Báo cáo khoa học: The mitochondrial permeability transition from in vitro artifact to disease target ppt

REVIEW ARTICLE The mitochondrial permeability transition from in vitro artifact to disease target Paolo Bernardi 1 , Alexandra Krauskopf 1, *, Emy Basso 1 , Valeria Petronilli 1 , Elizabeth Blalchy-Dyson 2 , Fabio Di Lisa 3 and Michael A. Forte 2 1 Department of Biomedical Sciences and CNR Institute of Neurosciences, University of Padova, Italy 2 Vollum Institute, L474, Oregon Health and Sciences University, Portland, OR, USA 3 Department of Biological Chemistry and CNR Institute of Neurosciences, University of Padova, Italy Introduction The mitochondrial permeability transition (PT) is an increase of mitochondrial inner membrane permeabil- ity to solutes with molecular masses up to  1500 Da. Under the conditions used in most in vitro studies, PT is accompanied by depolarization, matrix swelling, depletion of matrix pyridine nucleotides (PN), outer membrane rupture and release of intermembrane pro- teins, including cytochrome c [1,2]. The occurrence of swelling in isolated mitochondria, its stimulation by Ca 2+ , Pi and fatty acids, its inhibition by Mg 2+ , aden- ine nucleotides and acidic pH, and its detrimental effects on energy conservation, have been clearly recognized since the early studies on isolated mitochondria were carried out [3–13]. Ever since, the Keywords apoptosis; calcium; cancer; cell death; degenerative diseases; drugs; mitochondria; necrosis; permeability transition Correspondence P. Bernardi, Department of Biomedical Sciences, University of Padova, Viale Giuseppe Colombo 3, I-35121 Padova, Italy Fax: +39 049 827 6361 E-mail: bernardi@bio.unipd.it *Present address Center for Integrative Genomics, University of Lausanne, 1015 Lausanne-Dorigny, Switzerland (Received 6 January 2006, revised 1 March 2006, accepted 3 March 2006) doi:10.1111/j.1742-4658.2006.05213.x The mitochondrial permeability transition pore is a high conductance chan- nel whose opening leads to an increase of mitochondrial inner membrane permeability to solutes with molecular masses up to  1500 Da. In this review we trace the rise of the permeability transition pore from the status of in vitro artifact to that of effector mechanism of cell death. We then cover recent results based on genetic inactivation of putative permeability transition pore components, and discuss their meaning for our understand- ing of pore structure. Finally, we discuss evidence indicating that the per- meability transition pore plays a role in pathophysiology, with specific emphasis on in vivo models of disease. Abbreviations AAF, 2-acetylaminofluorene; ANT, adenine nucleotide translocator; CNS, central nervous system; CRC, Ca 2+ retention capacity; CsA, cyclosporin A; CyP, cyclophilin; Dp, proton electrochemical gradient; Dw m , mitochondrial membrane potential; MMC, mitochondrial megachannel; PARP, poly(ADP-ribose) polymerase; PBR, peripheral benzodiazepine receptor; PN, pyridine nucleotides; PT, permeability transition; PTP, permeability transition pore; TNF-a, tumor necrosis factor; ROS, reactive oxygen species; Ub0, ubiquinone 0; VDAC, voltage- dependent anion channel. FEBS Journal 273 (2006) 2077–2099 ª 2006 The Authors Journal compilation ª 2006 FEBS 2077 conditions for isolation, storage and incubation of mitochondria have been intentionally (albeit empiric- ally) designed to minimize its occurrence. The typical mitochondrial isolation and storage solu- tions are based on K + -free mannitol and ⁄ or sucrose. Incubation in these media promotes H + –K + exchange and matrix acidification, which in turn potently inhib- its the PT [14,15]. The presence of EGTA prevents Ca 2+ accumulation during homogenization, but also depletes the physiological pool of matrix Ca 2+ , which is an essential permissive factor for PTP opening. Finally, most in vitro studies have used succinate (in the presence of rotenone) as the energy source. From the bioenergetics point of view, this choice of substrate tends to provide the most consistent results because succinate oxidation does not require matrix PN, which can be lost via the PT during storage or incubation even when swelling is undetectable [16]. What was not appreciated until recently is that rotenone makes PT opening more difficult by blocking electron flux through Complex I and by preventing the oxidation of PN [17,18]. Having carefully selected conditions that minimize its occurrence, it should not be too surprising that the requirements to induce a PT in isolated mito- chondria may appear ‘extreme’ in what obviously rep- resents a circular argument. Yet, together with the detrimental effects of the PT on energy conservation, this state of affairs has significantly contributed to the widespread feeling that the PT was an in vitro artifact of little pathophysiological relevance. A notable exception was the proposal by Pfeiffer and coworkers that the ‘damaging’ effects of Ca 2+ had a physiological role in steroidogenesis. These authors showed that Ca 2+ induces a ‘transformation’ of adre- nal cortex mitochondria, allowing extramitochondrial PN to gain access to the otherwise impermeable mat- rix, and that NADPH entering in this way supports the 11-b hydroxylation of deoxycorticosterone [19–21]. These findings matched those of Vinogradov et al., who documented a Ca 2+ -dependent release of matrix PN through the otherwise impermeable inner mem- brane in liver mitochondria [16]. The term ‘permeabil- ity transition’, however, was introduced by Haworth & Hunter, who carried out a detailed characterization of its basic features in heart mitochondria. These authors provided a key insight that the PT was caused by reversible opening of a proteinaceous pore in the inner mitochondrial membrane – the permeability transition pore (PTP) – and proposed that it may serve an unde- fined physiological role [22–25]. It is fair to say that this proposal was not met by enthusiasm. In part, at least, this was an indirect conse- quence of the general acceptance of the chemiosmotic hypothesis, which had just been fully recognized with the award of the Nobel Prize in Chemistry to Peter Mitchell in 1978 [26]. As already noted [1], studies of mitochondrial ion transport were mostly carried out in the same laboratories involved in clarifying the mecha- nisms of energy conservation; and they tended to become tests of the predictions of the chemiosmotic theory, in particular about the existence of a membrane potential across the inner membrane. Given the view, prevailing well into the 1980s, that mitochondria did not possess cation channels [27] it is not too surprising that the existence of a large pore in the inner membrane appeared to contradict the basic tenets of chemiosmo- sis. As a result, and with very few exceptions [28–37], research in this area did not enjoy much popularity until a set of major findings began to attract a larger number of investigators. A turning point was the discovery that the PT could be inhibited by submicromolar concentrations of the immunosuppressant