Three Classes of Ubiquinone Analogs Regulate the Mitochondrial Permeability Transition Pore through a Common Site*

To identify the structural features required for regulation of the mitochondrial permeability transition pore (PTP) by ubiquinone analogs (Fontaine, E., Ichas, F., and Bernardi, P. (1998)J. Biol. Chem. 40, 25734–25740), we have carried out an analysis with quinone structural variants. We show that three functional classes can be defined: (i) PTP inhibitors (ubiquinone 0, decylubiquinone, ubiquinone 10, 2,3-dimethyl-6-decyl-1,4-benzoquinone, and 2,3,5-trimethyl-6-geranyl-1,4-benzoquinone); (ii) PTP inducers (2,3-dimethoxy-5-methyl-6-(10-hydroxydecyl)-1,4-benzoquinone and 2,5-dihydroxy-6-undecyl-1,4-benzoquinone); and (iii) PTP-inactive quinones that counteract the effects of both inhibitors and inducers (ubiquinone 5 and 2,3,5-trimethyl-6-(3-hydroxyisoamyl)-1,4-benzoquinone). The structure-function correlation indicates that minor modifications in the isoprenoid side chain can turn an inhibitor into an activator, and that the methoxy groups are not essential for the effects of quinones on the PTP. Since the ubiquinone analogs used in this study have a similar midpoint potential and decrease mitochondrial production of reactive oxygen species to the same extent, these results support the hypothesis that quinones modulate the PTP through a common binding site rather than through oxidation-reduction reactions. Occupancy of this site can modulate the PTP open-closed transitions, possibly through secondary changes of the PTP Ca2+ binding affinity.

Regulation of ion fluxes across the inner mitochondrial membrane is essential both for metabolic regulation and for energy conservation. The inner membrane possesses an intrinsically low permeability to ions and solutes, which allows energy conservation in the form of a proton electrochemical potential difference (1), and a set of channels and transporters that regulate ion fluxes and volume homeostasis (see Ref. 2 for a recent review). However, mitochondria in vitro can easily undergo a permeability increase to solutes with molecular masses of about 1,500 Da or lower, which is followed by deenergization, disruption of ionic homeostasis, and swelling, the so-called permeability transition (PT). 1 The PT has long been studied as a target for mitochondrial dysfunction in vivo, particularly in the context of ischemia-reperfusion injury (3)(4)(5). Interest in the PT as a mediator of cell death has more recently been extended to programmed cell death, a process in which mitochondria are involved through release of cytochrome c (6) and of other proteins like apoptosis-inducing factor (7), procaspases 3 and 9 (8,9), and through regulation of the levels of cellular ATP (10). It is widely assumed that the PT is due to opening of a proteinaceous pore, the PTP, whose molecular identity remains debated (see Refs. 2 and 11 for discussion). Recent results suggest that the PTP may open under physiological conditions (12), supporting the hypothesis that it may serve as a mitochondrial Ca 2ϩ release channel (13).
Ca 2ϩ is the single most important factor for PTP opening in vitro, but the PTP open-closed transitions are also modulated by a large variety of physiological factors and drugs. Among these, the PTP inhibitor CsA has become the standard diagnostic tool for the characterization of the PTP in isolated mitochondria, in living cells and organs, and in vivo (see Ref. 2 for review). However, CsA does not selectively act on mitochondria, and it inhibits other cellular functions that depend on Ca 2ϩ and/or on calcineurin (14). Furthermore, PTP inhibition by CsA is transient (15) and can be overcome by increasing Ca 2ϩ loads (16). Even in the most successful cases of in vivo protection by CsA in brain models of disease, exclusion of the drug by the blood-brain barrier may limit its usefulness (17)(18)(19).
