Loss of a Conserved Tyrosine Residue of Cytochrome b Induces Reactive Oxygen Species Production by Cytochrome bc1

Production of reactive oxygen species (ROS) induces oxidative damages, decreases cellular energy conversion efficiencies, and induces metabolic diseases in humans. During respiration, cytochrome bc1 efficiently oxidizes hydroquinone to quinone, but how it performs this reaction without any leak of electrons to O2 to yield ROS is not understood. Using the bacterial enzyme, here we show that a conserved Tyr residue of the cytochrome b subunit of cytochrome bc1 is critical for this process. Substitution of this residue with other amino acids decreases cytochrome bc1 activity and enhances ROS production. Moreover, the Tyr to Cys mutation cross-links together the cytochrome b and iron-sulfur subunits and renders the bacterial enzyme sensitive to O2 by oxidative disruption of its catalytic [2Fe-2S] cluster. Hence, this Tyr residue is essential in controlling unproductive encounters between O2 and catalytic intermediates at the quinol oxidation site of cytochrome bc1 to prevent ROS generation. Remarkably, the same Tyr to Cys mutation is encountered in humans with mitochondrial disorders and in Plasmodium species that are resistant to the anti-malarial drug atovaquone. These findings illustrate the harmful consequences of this mutation in human diseases.

In most organisms, the ubiquinol:cytochrome c oxidoreductase (cytochrome bc 1 or complex III) is a central enzyme for ATP production through oxidative phosphorylation that relies on the proton motive force (⌬H ϩ ) generated by the respiratory chain (1). Production of reactive oxygen species (ROS) leads oxidative damages of cellular components and in eukaryotes induces apoptosis (2,3). Most cellular ROS are thought to emanate from the respiratory NADH:dehydrogenase (i.e. complex I) and cytochrome bc 1 (see Fig. 1a for Rhodobacter capsulatus structure) under compromising physiological conditions (4,5). Cells use antioxidant enzymes (e.g. superoxide dismutase or glutathione peroxidase) to prevent oxidative damages, but upon extensive ROS generation, harmful damage occurs. Indeed, cellular redox homeostasis, regulated by the rate of electron flow through the respiratory chain and O 2 availability, is tightly coupled with the global metabolism (4,5). For example, recent studies show that cytochrome bc 1 is involved in stabilization and activation of hypoxia-induced factors like HIF-1␣ by mitochondria-generated ROS under hypoxic conditions (6,7).
Mitochondrial DNA mutations are known causes of clinical syndromes (e.g. LHON, Pearson syndrome, and exercise intolerance) or provide predisposition for inherited and common diseases (e.g. aging, cardiomyopathy, cancer, diabetes, and neurodegenerative diseases) (8,9). Mutations in mitochondrial or nuclear DNAs associated with mitochondrial fission and fusion (10), ascribed to ROS generation, lead to progressive dysfunction of mitochondria and loss of energy efficiency (11). For example, the Tyr to Cys mutation at position 278 (position 302 in R. capsulatus; see Fig. 1b) of cytochrome b of human mitochondrial cytochrome bc 1 (complex III) causes severe exercise intolerance and "multi-system disorder" (12). Remarkably, the same mutation in this conserved residue (Fig. 1c) is also encountered in cytochrome bc 1 of malarial parasites (e.g. Plasmodium falciparum or Plasmodium yoelii) that are resistant to the antimalarial drug atovaquone (13,14). These clinical findings highlight that specific amino acids of cytochrome bc 1 are critical for human mitochondrial diseases (12,15), for resistance of parasites to therapeutic agents (16), and for catalytic activity of the enzyme in bacteria (17) and yeast (18,19), yet our understanding of the mechanisms underlying these disparate defects remains elusive.
Cytochrome bc 1 oxidizes reduced quinone (QH 2 ) 4 via an unstable semiquinone intermediate at a catalytically active (i.e. Q o ) site. Its function depends on an elaborate mechanism with two "one-electron" transfer steps, one to a high potential chain (comprising the iron-sulfur protein and cytochrome c 1 ) and another to a low potential chain (formed of the hemes b L and b H of cytochrome b) to generate ⌬H ϩ across the membrane with minimal energy expenditure (20 -22). Several models (23)(24)(25) have been proposed to rationalize the efficient and safe occurrence of these electron transfer steps in the presence of O 2 .
