Electron Spin Resonance Investigation of the Cyanyl and Azidyl Radical Formation by Cytochrome c Oxidase*

Cyanide (CN−) is a frequently used inhibitor of mitochondrial respiration due to its binding to the ferric heme a 3 of cytochrome coxidase (CcO). As-isolated CcO oxidized cyanide to the cyanyl radical (⋅CN) that was detected, using the ESR spin-trapping technique, as the 5,5-dimethyl-1-pyrroline N-oxide (DMPO)/⋅CN radical adduct. The enzymatic conversion of cyanide to the cyanyl radical by CcO was time-dependent but not affected by azide (N3 −). The small but variable amounts of compound P present in the as-isolated CcO accounted for this one-electron oxidation of cyanide to the cyanyl radical. In contrast, as-isolated CcO exhibited little ability to catalyze the oxidation of azide, presumably because of azide’s lower affinity for the CcO. However, the DMPO/⋅N3 radical adduct was readily detected when H2O2 was included in the system. The results presented here indicate the need to re-evaluate oxidative stress in mitochondria “chemical hypoxia” induced by cyanide or azide to account for the presence of highly reactive free radicals.

Cyanide has been widely used as an inhibitor to study enzymatic hemoproteins such as oxidases and peroxidases (1). The unique ability of cyanide to inhibit aa 3 -type cytochrome oxidase is extremely important for investigating the enzyme complexes of the biological respiratory chain (2). Cyanide has long been known to serve as a ligand for the heme a 3 based on its ability to change the visible spectral characteristics of the heme environment (3,4). Kinetic and mechanistic questions relating to the reaction of cytochrome c oxidase (CcO) 1 with cyanide have been the subject of many recent experiments (5)(6)(7). In addition to inhibiting hemoprotein-catalyzed reactions, cyanide is actively metabolized to carbon dioxide in biological systems (8). The cyanyl radical ( ⅐ CN) can be generated by one-electron oxidation of cyanide and detected with spin trapping in chemical (9) and electrochemical (10,11) oxidizing systems. Previous studies (12,13) indicated that peroxidases such as horseradish peroxidase, lactoperoxidase, or chloroperoxidase can use cyanide as a substrate and catalyze the one-electron oxidation to the cyanyl radical in the presence of H 2 O 2 . A similar result was also reported in the study of lignin peroxidase H2 from the white rot fungus Phanerochaete chrysosporium (14). Pharmacologically, cyanide is used as a chemical model of ischemia (15)(16)(17)(18)(19) because of its ability to block oxidative phosphorylation and thereby induce the production of reactive oxygen species. Cyanide-induced neurotoxicity and oxidative stress have also recently received much attention (20,21). It has been proposed that cyanide can activate the N-methyl-D-aspartate receptors for signal transduction (22,23) and trigger the enzymatic pathways required for the generation of nitric oxide and reactive oxygen species, which are associated with lipid peroxidation in the nervous system.
Azide also exhibits inhibitory effects on the mitochondrial CcO (2) and other hemoproteins such as catalase (24,25). The inhibition of purified CcO by azide was moderately strong (26) and kinetically different from that caused by cyanide (27). The property of azide to selectively inhibit the CcO of the electron transport chain in mitochondria has prompted researchers to use azide infusion to investigate mitochondrial dysfunctionrelated diseases such as hypoxia/ischemia and Alzheimer's disease (28 -32). More recently, a high resolution x-ray structure of azide-bound CcO has been reported (33), providing information to help in understanding the mechanism of azide inhibition. Like the cyanyl radical, the azidyl radical ( ⅐ N 3 ) can be formed by one-electron oxidation of azide and detected by the ESR spin-trapping technique (34). Earlier research indicated that the oxidation of azide to the azidyl radical can be catalyzed by various peroxidases in the presence of H 2 O 2 (34). This azidyl radical was later demonstrated to be covalently incorporated into the prosthetic group of peroxidases, resulting in its inactivation (35).
In this investigation, we report for the first time the detection of cyanyl radical during the inhibition of CcO with KCN. We point out that as-isolated CcO was capable of oxidizing cyanide to cyanyl radical in the absence of added H 2 O 2 . In contrast, as-isolated CcO showed little capability of oxidizing azide to azidyl radical, but produced readily detectable amounts of azidyl radicals in the presence of H 2 O 2 .