We have long been involved in the study and characterization of PTP inducers and inhibitors, with the long term goals of defining the PTP regulatory features and molecular nature, and of developing better drugs for its modulation in vivo. In a series of recent studies, we have shown that the PTP is modulated by electron flux through respiratory complex I (20), which in turn led to the demonstration that the PT is inhibited by Ub 0 and decyl-Ub (21). Inhibition by these quinones could be specifically relieved by Ub 5 , which is inactive per se (see Ref. 11 for review). In order to identify the structural features involved in PTP modulation by quinones and to address the issue of mech-anism, we report here an analysis of the effects of quinone structural variants on the PT. We have identified three functional classes, i.e. PTP inhibitory quinones, PTP-inducing quinones, and PTP-inactive quinones, that are nonetheless able to counteract the effects of both inhibitory and inducing quinones. Since the ubiquinone analogs used in this study have a very similar midpoint potential and decrease mitochondrial ROS production to the same extent, these results support the idea that the effects of quinones on the PTP are exerted through binding to a common site rather than through redox reactions.

MATERIALS AND METHODS
Rat liver mitochondria were prepared according to standard differential centrifugation procedures in a medium containing 250 mM sucrose, 10 mM Tris-HCl (pH 7.4), and 0.1 mM EGTA-Tris.
Mitochondrial oxygen consumption was measured polarographically at 25°C using a Clark-type oxygen electrode. Extramitochondrial Ca 2ϩ concentration was measured fluorimetrically in the presence of 1 M Calcium Green-5N exactly as described (20) with excitation and emission wavelengths set at 506 and 532 nm, respectively, with either a PTI Quantamaster C61 or a Kontron SFM 23 spectrofluorometer. Changes of DCFDA fluorescence (8 M) were determined with the same instruments with excitation and emission wavelengths set at 506 and 521 nm, respectively. Mitochondrial volume changes were measured from the absorbance changes at 540 nm with a Kontron Uvikon 941 spectrophotometer. All instruments were equipped with magnetic stirring and thermostatic control, and the incubation medium is specified in the legend to Fig. 2. Ub 0 , Ub 5 , Ub 10 , decyl-Ub, and CsA were purchased from Sigma (L'Iles d'Abeau, France) while TMOH-Ub 5 , TM-Ub 10 , DMdecyl-Ub, OH-decyl-Ub, and DOH-undecyl-Ub were provided by Hoffmann-La Roche (Basel, Switzerland). All other chemicals were of the highest purity commercially available. The chemical structure of the quinones tested in this study is presented in Fig. 1. For the sake of clarity, quinones have been grouped in three classes based on their effects on the PTP: (i) PTP inhibitors (group I), (ii) PTP inducers (group II), and (iii) PTP-inactive analogs that counteract the effects of both group I and II quinones (group III; see "Results" and "Discussion" for details).

Effect of Ubiquinone Analogs on Mitochondrial Ca 2ϩ
Loading-In the experiments of Fig. 2, energized rat liver mitochondria were loaded with a train of 30 M Ca 2ϩ pulses at 1-min intervals. Although a slight day-to-day variability was ob-served, mitochondria typically took up and retained Ca 2ϩ until the load reached a threshold of about 120 nmol of Ca 2ϩ ϫ mg of protein Ϫ1 , a point at which mitochondria underwent a fast process of Ca 2ϩ release (trace a). This precipitous Ca 2ϩ release was due to opening of the PTP because (i) it was accompanied by depolarization and swelling; and (ii) the Ca 2ϩ load required for Ca 2ϩ release was dramatically increased by CsA (data not shown, but see Ref. 21). When the quinones listed in Fig. 1 were tested, a striking PTP inhibition was observed with DM-decyl-Ub (trace b), Ub 10 (trace c), and TM-Ub 10 (trace d). A similar inhibitory effect has been previously reported for Ub 0 and decyl-Ub (21), which together with DM-decyl-Ub, Ub 10 , and TM-Ub 10 thus define a class of PTP inhibitory quinones (group I of Fig. 1). Fig. 2 also shows that OH-decyl-Ub (trace e) and DOH-undecyl-Ub (trace f) instead decreased the Ca 2ϩ load required for PTP opening without affecting the uptake of the initial Ca 2ϩ pulse(s). This identifies the previously unrecognized existence of PTP-inducing quinones as well (group II of Fig. 1). Finally, the addition of TMOH-Ub 5 had no detectable effects on the Ca 2ϩ retention capacity (trace g, compare with trace a). As will become clear later, TMOH-Ub 5 , like Ub 5 (21), can remove the effects of group I quinones; these two compounds thus define a third class of PTP-inactive quinones that still demonstrably affect its function (group III of Fig. 1).