drug, cyclosporin A (CsA) [38– 41]. It was later shown that CsA inhibits the PTP after binding to matrix cyclophilin (CyP)-D, a peptidyl- prolyl cis-trans isomerase whose enzymatic activity is blocked by CsA in the same range of concentrations required to inhibit the pore [42–45]. The recent charac- terization of mitochondria from mice with genetic inactivation of the Ppif gene encoding CyP-D has allowed a better understanding of the role of CyP-D in PTP regulation [46–49], and will be discussed in some detail later in the review. A second major finding was made possible by the demonstration that mitochondria possess ion channels that can be studied by electrophysiology using the patch-clamp technique [50]. This seminal study was soon followed by the demonstration that the inner mitochondrial membrane is also endowed with a high- conductance ( 1 nS) channel, the ‘mitochondrial megachannel’ (MMC) [51,52]. The MMC is inhibited by CsA [53], and possesses all the basic regulatory fea- tures of the PTP [54,55], leaving little doubt that they are the same molecular entity [56]. Electrophysiology has greatly contributed to our understanding of the MMC-PTP, and to the acceptance of the pore theory of the PT [57]. A third contribution was the demonstration that the PTP is controlled by the proton electrochemical gradi- ent (Dp), the open–closed transitions being modulated by the mitochondrial membrane potential (Dw m ) and by matrix pH [14,15,58]. These findings have been fully confirmed by studies at the single channel level [59]. As the threshold voltage for PTP opening is affected by a large variety of pathophysiological effectors [60,61], PTP control by the Dp provided a conceptual The mitochondrial permeability transition P. Bernardi et al. 2078 FEBS Journal 273 (2006) 2077–2099 ª 2006 The Authors Journal compilation ª 2006 FEBS framework to accommodate a large number of individ- ual agents known to induce or inhibit the PT [62–64]. The hypothesis that PTP opening could be a factor in cell death, was put forward nearly 20 years ago [65]. A series of seminal studies published in the early 1990s provided experimental support for this hypothesis in hepatocytes subjected to oxidative stress [66,67], anoxia [68] or treatment with ATP [69], and in cardiomyo- cytes [70] and isolated hearts [71] exposed to ischemia followed by reperfusion. The recent surge of interest in the PT as an effector mechanism of cell death, how- ever, only followed the demonstration that in the course of apoptosis, cytochrome c is released into the cytosol [72], together with apoptosis-inducing factor [73] and a set of proteins involved in the effector phase of apoptosis [74–77]. A rigorous test of whether a PT takes place in intact cells, organs and living organisms remains a major challenge. This is caused by the intrinsic complexity of PTP modulation [1,2]; by the fact that occurrence of a PT must still be deduced by indirect means, which in turn generates major interpretative problems of the experimental results [78]; by the lack of both selectivity and persistence of action of CsA [79], which may yield ‘negative’ results even in conditions where the PTP is actually involved; and by the lack of a defined struc- ture for the channel itself, as should become clear from the discussion of the most studied candidate proteins [i.e. the adenine nucleotide translocator (ANT), the voltage-dependent anion channel (VDAC) and the peripheral benzodiazepine receptor (PBR)]. We have already pointed out the many sources of artifacts that have hampered research in this field [78], and we have identified outstanding mechanistic problems on the role of mitochondria in cell death that involve the PTP [80]. Here we will provide an account of the status of PTP research in the light of recent achievements based on genetic and pharmacological strategies, and on the study of relevant in vivo models of disease. A Medline search identified close to 2000 publica- tions on the PTP, a figure that demands a selection of the primary references that can be quoted. We apolo- gize in advance to those who could not find a place in our reference list, and we refer the reader to recent reviews discussing the possible role of the PTP in sev- eral paradigms of disease for a more detailed coverage [81–96]. Modulation of the PT As mentioned above, the PT is most easily observed after the matrix accumulation of Ca 2+ , and it is widely believed to be caused by the opening of a regulated channel, the PTP. The pore can be defined as a volt- age-dependent, CsA-sensitive, high-conductance chan- nel of the inner mitochondrial membrane. In the fully open state, its apparent diameter is  3 nm, and the pore open–closed transitions are highly regulated by multiple effectors that may converge on a smaller set of regulatory sites. We have classified factors that affect the PT into matrix and membrane effectors, and we refer the reader to a previous review for details [1]. Matrix effectors Pore opening is favored by matrix Ca 2+ through a site that can be competitively inhibited by other Me 2+ ions, such as Mg 2+ ,Sr 2+ and Mn 2+ , and by Pi through a still-undefined mechanism. Pore opening is strongly promoted by an oxidized state of PN and of critical dithiols at discrete sites [97,98], and both effects can be individually reversed by proper reductants. The dithiol–disulfide interconversions correlate with the redox state of glutathione and can be blocked by 1-chloro-2,4-dinitrobenzene, suggesting that the dithiol is in redox equilibrium with matrix glutathione. This finding accounts easily for the PT-inducing effects of both peroxides and redox-cycling agents, and for the corresponding inhibition with monofunctional thiol reagents, such as N-ethylmaleimide and monobromobi- mane [1]. PTP modulation by these redox-sensitive sites easily accommodates the inducing effects of p66Shc, which directly oxidizes cytochrome c to pro- duce superoxide anion and causes PTP-dependent cell death [99,100]. The PT is strictly modulated by matrix pH. In de-energized mitochondria, the pH optimum for open- ing is 7.4, while the open probability decreases sharply both below pH 7.4 (through reversible protonation of critical histidyl residues that can be blocked by diethyl- pyrocarbonate) [14,15] and above pH 7.4 (through an unknown mechanism). Histidyl residues (particularly His126 of CyP-A and His87 of the FK506-binding protein) have also been shown to play important roles in ligand binding and peptidyl prolyl cis-trans iso- merase catalysis by immunophilins [101]. Our recent finding, that PTP modulation by matrix pH between 6.0 and 7.0 is identical in mitochondria from Ppif null and wild-type animals, demonstrates that PTP-regula- tory histidines are not located on CyP-D [47], as already suggested based on the effects of diethylpyro- carbonate [102]. It is important to stress that the over- all effect of pH on the PTP can be dramatically affected by energization, because an acidic pH does not inhibit, but rather promotes, PTP opening in ener- gized