochrome c oxidase (13)(14)(15), can be used as a probe for the binuclear center has been suggested by recent results using enzyme in slow turnover (16 -18). In these experiments, an electron was ejected from the binuclear center, partially reducing cytochrome a. In addition, depletion of the intermediates P and F was also observed on mixing with NO (16,17). As suggested in these papers, these reactions could be mediated via reaction of NO with Cu B , as it has been shown that in addition to binding to reduced Fe a3 , NO binds to both Cu B 1ϩ and Cu B 2ϩ , albeit with different affinities (19 -21). On the other hand, we have suggested previously (15) that binding of NO to Cu B 1ϩ could be a key to understanding the mechanism of inhibition of CcO by NO, although this hypothesis was based solely on steady-state kinetic considerations.
A mechanism describing the interaction of NO with Cu B , leading to the partial reduction of cytochrome a, could be the reverse of that postulated for the reduction of nitrite to NO by non-heme nitrite reductases, which contain only Cu as a metal (see Ref. 22 for review). In this mechanism, depicted in Scheme 1 (solid arrows), Cu 1ϩ of nitrite reductase (A) binds nitrite, * This work was supported by the Wellcome Trust, the Royal Society, the Spanish government, and the Biotechnology and Biological Science Research Council, UK. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
‡ To whom correspondence should be addressed. Tel.: 44-1206-872538; Fax: 44-1206-872592; E-mail: wilsmt@essex.ac.uk. 1 The abbreviations used are: Fe a and Fe a3 , iron centers of cytochromes a and a 3 , respectively; CcO, cytochrome c oxidase; NO, nitric oxide; O, fully oxidized cytochrome c oxidase formed after reduction and reoxidation of the enzyme; O b , oxidized binuclear center of CcO; P and F, oxygen intermediates of CcO at an oxidation state two and three electrons more reduced, respectively, than the oxidized binuclear center; P M , species formed immediately after CO-photodissociation from mixed valence-CO in the presence of oxygen; P R , species formed immediately after CO-photodissociation from the fully reduced CO complex in the presence of oxygen.
forming a complex (B), which after abstraction of one oxygen atom gives an electrophilic nitrosonium (C). This species is analogous to the nitrosonium (Fe 2ϩ NO ϩ ), detected by Fourier transform infrared spectroscopy in the heme-containing cytochrome cd 1 nitrite reductase of Pseudomonas stutzeri, obtained by incubating the oxidized enzyme with NO (23). The donation of one electron from the metal gives Cu 2ϩ NO (D), which then dissociates as NO and oxidized Cu (E).
We suggest that in CcO, NO interacts with the oxidized enzyme at the copper present in the binuclear center, Cu B 2ϩ , forming a complex analogous to that shown at point D in Scheme 1. This eventually yields nitrite and Cu B 1ϩ (Scheme 1, dotted arrows, E through A). As depicted in Scheme 2, the reduced Cu B could be reoxidized by equilibration with the other redox centers, i.e. Fe a , Cu A , and Fe a3 . In those cases in which oxygen intermediates are present in the binuclear center, an obvious possible outcome of the internal electron transfer from Cu B to Fe a3 would be the transition of the oxygen intermediate to the next intermediate in the catalytic cycle, i.e. P 3 F or F 3 O b (O b represents the oxidized binuclear center of CcO, Fe a3 3ϩ Cu B 2ϩ ). Alternatively, Cu B 1ϩ could bind NO, forming a relatively stable complex, Cu 1ϩ NO. The formation of this complex could ultimately be responsible for the inhibition of the enzyme.
In this study, we have investigated in detail these reactions with a view to test the hypotheses formulated in Schemes 1 and 2. We provide evidence for the mechanisms depicted in these schemes, which are able to explain many of the features encountered in the reaction of NO with CcO.

MATERIALS AND METHODS
Cytochrome c oxidase was prepared by the method of Soulimane and Buse (24), which yields highly active enzyme (maximal turnover number, 600 s Ϫ1 ). The buffer used throughout was 0.1 M HEPES, 0.5% Tween 80, pH 7.4. Because of the sluggish reactivity of the binuclear center of this enzyme as preparation with CO, H 2 O 2 (used to form P and F, respectively), and NO, the enzyme was "pulsed" by reduction and reoxidation. This was necessary both to prepare the intermediates P and F and to investigate the reactivity of the oxidized enzyme with NO.
