Nitric oxide reacts with the single-electron reduced active site of cytochrome c oxidase.

The reduction kinetics of the mutants K354M and D124N of the Paracoccus denitrificans cytochrome oxidase (heme aa(3)) by ruthenium hexamine was investigated by stopped-flow spectrophotometry in the absence/presence of NO. Quick heme a reduction precedes the biphasic heme a(3) reduction, which is extremely slow in the K354M mutant (k(1) = 0.09 +/- 0.01 s(-1); k(2) = 0.005 +/- 0.001 s(-1)) but much faster in the D124N aa(3) (k(1) = 21 +/- 6 s(-1); k(2) = 2.2 +/- 0.5 s(-1)). NO causes a very large increase (>100-fold) in the rate constant of heme a(3) reduction in the K354M mutant but only a approximately 5-fold increase in the D124N mutant. The K354M enzyme reacts rapidly with O(2) when fully reduced but is essentially inactive in turnover; thus, it was proposed that impaired reduction of the active site is the cause of activity loss. Since at saturating [NO], heme a(3) reduction is approximately 100-fold faster than the extremely low turnover rate, we conclude that, contrary to O(2), NO can react not only with the two-electron but also with the single-electron reduced active site. This mechanism would account for the efficient inhibition of cytochrome oxidase activity by NO in the wild-type enzyme, both from P. denitrificans and from beef heart. Results also suggest that the H(+)-conducting K pathway, but not the D pathway, controls the kinetics of the single-electron reduction of the active site.

center, a prerequisite for the reaction with O 2 , occurs via intramolecular electron transfer from heme a, which in turn is reduced by Cu A , the metal center accepting electrons from cytochrome c. Protons (both scalar and vectorial) are made available in situ via two putative H ϩ -conducting pathways, identified in the crystallographic structure (3,4). These pathways, called K and D from the residues Lys-354 2 and Asp-124 of subunit I, play different roles in the mechanism, as extensively investigated by site-directed mutagenesis (see Refs. 1, 2, and 5 for reviews).
The catalytic cycle of cytochrome c oxidase can be divided into a reductive and an oxidative part. In the reductive part, two electrons are sequentially transferred to the fully oxidized heme a 3 -Cu B center called O, yielding the two-electron reduced site R via a single-electron reduced intermediate E. In the oxidative part, upon reaction with O 2 , R restores the fully oxidized enzyme O, by populating the O 2 intermediates P and F (depending on the redox state of heme a, two different P intermediates are formed, called P M and P R ). The idea that the O 3 R process is the rate-determining step in the overall catalytic cycle is gaining further support (6 -8).
Mutation of Lys-354 to M within the K pathway yields a virtually inactive enzyme, as shown for the Rhodobacter sphaeroides aa 3 (9 -11), the Escherichia coli bo 3 (12), and the Paracoccus denitrificans aa 3 (5). This mutation affects primarily, but not exclusively (see Ref. 13), the reductive part of the catalytic cycle (9,11,14); in the absence of O 2 and with a large excess of reductant, heme a 3 is reduced at an extremely low rate (time scale of several minutes) as compared with the wild type (time scale of tens of milliseconds). This reduction block is presumably due to an impaired H ϩ transfer in the K354M mutant, consistent with the loss of the millisecond phase in laser-triggered reverse electron transfer experiments observed with the analogous mutant of the R. sphaeroides enzyme (15). Recently, two groups (16 -18) reported time-resolved electrometric measurements on liposome-reconstituted mutants of the P. denitrificans CcOX by laser excitation of ruthenium(II) bispyridyl. According to Ruitenberg et al. (16), injection of a