EXPERIMENTAL PROCEDURES
Materials-K 13 CN was purchased from MSD Isotopes (Montreal, Canada), and 15 N 3 -labeled NaN 3 was from Stohler Isotope Chemicals (Waltham, MA). The 5,5-dimethyl-1-pyrroline N-oxide (DMPO) spin trap was purchased from Aldrich and was vacuum-distilled twice and stored under nitrogen at Ϫ70°C until needed. The cytochrome c was reduced with ascorbic acid and then passed through a Sephadex G-25 column before use. All reactions used 50 mM phosphate buffer at pH 7.4, Chelexed overnight, with 200 M DTPA added after Chelex treatment.
Preparation of Mitochondrial Cytochrome c Oxidase from Bovine Heart-Highly purified bovine heart mitochondrial CcO was isolated using the method developed by Yu et al. (36). Submitochondrial particles were used as starting material and subjected to ammonium sulfate fractionation in the presence of 1.5% sodium cholate. Optical spectra of CcO were recorded in a SLM AMINCO DW2000 spectrophotometer. The enzyme concentration (per aa 3 ) of CcO was calculated using an extinction coefficient of 24 mM Ϫ1 cm Ϫ1 for the dithionite reduced-minusoxidized form at 605 nm (36). The CcO, isolated as the fast form (5), contained 7-8 g of phospholipid and 10 -12 nmol of heme a/mg protein. The concentration of compound P (peroxy form) or compound F (ferryloxo form) of CcO was calculated from the difference spectrum [(compound P or compound F) Ϫ(as-isolated CcO)] using extinction coefficients of ⌬A 607-630 ϭ 11 mM Ϫ1 cm Ϫ1 for compound P and ⌬A 580 -630 ϭ 5.3 mM Ϫ1 cm Ϫ1 for compound F (7).
ESR Experiments-ESR experiments were carried out on a Bruker ESP 300 ESR spectrometer operating at 9.8 GHz with 100 kHz modulation frequency at room temperature. The reaction mixtures were transferred to a 17-mm quartz ESR flat cell, which was then positioned into the TM 110 microwave cavity. The sample was scanned using the following parameters: center field, 3480 G; modulation amplitude, 1.0 G; time constant, 0.33 s; scan time, 335 s for a 100-G scan; gain, 1 ϫ 10 5 ; and microwave power, 20 mW. The spectral simulations were performed using the WinSim program of the NIEHS public ESR software package.

DMPO Spin Trapping of the Cyanyl Radical in the Reaction
of CcO with KCN-The ESR spectrum recorded after 1 h from a reaction mixture containing 0.2 mM CcO, 10 mM KCN, and 100 mM DMPO was simulated as a combination of two DMPO radical adducts (Fig. 1A). The first adduct was assigned to DMPO/cyanyl radical adduct (DMPO/ ⅐ CN) based on the hyperfine coupling constants a N ϭ 15.43 and a H ␤ ϭ 18.90 G (Fig. 1B) (12). The second adduct was assigned to DMPO/hydroxyl radical adduct (DMPO/ ⅐ OH) based on the hyperfine coupling constants a N ϭ 15.0 and a H ␤ ϭ 14.75 G (Fig. 1C) (37). In the absence of CcO (Fig. 2B) or in the presence of heat-denatured (100°C for 30 min) CcO (data not shown), DMPO/ ⅐ CN was not detected. The nucleophilic addition of CN Ϫ to DMPO could form an ESR-silent hydroxylamine that would be expected to facilely air oxidize to DMPO/ ⅐ CN (38). This enzymatic dependence of DMPO/ ⅐ CN formation supports a direct role of CcO in the oxidation of CN Ϫ to ⅐ CN. The DMPO/ ⅐ OH adduct was detected in the absence of CcO and was eliminated when manganese-containing superoxide dismutase was added to the cyanide/DMPO solution (Fig. 2C). This indicated that the DMPO/ ⅐ OH in Fig. 2B is present as a decomposition product of DMPO/ ⅐ OOH or that hydroxyl radical formation was superoxidedependent. It should be noted that manganese-containing superoxide dismutase did not eliminate all of the DMPO/ ⅐ OH when the CcO was present in the system (data not shown). In addition, no DMPO/ ⅐ CN was detected when the KCN was excluded from the system (Fig. 2D). To test the possibility that the detectable cyanyl radical in the CcO/KCN system resulted from contamination with H 2 O 2 , CcO was incubated with catalase prior to the addition of KCN and DMPO. The resulting ESR spectrum did not significantly differ from that obtained The spectrum was recorded after a 1-h incubation