Pulsed Oxidized Enzyme O-CcO (ϳ45 M) was fully reduced by incubation with 1 mM sodium dithionite for 2 h at 4°C. Then, the enzyme was reoxidized by passage through a Sephadex G-25 column equilibrated with the same buffer but containing no dithionite. To ensure that the enzyme was fully reoxidized, the column was loaded, immediately before the addition of the enzyme, with a band of 20 mM potassium ferricyanide. Formation of peroxide upon reoxidation was avoided by addition of 50 l of 40 mM catalase to the reduced enzyme prior loading onto the column. Following this procedure, the maximum of the Soret band corresponding to the pulsed enzyme was at 423 nm, identical to the maximum observed in the oxidized enzyme as prepared. Formation of oxygen intermediates of the enzyme (e.g. peroxy), due to either incomplete reoxidation or the formation of peroxide, would induce a red shift of the Soret band. The absence of a significant shift in this band after pulsing is indicative of the absence of such oxygen intermediates. The pulsed oxidized enzyme (O) thus obtained was used to prepare the intermediates P and F.
Compounds P and F-Compound F was formed by incubating O (see above) in 1 mM H 2 O 2 , and the concentration of F was determined spectroscopically using ⌬⑀ 580 -630 ϭ 5,500 M Ϫ1 cm Ϫ1 (7) relative to O. Compound P was obtained by two different methods. In the first method, P was formed when CO gas was bubbled for a short time (ϳ10 s) through a solution containing typically ϳ4 M O (see above) as described previously (25). After about 10 min, formation of P reached a steady state, with conversion of typically ϳ50% of the enzyme to this form. In the second method, the oxidized enzyme (pulsing was not necessary) was incubated with CO overnight at 4°C to obtain the mixed valence-CO complex. After briefly degassing the solution, the sample was exposed to an intense flash of white light in the presence of oxygen. The amount of compound P formed using this method was typically ϳ80% of the total enzyme, as determined using the extinction coefficient ⌬⑀ 607-630 ϭ 11,000 M Ϫ1 cm Ϫ1 (7) relative to O.
Static Experiments-In the static experiments, an aliquot of a solution of NO (50 l of 2 mM NO) was added to 1.5 ml of a solution containing compound P (607 nm), F (580 nm), or O (fully oxidized enzyme). A difference spectrum relative to O was recorded immediately in a Cary 5E UV-Vis-NIR spectrophotometer. The amount of cytochrome a reduced was calculated as follows: ⌬⑀ 605 (reduced minus oxidized) ϭ 17,500 M Ϫ1 cm Ϫ1 .
Stopped Flow-Compounds P, F, and O were rapidly mixed with saturated (ϳ2 mM) or diluted solutions of NO, prepared by diluting the stock saturated solution with different volumes of anaerobic buffer, in a stopped flow apparatus (model SX-18MV Applied Photophysics, Leatherhead, UK). The spectra were collected using a photodiode array detector with a time resolution of 3.3 ms. A global fitting analysis program (Global Analysis, Applied Photophysics) using singular value decomposition of the data was used to model the reactions, the spectra of the kinetic components, and the corresponding rate constants. The spectra of the components constructed in this way were consistent with difference spectra obtained by subtracting spectra taken at different time points during the course of the reaction. The amount of Cu A reduced after mixing O with NO was obtained by comparing the amplitude of the change at 830 nm with that obtained after the addition of dithionite (18).
Nitric Oxide-Nitric oxide (NO) was obtained with a Kipps apparatus by mixing 1 M sulfuric acid with sodium nitrite, essentially as described by Torres and Wilson (26). The NO in the stock solution (usually ϳ2 mM) and in the dilutions was measured with an NO electrode (Iso-NO Mark II, World Precision Instruments). The level of nitrite in the NO stock solution was also monitored (usually less than 0.1 mM) in the same electrode, by measuring the NO formed after acidification with H 2 SO 4 in the presence of KI.