single electron into the oxidized enzyme is coupled to an H ϩ transfer through the K pathway, linked to reduction of heme a. In contrast, Verkhovsky et al. (18) proposed that an H ϩ uptake through the K pathway controls the single-electron reduction of heme a 3 -Cu B (O 3 E), impaired in the K354M mutant, whereas the formation of the two-electron reduced active site (E 3 R) would be coupled to an H ϩ uptake through the D pathway, as deduced from data on the inactive D124N mutant * This work was supported in part by a Vigoni Program grant from the Conferenza dei Rettori delle Università Italiane and Deutscher Akademischer Austauschdienst (to M. B., F. M., and B. L.), by a Programma di Ricerca Scientifica Interuniversitario Nazionale "Bioenergetica: aspetti genetici, biochimici e fisiopatologici" from the Ministero dell'Istruzione, dell'Università e della Ricerca of Italy (to P. S. and F. M.), and by Grant SFB 472 from the Deutsche Forschungsgemeinschaft (to B. L.). 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: Instituto di Biologia e Patologia Molecolari del Consiglio Nazionale delle Ricerche, c/o Dipartimento di Scienze Biochimiche "A. Rossi. Fanelli", Università di Roma "La Sapienza," Piazzale Aldo Moro 5, I-00185 Roma, Italia. Tel.: 39-06-4450291; Fax: 39-06-4440062; E-mail: alessandro.giuffre@ uniroma1.it. 1 The abbreviations used are: CcOX, cytochrome c oxidase; SVD, singular value decomposition; O, enzyme with oxidized heme a 3 -Cu B site; E, enzyme with a single-electron reduced heme a 3 -Cu B ; R, enzyme with a two-electron reduced heme a 3 -Cu B . (17). The first of the two protons taken upon reduction of the active site has been proposed to charge-compensate the reduction of Cu B in the single-electron reduced active site via protonation of a putative OH Ϫ bound to this metal in the oxidized state (1,17).
The effect of the K354M mutation is drastic in the reductive part of the catalytic cycle but much smaller in the oxidative part. As assessed by the flow-flash technique using the R. sphaeroides CcOX analogous to the K354M mutant, the fully reduced enzyme exposed to O 2 becomes fully oxidized within ϳ5 ms (15), although without the formation of P R (13). The loss of oxidase activity associated to the K354M mutation has been therefore assigned to the extremely slow formation of R (9,11,14), which is a prerequisite for the reaction with O 2 .
Differently from O 2 , NO has been suggested to bind not only to R (19) but also to a single-electron reduced intermediate E (20,21). This hypothesis, raised to account for the very low apparent K i for NO inhibition, although consistent with computer simulations (20,21), is not yet supported by direct experimental evidence. In this report, we provide evidence for the reaction of NO with E by studying the kinetics of reduction of the K354M and D124N mutants of P. denitrificans CcOX in the presence of NO.

EXPERIMENTAL PROCEDURES
Dodecyl-␤-D-maltoside was purchased from Biomol (Hamburg, Germany); ascorbate, glucose oxidase, and catalase were purchased from Sigma; and ruthenium(III) hexamine was purchased from Aldrich. Stock solutions of NO (Air Liquide, Paris, France) were prepared by equilibrating degassed water with the pure gas at 1 atm ([NO] ϭ 2 mM at 20°C).
The K354M and D124N mutants of cytochrome c oxidase from P. denitrificans were purified according to Ref. 22 and stored at Ϫ80°C. Before use, the enzymes were equilibrated by dialysis (at 4°C for at least 5 h) with the buffer used in the experiments (100 mM K ϩ /phosphate, pH 7.0, ϩ 0.1% dodecyl maltoside or 35 mM K ϩ /phosphate, pH 7.0, ϩ 50 mM KCl ϩ 0.1% dodecyl maltoside). Cytochrome oxidase concentration is expressed in terms of functional units (aa 3 ) using the extinction coefficient ⌬⑀ red-ox, 444 ϭ 156 mM Ϫ1 cm Ϫ1 .