at room temperature. B, same reaction mixture as in A, but without the CcO. C, same reaction mixture as B, but the system contained 333 units/ml of manganese-containing superoxide dismutase (SOD). D, same reaction mixture as in A, but KCN was omitted from the system. E, same as in A, except CcO was pretreated with 333 units/ml catalase prior to the addition of the KCN and DMPO. F, same as in A, except CcO was pretreated with 100 mM of NaN 3 prior to the addition of KCN and DMPO. G, same as in A, but the K 12 CN was replaced with K 13 CN. The DMPO/ ⅐ OH features are labeled with asterisks. H, the computer simulation of G. The instrumental settings are described under "Experimental Procedures." from the system without catalase (Fig. 2E). H 2 O 2 contamination can therefore be excluded. It should be noted that an incubation of catalase, KCN, and DMPO did not generate a detectable cyanyl radical (data not shown). Pretreatment of enzyme with azide, up to 500 eq per aa 3 , did not inhibit the formation of cyanyl radical (Fig. 2F). To confirm that the cyanyl radical is the carbon-centered radical trapped in the reaction of CcO with cyanide, K 13 CN (I ϭ 1/2) was used. As shown in Fig.  2G, the 6-line DMPO/ ⅐ 12 CN spectrum was completely converted to the 12-line DMPO/ ⅐ 13 CN spectrum (a N ϭ 15.43, a H ␤ ϭ 18.90, and a 13C ␤ ϭ 12.95 G), thus confirming the origin of the trapped carbon-centered radical as the cyanyl radical (12).
To test the possibility that the highly reactive ⅐ OH may be generated and could attack CN Ϫ thereby producing the cyanyl radical, we added the well known hydroxyl radical scavenger Me 2 SO (700 mM) to the system containing 0.2 mM CcO, 10 mM KCN, and 100 mM DMPO. No effect on the DMPO/ ⅐ CN concentration was found (data not shown). Me 2 SO scavenging of hydroxyl radical should also inhibit any DMPO/ ⅐ OH adduct formed by the trapping of hydroxyl radical. The inclusion of Me 2 SO in the CcO/KCN system only slightly decreased the production of DMPO/ ⅐ OH (less than 10%, data not shown), indicating that greater than 90% of DMPO/ ⅐ OH was formed independently of hydroxyl radical (37). To investigate the possibility that DMPO/ ⅐ OH was formed by the nucleophilic substitution of CN Ϫ from DMPO/ ⅐ CN by OH Ϫ (39), the experiment was repeated in buffer prepared in H 2 17 O. The CcO was precipitated by 30% saturated ammonium sulfate and then dissolved with 50 mM phosphate buffer in H 2 17 O, pH 7.4. The ESR spectrum from the reaction system of CcO (H 2 17 O)/KCN was identical to that of CcO/KCN ( Fig. 2A), demonstrating that the possible conversion of DMPO/ ⅐ CN to DMPO/ ⅐ OH via a nucleophilic substitution reaction did not occur to a significant extent (data not shown).
The formation of the cyanyl radical with CcO was time-dependent. As shown in Fig. 3A, the intensity of DMPO/ ⅐ CN gradually increased with time. Each spectrum shown in Fig. 3A was simulated using three species, DMPO/ ⅐ CN, DMPO/ ⅐ OH, and DMPO/ ⅐ ScysCcO protein-derived thiyl radical (a N ϭ 14.68,  a H ␤ ϭ 15.74 G, and line width ϭ 3.12 G) (40). The formation of DMPO/ ⅐ ScysCcO was prevented by the pretreatment of CcO with N-ethylmaleimide. The protein-derived thiyl radical was presumably formed from the reaction of Cys residues of CcO with highly reactive ⅐ CN. At higher concentrations of DMPO (up to 800 mM), the contribution of DMPO/ ⅐ ScysCcO relative to DMPO/ ⅐ CN was diminished, consistent with DMPO trapping the ⅐ CN as the primary radical and preventing the formation of ⅐ ScysCcO (data not shown). This DMPO/ ⅐ ScysCcO radical adduct was previously detected when H 2 O 2 was added to CcO (40) and did not contribute to the spectra shown in Figs. 1A or 2A obtained from the reaction mixture after a 1-h incubation. The percent radical adduct concentration derived from the simulations was multiplied by the area of the complete spectrum (obtained from double integration) to quantify the radical adduct concentration. The DMPO/ ⅐ CN concentration was plotted versus time as indicated in Fig. 3B.