Static Optical Spectroscopy
Reaction of NO with Species O-Addition of NO to O generated a difference spectrum (Fig. 1A) that, in the Soret region, exhibited a peak at 445 nm and a trough at 430 nm. At longer wavelengths, a positive band at 605 nm was also observed. Both the positions of the bands and the relative intensities are indicative of reduction of cytochrome a (ϳ40% of the total). These changes were accompanied by a bleached region from 620 to 660 nm.
Reactions of NO with Species F and P-Addition of NO to P or F resulted in the rapid depletion of these intermediates. The spectrum resulting from the addition of NO to F (Fig. 1B) has a peak at 415 nm and a trough at 430 nm in the Soret region, and the region from ϳ590 to 660 is bleached. Because the scheme presented above (see Scheme 1, dotted arrows) predicts the formation of nitrite in the binuclear center of CcO, we tested this prediction by adding nitrite to O and comparing the final spectrum to that obtained after addition of NO to F. Fig.  1C shows that exposure of O to a high concentration of nitrite (10 mM) generated the same spectral changes as obtained when NO (20 M) was added to F. The affinity of the enzyme for nitrite is low and required a relatively high nitrite concentration to induce spectral changes of measurable amplitude. The fact that the same effect was obtained with a much lower NO concentration suggests that when NO is added to F, nitrite is formed within the binuclear center and interacts with the oxidized binuclear center (O b ) formed as a result of the When NO was added to P (Fig. 1D), the depletion of this intermediate was accompanied, as for O, by the appearance of bands at 605 and 442 nm, indicative of cytochrome a reduction. However, the extent of the reduction was not, as with O, 40% of the total enzyme, but rather 40% of the P present before the addition of NO (e.g. for a sample containing 50% P, the extent of cytochrome a reduction would be ϳ20%).
The features observed when adding NO to F (Fig. 1C), assigned to the interaction of nitrite with the oxidized binuclear center, O b , were also present after addition of NO to the samples O and P (see Fig. 1E). Indeed, the spectrum resulting after addition of NO to O or P can be shown to be a superposition of two components: (a) interaction of nitrite with O b , and (b) the reduction of cytochrome a (i.e. transition Fe a 3ϩ 3 Fe a 2ϩ ). Accordingly, the subtraction of the spectral contribution of nitrite gives a residual identical to the difference spectrum corresponding to reduced minus oxidized cytochrome a (27) (data not shown). We postulate that NO converts P and F to O b , and NO reacts with O b , reducing cytochrome a (Fig. 1A). However, it is clear that we did not observe cytochrome a reduction after mixing F with NO (Fig. 1B). This was due to the presence of H 2 O 2 in the sample, because when we repeated the experiment adding catalase prior to mixing F with NO, we observed reduction of cytochrome a simultaneous with the decay of F (not shown). This indicates that H 2 O 2 either impedes the interaction of NO with the binuclear center or reoxidizes cytochrome a as soon as it becomes reduced. It is also worth mentioning that even though CO and H 2 O 2 are still present in solution, neither P nor F were reformed after the reaction of NO with these intermediates. This is consistent with the observed sluggish reactivity of the binuclear center of CcO with H 2 O 2 and with CO (which in certain conditions lead to formation of F and P, respectively; see "Materials and Methods") after NO has been added to O (not shown).
Thus, these experiments confirm three predictions derived from Schemes 1 and 2. First, the product of the reaction of NO with O, P, or F displays a spectrum consistent with the presence of nitrite interacting with the oxidized binuclear center

Reaction of Compound F with NO
The reaction of a sample containing compound F (ϳ85%) with NO led, as expected (see Fig. 1B), to the rapid decay of this compound (Fig. 2, curves a and b). The time course was fitted to a biexponential decay, and the rates of the two phases were essentially proportional to the NO concentration (see Fig. 2 legend). Difference spectra relative to O are shown in Fig. 2, inset. The final spectrum was found to be identical to that obtained in the static spectral experiments (Fig. 1B) and thus identical to that observed after the addition of nitrite to the oxidized enzyme (Fig. 1C). Although the time course could be fitted to a biexponential decay, the spectra corresponding to the two kinetic components were indistinguishable (not shown). This indicates that the components corresponding to the decay of compound F and the appearance of the features that we identify with the interaction of the binuclear center with nitrite cannot be separated in this reaction.

