Reaction Mechanism of Superoxide Generation during Ubiquinol Oxidation by the Cytochrome bc1 Complex*

In addition to its main functions of electron transfer and proton translocation, the cytochrome bc1 complex (bc1) also catalyzes superoxide anion (O2̇̄) generation upon oxidation of ubiquinol in the presence of molecular oxygen. The reaction mechanism of superoxide generation by bc1 remains elusive. The maximum O2̇̄ generation activity is observed when the complex is inhibited by antimycin A or inactivated by heat treatment or proteinase K digestion. The fact that the cytochrome bc1 complex with less structural integrity has higher O2̇̄-generating activity encouraged us to speculate that O2̇̄ is generated inside the complex, perhaps in the hydrophobic environment of the QP pocket through bifurcated oxidation of ubiquinol by transferring its two electrons to a high potential electron acceptor, iron-sulfur cluster, and a low potential heme bL or molecular oxygen. If this speculation is correct, then one should see more O2̇̄ generation upon oxidation of ubiquinol by a high potential oxidant, such as cytochrome c or ferricyanide, in the presence of phospholipid vesicles or detergent micelles than in the hydrophilic conditions, and this is indeed the case. The protein subunits, at least those surrounding the QP pocket, may play a role either in preventing the release of O2̇̄ from its production site to aqueous environments or in preventing O2 from getting access to the hydrophobic QP pocket and might not directly participate in superoxide production.

It has long been recognized that during mitochondrial respiration, there is a continuous release of electrons from the electron transfer chain to react with molecular oxygen to form a superoxide anion (O 2 . ) (1)(2)(3) . is produced when the membrane potential of mitochondria is high (6,7). In the past, most information concerning mitochondrial O 2 . generation sites was obtained from studies using intact heart mitochondria with selected electron transfer inhibitors by measuring the H 2 O 2 concentration in the suspending medium (8,9).
Two segments of the respiratory chain have been demonstrated to be responsible for the generation of O 2 . from oxygen.
One is located at the NADH-Q 3 oxidoreductase (complex I), and the other is at the cytochrome bc 1 complex (ubiquinolcytochrome c oxidoreductase). Production of O 2 . by complex I is either via auto-oxidation of the flavine radical in NADH dehydrogenase (10) or via a bound ubisemiquinone radical (11) or the center N-2 (12) of the complex. It was recently suggested (13) that the reversed electron transport through complex I produced more O 2 . than the forward transport. Two redox components of the bc 1 complex, ubisemiquinone at the Q P site (8) and the reduced cytochrome b 566 (9,14), have been implicated as electron donors for molecular oxygen to generate O 2 . .
The production of O 2 . by the bc 1 complex is greatly enhanced when the complex is inhibited by antimycin (14 -16).
In continuing our study of the structural and functional relationship of the cytochrome bc 1 complex, it is important to understand the reaction mechanism of superoxide generation in this complex. The fact that antimycin inhibits the electron transfer activity of the cytochrome bc 1 complex and stimulates the O 2 . -generating activity, together with the observation that both activities have a similar activation energy, led investigators to believe that both activities share, at least, a common intermediate (17). According to the Q-cycle mechanism (18 -20), during the catalytic reaction of the cytochrome bc 1 complex, Q-H 2 undergoes bifurcated oxidation by transferring its two electrons, sequentially or simultaneously (concerted), to the iron-sulfur cluster (ISC) of the iron-sulfur protein (ISP) subunit and heme b L of the cytochrome b subunit. In the sequential mechanism (21)(22)(23), ubiquinol transfers its first electron to the ISC to become low potential ubisemiquinone that contains a free electron and reduces heme b L instantly. The lack of a functional ubisemiquinone at the Q P site (21,24,25) undermines this mechanism substantially, although some radicals have been reported under abnormal conditions (26 -28). Recently, more compelling evidence against the existence of the semiquinone radical at the Q P site has been reported (29). No such radical in a heme b L knock-out mutant complex is detected upon reduction by quinol. Because heme b L is the designated electron acceptor of the semiquinone radical at the Q P site in the sequential Q-cycle mechanism, one would expect to see the accumulation of this intermediate when the acceptor is not available, but this is not the case. In the concerted mechanism (29 -31), no semiubiquinone is formed, and two electrons of Q-H 2 are transferred simultaneously to ISC and heme b L . This bifurcated electron transfer reaction provides a basis for the high efficiency of the bc 1 complex and is done inside the cytochrome b subunit buried in the membrane bilayer. It is thus expected that any compromise in the structural integrity of cytochrome b should lead to a decrease in the electron transfer efficiency and to an increase in the production of superoxide. The observation that mutants lacking heme b L or heme b H , respectively, show little electron transfer activity but have high superoxide-generating activity (25)
Enzyme Preparations and Activity Assays-Chromatophores, intracytoplasmic membrane, and the His 6 -tagged cytochrome bc 1 complexes, wild type (34) and mutants (25,35), were prepared as reported previously. Bovine heart mitochondrial cytochrome bc 1 complex was prepared according to the method developed in our laboratory (36,37).