mitochondria owing to an increased rate of Pi P. Bernardi et al. The mitochondrial permeability transition FEBS Journal 273 (2006) 2077–2099 ª 2006 The Authors Journal compilation ª 2006 FEBS 2079 uptake, an effect that may worsen PTP-dependent tis- sue damage in ischemic and postischemic acidosis [103]. PTP regulation by matrix CyP-D will be discussed below. Membrane effectors The inside-negative Dw m tends to stabilize the PTP in the closed conformation [14]. We have postulated the existence of a voltage sensor that decodes the changes of both the transmembrane voltage and of the surface potential into changes of the PTP open probability [60]. Such a sensor would easily account for pore opening following depolarization as such, and for the effects of a large variety of membrane-perturbing agents that can either inhibit or promote the PT. In general, amphipathic anions, such as fatty acids pro- duced by phospholipase A 2 , favor the PT with an effect that cannot be explained by depolarization. In particular, arachidonic acid appears to play a key role in apoptotic Ca 2+ -dependent apoptotic signalling through the PTP [104,105]. Conversely, polycations such as spermine, amphipathic cations such as sphing- osine and trifluoroperazine, and positively charged peptides, inhibit pore opening [64]. The putative volt- age sensor may comprise critical arginine residues, as suggested by modulation of the PTP voltage depend- ence by a set of arginine-selective reagents [106,107]. The PTP is regulated by electron flux within respir- atory chain complex I, with an increased open probab- ility when flux increases [17]. This finding led to the discovery that the PT is regulated by quinones, poss- ibly through a specific binding site, whose occupancy affects the open–closed transitions depending on the bound species [108]. Binding of ubiquinone 0 (Ub0) or decylubiquinone prevents Ca 2+ -dependent pore open- ing, irrespective of the inducing agent; and the inhibi- tory effect of Ub0 and decylubiquinone (but not that of CsA) can be relieved by pore-inactive quinones, such as ubiquinone 5 [108]. High-throughput screening of isolated mitochondria has recently identified a novel PTP inhibitor, Ro 68–3400, which apparently interacts with the same site as Ub0 [109]. This inhibitor will be discussed in greater detail in relation to the possible role of VDAC in PTP formation. Consequences of pore opening The only primary consequence of PTP opening is mito- chondrial depolarization. Unless single channel events are being recorded, however, PTP openings of short duration may be undetectable. Indeed, for short durations of the open time, repolarization follows, and the depolarization–repolarization cycle may not be detected by potentiometric probes. Furthermore, open- ing events are not synchronized for individual mito- chondria [110,111] and may be missed in population studies as a result of probe redistribution among indi- vidual mitochondria. The occurrence of PTP openings of different durations in mitochondria in situ, and their consequences on cell viability, have been addressed in a series of specific studies to whom the reader is referred for details [104,112,113]. For longer times of opening, depolarization can be easily measured both in isolated mitochondria and intact cells, and the PT may have consequences on res- piration that depend on the substrates being oxidized. With Complex I-linked substrates, PTP opening may be followed by respiratory inhibition owing to a loss of matrix PN [16]. With Complex II-linked substrates, the PT is rather followed by uncoupling. The conse- quences of a PT on respiration in vivo, and the related issue of production of reactive oxygen species (ROS) therefore depend on whether, and to what extent, PN are lost. Irrespective of whether respiration is inhibited or stimulated, collapse of the Dp will prevent ATP syn- thesis as long as the pore is open. ATP hydrolysis by the mitochondrial ATPase would then worsen ATP depletion, which, together with altered Ca 2+ homeo- stasis, is a key factor in various paradigms of cell death [80]. Persistent PTP opening is followed by equilibration of ionic gradients and of species that have a molecular mass of < 1500 Da, which may cause swelling, cristae unfolding and outer membrane rupture. The occur- rence of swelling can be prevented by pore-impermeant solutes, and for short open times solute equilibration may not occur at all. Thus, assessing whether swelling, outer membrane rupture, and release of cytochrome c and other proapoptotic factors follow a PT in situ needs to be verified in each experimental setting. An interesting mechanism that links PTP opening to release of cytochrome c in the absence of outer mem- brane rupture has recently emerged. Cytochrome c is compartmentalized within mitochondria in two pools that can be distinguished based on their redox interac- tions with the outer and inner membrane electron transfer systems [114]. About 15% of cytochrome c can be reduced by outer membrane NADH-cyto- chrome b 5 reductase, suggesting that it is located within the intermembrane space; while 85% can only be reduced by the inner membrane electron transfer chain [114] and probably resides within the intercristal compartment identified by tomographic reconstruction of mitocondria after high-voltage electron microscopy The mitochondrial permeability transition P. Bernardi et al. 2080 FEBS Journal 273 (2006) 2077–2099 ª 2006 The Authors Journal compilation ª 2006 FEBS [115]. During proapoptotic stimulation, prominent cris- tae remodeling occurs, which effectively increases the communication between the two pools of cytochrome c, and therefore the fraction that can be released through BAX ⁄ BAK channels on the outer membrane [116]. In summary, a rigorous assessment of the occurrence and of the consequences of PTP opening, in particular mitochondrial depolarization, swelling and outer membrane rupture, demands a careful measurement of several variables that can only be deduced by indirect means [78]. Whether the PTP has a physiological function other than taking part in cell death remains a matter of spe- culation. We, as well as others, have suggested that the pore may serve as a mitochondrial Ca 2+ -release chan- nel [117,118], and we refer the reader to previous reviews that specifically discuss this possibility in some detail [1,117]. Role of the ANT The PTP is strikingly modulated by ligands of the ANT. Atractylate, which inhibits the ANT and stabil- izes it in the ‘c’ conformation [119], favors PTP open- ing, while bongkrekate, which also inhibits the ANT but stabilizes it in the ‘m’ conformation [119], favors PTP closure. These findings led to the suggestion that the PTP may be directly formed by the ANT [120]. We have proposed an alternative explanation that is based on PTP modulation by the changes of the surface potential. Indeed, the transition of the translocase from the ‘m’ to the ‘c’ conformation is accompanied by a large decrease of the surface potential [121,122], an observation that may