Reactions of Compound P with NO
Reactions within the Dead Time (1.4 ms)-When a sample containing a high proportion of compound P (ϳ80%) was mixed with NO in the stopped flow spectrophotometer, some spectral changes took place in the dead time of the apparatus (ϳ1.4 ms), whereas the band at 607 nm remained unchanged in this time frame (Fig. 3, arrows). The final spectrum after ϳ1 s is consistent with that shown in Fig. 1D (static experiments), showing reduction of cytochrome a and depletion of P. The difference spectrum corresponding to the changes which occurred during the dead time is given in Fig. 4A, solid line (and also in Fig. 3, inset, solid line). This spectrum (Fig. 4A, solid line) shows two peaks in the Soret region at about 425 and 445 nm and a trough around 410 nm. At longer wavelengths, a positive band at ϳ580 nm, a shoulder at 600 nm, and a negative region centered at ϳ615 nm can also be observed. These features are reminiscent of the formation of compound F, possibly accompanied by some reduction of cytochrome a. However, both the changes in the Soret and the ␣ regions are slightly different from those expected for simultaneous formation of F and reduction of cytochrome a. For example, although a band at 445 nm is clearly present in the Soret region, there is no concomitant increase in the ␣ band at ϳ605 nm. Only a shoulder at 600 nm is observed. We can discount spectral contributions from a transition of the type P 3 F, because the increase at 580 nm is not paralleled by a decrease at 607 nm of an amplitude approximately twice as large (see the extinction coefficients given under "Materials and Methods") (Fig. 4A, solid line). In addition, even in case that reduction of cytochrome a (increase at 605 nm) compensates for the decrease at 607 nm, the change at 445 nm in the Soret region (which is 3 times larger in amplitude than at 605 nm; see Fig. 1A) would have to be about twice as large than observed in Fig. 4A.
Thus, a transition of the type O 3 F seems more plausible, except that the peak in the Soret region, which should appear at ϳ435 nm (9), appears to be shifted to 425 nm. These differences indicate that other changes, possibly attributable to the effect of nitrite (see Fig. 1C), distort the composite spectrum arising from the formation of compound F and the reduction of cytochrome a. In fact, when the contribution of the interaction of nitrite with O b (Fig. 4A, dotted line) was subtracted, the spectral features conformed more closely to those expected from the transitions O 3 F and Fe a 3ϩ 3 Fe a 2ϩ (Fig. 4B). Further, when the contribution of the reduction of cytochrome a was also subtracted (see Fig. 4 legend for details), the resulting spectrum was very similar to that of F relative to O (see Fig.  1B, solid line). This analysis reveals that when NO is mixed with samples containing P, spectral changes occur within the dead time of the stopped flow spectrometer consistent with the following events: (a) formation of species F from a species spectrally similar to O, not from P, (b) formation of nitrite within the binuclear center, and (c) reduction of some cytochrome a.
When P was generated from mixed valence-CO enzyme (ϳ80% P), the percentage of compound F formed in the dead time was only ϳ20% of the total enzyme. However, when P was generated by bubbling CO in the presence of oxygen (ϳ40% P), the percentage of F generated was larger (40 -50%) (28). This indicates that the precursor of F is not P itself. This finds further support from experiments in which compound P, generated from the mixed valence-CO preparation, was allowed to decay at room temperature (not shown) prior to mixing with NO. Addition of NO after a long period (ϳ1 h) produced a larger fraction of F formed (30 -40%) in the dead time. Only after about 2 h was a fast partial reduction of cytochrome a (k ϳ 100 s Ϫ1 ), which corresponded to ϳ20% of the enzyme (not shown), also observed, consistent with a proportion of the molecules having returned to O. This shows that the species absorbing at 607 nm does not decay directly to O but rather to another species that is O-like in spectral properties but that generates F rapidly on mixing with NO.