However, how the reactive intermediates of the Q o site are shielded from O 2 to avoid oxidative damages remains unknown (26). We hypothesized that specific amino acids at the Q o site might provide protective mechanism(s) against oxidative damage while ensuring catalytic efficiency. Using the bacterial enzyme, here we show that mutating a specific Tyr residue of cytochrome b subunit of cytochrome bc 1 greatly enhances ROS production. Remarkably, the Tyr to Cys substitution crosslinks together the cytochrome b and the iron-sulfur subunits of cytochrome bc 1 and renders the bacterial enzyme sensitive to O 2 by oxidative disintegration of its catalytic [2Fe-2S] cluster. These findings demonstrate the occurrence of "protective residue(s)" in cytochrome bc 1 , like this Tyr of cytochrome b that controls ROS generation. They also illustrate how specific mutations on such critical residues cause broad cellular defects extending from human mitochondrial diseases (12) to parasite resistance to therapeutics (16).

EXPERIMENTAL PROCEDURES
Bacterial Strains, Growth Conditions, and Genetic Techniques-Escherichia coli and R. capsulatus strains were grown as described earlier (27). Mutations at Tyr 302 of R. capsulatus cytochrome b were introduced by using the QuikChange sitedirected mutagenesis kit (Stratagene) with the plasmid pPET1 as a template and appropriate mutagenic primers. The 1.66-kb XmaI-StuI DNA fragment of pPET1 bearing the mutation was exchanged with its counterpart in pMTS1, yielding the plasmids pMTS1:Y302X (where X indicates Ala, Cys, Phe, Ser, or Val), which were crossed into the strain MT-RBC1 lacking cytochrome bc 1 , as described earlier (27)(28)(29)(30)(31). The presence of only the desired mutation was confirmed by DNA sequence analysis of plasmids.
Preparation of Aerobic and Anaerobic Membrane Samples-Aerobic chromatophore membranes were prepared in 50 mM MOPS (pH 7.5), 100 mM KCl, 100 mM EDTA, 1 mM caproic acid, and protein concentrations determined as described earlier (28,29). For anaerobic membrane samples, an anaerobic chamber (MBraun-USA Inc., Stratham, NH) at 20°C was used with degassed, argon-flushed buffers containing 5 mM sodium dithionite. The membranes were prepared under argon flushing and stored in gas-tight serum vials with septa (National Scientific Co.). For sulfhydryl alkylation experiments, 100 mM idoacetamide (IAM) or N-ethylmaleimide was added prior to cell disruption.
Biochemical and Biophysical Techniques-Protein samples were solubilized in 62.5 mM Tris-HCl (pH 6.8), 4% SDS, 100 mM dithiothreitol, 25% glycerol, and 0.01% bromophenol blue incubated at 60°C for 10 min, separated by SDS-polyacrylamide gel electrophoresis, and stained with Coomassie Brilliant R-250, and immunoblots were performed using monoclonal and polyclonal antibodies specific for R. capsulatus cytochrome bc 1 subunits as described in Ref. 29. Cytochrome bc 1 activity was measured using decylbenzo-hydroquinone (DBH 2 ):cytochrome c reductase assay as described in Refs. 27 and 29. For all of the strains, at least three independent cultures were used, three independent measurements per culture were performed, and the mean values Ϯ standard deviations are indicated. Optical difference spectra for b-and c-type cytochromes were recorded, and R. capsulatus cytochrome bc 1 was purified and stored at Ϫ80°C as described (28,29). Light-activated kinetic spectroscopy was performed as described in Ref. 30 using a dual wavelength spectrophotometer. In all cases, the membranes contained 0.3 M of reaction center as determined by a train of 10 light flashes separated by 50 ms at an E h of 380 mV, and an extinction coefficient ⑀ 605-540 of 29 mM Ϫ1 cm Ϫ1 . Transient cytochromes c (at 550 -540 nm) and b (at 560 -570 nm) reduction kinetics exhibited by membranes resuspended in 50 mM MOPS buffer (pH 7.0), 100 mM KCl, and 100 mM EDTA at an ambient redox potential of 100 mV were monitored. Antimycin A, myxothiazol, and stigmatellin (Stg), inhibitors of cytochrome bc 1 , were used at 5, 5, and 1 M, respectively. EPR spectroscopy was performed using a Bruker ESP-300E (Bruker Biosciences), equipped with an Oxford Instruments ESR-9 helium cryostat (Oxford Instrumentation Inc.) as described in Ref. 31. As needed, the membranes (25-30 mg of protein/ml) were reduced by incubation on ice for 10 min with 20 mM ascorbate in the presence or absence of 100 M Stg or 100 M famoxadone. Spectra recording conditions were for the [2Fe-2S] cluster: modulation frequency, 100 kHz; microwave power, 2 milliwatts; sample temperature, 20 K; microwave frequency, 9.443 GHz; modulation amplitude, 12 G, and for the b hemes: modulation frequency, 100 kHz; sample temperature, 10 K; microwave frequency, 9.59 GHz; modulation amplitude, 10 G.