Because the generation of the cyanyl radical from cyanide by peroxidases such as horseradish peroxidase, lactoperoxidase, chloroperoxidase (12,13), and lignin peroxidase H2 (14) was H 2 O 2 -dependent, it was important to understand how the asisolated CcO could independently catalyze the same reaction without H 2 O 2 . As reported previously by Fabian and Palmer (7,41), as-isolated CcO contains a small and variable amount of the peroxy form of the enzyme (compound P). To address this issue, the CcO from our preparation was converted to a mixture of its peroxy and ferryloxo forms by the addition of 30 eq of H 2 O 2 at pH 8.0 and examined by the differential spectrum using the as-isolated CcO as a reference. We consistently estimated the amount of compound P/compound F mixture to be 85-95% of theoretical (up to five batches of preparation), which is in line with the observation of Fabian and Palmer (7,41). To bleach the peroxy form present in the as-isolated CcO and convert it to the fully oxidized form, the enzyme was dialyzed against 50 mM phosphate buffer, pH 7.4, overnight. Sodium cholate (1%) was included in the dialysis buffer to prevent the aggregation of protein. The addition of sodium cholate had no effect on the formation of radical by CcO (data not shown). A significant decrease of DMPO/ ⅐ CN (75-95% decrease, based on the four batches of preparation) was observed when this CcO preparation was incubated with cyanide in the presence of DMPO (data not shown). The addition of H 2 O 2 to dialyzed CcO restored the ESR spectrum of DMPO/ ⅐ CN (data not shown).

FIG. 3. Time dependence of radical adducts from the reaction of cytochrome c oxidase and KCN.
A, the components in the system were the same as those described in the legend of Fig.  1A. B, the spectra in A were simulated using three species: DMPO/ ⅐ OH, DMPO/ ⅐ CN, and DMPO/ ⅐ ScysCcO, a spin-trapped, protein-derived thiyl radical (40). The parameters of the computer simulation for DMPO/ ⅐ OH and DMPO/ ⅐ CN are the same as those described in Fig. 1. The parameters for DMPO/ ⅐ ScysCcO are described under "Results." The concentrations of DMPO/ ⅐ CN from each spectrum shown in A were calculated based on the product of molar ratio (from the computer simulation) and the area (obtained by double integration) of the complete spectrum.
The cyanyl radical can also be formed from the oxidation of cyanide by horseradish peroxidase in the presence of H 2 O 2 (12,13). Similar results were also observed in the system of CcO/ H 2 O 2 /KCN (Fig. 4A). The DMPO/ ⅐ CN was readily detectable when the H 2 O 2 was included in the system of as-isolated CcO/ KCN/DMPO (Fig. 4A). The elimination of H 2 O 2 from the system yielded a spectrum (Fig. 4B) that is similar to the spectrum recorded at 3 min in Fig. 3A and depends totally on CcO (Fig.  4C). Only the 4-line spectrum of DMPO/ ⅐ OH (Fig. 4D) was observed when the CcO was omitted from the system. The replacement of K 12 CN with K 13 CN converted the 6-line spectrum to a 12-line spectrum (Fig. 4E), thus confirming that the carbon-centered cyanyl radical was produced in the system of CcO/KCN/H 2 O 2 (12).