Reactions Following the Dead Time-Following the dead time, species P (607 nm) and the newly formed species absorbing at 580 nm decayed simultaneously (Fig. 5), with a rate that was found to depend on the NO concentration (Fig. 5, inset). These changes were complete in ϳ800 ms at 1 mM NO, and together with those changes observed in the Soret region, they are compatible with the decay of compounds P and F. The amplitude of the decrease at 580 nm from 1.4 to 800 ms was found to be identical, within our experimental error, to the amplitude of the increase observed within the dead time (Fig. 3,  inset). This indicates that the compound F that decays from 1.4 to 800 ms corresponds to that formed within the dead time. The amplitude of the decrease at 607 nm, however, was smaller than the corresponding amplitude of the P present before mixing (Fig. 3). In fact, one of the troughs in Fig. 5 is centered at 612 nm, not at 607 nm. This apparent shift can be explained by a simultaneous decrease at 607 nm and an increase at 605 nm, which, together with the presence of a positive band at 442 nm in the difference spectrum, indicates that some reduction of cytochrome a takes place simultaneously with the decay of the P and F intermediates. 2 Accordingly, the reduction of cytochrome a and the decay of P and F resulted in a single kinetic transition in the global analysis (see "Materials and Methods"). In addition, internal differences between spectra collected at different time points show that the component corresponding to cytochrome a reduction is present at all time intervals (not shown). This indicates that both processes are interconnected, such that the rate of reduction of cytochrome a is limited by the rate of decay of the intermediates P or F. As in the static experiments (Fig. 1D), the final amount of cytochrome a reduced was about 40% of the P initially present.

Reactions of O with NO
As for compound P, when NO was mixed with O some changes occurred within the instrumental dead time (Ͻ1.4 ms) (Fig. 6A). The difference spectrum recorded (t ϭ 1.4 ms minus t ϭ 0 ms) is compatible with the appearance of approximately equal percentage (ϳ15%) of reduced cytochrome a and compound F. Although the reduction of cytochrome a observed may be expected (in ϳ1.4 ms, a process with k ϭ 100 s Ϫ1 completes ϳ15% of its total amplitude), the formation of F was not anticipated. We attribute the formation of F to heterogeneity in the sample, suggesting that preparations of O (after reduction and reoxidation; see "Materials and Methods") contain a small proportion of the O-like species 3 (described above). After the dead time, about 40% of cytochrome a became reduced (k ϳ 100 s Ϫ1 ), as indicated by bands at 442 and 605 nm (intensity ratio, 3:1) (Fig. 6B), in agreement with the results obtained in the static spectral experiments (Fig. 1A). Reduction of cytochrome a was accompanied by reduction of 15-20% Cu A (not shown) (18).
In a slower phase (Fig. 6C), a decay at 580 nm and a simultaneous increase at 605 nm was also observed (k ϳ 8 s Ϫ1 ). Both these changes and those detected in the Soret region are indicative of simultaneous decay of F (formed in the dead time) and reduction of cytochrome a. This slow process is similar to that seen when mixing P and NO, when reduction of cytochrome a 2 Because the spectra of P and reduced cytochrome a are similar in the ␣ band (P has a band at 607 nm and cytochrome a at 605 nm) the simultaneous decay of P and a smaller increase of reduced cytochrome a led only to a small perturbation in this region (see extinction coefficients under "Materials and Methods"). 3 Other authors have previously suggested the presence of heterogeneities even in "fast" oxidase preparations, based on the presence of two phases in the kinetics of ligand binding (32), especially formate.
FIG. 5. Spectral changes obtained after mixing CcO samples containing P with NO. CcO (2.35 M after mixing) containing ϳ80% P was mixed with NO (1 mM after mixing). Difference spectra relative to the first spectrum collected after mixing (1.4 ms) are shown at a number of times during the time course of the reaction. Time points: first spectrum, 4ms; spectra 2-10, ⌬t ϭ 10 ms; spectra 10 -18, ⌬t ϭ 100 ms. The arrows indicate the direction of the changes; the region 500 -800 has been expanded (ϫ5), and characteristic wavelengths are indicated. Inset, time courses of the absorbance changes monitored as ⌬A(430 -500 nm) after mixing P (as above) with 1 mM (a) or 20 M (b) NO. The time courses (--) were fitted to a monoexponential decay (E), and the rate constants for the decay are indicated.
accompanied the decay of the bands corresponding to P (607 nm) (Fig. 5). This can be clearly seen in Fig. 7, where the difference spectrum produced ϳ50 ms after mixing O with NO (i.e. at the end of the process with k ϳ 100 s Ϫ1 ) (see Fig. 6B), is compared with the difference spectrum of the final product (after ϳ800 ms) obtained when mixing P with NO (i.e. following the process with k ϳ 8 s Ϫ1 ) (from Fig. 3). It is clear from this figure that the same optical change is obtained in both cases, regardless of the initial species (P or O), although with different kinetics. Together with the results in Fig. 6C, these results indicate that the reduction of cytochrome a is rate limited by the decay of P or F when these oxygen intermediates are present.

