Selective Targeting of a Redox-active Ubiquinone to Mitochondria within Cells

With the recognition of the central role of mitochondria in apoptosis, there is a need to develop specific tools to manipulate mitochondrial function within cells. Here we report on the development of a novel antioxidant that selectively blocks mitochondrial oxidative damage, enabling the roles of mitochondrial oxidative stress in different types of cell death to be inferred. This antioxidant, named mitoQ, is a ubiquinone derivative targeted to mitochondria by covalent attachment to a lipophilic triphenylphosphonium cation through an aliphatic carbon chain. Due to the large mitochondrial membrane potential, the cation was accumulated within mitochondria inside cells, where the ubiquinone moiety inserted into the lipid bilayer and was reduced by the respiratory chain. The ubiquinol derivative thus formed was an effective antioxidant that prevented lipid peroxidation and protected mitochondria from oxidative damage. After detoxifying a reactive oxygen species, the ubiquinol moiety was regenerated by the respiratory chain enabling its antioxidant activity to be recycled. In cell culture studies, the mitochondrially localized antioxidant protected mammalian cells from hydrogen peroxide-induced apoptosis but not from apoptosis induced by staurosporine or tumor necrosis factor-α. This was compared with untargeted ubiquinone analogs, which were ineffective in preventing apoptosis. These results suggest that mitochondrial oxidative stress may be a critical step in apoptosis induced by hydrogen peroxide but not for apoptosis induced by staurosporine or tumor necrosis factor-α. We have shown that selectively manipulating mitochondrial antioxidant status with targeted and recyclable antioxidants is a feasible approach to investigate the role of mitochondrial oxidative damage in apoptotic cell death. This approach will have further applications in investigating mitochondrial dysfunction in a range of experimental models.

The mitochondrial respiratory chain is a major source of superoxide and, therefore, mitochondria accumulate oxidative damage more rapidly than the rest of the cell, contributing to mitochondrial dysfunction and cell death in degenerative diseases and in aging (1)(2)(3)(4)(5). Mitochondria are also central to activating apoptosis and oxidative damage can lead to cell death, however, the significance of mitochondrial oxidative damage for cell death is unclear (6 -8). One approach to this problem is to selectively target antioxidants to mitochondria (9 -11). This should allow the relative importance of mitochondrial and cytoplasmic oxidative stress for cell death to be distinguished, and also enable the contribution of mitochondrial damage to aging, diabetes, and cancer to be investigated in cell and animal models.
Derivatives of ubiquinol are promising antioxidants to target to mitochondria (11,12). In mammals ubiquinone comprises a 2,3-dimethoxy-5-methylbenzoquinone core with a hydrophobic 45-to 50-carbon chain at the 6 position (13,14). Mitochondrial ubiquinone is a respiratory chain component buried within the lipid core of the inner membrane where it accepts two electrons from complexes I or II becoming reduced to ubiquinol, which then donates electrons to complex III (14). The ubiquinone pool in vivo is largely reduced and ubiquinol is an effective antioxidant, as well as being a mobile electron carrier (15)(16)(17)(18). Ubiquinol acts as an antioxidant by donating a hydrogen atom from one of its hydroxyl groups to a lipid peroxyl radical, thereby decreasing lipid peroxidation within the mitochondrial inner membrane (18 -20). The ubisemiquinone radicals thus formed disproportionate to ubiquinone and ubiquinol (21), or react with oxygen to form superoxide and ubiquinone thereby transferring the radical to the aqueous phase for detoxification by superoxide dismutase and peroxidases (17,20). The respiratory chain then recycles ubiquinone back to ubiquinol to restore its antioxidant function. Vitamin E is another important antioxidant within the mitochondrial inner membrane, and the tocopheroxyl radical thus formed is regenerated to active vitamin E by reaction with ubiquinol or ubisemiquinone (15,17,20,22,23). Therefore, in vivo ubiquinol probably acts as an antioxidant by direct reaction with peroxyl radicals and by regenerating vitamin E (16,17,20).