Superoxide Measurements-The rate of superoxide production was determined using manganese-SOD-sensitive and Stgsensitive cytochrome c reduction assays (32). First, in the presence and absence of manganese-SOD (150 units/ml SOD from Sigma), (ϪSOD/ϪStg) and (ϩSOD/ϪStg) cytochrome c reduction rates were determined by using DBH 2 :cytochrome c reductase assays (27,29). Then (ϩSOD/ϩStg) and (ϪSOD/ϩStg) rates of cytochrome c reduction were determined using membranes inhibited by 10 M Stg to obtain cytochrome bc 1 -specific rates. The amounts of superoxide produced independently of cytochrome bc 1 were excluded by subtracting from (ϪSOD/ ϩStg) the (ϩSOD/ϩStg) values. The absence of enhancement of ROS production by the addition of antimycin A in the presence of Stg provided an extra control for cytochrome bc 1 -independent ROS generation. The ROS that is specifically produced by the cytochrome bc 1 was then determined by subtracting from [(ϪSOD-Stg) Ϫ (ϩSOD-Stg)] the [(ϪSODϩStg) Ϫ (ϩSODϩStg)] values. In this way, of the total cytochrome c reduction rates observed, only the portion that is sensitive to Stg and to manganese-SOD was attributed to superoxide originating specifically from cytochrome bc 1 Q o site. Antimycin A-induced ROS productions were measured in similar ways except that 20 M antimycin A was added.
The cumulative amounts of superoxide production were measured via H 2 O 2 formation using the Amplex Red-horseradish peroxidase method (Molecular Probes, Inc.). As above, DBH 2 :cytochrome c reduction assay mixtures with and without Stg were supplemented with 50 M Amplex Red, 150 units/ml manganese-SOD and 0.1 units/ml of horseradish peroxidase and incubated at room temperature for 1 min. The amount of the fluorescent resorufin formed from Amplex Red at the end of the incubation period was measured using a PerkinElmer Life Sciences LS-5B luminescence spectrophotometer and 530-and 590-nm excitation and emission wavelengths, respectively. Fluorescence production was linear during the incubation period, and the amount of H 2 O 2 production was calibrated using the standards provided in the Amplex Red kit. Of the total amounts of H 2 O 2 produced, only the portions that were sensitive to Stg were considered as superoxide generated by cytochrome bc 1 . Antimycin A-induced superoxide productions were measured in the same way except that 20 M antimycin A was added to the assay mixture (29).
Mass Spectrometry Analyses-SDS-PAGE samples were excised and subjected to in-gel digestions, whereas purified cytochrome bc 1 samples were digested in-solution using trypsin and, as needed, GluC, as described (33)(34)(35). Peptides were analyzed with a Thermo LCQ Deca XPϩ MS/MS spectrometer coupled to a LC Packings Ultimate Nano HPLC system controlled by Thermo Xcalibur 2.0 software. Thermo Bioworks 3.3 software was used to perform SEQUEST searches against the R. capsulatus protein database as described earlier (33)(34)(35). The results were filtered using standard values for X corr (1.5, 2.0, and 2.5 for m/z of ϩ1, ϩ2, and ϩ3, respectively) and ⌬C N Ն 0.1, and relevant spectra were inspected manually for validity of the assignments.
Chemicals-All of the chemicals were as described earlier (29).  (Fig. 1, b and c). This residue is part of the docking niche of the iron-sulfur protein subunit of the enzyme and is H-bonded to the backbone of a Cys residue of this subunit to position correctly its [2Fe-2S] cofactor at the Q o site. The role of this Tyr residue on QH 2 oxidation has been described earlier using Rhodobacter sphaeroides cytochrome bc 1 mutants where Tyr 302 was substituted with Phe, Leu, Gln, and Gly) (17). We had also examined its role during our studies related to malarial atovaquone resistance (16). A surprising finding was that in the closely related R. capsulatus species, the Y302C mutant exhibited robust anoxygenic (ϪO 2 ) photosynthetic growth that absolutely requires an active cytochrome bc 1 , whereas routinely prepared membranes from aerobically (ϩO 2 ) grown cells had poor cytochrome bc 1 activity (16) (Fig.  2a). Because the Y302C mutant has not been examined earlier (17), to further explore this observation, we studied the effects of R. capsulatus Y302X (where X indicates Ala, Cys, Phe, Ser, or Val) substitutions on cytochrome bc 1 activity. When cells were grown and membranes were prepared in the presence of O 2 , all of the mutants contained