DISCUSSION
The results presented above may be rationalized by reference to a simplified reaction mechanism presented in Scheme 3. The basic feature of this scheme is that the different reactions, observed upon mixing NO with derivatives of CcO, can all be explained by a single electron donation from NO to the binuclear center. Each step in Scheme 3 comprises a sequence of events analogous to those indicated by dotted arrows in Scheme 1, i.e. NO binds to Cu B 2ϩ and results in the formation of HNO 2 and reduction of Cu B . The latter would permit electron transfer from Cu B either to cytochrome a/Cu A or to oxygen intermediates bound to Fe a3 . A detailed reaction mechanism is given in Scheme 4, in which the two parts, A and B, differ in the chemical assignment of P. We have assumed in part A that the species P (607 nm) and F (580 nm) are the ferric peroxy and oxoferryl derivatives of Fe a3 , respectively, as suggested previously by Wikström and Morgan (7).
Because we propose that NO provides the binuclear center with one reducing equivalent, it follows that the precursor of the F formed during the dead time (spectrally O-like and labeled X in Scheme 4) is at the formal oxidation level of peroxide. Therefore, if P (P M ) is the ferric peroxy species, then X could contain either Cu B 3ϩ , with the iron as a ferryl oxo species (Fe 4ϩ ϭ O), or alternatively as Cu B 2ϩ , with Fe 5ϩ ϭ O (see Scheme 4). The presence of either of these species, suggested previously by other authors (9 -12, 29, 30), would explain the fast formation of compound F solely by reduction of Cu B 3ϩ to Cu B 2ϩ or Fe 5ϩ to Fe 4ϩ . Our data suggest that P M coexists with species X. possibly in an equilibrium, because they are at the same oxidation level.
An analogous and equally fast reaction may be expected to occur between NO and P M , generating a peroxy species with Cu B reduced. Such a species, postulated previously, has been termed P R in the literature (31). 4 The conversion from P M to P R would not result in observable optical changes, because these species have similar spectra (31). This is in agreement with our results, because the band at 607 nm corresponding to P remains unchanged on addition of NO during the dead time (Fig.  3). Thus, after mixing a sample containing P (P M and X) with NO, we obtained two different products: one was P R , from the transition P M 3 P R , and the other was F, from the transition X 3 F. The mechanism in Scheme 4 accounts, therefore, for the spectral changes that occur within the dead time after mixing P with NO, namely, the apparent nonreactivity of P M (Fig. 5) and the formation of F from a species spectrally similar to O.
Furthermore, our results show that the species F (absorbing at 580 nm) and P R (assumed to absorb at 607 nm) (31) decay simultaneously after mixing with NO (Fig. 5). This suggests that they are in rapid equilibrium, one of them reacting with NO and recruiting the other into the reactive form. From our experiments (Fig. 1B), it is clear that it is compound F that reacts with NO yielding O, and we thus incorporated this into Scheme 4. Thus, P R would react not with NO per se, but through an equilibrium established between P R and F. For the reaction between F and NO, a reaction analogous to that described for P M and species X was suggested, after which, an internal electron transfer from Cu B could facilitate the transi-4 P R is the form obtained after photolysis, in the presence of oxygen, of the Fe a3 OCO bond in CO-bound fully reduced oxidase. In this case, the second reducing equivalent for oxygen is provided directly from cytochrome a, and not by Cu B , which remains reduced. In contrast, the species obtained from CO-bound mixed valence enzyme, in which the second reducing equivalent for oxygen is provided through oxidation of Cu B , is called P M (31).

FIG. 7.