The low solubility of ubiquinone in water makes it difficult to use in vitro, and animals must be fed ubiquinone-enriched diets for several weeks to increase levels in subsequently isolated mitochondria (11,14). Therefore, to manipulate mitochondrial ubiquinone content in vitro we synthesized a ubiquinone analog selectively targeted to mitochondria by addition of a lipophilic triphenylphosphonium cation. Such lipophilic cations easily permeate lipid bilayers and accumulate in mitochondria within cells, driven by the large mitochondrial membrane po-tential (9,10,24). Here we report on the antioxidant and antiapoptotic properties of this mitochondrially targeted ubiquinone derivative.
Characterization of Products-Stock solutions containing a mixture of 4 and 5 in ethanol were stored at -80°C, and their concentrations were confirmed by 31 P NMR. Fully oxidized solutions were generated by incubation in basic 95% ethanol on ice (13) or with beef heart mitochondrial membranes at room temperature. Both procedures gave an extinc-tion coefficient of 10,400 M -1 cm -1 at 275 nm for the quinone, with shoulders at 263 and 268 nm corresponding to the triphenylphosphonium moiety (26,27) and a broad shoulder at 290 nm due to the quinone (13) (Fig. 1A). Reduction with NaBH 4 gave the quinol, which had local maxima at 290 nm (⑀ ϭ 1800 M -1 cm -1 ) and at 268 nm (⑀ ϭ 3000 M -1 cm -1 ) (27). The ⌬⑀ ox-red at 275 nm in 50 mM sodium phosphate, pH 7.2, was 7000 M -1 cm -1 . The quinone extinction coefficient (10,400 M -1 cm -1 at 275 nm) was slightly lower than that reported for other quinones (12,250 M -1 cm -1 ) in aqueous buffer (28). This difference was not due to an intermolecular interaction between the phosphonium and the quinone, because the absorbances of 2 and the simple phosphonium methyltriphenylphosphonium (TPMP) were additive when 50 M of each were mixed together in either ethanol or aqueous buffer. To prepare mitoquinol an ethanolic solution was diluted in ϳ0.5-1 ml of water and a few grains of NaBH 4 were added. After incubation on ice in the dark for 5 min, excess NaBH 4 was quenched with 5% HBr (0.2 ml) and the quinol was extracted into CH 2 Cl 2 (3 ϫ 0.5 ml). The extract was then washed with water and 2 M NaCl, then the CH 2 Cl 2 was removed under a stream of nitrogen. The pale yellow solid residue was dissolved in acidified 96% ethanol. The yield was typically 70 -80% and, as the quinol slowly oxidized in air, it was freshly prepared and stored on ice under argon in the dark. To determine partition coefficients, compounds were added to 2 ml each of 1-octanol-saturated PBS and PBSsaturated 1-octanol then shaken at 37°C for 30 min in the dark. After separation by centrifugation, the amounts in each phase were determined by absorption relative to standard curves in 1-octanol-saturated PBS or PBS-saturated 1-octanol.
Mitochondrial Preparations and Incubations-Rat liver mitochondria were prepared by homogenization followed by differential centrifugation (29). Beef heart mitochondria were isolated by standard procedures, and membrane fragments were prepared by sonication followed by centrifugation (30). Protein concentrations were determined by the biuret assay using bovine serum albumin as a standard (31). Endogenous ubiquinone was removed from lyophilized beef heart mitochondria by pentane extraction, and complete extraction was confirmed by the inability of these mitochondria to oxidize NADH in the absence of added Q 1 (13).