normal amounts of cytochrome b, cytochrome c 1 , and iron-sulfur protein subunits as indicated by optical difference spectra and SDS-PAGE analyses (Fig. 2, b and c). They also responded normally to the Q o site inhibitors Stg and famoxadone (Table 1). Thus, like the homologous R. sphaeroides (17) or yeast (18,19) mutants, all R. capsulatus Y302X mutants also assembled cytochrome bc 1 , but the Y302A, Y302S, and Y302C substitutions had comparatively lower activities than the Y302V and Y302F substitutions using membranes prepared in air from aerobically grown cells (Fig. 2a). Remarkably, however, these activities changed when cells were grown by anoxygenic photosynthesis, and the membranes were prepared in the absence of O 2 . Wild type cells showed little decrease under ϪO 2 conditions with respect to cytochrome bc 1 activity (ϳ70% of ϩO 2 activity, possibly because of total membrane protein content changes and because of anaerobic preparations). A similar pattern was also observed with Y302F and Y302V substitutions. In contrast, the Y302A, Y302S, and Y302C mutants displayed an opposite pattern with increased cytochrome bc 1 activity under ϪO 2 as compared with ϩO 2 conditions (Fig. 2a). In particular, the Y302C substitution showed a marked enhancement (ϳ2-3-fold increase from ϩO 2 to ϪO 2 conditions), but as expected, it did not reach wild type activity levels because of its partially defective Q o site (like all Y302X mutants). When corrected for the corresponding wild type activities observed under ϩO 2 and ϪO 2 conditions (15 and 56% of wild type, respectively), the Y302C enhancement appears even more (ϳ3-4-fold) pronounced.

Mutation of Cytochrome b Tyr 302 to Cys Renders
Performance of light-activated, single-turnover cytochrome c and b reduction kinetics (30) established that the O 2 -sensitive catalytic step in the Y302C mutant was the bifurcated electron transfer event during QH 2 oxidation. Although rapid reduction kinetics of cytochromes c and b were seen with ϩO 2 membranes of aerobically grown wild type and Y302V mutant, they were undetectable with similar ϩO 2 Y302C samples (Fig. 3a, upper and lower traces, respec- tively; see also the schematic drawings underneath of them). However, rapid kinetics were readily seen with Y302C membranes prepared in the absence of O 2 using anaerobically grown cells, demonstrating that Q o site catalysis became O 2 -sensitive in Y302C mutant.

Y302C Substitution Induces Oxidative Disruption of the [2Fe-2S] Cluster of Cytochrome bc 1 -Because the cytochromes b and
c contents of all Y302X mutants were similar (Table 1), EPR spectroscopy was used to examine their iron-sulfur protein [2Fe-2S] cluster (31). The mutants showed altered EPR spectra and g x and g y values different from those (1.80 and 1.89, respectively) of the wild type, revealing that substitution of Tyr 302 perturbed the Q o site and the environment of the [2Fe-2S] cluster (Fig. 3b, left panel). When samples prepared under ϩO 2 and ϪO 2 conditions using appropriately grown cells were compared, drastic changes were seen with the Y302C mutant in respect to the iron-sulfur protein [2Fe-2S] cluster amounts, as reflected by the EPR g y signals (Fig. 3b , middle and right panels). For the wild type, EPR spectra of the [2Fe-2S] cluster of ϪO 2 or ϩO 2 grown cells were similar to one another (and the observed amplitudes of the g x and g y signals were in agreement with the respective cytochrome bc 1 activities shown in Fig. 2a). On the other hand, the amplitudes of the iron-sulfur protein [2Fe-2S] cluster g y and g x signals of Y302C samples were very low under ϩO 2 , but significantly higher under ϪO 2 conditions (Fig. 3b, middle and right bottom panels). Based on the g y signal amplitudes, ϩO 2 and ϪO 2 grown Y302C cells contained ϳ14 and 53%, respectively, of the iron-sulfur protein [2Fe-2S] cluster as compared with the wild type enzyme (31). These percentage values roughly match the corresponding relative enzyme activities for Y302C (Fig. 2a, at 15 and 56% of wild type activities) obtained for ϩO 2 and ϪO 2 conditions, suggesting that a loss of cytochrome bc 1 activity correlates with a loss of the [2Fe-2S] cluster in Y302C mutant. This finding was unexpected, because the [2Fe-2S] cluster of wild type cytochrome bc 1 has a high redox midpoint potential, and its oxidative disruption does not occur readily upon exposure to air (31,39).