The experiments of Fig. 3 report the concentration dependence of the effects of these compounds on the Ca 2ϩ retention capacity of rat liver mitochondria. DM-decyl-Ub (panel A) and Ub 10 (panel B) increased the Ca 2ϩ retention capacity with an optimum of approximately 50 M, while the effect decreased at higher concentrations without, however, affecting the rate of Ca 2ϩ uptake (data not shown). TM-Ub 10 also improved the Ca 2ϩ retention capacity at concentrations up to 40 M, but it inhibited the rate of Ca 2ϩ uptake at higher concentrations (data not shown). Finally, OH-decyl-Ub (panel D) and DOHundecyl-Ub (panel E) decreased the Ca 2ϩ retention capacity at all concentrations, while TMOH-Ub 5 was without any detectable effects (panel F).
Effects of Combinations of Group I and Group III Quinones-In the experiments depicted in Fig. 4, onset of the PT was monitored from the changes of absorbance at 540 nm, which reflect mitochondrial permeabilization to sucrose. Mitochondria were challenged with a Ca 2ϩ load of 150 nmol ϫ mg of protein Ϫ1 , which caused a detectable PT (traces a in both panels). PTP opening was prevented by TM-Ub 10 (panel A, trace b), and this inhibition was relieved by Ub 5 (trace c), a group III quinone that does not affect the mitochondrial Ca 2ϩ retention capacity but can remove the inhibitory effects of Ub 0 and decyl-Ub but not of CsA (21). Very similar results were obtained when the PTP had been inhibited by other group I quinones (Ub 10 and DM-decyl-Ub; results not shown). Thus, Ub 5 effectively removed the effects of all available group I quinones (see also Ref. 21).
A similar experiment was performed with the group III quinone TMOH-Ub 5 in mitochondria where PTP opening had been inhibited with decyl-Ub (Fig. 4, trace b). It can be seen that TMOH-Ub 5 relieved the inhibitory effects of decyl-Ub (trace c) but not of CsA (results not shown). A similar effect of TMOH-Ub 5 was observed in mitochondria where the PTP had been inhibited by Ub 10 but not by the other group I quinones Ub 0 and TM-Ub 10 (results not shown). These data suggest that group III quinones may bind to the same site as group I qui-nones, and that the final effect on the PTP may depend on their relative affinities for the PTP regulatory site(s). Fig. 5, we tested whether group III quinones can prevent the effects of the PTPinducing group II quinones. In the experiments of panel A, rat liver mitochondria were challenged with a Ca 2ϩ load of 60 nmol ϫ mg of protein Ϫ1 , which is not sufficient for PTP opening under these experimental conditions (trace a). In the presence of OH-decyl-Ub, PTP opening readily followed the addition of the same Ca 2ϩ load (trace b), and this inducing effect on the PTP was prevented by Ub 5 (trace c). The experiments of panel B show that neither OH-decyl-Ub alone (trace b) nor the combination of OH-decyl-Ub and Ub 5 (trace c) affected the rate of Ca 2ϩ uptake prior to onset of the PT relative to untreated mitochondria (trace a), a finding that was confirmed in experiments with an appropriate time scale (omitted for clarity). It is noteworthy that the combination of OH-decyl-Ub and Ub 5 was unable to fully restore the Ca 2ϩ retention capacity of untreated mitochondria (compare traces c and a).

Effects of Combinations of Group II and Group III Quinones-In the experiments depicted in
Similar experiments were carried out with different combinations of group II and III inhibitors. We found that, at vari- ance from Ub 5 , TMOH-Ub 5 was unable to prevent the inducing effects of OH-decyl-Ub, and that neither Ub 5 nor TMOH-Ub 5 could prevent the PTP-inducing effects of DOH-undecyl-Ub (results not shown).