For respiration measurements, rat liver mitochondria (2 mg of protein/ml) were suspended in KCl medium (120 mM KCl, 10 mM Hepes, 1 mM EGTA, pH 7.2) at 25°C supplemented with respiratory substrates and 1 mM phosphate in a 3-ml oxygen electrode (Rank Brothers, Bottisham, Cambridge, UK). After measuring the rate of coupled respiration 200 M ADP was added, the rate of phosphorylating respiration was measured and then FCCP (300 nM) was added and the rate of uncoupled respiration determined. To measure membrane potential, rat liver mitochondria (2 mg of protein/ml) were incubated for 3 min in 0.5 ml of KCl medium supplemented with nigericin (1 g/ml), 5 mM each of glutamate and malate, 1 M TPMP, and 100 nCi/ml [ 3 H]TPMP. After incubation the mitochondria were pelleted by centrifugation, the radioactivity in the pellet and supernatant were measured by scintillation counting, and the membrane potential was calculated using the Nernst equation, assuming a mitochondrial volume of 0.5 l/mg of protein and that 60% of the intramitochondrial TPMP was membrane-bound (32,33). The uptake of [ 3 H]mitoQ by rat liver mitochondria was measured under the same conditions. Scanning spectra and kinetic measurements were made with an Aminco DW-2000 dual beam spectrophotometer using matched 1-ml quartz cuvettes at 20°C. Beef heart mitochondrial membranes and freeze-thawed yeast mitochondria were incubated in 50 mM potassium phosphate, pH 7.2. Rat liver mitochondria were incubated in KCl medium.
Oxidative Damage Assays-To measure thiobarbituric acid reactive species (TBARS), rat liver mitochondria (2 mg of protein/ml) were incubated at 37°C with shaking for 45 min in 100 mM KCl, 10 mM Tris-HCl, pH 7.6. Then 0.8-ml aliquots were mixed with 400 l of 0.5% thiobarbituric acid in 35% HClO 4 , heated at 100°C for 15 min, diluted with 3 ml of water, and extracted into 3 ml of n-butanol. TBARS were determined fluorometrically ( excite ϭ 515 nm; emission ϭ 553 nm) and expressed as nanomoles of MDA by comparison with standard solutions of 1,1,3,3-tetraethoxypropane processed as above. Prior to analyzing samples, their mitoQ contents were brought to the same concentration to eliminate differences in MDA formation during heating and processing. To measure the membrane potential after exposure to oxidative damage, mitochondria were pelleted by centrifugation and resuspended in KCl medium, and their membrane potentials were determined as described above.
To measure cis-parinaric acid oxidation, mitochondria (2 mg of protein/ml) were suspended in a 3-ml fluorimeter cuvette in 100 mM KCl and 10 mM Tris-HCl, pH 7.6, at 37°C. The oxidation of cis-parinaric acid (Molecular Probes) was monitored fluorometrically ( excite ϭ 324 nm; emission ϭ 413 nm). A control experiment showed that titration of cis-parinaric acid into a mitochondrial suspension initially increased the fluorescence as it partitioned into membranes, but the increase became nonlinear after about 3 M and declined above 10 M due to self quenching (34). Therefore, 3 M cis-parinaric acid was used for all experiments. Peroxynitrite was synthesized from acidified H 2 O 2 and NaNO 2 in a simple flow reactor as described previously (35), concentrated by freeze fractionation and stock solutions in 1.5 M NaOH quantitated [⑀ 302 ϭ 1.67 mM -1 .cm -1 (36)].