Loss of Tyr 302 Enhances ROS Generation During Q o Site Catalysis-Native cytochrome bc 1 produces insignificant amounts of ROS except under conditions that interfere with bifurcated electron transfer at the Q o site (e.g. in the presence of inhibitors like antimycin A or high membrane potential, ⌬ m ) (22,26,32,40). The unusual oxygen sensitivity of the [2Fe-2S] cluster of the iron-sulfur protein in Y302C seen above (i.e. the decreased g y signal in Fig. 3b) led us to examine ROS production by the Y302X mutants. Manganese-SOD-sensitive cytochrome c reduction assays were used to monitor Stg-sensitive Variations in the wild type cytochrome bc 1 activities under ϩO 2 and ϪO 2 conditions derive from the changing total membrane protein contents and different sample preparation methods and are not to be compared directly. b, dithionite-reduced (black) or ascorbate-reduced (gray) minus ferricyanide-oxidized optical difference spectra of aerobically prepared membrane fractions (0.4 mg of total proteins) from ϩO 2 grown WT and Y302X mutant strains. The 550-and 560-nm peaks correspond to the cytochromes c and b, respectively. c, SDS-PAGE and Western blot analyses of the same membrane fractions (50 g of total proteins) from WT and Y302X mutant cells. The cytochrome bc 1 -deficient strain MT-RBC1 was used as a negative control, and cytochrome b and iron-sulfur protein subunits were detected by Western blot analysis using monoclonal (a-cyt b) and polyclonal (a-Fe/S) antibodies, respectively. a Ps ϩ refers to anaerobic photosynthetic (ϪO 2 ) growth ability. b Cytochromes c and b contents of membranes were determined using optical redox difference spectra. c Cytochrome bc 1 activity refers to the 2,3-dimethoxy-5-methyl-6-decyl-1,4-benzohydroquinone (DBH 2 ): cytochrome c reductase activity expressed as a percentage of the wild-type activity, which was approximately 3.5 mol of cytochrome c reduced min Ϫ1 mg of protein Ϫ1 of membranes prepared in the presence of air using aerobically grown cells. Each assay was conducted at least in triplicate. d The data are the means Ϯ standard deviations.
(i.e. cytochrome bc 1 -specific) ROS generation. Under these conditions, native cytochrome bc 1 produced detectable amounts of ROS exclusively upon exposure to antimycin A (Fig.  4a, observed as a decrease of cytochrome bc 1 activity by ϳ12% upon manganese-SOD addition). In contrast, all of the Y302X mutants produced large amounts of ROS (6 -13% decrease of cytochrome bc 1 activity upon manganese-SOD addition) even in the absence of antimycin A (Fig. 4a). In all cases, this produc-tion was further enhanced (to ϳ10 -23%) by the addition of antimycin A.
In addition, cumulative ROS production by Y302X mutants was also tested using Amplex Red-based assays. The data demonstrated that Y302X mutants produced ROS in the absence of antimycin A at amounts similar to those seen with a native enzyme inhibited with antimycin A (Fig. 4b). Thus, loss of Tyr 302 led to electron leakage from the Q o site to O 2 in a way  (Fig.  3b). We examined ϪO 2 -grown Y302C cells after treatment with Stg to inactivate cytochrome bc 1 Q o site prior to O 2 exposure to elucidate how the [2Fe-2S] cluster was inactivated (Fig.  5a). Monitoring the Y302C [2Fe-2S] cluster as a function of exposure time to O 2 , using the EPR g y signal, revealed that the [2Fe-2S] cluster of uninhibited Y302C decreased by 50% of its initial value in 50 h of exposure to O 2 , whereas Stg-treated samples showed no effect. Thus, inhibiting Q o site catalysis and burying the iron-sulfur protein head domain into cytochrome b surface prevented oxidative disruption of cytochrome bc 1 [2Fe-2S] cluster in Y302C mutant.
Because no cysteine residue is present in R. capsulatus cytochrome b (41), we suspected that the newly gained thiol group in Y302C mutant might be responsible for the oxidative disruption of the [2Fe-2S] cluster. We treated ϪO 2 -grown Y302C cells with IAM to alkylate the thiolate anion at position 302 and monitored the [2Fe-2S] cluster disintegration by EPR g y signal as a function of exposure time to O 2 . Upon exposure to air, untreated Y302C lost the [2Fe-2S] cluster g y signal and the cytochrome bc 1 activity during the initial 50 h, whereas IAMtreated membranes retained ϳ80% of both of these features (Fig. 5, a and b). EPR spectra of the same samples exhibited no significant variations with respect to hemes b L and b H amounts, which were used as internal controls via their EPR g z transitions (Fig. 5c). Upon extension of the exposure time to O 2 up to ϳ250 h, the g y signal amplitude of untreated Y302C decreased to 30%, but IAM-treated membranes retained ϳ70% of their initial levels (data not shown). We noted that the rate of cytochrome bc 1 activity loss induced by IAM was more rapid than that of the EPR g y transition, possibly because of inhibitory hindrances inflicted by alkylation of the Y302C residue at the Q o site. Similar to IAM, treatment with the bulkier alkylating reagent N-ethylmaleimide also protected the [2Fe-2S] cluster from oxidative damage, but it induced a more rapid loss of cytochrome bc 1 activity (data not shown). We concluded that introduction of a thiol group at position 302 of cytochrome b rendered bacterial cytochrome bc 1 O 2 -sensitive via oxidative disruption of its iron-sulfur protein [2Fe-2S] cluster upon exposure to air.