Effects of Combinations of Group I and Group II Quinones-We next tested whether the group II, PTP-inducing quinones could contrast the effects of the group I, PTP-inhibiting quinones in assays based on the mitochondrial Ca 2ϩ retention capacity. Fig. 6 shows that (i) OH-decyl-Ub was able to counteract the increased Ca 2ϩ retention capacity elicited by Ub 10 (trace c, compare with trace b), yielding a result that came close to that of untreated mitochondria (trace a), and (ii) that DOH-undecyl-Ub likewise counteracted the large increase of Ca 2ϩ retention caused by Ub 0 , the most efficacious PTP inhibitor we found so far (trace e, compare with trace d). Similar results were obtained when mitochondria were treated with the other group I quinones decyl-Ub, DM-decyl-Ub, or TM-Ub 10 (data not shown). These experiments demonstrate that group II quinones can counteract the effects of all group I quinones, further suggesting an interaction at a common site.
Effects of Ubiquinone Analogs on Respiration and ROS Production-We have previously shown that there is no obvious correlation between the effects of quinones on respiration and on the PTP. Thus, both the group I Ub 0 and the group III Ub 5 inhibited uncoupled respiration, whereas the group I decyl-Ub had only marginal effects (21). We extended this analysis to the quinones tested in this study. Since PTP opening is accompanied by loss of NADH, which in turn causes respiratory inhibition with complex I substrates (20,22), these experiments were performed in the presence of CsA. Fig. 7A shows that the group I inhibitory quinones DM-decyl-Ub and Ub 10 had no effects on uncoupled respiration, while 50% inhibition was observed with TM-Ub 10 . Strong inhibition of respiration was observed with the group II quinones OHdecyl-Ub and DOHundecyl-Ub (panel B), while the group III quinone TMOH-Ub 5 inhibited maximal respiration only slightly (panel C). Since ROS may be relevant for PTP opening, we finally assessed whether quinones of group I, II, or III could modify mitochon-

FIG. 4. Effect of group III analogs on PTP inhibition by group I analogs.
Experimental conditions were the same as in Fig. 2 except that Calcium Green-5N was omitted. PTP opening was monitored as the absorbance decrease of the mitochondrial suspension at 540 nm. Experi drial production of ROS as assessed by the fluorescence changes of DCFDA. Fig. 8 shows that addition of DCFDA to respiring mitochondria led to a measurable fluorescence increase (trace a), which was slowed down by group I Ub 0 (trace b), group II OH-decyl-Ub (trace c), and group III Ub 5 (trace d). Similar effects were obtained with all the quinone derivatives described in this study (omitted for clarity). Thus, no obvious correlation exists between mitochondrial ROS production and effects of quinones on the PTP.

DISCUSSION
In this paper we have shown that ubiquinone analogs profoundly affect the mitochondrial PT. We have identified three classes of quinones: (i) group I, or PTP inhibitory quinones (Ub 10 , TM-Ub 10 , and DM-decyl-Ub), which mimic the effects we reported previously for Ub 0 and decyl-Ub (21); (ii) group II, or PTP-inducing quinones (OH-decyl-Ub and DOH-undecyl-Ub), which dramatically decrease the Ca 2ϩ load required for PTP opening and represent a previously unrecognized functional class of PTP inducers; and (iii) group III, or PTP-inactive quinones (TMOH-Ub 5 (this paper) and Ub 5 (Ref. (21)), which are nonetheless able to counteract the effects of group I and II quinones.