Mammalian Cell Culture-Human osteosarcoma 143B cells were cultured at 37°C under humidified 95% air/5% CO 2 in DMEM supplemented with penicillin (100 units/ml), streptomycin (100 mg/ml), and 10% fetal calf serum. For toxicity studies, cells were grown to confluence in 24-well tissue culture dishes and incubated for 24 h with DMEM/ serum containing the compound. The supernatants were then harvested, and the amount of LDH released was assayed and compared with that present in untreated wells lysed with 0.1% Triton. For uptake studies cells were suspended in 0.5 ml of DMEM supplemented with 10 mM HEPES and 5 M [ 3 H]mitoQ. After incubation, cells were pelleted by centrifugation, and the radioactivity in the pellet was quantitated by scintillation counting. For digitonin fractionation, cells were incubated as above, and then 500 l of the cell suspension was mixed rapidly with 1.2 ml of ice-cold 250 mM sucrose, 20 mM MOPS pH 6.7, 3 mM EDTA, and 1 mg of digitonin, then 1 ml was layered onto 350 l of oil (66% silicone oil/34% dioctyl pthalate) over 100 l of 500 mM sucrose, 0.1% Triton and separated into mitochondrial and cytoplasmic fractions by centrifugation. The two fractions were assayed for citrate synthase and LDH activity or for content of radioactivity by scintillation counting (39). The Jurkat human T lymphocyte line was grown at 37°C under humidified 95% air/5% CO 2 in RPMI 1640 supplemented with penicillin (100 units/ml), streptomycin (100 mg/ml), and 10% fetal calf serum. Apoptosis was induced by addition of hydrogen peroxide (8). Caspase activation in lysed cell pellets was measured fluorometrically by the cleavage of the peptide DEVD labeled with AMC (DEVD-AMC) and calibrated using an AMC standard curve (40). The proportion of cells undergoing apoptosis was quantitated by annexin V-fluorescein isothiocyanate staining followed by detection of annexin-positive cells using a Becton Dickinson FACScan flow cytometer.
General Procedures-Column chromatography was on Silica Gel type 60, 200 -400 mesh, 40 -63 m (Merck). TLC was on Silica Gel 60F 254 (Merck) or on C-18 silica (Whatman). Nuclear magnetic resonance spectra were acquired on Varian 500 MHz or Varian 300 MHz instruments in CDCl 3 . Chemical shifts are in ␦ units (ppm) downfield from tetramethylsilane for 1 H NMR and 13 C NMR and 85% phosphoric acid for 31 P NMR. In some cases the total integral ratios in the 1 H NMR were not precisely as expected; however, all other structural data are fully consistent with the proposed structures. Infrared absorption spectra were acquired using a PerkinElmer 1600 FTIR spectrometer. Phosphonium salts were examined by applying a concentrated deuterochloroform solution to NaCl discs followed by evaporation of the solvent in a nitrogen stream. Other samples were examined neat or as Nujol mulls between NaCl discs. Mass spectra were obtained from the Chemistry Department, University of Canterbury. Data are presented as m/z values for the parent molecular ion. Fluorescence measurements were made using a PerkinElmer MPF-3L fluorescence spectrophotometer.
Electron Transfer between Mitochondria and mitoQ-To determine whether the respiratory chain could reduce mitoquinone, we incubated it with beef heart mitochondrial membranes and recorded its spectrum (Fig. 1B, t ϭ 0). Addition of the respiratory substrate succinate reduced mitoquinone to mitoquinol (Fig. 1B). Mitoquinone-mitoquinol interconversion was then measured continuously by monitoring mitoquinone at 275 nm (Fig. 2). Mitoquinone was reduced by beef heart mitochondrial membranes and succinate, and this reduction was blocked by the complex II inhibitor malonate ( Fig. 2A). Chemically reduced mitoquinol was also oxidized by membranes, and this oxidation was blocked by the complex III inhibitor myxothiazol (Fig. 2B). Mitoquinol and mitochondrial membranes reduced ferricytochrome c and myxothiazol inhibited this reduction by 60 -70% (data not shown). Rat liver mitochondria respiring on succinate or glutamate/malate reduced mitoquinone, and this activity was blocked by the respiratory inhibitors malonate (Fig. 2C) or rotenone (Fig. 2D), respectively. Dissipation of the membrane potential with the uncoupler FCCP also eliminated the reduction of mitoquinone by preventing its uptake into mitochondria (Fig. 2, C and D).