easily explain pore opening by atractylate and pore closure by bongkrekate within the framework of PTP modulation by the membrane potential [1]. The ANT reconstituted in giant liposomes exhibits high-conductance (but CsA-insensitive) channel activ- ity that is stimulated by Ca 2+ [123]. The channel dis- plays a marked voltage dependence with prominent gating effects that are consistent with the reported voltage dependence of the pore in intact mitochondria [123]. In addition, after reconstitution in liposomes, the ANT catalyzes Ca 2+ -dependent malate transport that is inhibited by ADP and favored by atractylate [124]. By using chromatography of mitochondrial extracts on a CyP-D affinity matrix, specific binding of the ANT has been demonstrated by two laboratories, while there is a discrepancy as to whether VDAC is also essential for reconstitution of PTP activity [125,126]. Although intriguing, we think that the rele- vance of these observations to PTP regulation remains unclear. Indeed, in the work of Woodfield et al. [125], CyP-D also bound many other proteins besides ANT, some of them with high affinity and in a CsA-inhibita- ble manner. Moreover, CyP-D bound equally well to ANT purified from rat liver or from yeast, despite the fact that the PT is not inhibited by CsA in yeast mito- chondria [127]. In the work of Crompton et al., CyP-D column eluates containing both ANT and VDAC con- ferred CsA-sensitive permeabilization to proteolipo- somes that had been treated with Ca 2+ plus Pi, yet it is difficult to exclude that permeabilization was caused by other species represented less than the abundant ANT and VDAC [126]. Conclusive evidence that the ANT is not essential for PTP formation was obtained in a detailed analysis of mitochondria lacking all ANT isoforms, which revealed that a Ca 2+ -dependent PT took place [128]. Of note, the PT of ANT null mitochondria was fully inhibitable by CsA and could be triggered by H 2 O 2 and diamide, indicating that the ANT is neither the obligatory binding partner of CyP-D nor the site of action of oxidants [128]. In addition, hepatocytes pre- pared from control and ANT-deficient livers showed identical responses to activation of receptor-mediated apoptotic pathways initiated by tumor necrosis factor-a (TNF-a) and Fas [128]. It has been argued that a low, undetectable level of ANT expression could have been present, producing the PTP observed in ANT-null mitochondria [129]. As the PTP was insensitive to opening by atractylate and to closure by ADP, it is very difficult to envisage how any ANT molecule in mutant mitochondria would not respond to atractylate and ADP, and yet be able to promote a CsA-sensitive PT. Another relevant obser- vation is that mitochondria from the anoxia-tolerant brine shrimp, Artemia franciscana, do not undergo a PT, despite a remarkable Ca 2+ -uptake capacity and the presence of ANT, VDAC and CyP-D [130]. Role of the VDAC Early studies demonstrated that the PT induced by sul- fhydryl reagents is not observed in mitoplasts, suggest- ing that inner membrane permeability changes require the outer mitochondrial membrane as well [34]. Several lines of evidence suggest that the outer membrane protein involved in PTP formation may be VDAC, namely that (a) purified VDAC incorporated into pla- nar phospholipid bilayers forms channels with a pore diameter of 2.5–3.0 nm whose electrophysiological properties are strikingly similar to those of the PTP [131,132], (b) the VDAC channel properties are modu- lated by the addition of NADH, Ca 2+ , glutamate P. Bernardi et al. The mitochondrial permeability transition FEBS Journal 273 (2006) 2077–2099 ª 2006 The Authors Journal compilation ª 2006 FEBS 2081 [133–135] and by binding of hexokinase [136–138], all conditions that also modulate the activity of the PTP [17,97,139], and (c) as mentioned above, chromato- graphy of mitochondrial extracts on a CyP-D affinity matrix allowed purification of VDAC and the ANT, which in the presence of CyP-D catalyzed CsA-sensi- tive permeabilization of liposomes to solutes [126]. The inner and outer mitochondrial membranes appear to interact in specialized regions called the ‘contact sites’, which are enriched in ANT and VDAC bound to cytosolic hexokinase I [140], leading to the idea that the PTP may be formed by interacting ANT and VDAC molecules at these sites. Fractions of deter- gent-solubilized mitochondria containing hexokinase activity were assessed by western blotting and found to contain a variety of additional proteins, including VDAC, the ANT and CyP-D. When reconstituted into planar lipid bilayers or liposomes, these fractions gave rise to channels with properties resembling those of the PTP [140,141]. While these studies establish that the PTP activity can be reconstituted after detergent extraction of mitochondria, whether VDAC is essential in this activity is far less clear (e.g. Fig. 1 in ref. 142). Indeed, further purification of such extracts by gel fil- tration chromatography resulted in fractions contain- ing hexokinase in which the presence of VDAC (and the ANT) could not be demonstrated, yet remained capable of forming PTP-like pores on incorporation into liposomes [140]. Thus, it appears legitimate to wonder what component(s) of these extracts are, in fact, responsible for the formation of CsA-inhibitable pores, an argument that applies to all attempts to identify the PTP components based on purification of hexokinase activity [140–142]. More recent biochemical attempts to place VDAC in the PTP complex have been based on the characterization of proteins that co-immunoprecipitate with the ANT, constituting the ‘adenine nucleotide translocase interactome’ [143]. While VDAC, as well as a wide variety of cytosolic, endoplasmic reticulum, inner membrane and matrix proteins are present in these immunoprecipitates, their direct relation to the PTP has yet to be established. Unfortunately, much of the data used to support the presence of VDAC in the PTP is based on a tenuous logical generalization – properties of VDAC in bilayers correspond (closely or not) to properties of the PTP observed in mitochondria, therefore VDAC must be part of the PTP. This generalization has been embellished in most recent reports on the involvement of VDAC in the PTP. Thus, data demonstrating the VDAC activity in bilayers is modulated by ruthenium red and La 3+ [134], arsenic trioxide [144], hexokinase [145], protein cross-linkers [146] and fluoxetine [147] and may reflect in vitro alterations in VDAC activity by these treat- ments, but cannot formally be extended to reflect the involvement of VDAC in the PTP, either in mitochon- dria or in a cellular context. These considerations obviously do not exclude a possible role for VDAC in apoptotic pathways not dependent on the PTP. As been demonstrated in the case of CyP-D, poten- tially the most convincing data on the role of VDAC in PTP formation could be generated through an examination of PTP activity in mitochondria prepared from tissues in which VDAC has been genetically