To assay the cytochrome bc 1 complex activity, purified complexes were diluted with 50 mM Tris-Cl, pH 8.0, containing 200 mM NaCl and 0.01% DM to a final concentration of cytochrome c 1 of 1 M. Appropriate amounts of the diluted samples were added to 1 ml of assay mixture containing 100 mM Na ϩ /K ϩ phosphate buffer, pH 7.4, 300 M EDTA, 100 M cytochrome c, and 25 M Q 0 C 10 BrH 2 . Because Q 0 C 10 BrH 2 is easily auto-oxidized at neutral or higher pH, the stock solution is kept in 95% ethanol containing 1 mM HCl and diluted in the buffer before use. Activities were determined by measuring the reduction of cytochrome c (the increase of absorbance at 550 nm) in a Shimadzu UV 2101 PC spectrophotometer at 23°C, using a millimolar extinction coefficient of 18.5 for the calculation. The non-enzymatic oxidation of Q 0 C 10 BrH 2 , determined under the same conditions in the absence of the enzyme, was subtracted from the assay. Preparation of Phospholipid Vesicles-Phospholipid vesicles were prepared by the cholate dialysis method (38). Asolectin was dissolved in chloroform and dried as a thin film against the tube by flushing with nitrogen gas while the tube was rotating. The phospholipid was then suspended in 50 mM potassium/ sodium phosphate buffer, pH 7.4, containing 1% sodium cholate. The mixture was subjected to sonification intermittently for 30 min until the solution become clear and then dialyzed against the same buffer overnight, with three changes of buffer.
Determination of Superoxide Production-Superoxide production was determined by measuring the chemiluminescence of MCLA-O 2 . adduct (39) in an Applied Photophysics stoppedflow reaction analyzer SX.18MV-R (Leatherhead, UK) by leaving the excitation light off and registering light emission (40,41). Reactions were carried out at 23°C by mixing 1:1 of solutions A and B. For the determination of O 2 . production by the native, heat-inactivated, or proteinase K-digested cytochrome bc 1 complexes, Solution A contains 100 mM Na ϩ /K ϩ phosphate buffer, pH 7.4, 1 mM EDTA, 1 mM NaN 3 , 0.1% bovine serum albumin, 0.01% DM, and 5.0 M cytochrome bc 1 . Solution B contains 125 M Q 0 C 10 BrH 2 and 4 M MCLA in the same buffer. Once the reaction started, the produced fluorescence, in voltage, was consecutively monitored for 2 s. One volt from the Applied Photophysics stopped-flow reaction analyzer SX.18MV-R equals the chemiluminescence (maximum peak height of light intensity) generated by 0.5 unit of xanthine oxidase using 100 M hypoxanthine as a substrate. Because MCLA is a neutral, relatively non-polar molecule, it is possible that the efficiency of O 2 . production detected by chemiluminescence may be affected by the presence of different detergent micelles, thus complicating the determination of the micelle effect of different detergents on superoxide production. To avoid this possible complication, superoxide produc-tions in the presence of different detergent, micelles were compared by measuring the reduction of acetylated cytochrome c (42) because different detergent micelles do not show significant effect on superoxide production by xanthine oxidase and hypoxanthine determined by the acetylated cytochrome c method. Reduction of acetylated cytochrome c was followed by the increase of absorption at 550 nm in the same stopped-flow reaction analyzer in the normal way. A millimolar extinction coefficient of 18.5 was used for the concentration calculation. Solution A contains 100 mM Na ϩ /K ϩ phosphate buffer, pH 8.0, 10 M acetylated cytochrome c, and various amounts of detergents. Solution B contains 250 M Q 0 C 10 BrH 2 in 0.5 mM Na ϩ /K ϩ phosphate buffer, pH 4, in the presence or absence of 300 units/ml superoxide dismutase.

RESULTS AND DISCUSSION
The Inverse Relationship between Superoxide-generating and Electron Transfer Activities in the Cytochrome bc 1 Complex- Table 1    The notion that an intact protein subunit structure is not required for O 2 .