DMPO Spin Trapping of Azidyl Radical in the Reaction of NaN 3 with CcO in the Presence of H 2 O 2 -
The azidyl radical was detected during an investigation of the inhibitory effect of azide on catalase (34). In addition to catalase/H 2 O 2 , the oneelectron oxidation of azide to the azidyl radical can be catalyzed by various peroxidase/H 2 O 2 systems (34). More recently, azidyl radical was detected in the succinate-driven respiration of azide-inhibited submitochondrial particles (42). It was not clear whether the azidyl radical detected in the system of NaN 3 / submitochondrial particles arose from hydroxyl radical or from membrane-bound components with peroxidase activity. However, the inclusion of the CcO in the mixture of NaN 3 , H 2 O 2 , and DMPO resulted in an ESR spectrum consisting of a quartet of triplets (a NO N ϭ 14.8, a H ␤ ϭ 14.2, and a 14N ␤ ϭ 3.1 G) assigned to DMPO/ ⅐ N 3 (Fig. 5A). DMPO/ ⅐ N 3 formation was dependent on H 2 O 2 (Fig. 5B), apparently because of the inability of the asisolated CcO to oxidize NaN 3 . Removal of the enzyme from the system resulted in failure to produce the azide radical (Fig. 5C). The computer simulation of the spectrum in Fig. 5A indicated a mixture of azidyl radical and hydroxyl radical adducts. The features of the DMPO/ ⅐ OH radical adduct are not resolved in Fig. 5A because of coincidental overlap with the DMPO/ ⅐ N 3 radical adduct signal. To simulate the spectrum more accurately, 15 N 3 -azide (I ϭ 1/2) was used, which allowed the features of DMPO/ ⅐ OH to be clearly visible in the composite spectrum (Fig. 5D) (34). Simulation of the spectrum shown in Fig.  5D indicated a mixture of azidyl radical and hydroxyl radical adducts in an 84:16 ratio. The expected spectrum with a quartet of doublets (a NO N ϭ 14.8, a H ␤ ϭ 14.2, and a 15N ␤ ϭ 4.3 G) obtained from 15 N 3 -labeled sodium azide was found. The hyperfine coupling constants were essentially the same as those published (34). The ESR spectrum was recorded immediately (roughly 3 min) after reaction was initiated at room temperature. B, same as A, but the H 2 O 2 was omitted from the system. C, same as B, but the CcO was omitted from the system. D, same as A, but the CcO was omitted from the system. E, identical to A, except that the K 12 CN was replaced with K 13 CN. The hyperfine coupling constants for DMPO/ ⅐ N 3 and for DMPO/ ⅐ OH are described under "Results." B, identical to A, but with H 2 O 2 omitted. C, identical to A, but with CcO omitted. D, identical to A, except that NaN 3 was replaced with Na 15 N 3 . The computer simulation (dashed line) is superimposed upon the experimental spectrum (solid line). The hyperfine coupling constants for DMPO/ ⅐15 N 3 and for DMPO/ ⅐ OH are described under "Results." The DMPO/ ⅐ OH features are labeled with asterisks.

DISCUSSION
It had been pointed out by Yamazaki and Piette (43) as early as 1963 that the oxidation of cyanide by compound I of hemoproteins should be considered when cyanide is used as an inhibitor and irregular effects are found. The results presented in our study clearly demonstrate that the reaction of CcO with cyanide leads to a detectable cyanyl radical. Therefore, in addition to forming a cyanide complex, CcO also catalyzes the one-electron oxidation of cyanide to cyanyl radical. As previously mentioned, peroxidases such as horseradish peroxidase (HRP) can catalyze a similar reaction only in the presence of H 2 O 2 (12,13,34), and both compound I and compound II were proposed to account for the production of cyanyl radical in the KCN/HRP/H 2 O 2 system.
Both peroxidase compound I-and II-like analogs have been suggested to participate in the catalytic mechanism of CcO (43,44). It is generally thought that the catalytic cycle is initiated with the fully oxidized CcO. Cytochrome c provides two electrons to the Cu B /heme a 3 binuclear center via Cu A and heme a; the enzyme is converted to its two-electron reduced form, socalled "species R," which has a high affinity for dioxygen. The reaction of dioxygen with species R results in a transient formation of compound P (peroxy, "607 nm" form). Because H 2 O 2 carries two reducing equivalents, its addition to the fully oxidized CcO also forms the compound P species (8,44,45,53). The addition of a third electron to compound P leads to the formation of compound F (ferryloxo, "580 nm" form). The ferric form of CcO is then regenerated from compound F by the addition of a fourth electron accompanied by the release of H 2 O. The structure of compound P is not clear and is still under debate (44 -46). In a recent report, the reaction of the twoelectron reduced form of CcO with dioxygen was studied using time-resolved resonance Raman techniques. Those data indicate the involvement of a heme a 3 ferryloxo species in the structure of compound P (47). Compound F, as supported by substantial evidence (48 -52), is structurally related to compound II. It is proposed that compounds P and F would be the equivalents of compounds I and II of peroxidases if the oxygen chemistry of CcO were similar to that of a peroxidase (44 -46). The results conveyed here support this hypothesis. The ferric form of CcO (Fe a3 3ϩ Ϫ Cu B 2ϩ ) presumably cannot catalyze the one-electron oxidation of cyanide to the cyanyl radical without additional oxidizing equivalents from H 2 O 2 . Hence, the addition of H 2 O 2 yields compounds P and/or F that are capable of catalyzing the production of cyanyl radical from cyanide (Fig.  4A). Based on the observations of this investigation and the discussion above, the mechanism by which the cyanyl radical was produced from the reaction of cyanide with CcO is proposed in the following reactions.