elim- inated. The difficulty with these studies stems from the fact that in mammals three genes encode VDAC iso- forms (for a review of the genetics of VDAC see ref. 148) and each VDAC isoform is able to form channels when incorporated into planar bilayers, albeit with somewhat different characteristics [149]. Mice have also been created in which genes encoding individual VDAC isoforms have been eliminated by ‘knockout’ strategies. Mice missing VDAC1 and VDAC3 are viable, but show isoform-specific phenotypes [150], while the unconditional elimination of VDAC2 results in embryonic lethality [151]. Thus, the involvement of VDAC in PTP function has been difficult to assess in these animals because it is likely that any VDAC iso- form may potentially compensate for the absence of any other isoform, given that all cells appear to express each isoform. The idea that VDAC is a component of the PTP has been considerably reinforced by Cesura et al. who used a functional assay based on swelling of isolated mitochondria to screen a chemical library for inhibi- tors of the PTP [109]. A high-affinity inhibitor was identified (Ro 68–3400), which was then used to label mitochondria and identify a protein, of  32 kDa, as VDAC1 [109]. More recently, we have found that the PTP in mitochondria prepared from VDAC1-null mice is fully sensitive to inhibition by Ro 68–3400, and that the inhibitor labels a 32 kDa protein that is indistin- guishable from the species labeled in wild-type mito- chondria. Of note, we have been able to separate the labeled 32 kDa protein from all VDAC isoforms in both VDAC1-null and wild-type mitochondria, unam- biguously proving that the latter are not the targets for inhibition by Ro 68–3400 (A. Krauskopf et al., unpub- lished results). In the end, the involvement of VDAC in the PT remains reasonable, but convincing data directly impli- cating this protein in PTP formation in an in vivo, phy- siological context are absent. Given the redundancy of VDAC genes in mammals, the application of more advanced genetic approaches [e.g. loxP versions of individual VDAC genes or the use of small interfering The mitochondrial permeability transition P. Bernardi et al. 2082 FEBS Journal 273 (2006) 2077–2099 ª 2006 The Authors Journal compilation ª 2006 FEBS (siRNA)] may allow the generation of more reliable data. Alternatively, the development of pharmacologi- cal agents that specifically target all VDAC isoforms with high affinity may also allow the involvement of VDAC in the PTP to be convincingly established. Until such studies are available, the involvement of VDAC in the PTP remains unproven. Role of the PBR The PBR is an 18 kDa, highly hydrophobic protein located in the outer mitochondrial membrane [153] and was initially identified as a binding site for ben- zodiazepines in tissues that lack 4-aminobutyrate receptors, the clinical target of benzodiazepines in the central nervous system (CNS). The PBR shares no amino acid homology with CNS 4-aminobutyrate receptors, and can be distinguished pharmacologically from CNS receptors by its binding to a variety of high-affinity ( nm), PBR-specific ligands [154,155], notably the benzodiazepine, Ro5-4864, and the iso- quinoline carboxamide, PK11195. These, and a num- ber of other compounds, have been used extensively in the biochemical and physiological characterization of the PBR in vivo and in vitro [156]. The PBR is found in a wide variety of tissues at varying levels and is especially abundant in cells producing steroid hor- mones, such as adrenal cortex and Leydig cells of the testis [157,158]. In these cells, PBR promotes the trans- port of cholesterol into the mitochondrial matrix, a rate-limiting step in steroid synthesis [159]. The PBR also binds some porphyrins, including protoporphyrin IX, a potent inducer of the PTP [160], and is thought to be involved in transport of porphyrins into the mitochondrion [156,161]. Involvement of the PBR in PTP function was ini- tially suggested following biochemical isolation of the PBR, which indicated a close association of this pro- tein with VDAC and the ANT [162]. However, expres- sion of the 18 kDa protein alone in bacterial cells demonstrated that binding of high-affinity ligands to the PBR does not require either of these proteins [163]. Additional evidence that the PBR plays a role in PTP function was obtained following patch-clamp analysis of mitoplasts. Treatment with PBR ligands affected the channel activity of the MMC (i.e. the PTP) [164], but it is difficult to see how an inner membrane chan- nel could be affected by the outer membrane PBR receptor. Indeed, it appears unlikely that the patched membrane always contains outer membrane fragments. Experiments directly aimed at testing the activity of PBR ligands on PTP activity in isolated mitochondria have demonstrated that the effect of these drugs in promoting the opening of the PTP, and of apoptosis, depends on the cell type examined and the concentra- tions tested [132,155,160,165–167]. Other studies have shown that PBR ligands inhibit, rather than promote, the PTP and apoptosis. For example, rat forebrain mitochondria underwent PTP-dependent swelling and cytochrome c release following treatment with platelet activating factor. Here, swelling was inhibited by pretreatment of the mitochondria with Ro5-4864 or PK11195, as well as by treatment with CsA or the platelet activating factor inhibitor, BN50730 [168]. Also, rat heart mitochondria underwent a decrease in phosphorylation rate when treated with H 2 O 2 , but were restored to their normal rate when treated with Ro5-4864 [169]. While these apparently contradictory effects of PBR ligands can be rationalized on the basis of dif- ferences in cell type, additional confusion arises from studies showing that individual PBR ligands can have different effects on the same cell. For example, in U937 cells, Ro5-4864 counteracted TNF-a-mediated apoptosis, while PK11195, in a similar concentration range, enhanced apoptosis. Indeed, the addition of Ro5-4864 could overcome the effect of PK11195 [165]. This study also showed that Jurkat T cells, which contain little or no PBR, became more sensi- tive to cell death caused by TNF-a after they had been transfected with a gene expressing the PBR pro- tein. As would be predicted from the studies men- tioned earlier, Ro5-4864 protected the transfected cells from apoptosis [165]. While PBR-specific drugs have also been used to examine the role of the PBR in in vivo disease models [169–171], extension of these results, as well as many of those outlined above in the context of cultured cells and mitochondria, to PTP function depends, to some degree, on the ability of these drugs to specifically tar- get the PBR. Thus, recent results suggesting that PBR ligands can generate cellular phenotypes, independently of the PBR, have further muddled the physiological role of the PBR and its involvement in PTP activity [172–174]. Clearly, the examination of mitochondria prepared from animals in which the expression of PBR had been eliminated by ‘knockout’ strategies would greatly help to clarify these issues. Unfortunately, ini- tial attempts to generate such animals has indicated that nonconditional elimination of PBR expression results in embryonic lethality [159]. Thus, although it cannot be ruled out that the PBR is part of the PTP, conclusions based on the exclusive use of PBR ligands should, at this point, be viewed and interpreted with some caution until genetic tools are generated that will allow these questions to be addressed directly. P. Bernardi et al. The mitochondrial permeability transition FEBS Journal 273 (2006) 2077–2099 ª 2006 The Authors Journal compilation ª 2006 FEBS 2083 Role of CyP-D Opening of the PTP is inhibited by CsA after binding to CyP-D, a matrix peptidyl-prolyl cis-trans isomerase [175]. The relevant binding sites for CsA display a very high affinity, the estimated K d being between 5 and 8nm [175–177]. Early indications that CyP-D is involved in modulation of the PTP affinity for Ca 2+ (and conversely that Ca 2+ modulates the efficacy of PTP inhibition by CsA) included the demonstration that Ca 2+ displaced CsA from high-affinity binding sites in rat liver mitochondria [176] and that higher concentrations of CsA were required to inhibit spread- ing of the PTP to a population of mitochondria when the Ca 2+ load was increased [54]. The immunosuppressive effects of CsA are caused by the Ca 2+ -calmodulin-dependent inhibition of cal- cineurin (a cytosolic phosphatase) by the complex of the drug with cytosolic CyP-A [178]. In turn, this pre- vents dephosphorylation and nuclear translocation of nuclear factor of activated T cells and other transcrip- tion factors that are essential for the activation of T cells [179]. Available evidence suggests that calcineurin is not involved in the effects of CsA on the PTP because CsA derivatives have been described that bind CyP-D and desensitize the pore, but do not inhibit cal- cineurin [44,180,181]. On the other hand, calcineurin may affect mitochondrial function through the de- phosphorylation of BAD and the release of apopto- genic proteins [182]. The most conclusive results on the role of CyP-D in regulation of the PTP were obtained after inactivation of the Ppif gene, which encodes CyP-D in the mouse [46–49]. In three studies, the ablation of CyP-D approximately doubled the Ca 2+ -retention capacity (CRC) (i.e. the threshold Ca 2+ load required to open the PTP), which became identical to that of CsA-trea- ted, strain-matched wild-type mitochondria, while no effect of CsA was observed in Ppif – ⁄ – mitochondria [46,47,49]. Nakagawa et al. reported a much higher increase of the CRC after ablation of CyP-D, which matched an unusually high CRC after treatment of control mitochondria with CsA [48]. The basis for this discrepancy is not clear, but it may depend on the F2 mosaic mice used in this study. Indeed, it has been noted that a mixed genetic background can probably also account for many of the discrepancies already des- cribed for cytochrome c, APAF-1 and caspase 9 null mice [183]. Taken together, these findings demonstrate that CyP-D is a regulator, but not a component, of the PTP, whose structure is unlikely to be altered by the absence of CyP-D. A further implication of these results is that the effect of CsA is best described as ‘desensitization’ rather than inhibition, of the PTP, because its effects (similarly to the lack of CyP-D) can be overcome by a moderate increase of the mitochond- rial Ca 2+ load [54]. We would like to stress that, at variance from the conclusions of recent influential reviews [183,184], the in vivo studies on Ppif – ⁄ – mice can only be interpreted in terms of the role of CyP-D, not of the PTP, in cell death. Indeed, all studies agree that the PTP can form and open in the absence of CyP-D, provided that a permissive Ca 2+ load is accu- mulated [46–49]. Furthermore, these results cannot be used to conclude that the PTP only plays a role in nec- rotic, rather than apoptotic, responses [183,184] because it should not be surprising that matrix CyP-D does not play a role in cytochrome c release by tBID and BAX added to isolated mitochondria [46], a proto- col that directly permeabilizes the outer mitochondrial membrane. Thus, the inference that PTP opening does not take place because CyP-D is absent has not been documented in vivo, an issue that questions the conclu- sion that the PTP participates in cell death pathways only in response to a restricted set of challenges. Ppif – ⁄ – pups were born at the expected Mendelian ratio, and were otherwise indistiguishable from wild- type animals, suggesting that CyP-D is dispensible for embryonic development and viability of adult mice. The lack of an overt phenotype could be a result of adaptive responses bypassing the decreased sensitivity of the PTP to Ca 2+ (like the increased response to oxi- dative stress [47], or to effectors that are not detectable in isolated mitochondria). Furthermore, CyP-D overexpression desensitizes cells from apoptotic stimuli, indicating that CyP-D may play an additional role as a survival-signaling molecule acting on target(s) other than the PTP [185]. This dual function could lead to a balance of the pro-apoptotic and anti-apoptic effects of the protein in animals lacking CyP-D. Structure of the PTP As should be clear from the above analysis, to date, none of the candidate pore components stood rigorous genetic testing. He & Lemasters have proposed a model of PTP formation and gating in which the pore forms by aggregation of misfolded integral membrane proteins damaged by oxidant and other stresses [186], which is reminescent of an earlier model suggesting that the PT is not a consequence of the opening of a preformed pore, but rather the result of oxidative dam- age to membrane proteins [187]. In the model of He & Lemasters, conductance through these misfolded protein clusters would be normally blocked by chaper- The mitochondrial permeability transition P. Bernardi et al. 2084 FEBS Journal 273 (2006) 2077–2099 ª 2006 The Authors Journal compilation ª 2006 FEBS one-like proteins, including CyP-D, and it would be modulated by Ca 2+ in a CsA-sensitive manner. When protein clusters exceed chaperones available to block conductance, opening of ‘unregulated’ pores would occur, which would no longer be sensitive to CsA [186]. While interesting, the model fails to account for PTP regulation by the voltage and by matrix pH, which is not easy to reconcile with a permeability pathway created by a heterogeneous set of denatured proteins. Furthermore, Ro 68-3400 inhibits the PTP in the submicromolar range without affecting CyP-D activity, and under conditions of full pore inhibition it binds to a 32 kDa protein rather than to the large set of proteins that would be reasonable to find based on this model of the PTP [109]. Finally, it should be men- tioned that Ca 2+ -dependent, CsA-insensitive PT-like activities have been described that are formed or acti- vated by fatty acids [188,189] or 3-hydroxybuty- rate ⁄ polyphosphate [190]. Irrespective