Reduction of Mitoquinone by Respiratory Complexes-To distinguish between mitoQ reactions with respiratory complexes and the endogenous ubiquinone/ubiquinol pool, we extracted beef heart mitochondrial membranes with pentane to remove endogenous ubiquinone. These mitochondria still oxidized mitoquinol and, in the presence of succinate, reduced mitoquinone by a malonate-sensitive pathway (Fig. 3A). This strategy was extended to investigate yeast that entirely lacked endogenous ubiquinone due to inactivation of ubiquinone biosynthesis. These yeast did not grow on the nonfermentable carbon source YPEG until the short-chain ubiquinone analog Q 2 was added, but addition of mitoQ did not lead to cell growth (Fig. 3B). When respiratory chain activity was measured in mitochondria isolated from these yeast, there was no electron flux from NADH or succinate to cytochrome c until Q 2 was added (Fig. 3,  C and D). MitoQ did stimulate myxothiazol-sensitive cytochrome c reduction, but less so than Q 2 (Fig. 3, C and D). When ubiquinone reduction was analyzed directly, mitochondria lacking ubiquinone reduced both Q 2 and mitoquinone, however, the rate for mitoquinone was slower than for Q 2 (Fig. 3E). The rate of reduction of mitoquinone by mitochondria was FIG. 1. Absorption spectra of mitoquinol and mitoquinone. A, mitoQ (50 M) was incubated with beef heart mitochondrial membranes (20 g of protein/ml) for 1 h to oxidize it to mitoquinone. Reduction with NaBH 4 (ϳ250 g) gave mitoquinol. B, 50 M mitoQ was incubated with beef heart mitochondrial membranes, and the spectrum of mitoquinone was recorded (t ϭ 0). Then antimycin (5 M) and succinate (5 mM) were added, and further spectra were acquired at 5-min intervals (t ϭ 5-25).  2. Reduction of mitoquinone and oxidation of mitoquinol by mitochondria. A, beef heart mitochondrial membranes (20 g of protein/ml) were incubated with rotenone (8 g/ml) and mitoQ (50 M). A 275 was monitored continuously and succinate (5 mM) and malonate (20 mM) were added where indicated. B, mitochondrial membranes were incubated as above with mitoquinol (50 M), rotenone (8 g/ml), and malonate (20 mM). Myxothiazol (10 M) was added where indicated. When this experiment was repeated in the presence of ferricytochrome c (50 M) and KCN (200 M) the ferricytochrome c was reduced and this reduction was decreased by 60 -70% by myxothiazol (data not shown). C and D, rat liver mitochondria (100 g of protein/ml) were incubated in KCl, medium, and A 275 was monitored. For the experiments shown in C, rotenone (8 g/ml) and succinate (5 mM) were present and mitoQ (20 M) was added where indicated. This experiment was repeated in the presence of malonate (20 mM) or FCCP (300 nM: not shown but identical to the experiment in the presence of malonate). For the experiments shown in D glutamate and malate (5 mM of each) were present and mitoQ (20 M) was added where indicated. This experiment was repeated in the presence of rotenone (8 g/ml) or FCCP (300 nM; not shown but identical to the experiment in the presence of rotenone). similar in the presence or absence of endogenous ubiquinone (Fig. 3, E and F). We conclude that mitoQ can be reduced and oxidized by the mitochondrial respiratory chain and that this is primarily through the active sites of respiratory complexes rather than via the endogenous ubiquinone pool (14,28).
Uptake of mitoQ by Isolated Mitochondria-Tritiated mitoQ was taken up rapidly by energized mitochondria, and addition of the uncoupler FCCP caused its immediate efflux (Fig. 4A). This FCCP-sensitive accumulation of mitoQ was substantial up to 20 M mitoQ (Fig. 4B). The mitoQ accumulation ratios were measured relative to the simple lipophilic cation TPMP over a range of membrane potentials (Fig. 4C). At low membrane potential, the mitoQ accumulation ratio was greater than for TPMP, but at high potentials the maximum accumulation ratio for mitoQ (ϳ500 -600) was slightly less than that for TPMP (ϳ1300), suggesting that the greater hydrophobicity of mitoQ decreases its uptake slightly relative to TPMP.