Stopped-flow experiments were carried out with a DX.17MV Applied Photophysics instrument equipped with a diode array (Leatherhead, UK). The mixing apparatus allows rapid mixing of equal volumes of solutions either in a simple or a sequential mode; in the latter mode, two solutions are premixed, and after a preset delay, they are mixed again with another solution. The instrument has a 1-cm light path and can acquire absorption spectra with an acquisition time of 2.5 ms. In a typical experiment, ascorbate (80 mM) and ruthenium hexamine (4 mM) are premixed with N 2 -equilibrated buffer (with or without NO), and the resulting solution is mixed after a 100-ms delay with degassed oxidized CcOX at 20°C. This protocol prevents prolonged incubation of reductants with NO. Contaminant oxygen was scavenged with glucose and catalytic amounts of glucose oxidase and catalase. After the second mixing, absorption spectra were collected as a function of time according to a logarithmic scale. Data analysis was carried out using the software MATLAB (The MathWorks, South Natick, MA). Spectral smoothing and deconvolution were performed by using the singular value decomposition (SVD) algorithm according to Henry and Hofrichter (23) or by the pseudoinverse algorithm.
The stoichiometry of NO binding to the K354M CcOX has been measured according to Stubauer et al. (24) by using a NO-selective Clark-type electrode (ISO-NO, World Precision Instruments). The electrode is calibrated using aliquots of NO-saturated water added to the degassed buffer and, after the addition of CcOX, the concentration of NO in solution is monitored.

RESULTS
In the present investigation, we studied by stopped-flow spectrophotometry the kinetics of reduction of the K354M and the D124N mutants of the P. denitrificans CcOX both in the presence and in the absence of NO. As shown in Fig. 1, the K354M mutation yields a dramatic decrease in the rate of heme a 3 reduction, consistent with the literature (9,11,14,17). Upon mixing anaerobically oxidized K354M CcOX with a large excess of ascorbate and ruthenium hexamine, heme a reduction is very fast (ϳ7 ms), whereas reduction of heme a 3 is extremely slow (Ͼ500 s, Fig. 1A). SVD analysis of the latter process shows a single significant optical component (corresponding to the reduced minus oxidized heme a 3 spectrum), displaying a biphasic time course (Fig. 1C). Best fit of the time course yields k 1 ϭ 0.09 Ϯ 0.01 s Ϫ1 and k 2 ϭ 0.005 Ϯ 0.001 s Ϫ1 for the two phases, accounting for ϳ30 and 70% of the total amplitude, respectively. We cannot exclude that the intrinsic reduction rate might be even slower than observed given that over the very long time scale explored (500 s), the high intensity light beam of the diode array instrument causes partial enzyme reduction even in the absence of reductants (data not shown). In agreement with others (5,9,11), we conclude that the very slow heme a 3 reduction may account for the extremely slow turnover observed with O 2 (ϳ0.02 mol of O 2 /mol of CcOX ϫ s at 170 M cytochrome c) A different scenario is observed when the reduction of the K354M CcOX is carried out in the presence of NO (Fig. 1B). In these experiments, the stopped-flow apparatus was used in the sequential mixing mode to prevent the prolonged incubation of NO with the reductants in the stopped-flow syringe to avoid NO loss (see "Experimental Procedures"). At 500 M NO (concentration after mixing), heme a reduction is again complete within a few milliseconds. In this case, however, the end point species (i.e. the fully reduced enzyme with NO bound to heme a 3 ) is already populated after about 10 s, indicating a much In the absence of NO, heme a 3 is reduced very slowly with rate constants of k 1 ϭ 0.09 s Ϫ1 (30% total amplitude) and k 2 ϭ 0.005 s Ϫ1 (70% total amplitude). In the presence of NO, heme a 3 reduction is much faster with rate constants of k 1 ϭ 8.9 s Ϫ1 (50% total amplitude) and k 2 ϭ 0.58 s Ϫ1 (50% total amplitude).
faster internal electron transfer in the presence of NO. SVD analysis of the absorption spectra collected from 7 ms up to 10 s reveals only a single optical component corresponding to the [heme a 3 2ϩ -NO]-[heme a 3 3ϩ ] difference spectrum, indicating that NO binding is rate-limited by (and thus, apparently synchronous with) heme a 3 reduction. Analysis of the time course in Fig. 1C shows that the time course of heme a 3 2ϩ -NO formation is biphasic with rate constants of k 1 ϭ ϳ8.9 s Ϫ1 and k 2 ϭ ϳ0.6 s Ϫ1 , the two phases having similar amplitude. Thus, NO seems not to interfere with heme a reduction (very fast both with and without NO) but clearly drives heme a 3 reduction, which occurs in the presence of NO at least 100-fold faster than in its absence (Fig. 1C).
