• Open Access

Mitochondria and aging: a role for the permeability transition?


Professor Martin Crompton, Department of Biochemistry and Molecular Biology, University College London, Gower St., London WC1E 6BT, UK. Tel.: +44 20 76792251; fax: +44 20 76797193; e-mail: m.crompton@biochemistry.ucl.ac.uk


When mitochondria are subjected to oxidative stress and relatively high [Ca2+], they undergo a ‘permeability transition’ in which the inner membrane becomes freely permeable to low-molecular-weight solutes. This phenomenon reflects reversible deformation of the adenine nucleotide translocase, the loss of its native gating properties and the stabilization of the deformed state by cyclophilin-D. The permeability transition may be a factor in cell dysfunction associated with aging. This can manifest in a number of ways ranging, in the most severe, from impaired energy transduction and compromised viability to more subtle influences on the propagation of Ca2+ signals. This article critically examines data relevant to this issue.

Oxidative stress (Ames et al., 1993; Beal, 1995) and impaired Ca2+ homoeostasis (Khachaturian, 1994; Thibault et al., 1998; Verkhratsky & Toescu, 1998) are believed to be important factors in the degenerative processes associated with aging. Although there is a wide range of literature on each of these topics, clues to the nature of things may often be found at the interface, and this article is restricted to one small aspect of the interface between them. Ca2+ metabolism and oxidative stress are interlinked, and the link is most evident in mitochondria. Mitochondria not only contribute to Ca2+ metabolism and are the major source of reactive oxygen species (ROS), but can also become severely dysfunctional via the permeability transition under the synergistic effects of oxidative stress and deregulated cytosolic free Ca2+. It is instructive then to consider whether the permeability transition is at all relevant to the aging process. This article is prompted by a collection of key papers relevant to this issue.

If one were to pick a pointer to the general ‘health’ of mitochondria in situ, then a high inner membrane potential (ΔψM) would seem a reasonable criterion. ΔψM is the main driving force for mitochondrial ATP synthesis and export to the cytosol, and a diminished ΔψM would be expected to have a significant bearing on cell viability and capacity to withstand stress. In fact, there are a number of reports that old cells display low ΔψM, as measured by uptake of fluorescent cationic probes, e.g. lymphocytes (Mather & Rottenberg, 2002), neurons (Xiong et al., 2002) and hepatocytes (Hagen et al., 1997). ΔψM results from respiratory chain activity, consumption of the proton motive force by ATP synthesis and other mitochondrial energy-dependent processes, and ionic leaks; in principle, an age-related deterioration in any of these may manifest as low ΔψM. Indeed, age-related oxidative damage to the proteins of the electron transport chain and to mitochondrial DNA encoding these proteins is documented (Beal, 1995; Wallace et al., 1995). Intriguingly, however, Mather & Rottenberg (2002) report that the disparity between ΔψM in populations of old and young lymphocytes is abolished in the presence of cyclosporin A (CSA). This is a stunning observation, because it cuts through numerous possible lines of reasoning and experimentation, and focuses attention on a CSA binding protein as a key player in age-related mitochondrial dysfunction.

First, however, are fluorescent indicators a reliable means of estimating age-related changes in ΔψM? The p-glycoprotein, which pumps foreign, hydrophobic substances from cells, is blocked by CSA, and is expressed more in old lymphocytes (Eisenbraun & Miller, 1999). Because rhodamine and other fluorescent indicators of ΔψM are also substrates of the p-glycoprotein, enhanced activity of this protein in old cells could lead to an underestimate of ΔψM. Yet the difference in rhodamine uptake by old and young lymphocytes was not significantly altered by verapamil, another p-glycoprotein inhibitor. In addition, the capacity of isolated mitochondria (liver, brain) to accumulate Ca2+ (ΔψM-dependent) deteriorated with age of the animal and, again, the age-related disparity was attenuated by CSA (Mather & Rottenberg, 2000). These data draw attention to mitochondrial cyclophilin-D as a possible age-related CSA target.

Cyclophilin-D is a component of the permeability transition (PT) pore. The basic molecular aspects of the PT are now clear. Under conditions of high intramitochondrial Ca2+ and oxidative stress, the adenine nucleotide translocase (ANT) in the mitochondrial inner membrane deforms from its native state as a gated pore (mediating ADP/ATP exchange) into a non-selective pore, allowing free permeation of small ions and metabolites across the inner membrane. ANT deformation is catalysed and/or stabilized by cyclophilin-D in the mitochondrial matrix. This action of cyclophilin-D is blocked by its ligand CSA so that, in the presence of CSA, the native gating properties of ANT are preserved even in the face of Ca2+ overload and oxidative stress. Destabilization of ANT into the PT pore is a pathological phenomenon, which uncouples oxidative phosphorylation and collapses ΔψM, so that mitochondria hydrolyse ATP rather than synthesize it. Enduring PT pore formation has been linked to lethal cell injury associated with ischaemia and reperfusion (reviewed in Crompton, 1999). However, transient (< 1 s) PT pore formation, restricted to a few mitochondria per cell, has been detected in viable cells under oxidative stress, and is evidently tolerated (Jacobson & Duchen, 2002).