Location of mitoQ within Mitochondria-Alkyltriphenylphosphonium cations equilibrate between the bulk phase and a potential energy well on the membrane surface where they adsorb as a monolayer close to the carbonyl groups of the phospholipid fatty acids (41)(42)(43). This adsorption is described by ␤, the ratio of the surface density of adsorbed ions to their volume density in the bulk phase (42). We measured the nonspecific adsorption of mitoQ and TPMP to de-energized mitochondria over a range of concentrations and determined the ratio of adsorbed to free cations from double-reciprocal plots (Fig. 4D). From these and the surface area of rat liver mitochondria (155 and 520 cm 2 /mg of protein for the outer and inner membranes, respectively (44)) we calculated ␤ ϭ 22 ϫ 10 Ϫ6 cm for TPMP and ␤ ϭ 57 ϫ 10 Ϫ6 cm for mitoQ, values which are comparable to those of tetraphenylphosphonium adsorption to simple phospholipid bilayers (␤ ϭ 4.2 ϫ 10 Ϫ6 cm (42)). An earlier study showed that the greater adsorption of decyltriphenyl phosphonium relative to TPMP was due to the increased entropy change for insertion of the alkyl group into the lipid bilayer, whereas the enthalpy change for cation adsorption was unaffected (42). This suggests that mitoQ adsorbs to membranes about three times more strongly than TPMP because of insertion of its hydrophobic side chain into the lipid bilayer (41). We conclude that mitoQ is taken up into energized mitochondria and is then largely adsorbed to the matrix face of the inner membrane with the phosphonium cation at the level of the fatty acid carbonyls while the hydrophobic side chain inserts into the lipid interior of the membrane.

Uptake of mitoQ by Mitochondria within Cells-Tritiated
mitoQ incubated with a suspension of 143B osteosarcoma cells was taken up over 20 -40 min, and this uptake was decreased by disrupting the mitochondrial membrane potential (Fig. 5A).
To determine the location of mitoQ within cells, we disrupted the plasma membrane of cells with the detergent digitonin and pelleted the mitochondria by centrifugation through oil (Fig. 5, B and C). This separated cells into mitochondrially enriched and cytosolic fractions, as confirmed by the distribution of the mitochondrial and cytosolic marker enzymes citrate synthase and lactate dehydrogenase (Fig. 5B). About half the mitoQ within the cell was found in mitochondrial fraction, similar to the proportion of mitoQ uptake sensitive to FCCP (Fig. 5C). Therefore, substantial amounts of mitoQ are taken up by both isolated mitochondria and mitochondria within cells, driven by the mitochondrial membrane potential.
Low Toxicity of mitoQ to Mitochondria and Cells-The toxicity of mitoQ was investigated in mitochondria and cells. Up to 10 M mitoQ had little effect on the membrane potential of isolated mitochondria but at 25 M and above the potential decreased (Fig. 6A). For mitochondria respiring on either succinate or glutamate/malate, 10 M mitoQ had little effect on uncoupled or phosphorylating respiration, but was inhibitory at 25-50 M (Fig. 6, B and C). MitoQ stimulated coupled respiration by increasing the proton leak though the inner membrane; this effect was minimal for succinate at 10 M but was noticeable at 10 M for glutamate/malate (Fig. 6C). MitoQ toxicity to human 143B cells was determined from the release of lactate dehydrogenase into the culture medium over 24 h (Fig. 6D). MitoQ up to 10 M did not affect cell viability, and concentrations of 25-50 M were required for substantial cell death (Fig. 6D). In summary, mitoQ concentrations up to 10 M do not disrupt mitochondrial or cell function and, therefore, concentrations of 1-5 M were used in subsequent experiments.