The experiments reported above were extended to the D124N mutant (Fig. 2). In the latter mutant, similarly to the K354M mutant, the reduction of heme a is fast both in the presence and in the absence of NO, being complete within a few milliseconds after mixing with reductant (Fig. 2, A and B). In agreement with the literature (17), the enzyme is completely reduced within a few seconds even in the absence of NO (compare Figs. 1A and 2A), and thus, heme a 3 reduction is much faster than in the case of the K354M mutant. Also, for the D124N mutant, heme a 3 reduction is biphasic with k 1 ϭ 21 Ϯ 6 s Ϫ1 and k 2 ϭ 2.2 Ϯ 0.5 s Ϫ1 (relative amplitudes ϳ70 and 30%, respectively). Complete reduction of the D124N mutant is accelerated in the presence of NO (Fig. 2B) and, for instance, at 500 M NO, the formation of the heme a 3 2ϩ -NO complex proceeds at k 1 ϭ 74 Ϯ 11 s Ϫ1 and k 2 ϭ 3.6 Ϯ 1.6 s Ϫ1 (relative amplitudes ϳ60 and 40%, respectively). Thus, we conclude that in both mutants, the addition of NO increases the rate of internal electron transfer; however, this increase corre-sponds to a factor of ϳ5 for the D124N mutant and to Ͼ100-fold in the K354M mutant. This result is better visualized in Fig. 3, in which the two observed rate constants of heme a 3 reduction for both mutants are reported at different NO concentrations (from 0 to 500 M). The data show that internal electron transfer in the D124N mutant is faster than in the K354M mutant. Moreover, in both mutants, the two rate constants (relative to the fast and the slow kinetic phases) depend on the NO concentration. Although the dependence is much less pronounced in the D124N mutant, all rate constants become essentially independent of [NO], reaching plateau values at k 1 ϭ 13 Ϯ 6 s Ϫ1 and k 2 ϭ 0.7 Ϯ 0.3 s Ϫ1 in the K354M mutant and k 1 ϭ 110 Ϯ 16 s Ϫ1 and k 2 ϭ 3.5 Ϯ 1.5 s Ϫ1 in the D124N mutant. All asymptotic values are independent of reductant concentration, as assessed by experiments at 500 M NO and variable ruthenium hexamine concentration; at ruthenium hexamine concentrations above 1 mM (after mixing), the observed rate constants were independent of reductant concentration.
It is worth noticing that the faster heme a 3 reduction observed with both mutants in the presence of NO is not due to direct reduction of the oxidized binuclear center O by NO. As demonstrated with beef heart CcOX and confirmed with the P. denitrificans wild-type CcOX, the reaction of NO with O occurs rapidly with the chloride-free enzyme, yielding nitrite-bound heme a 3 and reduced heme a (25), but it is prevented with the chloride-bound oxidase (26). Spectrophotometrically, we did not detect a reaction after mixing both oxidized mutants with NO (1 mM after mixing, not shown). Further in this respect, the reactivity of NO with the oxidized K354M oxidase was probed by amperometry by using a NO-selective Clark-type electrode (24). If the oxidized enzyme is anaerobically added to a degassed NO-containing solution, a reaction would be detected as a decrease in the NO concentration. Upon the addition of oxidized K354M CcOX (0.4 M) to a solution containing NO (1 M), only a small decrease in the NO concentration (ϳ0.2 mol of NO/mol of oxidase) was detected; in contrast, a stoichiometric (1:1) NO binding was observed after the addition of the fully reduced enzyme, as shown for the mammalian CcOX (24). We therefore conclude that both the P. denitrificans mutants  Fig. 1. A and B, similar to the K354M aa 3 , heme a reduction is complete within a few milliseconds, independently of NO. Afterward, complete reduction of the enzyme is achieved within a few seconds in the absence of NO (A) and on an even shorter time scale in the presence of NO (B). C, time courses of heme a 3 reduction as obtained by the pseudoinverse analysis. In the absence of NO, heme a 3 is reduced with rate constants of k 1 ϭ 15 s Ϫ1 (60% total amplitude) and k 2 ϭ 2.7 s Ϫ1 (40% total amplitude). At 500 M NO, heme a 3 reduction is faster and proceeds with rate constants of k 1 ϭ 78 s Ϫ1 (60% total amplitude) and k 2 ϭ 5.1 s Ϫ1 (40% total amplitude).