An Intersubunit Disulfide Bond Inactivates Cytochrome bc 1 and Destabilizes Its [2Fe-2S] Cluster-To further investigate the basis of O 2 sensitivity described above, native, O 2 -sensitive Y302C, and O 2 -tolerant Y302V cytochrome bc 1 enzymes were purified and subjected to comparative tandem mass spectrometry analyses. Appropriate protease digestions allowed detection of the peptide 296 WYFLPFXAILR 306 that encompasses position 302 of cytochrome b (Fig. 6) As expected, this position corresponded to Tyr and Val in the native and Y302V enzymes, respectively, but no such peptide could be detected in Y302C samples, despite searches for ROS-induced cysteine modifications (e.g. sulfenic, sulfinic, or sulfonic acids) (42). However, a shorter peptide ( 296 WYFLPF 301 ) corresponding to a cleavage product immediately before position 302 was observed in all cases (Fig. 6). This suggested that this position might be modified in Y302C mutant enzyme. . Rate of superoxide production, and cumulative superoxide production, by native (WT) and Y302X variants of cytochrome bc 1 in the absence and presence of antimycin A. a, rate of superoxide production by native (WT) and Y302X ϭ Ala, Cys, Phe, Ser, or Val variants of cytochrome bc 1 in the absence (filled bars) and presence (open bars) of antimycin A was monitoring via the manganese-SOD-induced decrease of DBH 2 :cytochrome c reduction rates. In all cases, cytochrome c reduction rates were corrected for Stg-insensitive (i.e. cytochrome bc 1 -independent) rates, and the difference observed between the corrected cytochrome c reduction rates seen in the absence and presence of manganese-SOD was attributed to ROS generated by cytochrome bc 1 Q o site, as described under "Experimental Procedures" and Fig. 1 legend. b, cumulative superoxide production was determined using membranes prepared in the presence of air from aerobically grown cells, incubated for 1 min under the DBH 2 :cytochrome c reduction assay conditions, and accumulated amounts of ROS produced were measured by the Amplex red assay. In this assay, H 2 O 2 formation in the presence of excess manganese-SOD is coupled to horseradish peroxidase-mediated Amplex Red conversion to resorufin. Of the total amounts of H 2 O 2 produced, only the portions that were sensitive to Stg are considered as superoxide generated by cytochrome bc 1 . Next, the possibility of an intersubunit disulfide bond formation between the iron-sulfur protein, which is located structurally close to position 302 of cytochrome b (37,38), was considered. Such an intersubunit cross-link was suspected to be incomplete based on the EPR data related to the destruction of the [2Fe-2S] cluster (Fig. 5a). Purified native and Y302C cytochrome bc 1 samples were extracted with Triton X-114 for cytochrome b enrichment and subjected to SDS-PAGE, mass spectrometry, and Western blot analyses. Only with the Y302C samples under nonreducing conditions, a 70-kDa protein band that disappeared upon treatment with DTT was found (Fig. 7,  left panel). Using both mass spectrometry (Fig. 7, right panel) and specific antibodies (not shown), the DTT-sensitive 70-kDa band was shown to contain both the 22-kDa iron-sulfur protein and 48-kDa cytochrome b subunits.
Furthermore, extensive MS/MS analyses found additional modified cysteine containing peptide fragments of the ironsulfur protein (supplemental Fig. S1). The 125 WLVMLGVC*T-HLGCVPMGDK (or 125 WLVMLGVCTHLGC*VPMGDK) fragment of the iron-sulfur protein contained a chemically modified Cys by 299 Da (C*), which is compatible with a disulfide bond-linked 302 CAI peptide fragment (expected mass of 303 Da) of cytochrome b. Similarly, the 145 SGDFGGWFC*P-CHGS (or 145 SGDFGGWFCPC*HGS) fragment of the ironsulfur protein contained a chemically modified Cys by 817 Da (C*), which is compatible with a disulfide bond-linked 300 PFCAILR peptide fragment (expected mass of 816 Da) of cytochrome b, was found (supplemental Fig. S1). Although these fragments had reliable X corr and ⌬C N scores, manual inspection of these spectra indicated that these assignments were not unambiguous, only suggesting the occurrence of a disulfide bond involving Cys 302 of cytochrome b.