of the molecular nature of the pore, a comment is in order on the popular ideas that the pore (a) forms at ‘contact sites’ between the inner and outer mitochondrial membranes and (b) that it spans both membranes [191]. As discussed in the paragraphs on the ANT and VDAC, the idea that the PTP forms at contact sites is based on a set of assumptions rather than on established facts, and should be considered with great caution also because the very existence of points of fusion between the outer and inner mem- branes has been questioned by tomography of unfixed mitochondria [192]. The second point does not take into account that the permeability pathway resulting from a pore spanning both membranes would directly connect the matrix with the cytosol, resulting in the release of matrix solutes, but not of cytochrome c and of other intermembrane pro-apoptotic proteins. We would like to stress that the PT is primarily an inner membrane event that may cause secondary outer membrane changes, and that in an in vivo setting, the outer membrane can affect the probability of pore opening through protein–protein interactions, as exem- plified in the scheme of Fig. 1. Interaction of an outer membrane protein (e.g. VDAC or the PBR) with the PTP might depend on a specific conformation, which could be conferred by cytosolic regulator(s) after modi- fication by upstream signaling pathway(s). Binding would be followed by a conformational change that allows interaction with the PTP and its stabilization in the closed conformation (panel 1; note that the ligand- dependent change could instead favor the open confor- mation of the PTP and that intermembrane factors could also play a role, possibilities that are omitted for clarity). In the absence of outer membrane interac- tions, the PTP could flicker between the closed state (panel 2) and the open state (panel 3) under the effect of inner membrane and matrix modulators such as the Dw m , pH, CyP-D, Ca 2+ and PN. Stabilization of the open conformation could lead to the rearrangement of cristae structure and, eventually, to outer membrane rupture (panel 4). This scheme is meant as an example of how the outer membrane could confer regulatory features to the PTP without necessarily providing a permeability pathway for solute diffusion, but it should by no means be taken literally. As a matter of fact, and despite our detailed knowledge of PTP regu- lation, with the exception of CsA it is currently impossible to assign any pore effectors to a particular site, a key issue that will have to await PTP identifica- tion. We are developing new tools for the identification of pore component(s) through screening of chemical libraries, a program that is well underway and should soon provide novel clues about PTP structure and function. The PT in pathophysiology Occurrence of a PT has been amply documented in a variety of cell culture models. The number of such studies is so large that we must refer the reader to reviews covering both general and specific aspects of the problem [81–95]. We have already discussed the major sources of artifacts, particularly those arising from the use of fluorescent potentiometric probes [78]. Largely through the use of CsA, however, and more recently through the study of Ppif – ⁄ – mice, key sro t aluger cil o sotyC ) s( rotpada enarb mem re t u O P TP deso l C .m. o .m.i . m.o .m.i .m.o .m.i . m.o .m.i 1 2 3 4 P T P nepO Fig. 1. Model for permeability transition pore (PTP) regulation by outer membrane proteins. Hypothetical model of inner membrane (i.m.) PTP modulation by interaction with outer membrane (o.m.) proteins, which could be the target of cytosolic effector molecules. Broken lines denote outer membrane rupture following PTP open- ings of long duration. For explanation see the text. P. Bernardi et al. The mitochondrial permeability transition FEBS Journal 273 (2006) 2077–2099 ª 2006 The Authors Journal compilation ª 2006 FEBS 2085 advances have been made in understanding the role of CyP-D, and to some extent of the PTP, in organ and in vivo models of disease. In this section we will focus mostly on systems where an involvement of the PTP has been investigated in vivo. Myocardial ischemia-reperfusion The relevance of mitochondrial dysfunction to the onset of irreversible injury of the heart has prompted numerous studies aimed at defining the involvement of the PTP, especially in the setting of myocardial ische- mia-reperfusion. As already mentioned, the role of mitochondria in cell death extends beyond the shortage in ATP supply and involves the generation of ROS, the impairment of ion homeostasis, the accumulation and release of potentially harmful metabolites, and the release of pro-apoptotic proteins. These derangements are prominent in the heart where, paradoxically, mit- ochondrial function is also required for the sudden onset of cell death that occurs when coronary flow is re-established after a prolonged ischemic episode (i.e. postischemic reperfusion) [82,192,193]. Under these conditions, the partial recovery of mitochondrial func- tion generates an amount of ATP that is sufficient for contraction, but not for relaxation, resulting in hyper- contracture and sarcolemma rupture [4–6]. This dra- matic sequence of events is prevented by respiratory chain inhibitors or mitochondrial uncouplers [194,195], as well as by inhibition of ATP utilization by myosin ATPase [196]. Obviously, these findings are not directly amenable to clinical application, yet they indi- cate that the ischemic heart could be protected by pre- venting mitochondrial dysfunction. As first proposed at the end of the 1980s [197,198], and now documented by numerous reports [199,200], the PTP represents an ideal target for cardioprotection. While ischemia per se does not appear to cause PTP opening, probably because of the protective effects of intracellular acidosis [201,202], it creates the conditions for PTP opening at reperfusion. Indeed, recovery of respiration in the presence of increased intracellular [Ca 2+ ] and Pi provides an ideal scenario for promoting PTP opening, which would be further favored by the overproduction of ROS and the recovery of neutral pH. Initial supporting evidence that the PTP plays a role in reperfusion injury was obtained in experiments on isolated cardiomyocytes and perfused hearts, where CsA administration reduced the occurrence of con- tractile impairment and irreversible damage [203,204]. More direct evidence was subsequently obtained by methods allowing the detection of PTP opening in intact cells and tissues [71,112,205,206]. Occurrence of PTP opening was assessed through the (re)distribution of molecules that are not able to cross the inner mito- chondrial membrane unless a PT occurs. While calcein has been utilized to investigate the relationship between PTP and apoptosis in intact cells [78], the mitochondrial uptake of the otherwise impermeant 6-phosphodeoxyglucose provided an elegant demon- stration that PTP opening occurs in isolated hearts only upon postischemic reperfusion [71,82]. We recently assessed PTP opening in heart ischemia- reperfusion through the redistribution of an endog- enous ‘probe’, the pool of mitochondrial PN, which does not readily permeate the inner membrane unless a PT occurs [16]. We demonstrated that in isolated hearts subjected to ischemia-reperfusion, PN are released from the mitochondrial matrix into the inter- membrane and the cytosolic spaces, where they become the substrate of a wide array of NAD + util- izing enzymes [206]. The release of NAD + is not only a tool for PTP detection. Formation of cyclic nucleo- tides (such as cADP ribose) from NAD + may further increase the probability of PTP opening through the release of Ca 2+ from sarcoplasmic reticulum, and the released NAD + can be utilized by poly(ADP-ribose) polymerase (PARP) for DNA repair [207]. It has been proposed that uncontrolled PARP acti- vation might deplete intracellular NAD + and ATP, resulting in mitochondrial depolarization and eventu- ally cell death [208,209]. As the major fraction of PN is compartmentalized within the mitochondrial matrix, we think that pore opening must precede rather than follow PARP activation. Consistent with our view, compounds and procedures that cause DNA damage and PARP activation are also powerful PTP agonists. We have recently demonstrated that N-methyl- N¢-nitro-N-nitrosoguanidine, the reference compound utilized for activating PARP, causes PTP opening and cell death that are prevented by CsA [210]. PARP acti- vation may certainly hasten the progression towards cell death by rapidly consuming the NAD + released by mitochondria, yet it is unlikely to be a primary cause of mitochondrial dysfunction. The results obtained with both the deoxyglucose and NAD + distribution techniques demonstrate that myocyte viability is maintained when PTP opening is prevented [71,206]. This causal relationship between PTP-dependent mitochondrial dysfunction and cell death rules out the alternative possibility that PTP opening may be a secondary consequence of the mas- sive intracellular Ca 2+ overload that follows sarcolem- ma rupture. The relevance of these concepts to a clinical setting has been substantially strengthened by The mitochondrial permeability transition P. Bernardi et al. 2086 FEBS Journal 273 (2006) 2077–2099 ª 2006 The Authors Journal compilation ª 2006 FEBS [...]... relative to the hepatotoxic treatment ranged from pretreatment with a single dose to repeated administra- The mitochondrial permeability transition tions of the same dose at different time intervals The variability of dose and timing reflects the basic uncertainty of whether the PTP is actually inhibited after the administration of CsA in vivo, and of whether more than a single dose is necessary for the inhibitory... (2002) Mitochondrial permeability transition as a novel principle of hepatorenal toxicity in vivo Apoptosis 7, 395–405 230 Masubuchi Y, Suda C & Horie T (2005) Involvement of mitochondrial permeability transition in acetaminophen-induced liver injury in mice J Hepatol 42, 110– 116 231 Hirakawa A, Takeyama N, Nakatani T & Tanaka T (2003) Mitochondrial permeability transition and cytochrome c release in. .. regulation of the mitochondrial permeability transition pore; a cyclosporin-sensitive pore in the inner mitochondrial membrane J Bioenerg Biomembr 26, 509–517 The mitochondrial permeability transition 65 Crompton M & Costi A (1988) Kinetic evidence for a heart mitochondrial pore activated by Ca2+, inorganic phosphate and oxidative stress A potential mechanism for mitochondrial dysfunction during cellular... regulatory mechanism in mitochondrial function Biochem J 358, 147–155 Gincel D & Shoshan-Barmatz V (2004) Glutamate interacts with VDAC and modulates opening of the mitochondrial permeability transition pore J Bioenerg Biomembr 36, 179–186 Pastorino JG, Shulga N & Hoek JB (2002) Mitochondrial binding of hexokinase II inhibits Bax-induced cytochrome c release and apoptosis J Biol Chem 277, 7610–7618 Pastorino... for the complex toxic and mutagenic effects of AAF These include 2-nitrosofluorene, a redox cycling compound that drains electrons from the respiratory chain [236] and causes opening of the PTP in isolated mitochondria [237] Our studies have shown that very early into AAF feeding an adaptative response takes place, which desensitizes the PTP and the liver mitochondrial apoptotic pathway to TNF-a in vivo... benzodiazepine receptor (PBR) ligand cytotoxicity unrelated to PBR expression Biochem Pharmacol 69, 819–830 175 Halestrap AP & Davidson AM (1990) Inhibition of Ca2+-induced large-amplitude swelling of liver and heart mitochondria by cyclosporin is probably caused by the inhibitor binding to mitochondrial- matrix peptidyl-prolyl cis-trans isomerase and preventing it interacting with the adenine nucleotide... Fluctuations in mito¨ chondrial membrane potential caused by repetitive gating of the permeability transition pore Biochem J 343 Part 2, 311–317 112 Petronilli V, Miotto G, Canton M, Brini M, Colonna R, Bernardi P & Di Lisa F (1999) Transient and longlasting openings of the mitochondrial permeability transition pore can be monitored directly in intact cells by changes in mitochondrial calcein fluorescence... is an effector mechanism that can explain the involvement of mitochondria in many pathological conditions and high-prevalence diseases CsA proved essential for the development of the field and for testing the role of CyP-D (and to some extent of the PTP) in pathophysiology in vivo However, the demonstration that a PT can occur in the absence of CyP-D must induce some caution in interpreting experimental... Postconditioning inhibits mitochondrial permeability transition Circulation 111, 194–197 Hausenloy DJ, Maddock HL, Baxter GF & Yellon DM (2002) Inhibiting mitochondrial permeability transition pore opening: a new paradigm for myocardial preconditioning? Cardiovasc Res 55, 534–543 The mitochondrial permeability transition 213 Clarke SJ, McStay GP & Halestrap AP (2002) Sanglifehrin A acts as a potent inhibitor... necessary for the inhibitory effect to persist We systematically investigated this problem in the rat, and found that the PTP is maximally inhibited in the liver between 2 and 9 h of intraperitoneal (i.p.) injection of 5 mg CsAÆkg)1 body weight The inhibitory effect returned to the basal level within 24 h [223] By using proper times of treatment we achieved full protection from the otherwise lethal effects . a rate-limiting step in steroid synthesis [159]. The PBR also binds some porphyrins, including protoporphyrin IX, a potent inducer of the PTP [160], and is thought to be involved in transport of porphyrins. protein located in the outer mitochondrial membrane [153] and was initially identified as a binding site for ben- zodiazepines in tissues that lack 4-aminobutyrate receptors, the clinical target. interest in the PT as an effector mechanism of cell death, how- ever, only followed the demonstration that in the course of apoptosis, cytochrome c is released into the cytosol [72], together with