Antioxidant Properties of mitoQ-To investigate the antioxidant efficacy of mitoQ, we incubated mitochondria with cisparinaric acid (Fig. 7A). This fatty acid fluoresces within a lipid environment, and its conjugated double bond fluorophore is susceptible to oxidation. Consequently, the disappearance of fluorescence is a measure of lipid peroxidation (34). In the presence of mitochondria, cis-parinaric acid fluoresced due to its insertion into lipid bilayers (Fig. 7A). MitoQ prevented the oxidation of cis-parinaric acid by hydrogen peroxide and ferrous iron, demonstrating that mitoquinol is an antioxidant (Fig. 7A). To quantitate the antioxidant efficacy of mitoQ, mitochondria were incubated with ferrous iron, and the accumulation of MDA was measured as a marker of lipid peroxidation (Fig. 7B). This oxidative damage also disrupted mitochondrial function as indicated by a decrease in the membrane potential (Fig. 7C). Incubation with mitoQ prevented both the accumulation of MDA and the disruption to mitochondrial function caused by oxidative stress (Fig. 7, B and C). To determine whether mitoquinone or mitoquinol was the effective antioxi-dant, we oxidized mitoQ to mitoquinone and prevented its reduction by the respiratory chain by including malonate and rotenone in the incubation. Under these conditions, mitoquinone did not block lipid peroxidation (Fig. 7D). In contrast, when mitoQ was reduced to mitoquinol by the respiratory chain, oxidative damage was prevented (Fig. 7D). The simple lipophilic cation TPMP did not prevent lipid peroxidation (Fig.  7D). Therefore, the antioxidant activity of mitoQ is due to its ubiquinol moiety.
Recycling of mitoQ by the Respiratory Chain-To determine whether mitoquinol was recycled by the respiratory chain after detoxifying a reactive oxygen species, we studied its interaction with peroxynitrite, a biologically significant oxidant formed from nitric oxide and superoxide (45)(46)(47). Because the half-life of peroxynitrite is only 1-2 s, mitoquinol regeneration from mitoquinone can be studied after all the added peroxynitrite has decomposed (45). Mitoquinol was rapidly oxidized to mitoquinone by peroxynitrite, however, mitoquinone was only detected when its reduction by the respiratory chain was prevented by malonate (Fig. 8A, upper trace). Continuous monitoring of the mitoquinone concentration showed that peroxynitrite rapidly oxidized the mitoquinol to mitoquinone, which was then reduced to mitoquinol by the respiratory chain (Fig. 8B). When malonate was present, this reduction by the respiratory chain was prevented (Fig. 8B). We conclude that mitoQ is an effective antioxidant that can be recycled to its active form by the respiratory chain after detoxifying a reactive oxygen species.
Prevention of Apoptosis by mitoQ-A range of stimuli induce apoptosis by releasing cytochrome c from mitochondria into the cytoplasm, where it activates caspases. These include the oxidant hydrogen peroxide (8), the protein kinase C inhibitor staurosporine (40), and tumor necrosis factor-␣ (48). The mechanisms by which these stimuli cause cytochrome c release from mitochondria are unclear, but some or all may involve increased mitochondrial oxidative stress. Of particular interest is whether cytochrome c release induced by hydrogen peroxide is caused directly by mitochondrial oxidative damage or is a secondary consequence of cytoplasmic redox changes (8). A mito- FIG. 7. Antioxidant efficacy of mitoQ. A, rat liver mitochondria were incubated with 10 mM succinate, 8 g/ml rotenone, 50 M ferrous sulfate, 100 M ascorbic acid, and 1 mM hydrogen peroxide in the presence or absence of 5 M mitoQ. Where indicated, 3 M cis-parinaric acid was added and the fluorescence measured. Data are from a typical experiment repeated on three separate mitochondrial preparations. TPMP (5 M) did not block cis-parinaric acid oxidation (data not shown). B, mitochondria were preincubated with 10 mM succinate and mitoQ for 5 min, and then a sample was taken for TBARS analysis (zero time). Ferrous sulfate (100 M) and 300 M ascorbic acid were then added, and 40 min later MDA formation was quantitated. Data are means Ϯ range of duplicate determinations and are typical of experiments repeated on three separate mitochondrial preparations. C, mitochondria were incubated as described for B, isolated by centrifugation, and their membrane potential determined from the uptake of TPMP while respiring on glutamate/malate. Data are means Ϯ range of a typical experiment repeated on three separate mitochondrial preparations. D, mitochondria were incubated as in B supplemented with 8 g/ml rotenone. Succinate (5 mM) and mitoQ or TPMP (5 M) were added to some incubations, whereas for the other incubations malonate (20 mM) was present and mitoQ was oxidized completely to mitoquinone before addition by incubation at basic pH. After preincubation for 5 min, ferrous sulfate (50 M) was added, and 40 min later MDA formation was quantitated. Data are means Ϯ range of duplicate determinations. shows that this treatment reduced all of the mitoQ to mitoquinol. Then malonate (20 mM) was added to prevent further reduction, and finally, peroxynitrite (200 M) was added and the spectrum of mitoquinone was acquired (ϩperoxynitrite). When peroxynitrite was added without malonate the spectrum obtained was similar to that of mitoquinol (Ϫperoxynitrite). B, beef heart mitochondrial membranes were incubated with 50 M mitoQ and succinate (5 mM) as above, and the mitoquinone concentration monitored continuously at 275 nm. Where indicated peroxynitrite (500 M) was added alone (Ϫmalonate), or in the presence of 20 mM malonate (ϩmalonate; dashed line).