Comparison between the difference spectra obtained on mixing NO with P or O. Difference spectra with respect to the oxidized enzyme O obtained on mixing with 1 mM NO either O (-----) or a sample containing ϳ80% P (--). The spectra were collected 55 ms and 1 s after mixing NO with O and P, respectively. CcO, 4 M after mixing. SCHEME 3 tion F 3 O b . Support for the role of Cu B in this reaction comes from the fact that although F is known to react with CO (33) and other ferryl groups (e.g. in myoglobin) react with NO (34), these processes are much slower than those presented here. After the formation of species O, a further electron donation from NO would then lead to the observed reduction of cytochrome a. In agreement with this model, our results show that this process is fast when mixing O with NO (k ϭ 100 s Ϫ1 ) but is rate-limited by the decay of P and F (ϳ8 s Ϫ1 ) when NO reacts with these intermediates. This is expected if the substrate for this reduction (O b ) is the product of the reaction of P and F with NO, as Scheme 4 shows.
The Identity of Compound X-It seems generally agreed that F (580 nm) has an oxo ferryl structure, containing Cu B 2ϩ , but the assignment of P (607 nm) is more problematic. Basically, the discussion in the literature revolves around whether the oxygen OOO bond can be broken following addition of only two electrons to the enzyme. This could be achieved by means of an additional electron, recruited from elsewhere in the protein, to produce an oxo ferryl derivative. Four main sources for this electron have been suggested, i.e. the porphyrin ring, an amino acid, the copper atom Cu B , or the iron atom itself. The donation of an electron from the porphyrin ring or an amino acid, leading to the formation of a radical, as observed in other enzymes (i.e. peroxidases and catalases), seems to be unlikely (9,29,31), although a radical has been observed to occur in the enzyme in turnover (35). The assignment of the 607 nm form to an oxoferryl Fe 4ϩ ϭ O structure with Cu B 3ϩ has been postulated by other authors (5,9,30), and this unusually high oxidation state for Cu B has been recently detected in an inorganic model (36). Proshlyakov et al. (10) also associate the 607 nm absorbing species to an iron oxo group on the basis of Resonance Raman 16 O-18 O mixed-isotope experiments with 607 nm excitation. These authors, however, attribute the band that appears at 804 cm Ϫ1 to Fe 5ϩ ϭ O (and Cu B 2ϩ ). We propose that X could be represented by either of these two species. In our model, species X has a spectrum more similar to O than to P. This is in contrast with the conclusions of Morgan et al. (31), who argued against the formation of Cu B 3ϩ at the peroxide level. However, their conclusions were based on the assumption that the spectrum of the compound Fe 4ϩ ϭ O Cu B 3ϩ would be identical to that of P (607 nm). In addition, the parallel increase of the Raman band at 804 cm Ϫ1 and the optical band at 607 nm (11), could correspond to two different species, the parallel increase of these two signals being due to an equilibrium between X (optically O-like) and P M (607 nm) (see Scheme 4, top equilibrium).
The postulated interconversion of the species P R and F may also be considered in the context of the Raman data, e.g. the assignment of the band at 356 cm Ϫ1 . Proshlyakov et al. (37) assigned this band to a bending mode of the unusual Fe 5ϩ ϭ O species (referred to above) on the basis that the oxo oxygen could be exchanged with the oxygen in water through a reversible formation of ferric hydroxy (Fe 3ϩ OH Ϫ ). An alternative view is that the species absorbing at 356 cm Ϫ1 is analogous to either P R (38,39) or a ferrous peroxy/cupric species (Fe 2ϩ O 2 2Ϫ Cu B 2ϩ ) (5), as depicted in Scheme 4 (part A). These species are in equilibrium with F, explaining the oxygen exchange data. Interestingly, a fast equilibrium between species P R and F, as depicted in Scheme 4, has recently been suggested to exist during the transition, initiated by photolysis at room temperature, from fully reduced CO-bound CcO to the oxidized enzyme (40). Obviously, this equilibrium implies that the splitting of the OOO bond is reversible in these conditions, and this poses a problem from the thermodynamic point of view.
This problem can be avoided as shown in Scheme 4 (part B), where P M and P R are assigned the structures Fe 5ϩ ϭ O, with Cu B 2ϩ and Cu B 1ϩ , respectively, instead of ferric peroxide species. In this case, P R and F are also in the same oxidation state, but there is no need for reversible OOO cleavage. Clearly, this model lacks the ferric peroxy form, in contradiction of previous assignments and observations made using synthetic model compounds in which a peroxy species was detected during the the reduction of oxygen to water (41).