How might aging increase the likelihood of ANT deforming into the PT pore? ANT contains three cysteine residues on loops projecting into the matrix compartment. These are susceptible to oxidation if the high intramitochondrial GSH/GSSG ratio is not maintained during periods of oxidative stress (Hashimoto et al., 1999, and references therein). Thus increased ROS production with age may be an important factor in ANT destabilization. Mather & Rottenberg (2002), for example, found that superoxide production was two- to three-fold higher in lymphocytes from senescent mice than in those from adults. In addition, ROS oxidize polyunsaturated fatty acids to generate protein-reactive fragments. Lucas & Szweda (1998) investigated the modification of mitochondrial proteins by 4-hydroxy-2-nonenal (HNE), a major product of lipid peroxidation, during the reperfusion of hearts from 8- and 24-month-old rats. HNE formed adducts with two proteins in the older, but not the younger, heart, and one of the adducts had a similar molecular weight to ANT (30 kDa). Unfortunately, the identity of this protein was not investigated. Nevertheless, senescent hearts are evidently more vulnerable to protein modification by peroxidized lipids, and the study may be considered alongside others showing that ANT is particularly liable to this type of modification (Giron-Calle & Schmid, 1996; Brustovetsky et al., 2002). The susceptibility of ANT possibly reflects its close association with cardiolipin. Purified ANT has six or more cardiolipins tightly bound. Their removal inactivates the protein, but activity is restored when cardiolipin is re-added (Hoffman et al., 1994). It is believed that the negatively charged head groups of cardiolipin in the inner leaflet of the inner membrane interact electrostatically with lysine side chains of ANT and, in this way, stabilize its native conformation as a gated pore. Cardiolipin contains a relatively high proportion of polyunsaturated fatty acids, so that ROS would be expected to generate reactive lipid fragments in the immediate vicinity of ANT, thereby facilitating adduct formation with the lysine side chains. These reactions, vic-thiol oxidation and adduct formation, provide two possible mechanisms whereby oxidative stress might lead to PT pore formation in old animals.

In broad agreement, Jahangir et al. (2001) found that heart mitochondria isolated from senescent rats are less able to accumulate added Ca2+ than those isolated from adults, and that they are more susceptible to Ca2+-induced swelling. These characteristics point to enhanced PT pore activation in senescent heart mitochondria, although the inhibitory capacity of CSA was not tested. Evidently, the factor(s) that bring about dysfunctional behaviour in mitochondria from aged tissue are retained after mitochondrial isolation. This may well be increased ROS production; endogenous ROS production is certainly one determinant of the PT, because hyperoxia, which increases superoxide production, promotes Ca2+-induced PT in isolated mitochondria (Crompton & Andreeva, 1993). But this needs to be tested. In addition, analyses of mitochondrial phospholipids by Pepe et al. (1999) revealed a major age-related change in cardiolipin, which decreased by 30% in senescence. Because cardiolipin stabilizes the native conformation of ANT, the losses of cardiolipin in senescence may also be significant in terms of ANT stability. The finding (discussed below) that mitochondrial cardiolipin content can be manipulated according to diet might also allow this to be tested in isolated mitochondria.

Ca2+ is the key trigger for PT pore formation, but exactly how Ca2+ acts is unresolved. Possibly, accumulated Ca2+ binds to the negatively charged cardiolipin head groups on the inner face of the inner membrane, thereby disrupting the stabilizing interactions between cardiolipin and ANT (Hoffman et al., 1994; Brustovetsky et al., 2002). But does mitochondrial Ca2+ change in senescence? In fact, in neurons, mitochondrial Ca2+ appears to decrease with age (discussed below). In hearts (normoxic), Pepe et al. (1999) observed no change in endogenous mitochondrial Ca2+ in senescence. However, heart mitochondrial Ca2+ increased substantially (> three-fold) following ischaemia and reperfusion, and this increase was about two-fold greater in senescence. Taken together with the evidence for increased lipid peroxidation (HNE formation, above) in senescent hearts on reperfusion, it seems that aging hearts are increasingly vulnerable to reperfusion damage brought about by increased mitochondrial Ca2+ and oxidative stress, the twin triggers for PT pore formation (discussed in Pepe, 2000).