Several possible structures including the Fe a3 IV ϭ O porphyrin cation radical (Por ⅐ ϩ Fe a3 IV ϭ O) were suggested for compound P depending on how the O-O bond of molecular oxygen is cleaved (45). We believe that compound P would be able to convert cyanide to cyanyl radical if the Fe a3 IV ϭ O porphyrin cation radical or its functional equivalent were formed. Our results also help explain the previous observation of Fabian and Palmer (7) in which the loss of the 607 nm band (compound P) and an increase in the 580 nm band (compound F) is detected in the reaction of cyanide with compound P.
We propose the reaction of fully oxidized CcO (Fe a3 3ϩ Ϫ Cu B 2ϩ ) with CN Ϫ in the presence of H 2 O 2 (Fig. 4A) to involve the following reactions.
The reaction converting cyanide to cyanyl radical in this system is much faster than that of the as-isolated CcO/cyanide system, presumably because the reactive enzyme intermediates are not limiting. A peroxidase compound I-like species containing Fe a3 IV ϭ O porphyrin cation radical at heme a 3 was preferably suggested to be involved in the reaction of fully oxidized CcO with H 2 O 2 (45,53). Visible spectroscopic analysis indicated that the 607 nm species, referred to as compound P (7, 41, 44, 45, 49 -53), was dominant in this system (data not shown). This evidence supports the possibility of a ferryloxo state of compound P as suggested in studies led by Proshlyakov (47,51) and Fabian (41). It is worth noting that a proteincentered radical (protein radical ϩ Fe a3 IV ϭ O) (50) was detected in the reaction system of CcO/H 2 O 2 using either DMPO or nitroso spin traps (40). We do not rule out the possibility that a protein-centered radical can generate the cyanyl radical.
Cyanide has been used to chemically stimulate ischemia (15)(16)(17)(18)(19). The blockade of oxidative phosphorylation, changes of membrane potential, and lipid peroxidation were generally proposed to account for the effects of cyanide-induced ischemia. Reactive oxygen species were detected in this system, leading to oxidative stress. The results reported here imply that the generation of the cyanyl radical may serve as an important factor in triggering these effects. Because the chemistry of the cyanyl radical is extremely oxidative, it may be reactive enough to initiate lipid peroxidation via the abstracting of H ⅐ from lipid. Cyanide-stimulated generation of oxidative stress has recently been reported in the study of cerebellar granule cells (20,21).
Unlike the reactivity with cyanide, as-isolated CcO had little ability to oxidize azide to a detectable azidyl radical (Fig. 5B). CcO exhibited the same behavior as catalase and various peroxidases (34) in that it did not produce a detectable azidyl radical in the absence of added H 2 O 2 (Fig. 5A). Currently, there is no evidence to indicate that the relative substrate affinity of cyanide to compound P or HRP-compound I is higher than that of azide. However, two groups (54, 55) employed infrared spectroscopy to investigate the interaction of azide with the heme a 3 -Cu B center of CcO. The infrared spectrum characteristic of azide-bound CcO can be eliminated by cyanide and replaced with that of the cyanide-bound CcO, thus leading to the conclusion that cyanide has a higher relative affinity of the ligand for CcO than azide and readily displaces azide. The result of this investigation indicated that the enzymatic conversion of cyanide to cyanyl radical by CcO was not affected by azide (Fig.  2F), regardless of whether the enzyme was treated with azide before or after the addition of cyanide. Furthermore, the dissociation constant for azide from CcO was reported to be 65-250 M (2,4,54), and the dissociation constant for cyanide under similar conditions was reported to be 1-4 M (2, 4, 27, 56). Hence, the lower affinity of CcO for azide might contribute, in part, to its humble ability to oxidize azide.
Azidyl radical was also detected in the system of azideinhibited submitochondrial particles (42). It was proposed that the azide inhibition of CcO in the succinate-driven respiration prompted the accumulation of superoxide, leading to H 2 O 2 formation that, together with an endogenous peroxidase, oxidizes azide to the azidyl radical. Our results indicate that CcO might be the proposed endogenous peroxidase in submitochondrial particles. In conclusion, the highly reactive cyanyl and azidyl radicals should be considered in the interpretation of data when either inhibitor is used with CcO, mitochondria, or cells.