-generating activity of the bc 1 complex is further supported by the observation that mutants H198N and H111N, which lack heme b L and heme b H , respectively, have very little electron transfer activity but show O 2 . -generating activity equal to that of the antimycin-treated wild-type complex (25) ( Table  1). The increased rates of O 2 . formation when in the heme b L knock-out complex is strong evidence that although the heme b L may react with oxygen when it is present, a O 2 . can also be formed by a route other than by reaction with the heme b L . The loss of either heme b L or heme b H would be expected to have a strong impact on the overall structural integrity of the bc 1 complex, leading to the distortion of the Q P pocket environment and presumably increasing the accessibility of O 2 to the site. It is as expected that the incubation of these two heme b-lacking mutant complexes at 37°C to denature the protein subunits does not further increase the O 2 . -generating activity because the structural integrity of these two complexes has already been deteriorated by mutation. The Superoxide Anion-generating Activity in Proteinase K-digested Complex-To further confirm that superoxide-generating activity is independent of the presence of an intact protein structure, the bc 1 complex was subjected to proteinase K digestion and electron transfer, and superoxide-generating activities were measured during the course of digestion. As shown in Fig. 2, the electron transfer activity diminishes, whereas the O 2 .generating activity increases as the digestion time increases. Maximum superoxide-generating activity is observed when electron transfer activity is completely abolished. SDS-PAGE analysis of the proteinase K-digested complex reveals no intact subunits of cytochromes b, c 1 , or ISP (Fig. 3). The largest peptide band presence in the digested complex has an apparent molecular mass of less than 7 kDa. These results further support our suggestion that the intact protein components of the complex or the intact complex play no direct role in O 2 . generation. A Western blotting experiment using anti-ISP indicated that at least a part of this peptide band is derived from ISP. In other words, the ISC detected by EPR is housed in this peptide. If this notion is correct, then what remaining elements in the proteinase K-digested complex contributed to superoxide production from Q-H 2 ? A hydrophobic environment and a high potential electron acceptor ISC in the digested system could contribute to the superoxide formation in the presence of molecular oxygen. The presence of intact ISC in the proteinase K-digested complex was confirmed by the presence of EPR signals of ISC in the digested complex (data not shown). Thus, it is possible that ISC serves as a high potential electron acceptor and oxygen serves as a low potential electron acceptor during bifurcated oxidation of Q-H 2 in a hydrophobic environment of the Q P pocket to produce superoxide. Because there are no free metal ions in the purified cytochrome bc 1 complex and the iron in the heme and iron-sulfur cluster is not released during heat treatment or proteinase digestion, as indicated by absorption and EPR spectra of the tested samples, the possibility that the observed O 2 . production is due to free iron can be eliminated.

Generation of Superoxide Anion upon Oxidation of Ubiquinol by Cytochrome c or Ferricyanide in the Presence of Phospho-
lipid Vesicles-Although the addition of ferricytochrome c to the proteinase K-digested complex can increase its superoxide production, oxidation of ubiquinol by ferricytochrome c in the aqueous solution at neutral pH is a very slow reaction with a reduction rate constant K 1 of 0.24/s, and little O 2 . formation is detected, suggesting that a hydrophobic environment provided by the digested complex is required for superoxide production. Similar results were obtained when potassium ferricyanide was used to replace ferricytochrome c.
To confirm that a hydrophobic environment is needed for O 2 . production, varying amounts of phospholipid vesicles were added to the mixture containing constant amounts of ubiquinol and cytochrome c, and the superoxide anion production was measured by the reduction of acetylated cytochrome c method.
In the presence of phospholipid vesicles, the rate of the formation of O 2 . is proportional to the amount of vesicles added up to 0.3% (Fig. 4)   The detergent micelle solution (or phospholipid vesicles), which facilitates superoxide anion production, does not prevent dismutation of the superoxide anion generated by hypoxanthine/xanthine oxidase. Therefore, the superoxide generation-facilitating activity observed is real and not due to the decrease of dismutation. Fig. 5 shows the effect of detergent concentration on the superoxide production during quinol oxidation by cytochrome c. As expected, little O 2 . generation is observed when the concentration of detergent used is below its critical micelle concentration because no hydrophobic environment is available. When the concentration of detergent used is higher than its critical micelle concentration, the O 2 .