In the native iron-sulfur protein subunit, the Cys 138 and Cys 155 residues form an intramolecular disulfide bond that increases the redox midpoint potential of the [2Fe-2S] cluster (31,39). Structures of cytochrome bc 1 indicate that this disulfide bond is near the position 302 of cytochrome b when the iron-sulfur protein is docked at the Q o site (37,38). In R. capsulatus structure (IZRT), the distance between ␣-Cys 153 or ␣-Cys-155 and O-Tyr 302 is ϳ5 or 11 Å, respectively. Similarly, in the case of yeast structure (Protein Data Bank code 3CX5), the corresponding distances are ϳ4 and 10 Å (supplemental Fig.  S2). Cys 133 and Cys 153 of the iron-sulfur protein are the ligands of the [2Fe-2S] cluster and of the intramolecular disulfide bond forming cysteines (i.e. Cys 138 and Cys 155 ), Cys 155 is closer to Tyr 302 of cytochrome b. In the light of these data, we tentatively propose that in the Y302C enzyme, upon exposure to O 2 , possibly activated by ROS production via the Q o site, cytochrome b Y302C attacks the Cys 138 -Cys 155 disulfide bond of the ironsulfur protein to yield an intersubunit disulfide bond between these subunits. The resulting intersubunit disulfide bond then self-inactivates cytochrome bc 1 , and loss of the intramolecular disulfide bond of the iron-sulfur protein compromises its [2Fe-2S] cluster stability to induce its oxidative disruption (Fig. 8). FIGURE 6. Mass spectrometric identification of cytochrome b 296 WYFLPFXAILR 306 peptide that encompasses position 302 in native (WT) and Y302X variants of cytochrome bc 1 . a, native, Y302V, and Y302C variant cytochrome (cyt) bc 1 proteins were purified using Biogel A anionic exchange and Sephacryl S400 chromatography as described previously (29). Purity of the samples was checked by SDS-PAGE/Coomassie Blue staining (insets), and samples were digested in-solution with GluC and trypsin proteases and subjected to nanoLC-MS/MS spectrometry. FeS, iron-sulfur. b, MS/MS fragmentation spectra identified cytochrome b 296 WYFLPFYAILR 306 (z ϭ 2, X corr ϭ 3.18, ⌬C N ϭ 0.54) and 296 WYFLPFVAILR 306 (z ϭ 2, X corr ϭ 3.23, ⌬C N ϭ 0.64) peptides with the native and Y302V samples, respectively. No such peptide was identified using Y302C variant, although a shorter peptide corresponding to 296 WYFLPF 301 (z ϭ 1, X corr ϭ 1.36, ⌬C N ϭ 0.75) (right panel) was observed with all three samples. Only singly charged b and y ions are listed for simplification.
We emphasize that this attractive proposal should remain hypothetical until the resolution of the structure of Y302C enzyme.

DISCUSSION
Respiratory energy conservation by cytochrome bc 1 depends on an elaborate bifurcated electron transfer reaction that occurs in the presence of O 2 (22,43). Unless the catalytic steps associated with oxidation of reduced quinone are compromised, the native enzyme safely carries out this reaction without any ROS production. Various factors, like inhibitors that disable reduction of b hemes (32) or increased ⌬ m (44) and changes in pO 2 (45), trigger ROS production by cytochrome bc 1 . Serendipitously, studying cytochrome bc 1 of a facultative anoxygenic photosynthetic bacterium, R. capsulatus, we found that the Y302C mutant preserved its activity in the absence of O 2 but lost it progressively upon exposure to O 2 because of oxidative disintegration of its [2Fe-2S] cluster cofactor. This finding illustrates that specific amino acid(s) of cytochrome bc 1 , like Tyr 302 of cytochrome b, might play key roles in preventing undesirable electron leakage to O 2 during Q o site catalysis. In the case of R. capsulatus, substituting Tyr 302 by other amino acids decreased cytochrome bc 1 activity and enhanced ROS production, but only the Tyr to Cys mutation self-inactivated cytochrome bc 1 . In the latter case, we tentatively propose that formation of an intersubunit disulfide crosslink, probably via ROS-induced cysteine redox chemistry (42), destabilizes the iron-sulfur protein [2Fe-2S] cluster by elimi-nating its own stabilizing intramolecular disulfide bridge (Fig.  8, top panel). The role of Tyr 302 of cytochrome b in correctly positioning the iron-sulfur protein [2Fe-2S] cluster at the Q o site to confer optimal catalytic activity has been discussed in earlier studies for bacterial (16,17) and yeast (18,19) cytochrome bc 1 . However, in these studies, oxygen sensitivity of the enzyme or oxidative disintegration of its iron-sulfur protein [2Fe-2S] cluster has not been described. Although no Y302C mutant was examined in R. sphaeroides (17), it was reported that the homologous yeast Y279C mutant produced superoxide (18), but oxidative disruption of its iron-sulfur protein [2Fe-2S] cluster was not described (18,19). It is noteworthy that R. capsulatus native cytochrome b has no Cys residue, whereas the yeast homologue has several, and one of them (Cys 342 ) is located near Tyr 279 of cytochrome b (supplemental Fig. S2, bottom panel). The distance that separates Tyr 279 and Cys 342 is large (ϳ10 Å) in the yeast native cytochrome bc 1 structure. However, neither the corresponding distance in the Y279C mutant nor the occurrence of an intracytochrome b disulfide bond not affecting the [2Fe-2S] cluster stability is currently known.