chondrially targeted antioxidant can help elucidate the role of mitochondrial oxidative damage in apoptosis, because it would only block those apoptotic signals that require mitochondrial oxidative stress. Therefore, we investigated the effect of mitoQ on apoptotic cell death.
Addition of hydrogen peroxide to Jurkat cells led to caspase activation and induction of apoptotic cell death 4 -6 h later (Fig. 9). Preincubation with 1 M mitoQ completely blocked caspase activation (Fig. 9, A and B) and substantially decreased apoptotic cell death, as determined by the proportion of annexin-positive cells (Fig. 9, C and D). To determine whether the mitochondrial localization of mitoQ was required for this protective effect, we compared mitoQ with Q 1 , a quinol antioxidant that distributes evenly throughout the cell, and found that Q 1 did not block caspase activation (Fig. 9B). In contrast to the situation with hydrogen peroxide, mitoQ did not prevent apoptosis in Jurkat cells treated with staurosporine (40) or in WEHI 164 cells treated with tumor necrosis factor-␣ (49) (data not shown). We conclude that mitoQ blocks apoptosis induced by hydrogen peroxide. Because mitochondrial localization of the antioxidant is required to prevent apoptosis, mitochondrial oxidative stress may be a critical step in hydrogen peroxideinduced apoptosis but not for apoptosis following treatment with staurosporine or tumor necrosis factor-␣.
Conclusion-To provide new approaches to investigate the role of mitochondrial oxidative damage in cell death, we synthesized a mitochondrially targeted antioxidant, mitoQ, comprising a ubiquinone attached to a triphenylphosphonium lipophilic cation. The ubiquinone moiety was found to cycle between its oxidized (mitoquinone) and reduced (mitoquinol) forms by exchanging electrons with the respiratory chain. Mitoquinol was an effective antioxidant protecting mitochondria from oxidative damage and was rapidly regenerated by the respiratory chain after detoxifying a reactive oxygen species. As anticipated, the triphenylphosphonium cation led to the rapid and reversible accumulation of mitoQ by isolated mitochondria and by mitochondria within cells. Therefore, mitoQ is a mitochondrial-specific antioxidant.
We then used mitoQ to help elucidate the role of mitochondrial oxidative damage in apoptotic cell death. As a first step we showed that mitoQ prevented apoptosis caused by hydrogen peroxide but not that caused by staurosporine or tumor necrosis factor-␣. This suggests that mitochondrial oxidative damage plays an important role in hydrogen peroxide-induced apoptosis but is not required for apoptosis induced by staurosporine or tumor necrosis factor-␣. Further work using these and other mitochondrially targeted compounds to dissect out the role of mitochondrial oxidative changes in hydrogen peroxide-induced apoptosis is ongoing. The findings reported here demonstrate that mitochondrially targeted antioxidants such as mitoQ can be used to investigate the role of mitochondrial oxidative stress in cell death. This strategy also has potential for unraveling the contribution of oxidative stress to other pathologies involving mitochondrial dysfunction.