production rate increases as the detergent micelle concentration in the system increases. Interestingly, SC or DOC is much more effective in promoting superoxide generation than OG, DM, or LDAO. during the oxidation of quinol. First, the negative charges on the micelle surface may facilitate deprotonation of the 1-hydroxyl group of quinol. Second, the negatively charged deter-gent micelle surface may attract the positively charged ferricytochrome c better than the neutral or positively charged micelles. Thus, the high potential electron acceptor, ferricytochrome c, has a better accessibility to quinol, which is located near the surface of the detergent micelles. When potassium ferricyanide is used as the high potential electron acceptor, the effect on O 2 . generation by the difference in the charge of the micelle surface is less apparent.  . is produced during Q-H 2 oxidation by cytochrome c or ferricyanide in the presence of phospholipid vesicles or a detergent micelle and that the O 2 . is produced by a heat-inactivated or proteinase K-digested complex, a reaction mechanism for O 2 . production in the bc 1 complex is proposed. In this proposed mechanism, four elements are directly involved in O 2 . generation. These are: a hydrophobic environment, a high potential electron acceptor, a low potential electron acceptor, and an electron donor. In the intact cytochrome bc 1 complex, the protein subunits of the complex directly involved in the superoxide production are ISP, which houses a high potential electron acceptor ISC, and cytochrome b, which provides a hydrophobic environment, a Q P pocket, and a low potential electron acceptor heme b L . Based on the concerted Q-cycle mechanism (29 -31), quinol undergoes bifurcated oxidation in the Q P pocket by simultaneously transferring its two electrons to ISC and heme b L . The molecular oxygen can also receive a hydrogen from quinol to produce a protonated superoxide (O 2 H), which generates O 2 .

Superoxide Anion Generation Is High Potential Oxidantand Ubiquinol Concentration-dependent-Generation
upon deprotonation. It is also likely that the reduced heme b L can transfer an electron to the molecular oxygen to form O 2 . , particularly when the oxidant (heme b H ) of reduced heme b L is limited or unavailable, such as in the presence of antimycin A (Fig. 8A) . generation, they form a barrier for the Q P pocket to limit accessibility of molecular oxygen to the pocket. Destruction of protein structural integrity by heme b L or b H deletion, heat inactivation, or proteinase K digestion leads to a loosening of the structural integrity of the Q P pocket to facilitate the molecular oxygen to get access to the Q P pocket and to ease the release of produced O 2 . to the aqueous medium.
In addition, it is possible that in the cytochrome bc 1 complex, an oxygen molecule is located between the 4-HO Ϫ group of Q-H 2 and heme b L to mediate the electron transfer between them in the hydrophobic environment of the Q P pocket. The poor hydrogen-bonding nature of the oxygen molecule seems to work against this speculation. However, if this were the case, in a hydrophobic environment, one would expect to see a higher presteady state reduction rate of cytochrome b by ubiquinol in the presence of oxygen than in the absence of it. Our preliminary results seem to support this speculation. Further investigation on the role of oxygen in the reduction of cytochrome b by ubiquinol is currently in progress in our laboratory.
In the protein-free system, our results appear to suggest that the benzoquinol ring of Q-H 2 is located at or near the surface of the lipid vesicles or detergent micelles with its 1-hydroxy group extended into the water phase and the 4-hydroxy group together with its alkyl side chain located inside the bilayer or micelle (Fig. 8, B and C). When the ISC is used as an electron acceptor, we speculate that a hydrogen is transferred from the 1-hydroxyl group to the N-3 of the imidazole ring of histidine residue, which is a ligand of the ISC. At the same time, a hydrogen is transferred concurrently from the 4-hydroxy group to a molecule of oxygen, which is dissolved inside the lipid bilayer of  vesicles or in the hydrophobic interior of detergent micelles, to generate a protonated superoxide (HO 2 ), which then diffuses to the water phase to become a superoxide anion upon deprotonation.
The high potential electron acceptor ISC can be substituted with cytochrome c or ferricyanide. In this case, we surmise that the 1-hydroxy group of ubiquinone is deprotonated and releases a proton to the water phase before the electron is transferred to ferricyanide or cytochrome c. At the same time, the 4-hydroxy group transfers its hydrogen atom to a molecular oxygen.
In the phospholipid vesicles or detergent micelle systems, we suggest that the electron transfer between cytochrome c and Q-H 2 takes place at the surface of the lipid bilayer and that the transfer between Q-H 2 and O 2 occurs concurrently inside the bilayer. This suggestion is consistent with the fact that the solubility of oxygen is much higher in the hydrophobic environment than that in the aqueous medium and that the rate of reduction of cytochrome c by Q-H 2 is much lower under anaerobic conditions than when in the presence of oxygen.