Structure-Function Correlation-An interesting issue is represented by the structure-function correlation for the specific effects of quinone analogs on the PTP. Inspection of the structures (Fig. 1), as well as comparison with the results of a previous study (21), suggests that seemingly minor changes can profoundly affect the quinone interactions with the PTP. Thus, (i) a methyl group in position 5 of the benzoquinone ring is not required for PTP inhibition since DM-decyl-Ub is active, (ii) the radical in position 6 affects both the potency and the quality of the effects of quinones on the PTP, and (iii) a methoxy or a methyl group in positions 2 and 3 appears to be required for PTP inhibition (group I) or for displacement (group III) because 1,4-benzoquinone and 2-methoxy-1,4-benzoquinone are not effective (21). Several problems must be kept in mind, however, when these complex results are analyzed. Although there is no direct correlation between the effects of quinones on the PTP and on respiration, early respiratory inhibition may preclude the analysis of potential PTP inhibitory quinones in protocols based on energy-dependent Ca 2ϩ uptake, thus limiting the range of compounds that can be tested. Also, it should be noted that a lack of demonstrable effects of quinones does not necessarily mean that they are not interacting with the PTP. Since PTP-inactive, group III quinones can only be recognized by competition experiments, quinones binding the PTP with low affinity could escape detection. Finally, we have not been able to conclusively address the issue of whether the interactions of quinones of the three groups is competitive or non-competitive in nature. The apparent EC 50 values obtained in this type of experiments were not entirely reliable either because of the biphasic nature of the response, and/or because increasing concentrations of quinones caused inhibition of Ca 2ϩ uptake. We therefore consider the present results as the starting point for more detailed mechanistic studies that will be addressed through the synthesis and the analysis of further structural variants.
Mechanistic Aspects-A key issue is the mechanism through which quinones modulate the PTP. The analogs used in this study have similar midpoint potentials, and this makes a mechanism of action based on oxidation-reduction events extremely unlikely. Furthermore, we found no correlation between the effects of quinones on the PTP and on ROS production, which rather appears to be decreased by all quinones tested in this study (Fig. 8). In principle, the existence of inducing and inhibiting quinones could be explained by the existence of two quinone binding sites, one responsible for inhibition and one for activation. Assuming that group I quinones may also bind to the latter site with lower affinity, the biphasic response observed with some quinones could be easily explained because high concentrations of group I quinones could indeed open the pore through the second site. This would be consistent, however, with the response to Ub 10 (Fig. 3, panel  B) but not to all other group I quinones (Fig. 3, panel A and data not shown). Although the existence of two sites cannot be conclusively ruled out at present, we favor a model where the competition between group I, II, and III quinones is mediated by quinone binding to a common site, whose occupancy would in turn modulate the PTP open-closed transitions through secondary changes of the PTP Ca 2ϩ binding affinity. In this scenario, addition of a small Ca 2ϩ load can cause the transition from the closed state to the open state only if the PTP is liganded with inducing agents. A larger Ca 2ϩ load would be required to access the Ca 2ϩ binding site(s) in the conformation liganded with inactive, group III quinones and a still larger Ca 2ϩ load in the state liganded with inhibitory, group I quinones. When several quinones are present, they would compete with each other according to their relative membrane concentration and to their binding affinities, resulting in turn in a subtle modulation of the accessibility to Ca 2ϩ and therefore of the PTP open-closed transitions.
The biphasic nature of the response observed with the group I quinones (Fig. 3 and Ref. 21) remains difficult to explain at present. We note, however, that due to their high hydrophobicity quinones tend to accumulate in membranes but can also organize in non-monomeric states in water, the critical micelle concentration depending on numerous parameters such as the ionic strength, the presence of organic solvent, the redox state of quinones, etc. (23). Since the properties of quinones are likely to change when they are in non-monomeric states, we suspect that non-monomeric quinones may bind to the pore and compete with quinones in monomeric state. PTP-inactive non-monomeric quinones would displace the PTP-inhibitor monomers, thus abolishing their protective effects, whereas PTP-inducing non-monomeric quinones would trigger PTP opening. Although entirely hypothetical, this explanation has the merit to account for the biphasic behavior of group I quinones.
Implications and Perspectives-The identification and characterization of specific quinone features involved in PTP modulation represents a significant advance that offers great perspective for the development of better drugs specifically acting on mitochondria. These findings can also shed new light on the role of quinones in aging and disease. The protective effects of Ub 50 reported in the literature for a variety of models of disease (24 -34) are generally related to its free radical scavenging activity (35), but our results offer an additional explanation. Indeed, the finding that several exogenous quinones modify the Ca 2ϩ retention capacity of isolated mitochondria indicates that the putative quinone binding site of the PTP is not saturated, and therefore that the PTP can in principle be affected by modifications of the amount or composition of quinones in vivo.