FIG. 3. Effect of NO concentration.
Rate constants of heme a 3 reduction, relative to the fast (top) and the slow (bottom) phases, were measured at varying NO concentrations. Experimental conditions were as described in the legend for Fig. 1. Under all conditions, internal electron transfer in the D124N mutant is faster than in the K354M mutant. Both rate constants depend on the NO concentration, although the dependence is much less pronounced in the D124N mutant. tested in this study are in the chloride-bound state, due to the presence of chloride in the buffers used during both the purification and the experiments. DISCUSSION NO is a very efficient, yet reversible, inhibitor of cytochrome c oxidase activity (27,28), leading to the proposal that it may act as a physiological modulator of cell respiration (29). Since both NO and O 2 react with the fully reduced heme a 3 -Cu B center R with high affinity and similar rates (19), the small inhibition constant K i determined with mitochondrial CcOX in turnover (K i ϭ 270 nM NO at [O 2 ] ϭ 140 M, (27)) was somewhat puzzling. To account for this observation, it was proposed that NO can react with a single-electron reduced active site E (20 -21), which is known to be unreactive toward O 2 . Such a hypothesis is consistent with computer simulations (20 -21) but has never been demonstrated experimentally. The kinetics of reduction of the K354M mutant of P. denitrificans in the presence of NO, reported above, provides evidence that this hypothesis is correct.
The K354M mutation is associated with the loss of oxidase activity (5,9,11), although this mutant in the fully reduced state (R) is very quickly (ϳ5 ms) oxidized by O 2 , as reported for the R. sphaeroides CcOX (15, but see also Ref. 13). The same mutation has a dramatic effect on the reductive part of the catalytic cycle, and the extremely low rate of reduction of heme a 3 was correlated to the marginal turnover rate (9,11,14). It was therefore assumed that in this mutant, the turnover with O 2 is rate-limited by the extremely slow formation of R, which is an obligatory intermediate in the catalytic cycle. This is consistent with the widely accepted idea that O 2 can react exclusively with the two-electron reduced heme a 3 -Cu B site, like CO.
The novel result reported in this study on the K354M mutant of P. denitrificans CcOX is that, in the presence of NO, the reduction of heme a 3 occurs at a rate much faster (Ͼ100-fold) than in its absence and much faster than the extremely low turnover rate of this mutant with O 2 (Fig. 1). This effect depends on NO concentration and is maximal at [NO] Ͼ100 M (Fig. 3), at which the reduction of heme a 3 proceeds at rates (k 1 ϭ 13 Ϯ 6 s Ϫ1 ; k 2 ϭ 0.7 Ϯ 0.3 s Ϫ1 ) both remarkably larger than the turnover rate (ϳ0.02 mol of O 2 /mol of CcOX ϫ s at [O 2 ] Ͼ 250 M). This finding is diagnostic of a different reactivity of O 2 and NO with CcOX. It is indeed difficult to account for this result, assuming that NO, similarly to O 2 , can react exclusively with R, which combines with very high affinity and second order rate constants with both ligands. If this were the case, the formation of the reduced NO-bound heme a 3 would be much slower, being rate-limited by the formation of R, which in turn accounts for the extremely low oxidase activity. Therefore this result implies that NO can react not only with the two-electron reduced heme a 3 -Cu B site R (19) but also with the singleelectron reduced site E, whose occurrence in this mutant was already documented (11). We do not have a valid explanation for the observed heterogeneity in the reduction of heme a 3 , but we notice that a similar biphasic behavior was reported also for the beef heart enzyme (7,8).