Interestingly, the bovine and chicken native cytochrome b subunits also contain several Cys residues, but they lack a homologue of Cys 342 or any other Cys residue near its Tyr 278 (Fig. 8, bottom panel). Thus, as in R. capsulatus, in bovine, chicken, and possibly human cytochrome bc 1 , the iron-sulfur protein Cys residues remain as the likely candidates to yield an   How a Tyr at a specific position of cytochrome bc 1 interferes with ROS production is not obvious. We note that replacing Tyr with other amino acids (including Phe) eliminates a key hydroxyl group at position 302 of cytochrome b and enhances antimycin A-independent ROS production by the Q o site. The high resolution structures of cytochrome bc 1 indicate that Tyr 302 is mobile, and its position depends on that of the ironsulfur protein at the Q o site. Moreover, this residue is solvated, with its hydroxyl group within H-bonding distances from a cluster of H 2 O molecules (38). It is possible that elimination of this H 2 O cluster in Y302X mutants enhances unproductive encounters between O 2 and dangerous one-electron Q o site intermediates. Mutants losing the hydroxyl group would exhibit decreased cytochrome bc 1 activity, presumably because of incorrect positioning of the iron-sulfur protein [2Fe-2S] cluster at the Q o site and also produce more ROS to endanger host cell viability. Indeed, permanent ROS production induces lethal cellular damages but hypoxia-induced transient ROS generation by cytochrome bc 1 could act as a signal for oxidative stress conditions (4,7). Reversible chemical modifications of residues such as this Tyr might be of physiological significance.
Besides the bacterial Y302C, homologous human Y278C and malarial Y268C cytochrome bc 1 mutants also exhibit decreased Q o site activities (12,13). However, whether they produce ROS remains unknown. Extrapolation of the information gained using the bacterial cytochrome bc 1 to the evolutionarily conserved organelle homologues suggests that human mitochondrial cytochrome b Tyr 278 mutants might produce ROS, and cytochrome bc 1 variants with Tyr to Cys mutations might selfinactivate via oxidative disruption of their [2Fe-2S] clusters. Although this inference awaits experimental validation, it also suggests that resistance to the antimalarial drug atovaquone by the parasite might be acquired at the expense of increased O 2 sensitivity and concomitant decrease of energetic efficiency of the Q o site (16). Recent findings indicate that malarial cytochrome bc 1 might have different selective pressure than most other systems (46) to allow the survival of O 2 -compromised atovaquone resistant parasites in microaerobic human host environment. Regarding the human cytochrome bc 1 , all Y278X but C substitutions would have decreased activity and kill host cells by continuous ROS production. Only the self-inactivating Y278C allele of cytochrome bc 1 would cease ROS production via oxidative disruption of the [2Fe-2S] cluster and allow somewhat better host cell viability. However, even in this case, the hampered catalytic efficiency of the Y278C variant, combined with progressive loss of heteroplasmic state and differing energy threshold requirements of various tissues, would cause progressively increasing mitochondrial dysfunctions. These include exercise intolerance (15), ischemic cardiomyopathy (47), reperfusion injury (48), and multi-system disorder (12). In the Y302C variant of cytochrome bc 1 , because of ROS production by the Q o site, the thiol group of cytochrome b Cys 302 is thought to reduce the Cys 138 -Cys 155 disulfide bond of the iron-sulfur protein and destabilize the [2Fe-2S] cluster to induce oxidative inactivation of the enzyme. The intersubunit disulfide bond is tentatively proposed to occur between Cys 302 and Cys 155 (instead of Cys 302 and Cys 138 ) because of the shorter distance separating them in the native structure. However, available mass spectrometry data does not discriminate between these two possibilities, and no structure is yet available for R. capsulatus Y302C cytochrome bc 1  Clearly, reversible modification of a critical hydroxyl group of cytochrome bc 1 might be beneficial as a ROS signal initiator, but its permanent loss by mutation would steer mitochondria to a disastrous destiny via continuous oxidative damage and cell death.