The results presented indicate that a species (X) spectrally similar to O is in equilibrium with the 607 nm species in the sample containing compound P. In our view, this could clarify some of the apparent inconsistencies concerning the structural assignments of the intermediates P and F. For example, the cytochrome iron structure in compound P has been described either as a ferric peroxy or an iron oxo (Fe 4ϩ ϭ O or Fe 5ϩ ϭ O) by different authors (see references under "Introduction"). This discrepancy could be explained if a ferric peroxy and an iron oxo (either Fe 4ϩ ϭ O or Fe 5ϩ ϭ O, as shown in Scheme 4, part SCHEME 4 A), or two iron oxo species (e.g. Fe 4ϩ ϭ O and Fe 5ϩ ϭ O, as in Scheme 4, part B) coexisted at the peroxide level. Thus, in Scheme 4, part A, the ferric peroxy (P M ) would be the species detected optically (at 607 nm), whereas either of the postulated as species X (Fe 4ϩ ϭ O with Cu B 3ϩ or Fe 5ϩ ϭ O with Cu B 2ϩ ) could be detected at 804 cm Ϫ1 by Raman. Alternatively, in Scheme 4, part B, either species X (Fe 4ϩ ϭ O with Cu B 3ϩ ) or P M (Fe 5ϩ ϭ O with Cu B 2ϩ ) could be detected by the iron oxo stretching Fe ϭ O vibration at 804 cm Ϫ1 with Raman spectroscopy, but it is possible that Fe 5ϩ ϭ O may also be detected optically by difference spectroscopy (at 607 nm). In addition, it is worth noting that the assignment of species X as Fe 4ϩ ϭ O with Cu B 3ϩ is compatible with both parts A and B in Scheme 4, and as such, it may appear to be the more likely candidate. On the other hand, as discussed above, the experimental results suggesting that P R and F are in rapid equilibrium are better explained if P R and P M are both iron oxo (Fe 5ϩ ϭ O), supporting the assignments for X and P M made in part B of Scheme 4.
A further question concerns the fate of the electrons ejected from the binuclear center. In our experiments with the oxidized enzyme, O, approximately 50% of cytochrome a and ϳ15-20% of Cu A became reduced (18). Furthermore, approximately 20% of F was formed within the dead time. This suggests that around 90% of the molecules became reduced with one electron, and this is in agreement with recent experiments that show that only three electrons could be donated from cytochrome c to CcO prior to its inhibition by NO (42).
Physiological Implications of the Reactions Observed between NO and CcO-Our finding that nitrite is rapidly formed in the binuclear center of CcO suggests that Scheme 1 is likely to be one of the pathways through which NO is metabolized and detoxified. In addition, this mechanism is consistent with the fact that mitochondrial NO metabolism by CcO also occurs anaerobically (43), because the mechanism in Scheme 1 does not require oxygen. These results also have implications regarding the mechanism of inhibition of CcO by NO. It has been suggested that this process can be explained by reversible binding of NO to ferrous cytochrome a 3 (42), and this is consistent with the competitive nature of the inhibition with respect to oxygen. However, consideration of the "on" and "off" rates of NO with ferrous a 3 (44,42) is not in agreement with the observed low K i for the inhibition of cellular respiration by NO (43). This apparent conflict can be explained if, as we have shown, an additional interaction of NO with the binuclear center (i.e. with Cu B 2ϩ ) takes place, thus decreasing the K i of NO for CcO. Thus, although formation of ferrous a 3 -nitrosyl complex has been detected in the inhibitory process (15) and has been suggested as the primary event in the inhibition (42), this work suggests that formation of ferrocytochrome a 3 may be a consequence rather than a cause of the inhibition. A possible mechanism by which inhibition could take place is suggested by the fast reduction of Cu B 2ϩ to Cu B 1ϩ by NO. The competition of NO and oxygen would not be at the ferrocytochrome a 3 but instead at Cu B 1ϩ . Furthermore, binding of NO to Cu B 1ϩ could block the redox state of this metal, which would then constitute the real inhibitory site. Alternatively, NO may transfer rapidly from Cu B 1ϩ to cytochrome a 3 on addition of an electron to this site in turnover.