Working with beef CcOX, it was shown that the reaction of NO with Cu B in the oxidized binuclear site (O) occurs rapidly only with the chloride-free enzyme, leading to reduced heme a and nitrite-bound oxidized heme a 3 (25) but is prevented by the binding of chloride (26). This behavior was reproduced with the P. denitrificans wild-type enzyme. On mixing either of the mutants in the oxidized state with a large excess of NO, we did not observe spectrophotometrically either heme a reduction or nitrite formation. Moreover, we observed by amperometry only a small reaction between NO and oxidized K354M (similar to that generally detected even with the beef heart enzyme in the chloride-bound form, see Ref. 26). We therefore conclude that the two mutants employed in this study are in the Cl Ϫ -bound form since chloride is present in the buffers used during the purification and the experiments. This further implies that the enhanced reduction of heme a 3 observed in the presence of NO cannot be assigned to the direct reaction of NO with O.
It is noteworthy that the dependence on [NO] reported in Fig.  3 is consistent with the idea that the K354M mutation impairs the O 3 E electron transfer step (17,18). Assuming that NO binds to heme a 3 2ϩ in the E and the R states with very high affinity and second order rate constants, one may expect that NO should efficiently drive thermodynamically the reduction of heme a 3 in the O 3 E step. On this basis, the NO concentration dependence of the apparent rate constants measured for the K354M mutant (Fig. 3) seems diagnostic of a relatively slow forward electron transfer (maximal value of 13 Ϯ 6 s Ϫ1 ) as compared with an unusually fast reverse electron transfer (heme a 3 /Cu B 3 heme a) caused by the mutation. In this context, it is interesting to notice that the D124N mutant shows a kinetic behavior remarkably different from the one displayed by the K354M mutant. In the D124N mutant, (i) heme a 3 reduction in the absence of NO is much faster than in the K354M mutant, in agreement with Wikström et al. (17), and (ii) NO increases this rate at most by a factor of ϳ5 (ratio of the rate constant at saturating [NO] over the value measured in the absence of NO), i.e. much less than the over 100-fold increase observed with the K354M mutant. These results are fully consistent with the hypothesis that the K pathway, but not the D pathway, controls the first electron transferred to the oxidized heme a 3 -Cu B site (16 -18).
The analogous K354M mutant of the R. sphaeroides CcOX has been reported to display cytochrome c-peroxidase activity with a K m value of ϳ50 mM H 2 O 2 and a V max value of ϳ25 s Ϫ1 (14,30). The maximal turnover with H 2 O 2 was much faster than the turnover with O 2 , and this result was interpreted as an evidence that H 2 O 2 reacts with O, yielding the intermediate P, i.e. bypassing the whole reductive part of the catalytic cycle (14). Later on, it was proposed that H 2 O 2 might react with E in the K354M mutant, yielding directly the F intermediate, thus bypassing the formation of both R and P (2,31). We wish to point out that according to our results, the latter interpretation has to be favored, and in this respect, H 2 O 2 and NO behave similarly since they are both assumed to react with E. We notice that the maximal turnover number for the peroxidase activity (25 s Ϫ1 , see Ref. 14) is not inconsistent with the faster rate constant for the O 3 E process that we estimate from our data on the K354M mutant at saturating [NO] (13 Ϯ 6 s Ϫ1 ; Fig.  3, top panel). In other words, we propose that in the K354M mutant, the rate constant of heme a 3 reduction measured at saturating concentrations of NO and ruthenium hexamine is the forward rate constant for the single-electron reduction of the heme a 3 -Cu B site, which is also rate-limiting the cytochrome c-peroxidase activity at saturating [H 2 O 2 ].
In conclusion, our results provide direct evidence that NO, differently from O 2 , can react with a single-electron reduced heme a 3 -Cu B in CcOX. This finding validates the original hypothesis (20,21) raised to account for the finding that NO is a potent inhibitor of CcOX (27,28).