Pepe et al. (1999) also reported that senescence brought about a decrease in n-3 polyunsaturated fatty acids (PUFA) in mitochondrial phospholipids, and an increase in n-6 PUFA. Moreover, mitochondrial Ca2+ levels could be manipulated by feeding PUFA. Dietary n-6 PUFA raised mitochondrial Ca2+, whereas n-3 PUFA decreased it. These dietary- and age-related increases in mitochondrial Ca2+ were very largely abolished by perfusion with ruthenium red, which blocks mitochondrial Ca2+ uptake. Using these interventions it was found that the efficiency of working hearts (work performed per unit of O2 consumed) was inversely related to mitochondrial Ca2+. Thus, although there are probably multiple contributors to increased energy dissipation in n-6 PUFA-fed hearts, mitochondrial Ca2+ seems to be particularly significant in some way. It was suggested that an increased rate of Ca2+ cycling across the mitochondrial inner membrane via the uniporter (influx) and Na+–Ca2+ carrier (efflux) might account for the increased energy dissipation. Certainly, the observed increase in mitochondrial Ca2+ is consistent with increased uniporter activity, and, because the rate of Ca2+ cycling is fixed by the activity of the uniporter (discussed in Crompton, 1999), there may well be an increase in mitochondrial Ca2+ cycling. However, calculations based on the maximal rates of mitochondrial Ca2+ cycling in isolated heart mitochondria suggest that Ca2+ cycling probably contributes negligibly to energy dissipation in the heart, even under pathological conditions of high intracellular Ca2+ (ischaemia/reperfusion; Crompton, 1999). Whether the lowered efficiency associated with increased mitochondrial Ca2+ reflects a degree of PT pore opening could be tested. This is particularly so because n-6 PUFA feeding brought about a decrease in cardiolipin, which stabilizes ANT (see above). Experiments with CSA to block pore opening might help to resolve this question.

From one aspect, mitochondria seem to play only a minor role in the control of cytosolic Ca2+. Whereas Ca2+ fluxes into and out of the cytosol across the plasma membrane and membranes of the endo/sarcoplasmic reticulum are tightly regulated co-ordinated events, no physiological regulators of mitochondrial Ca2+ transport have yet been identified. In the absence of any such regulators, mitochondrial [Ca2+] will be controlled simply by the profile of change of [Ca2+] in the cytosol as established by the plasma membrane and endo/sarcoplasmic reticulum. Nevertheless, even such a passive, secondary role may exert subtle influences on Ca2+ metabolism. Although most mitochondria will respond to global changes in cytosolic [Ca2+], others lie close to the Ca2+ release channels of the endo/sarcoplasmic reticulum (Rapizzi et al., 2002) and to the store-operated channels (SOC) of the plasma membrane (Parekh, 2003). Mitochondria juxtaposed to these channels experience high localized [Ca2+] on cell activation and accumulate relatively high levels of Ca2+, at least transiently. There is evidence, from imaging ΔψM of individual mitochondria in situ, that channel-juxtaposed mitochondria are more susceptible to the PT, presumably as a result of the increased [Ca2+] (Jacobson & Duchen, 2002). The question posed here is what impact this may have on Ca2+ signalling, in particular in aged cells, which may be more disposed to the PT.

Ca2+ influx via plasma membrane SOC is triggered by the loss of Ca2+ from adjacent endoplasmic reticulum (ER) and is subject to feedback inhibition by cytosolic Ca2+. It appears that Ca2+ uptake by mitochondria located very close to these events promotes SOC activity in two ways: first, by partially relieving Ca2+-dependent slow inactivation of the channels and, second, by competing with the ER Ca2+ pumps and thereby prolonging the loss of Ca2+ from the ER (Parekh, 2003). For example, depletion of ER Ca2+ stores by thapsigargin-inhibition of the ER Ca2+ pump or by receptor stimulation yields a sustained elevation of cytosolic Ca2+, but the rise is attenuated by inhibition (antimycin, a plus oligomycin) or uncoupling (CCCP) of mitochondrial energy transduction and by inhibition (ruthenium red) of mitochondrial Ca2+ uptake (Gilabert & Parekh, 2000; Hoth et al., 2000) (in these experiments, [ATP] was maintained glycolytically or via the recording pipettes). It has been reported (Morgan & Jacob, 1994) that low concentrations of the Ca2+ ionophore ionomycin, insufficient to conduct significant amounts of Ca2+ across the plasma membrane directly, also stimulate SOC-mediated Ca2+ entry. Using this experimental system, Mather & Rottenberg (2002) report that the fraction of lymphocytes with high cytosolic free [Ca2+] after ionomycin decreased with age from 75% (3-month-old mice) to < 25% (26-month-old mice). As this decrease correlated with an increased percentage of cells with low ΔψM it was attributed to impaired Ca2+ uptake by SOC-juxtaposed mitochondria in aged cells, owing to enhanced PT pore activity. In agreement, CSA reversed the age-related defects in both ΔψM and Ca2+ signalling. This is the first report on the possible consequences of age-related PT pore opening on the intracellular transmission of Ca2+ signals. It is important to extend these observations to other systems using physiological (rather than ionomycin) means of cell activation. Ca2+ wave propagation within cells relies on Ca2+ (cytosolic)-dependent activation of ryanodine and IP3 receptors in sarco/endoplasmic reticulum, so that mitochondria juxtaposed to these channels would be expected to slow the propagation of Ca2+ waves, which they do (e.g. Boitier et al., 1999). In other cell types, however, mitochondria appear to promote Ca2+ wave progression, possibly by relieving Ca2+-inactivation of type 2 IP3 receptors at high [Ca2+] (Simpson & Russell, 1998). Thus, any consequences of localized PT pore opening on Ca2+ wave progression may depend on the dominant receptor isoforms in that particular cell type.

Neuronal Ca2+ homoeostatic mechanisms also deteriorate with age. In particular, the active transport systems used to extract Ca2+ from the cytosol decline in activity, i.e. the Na+–Ca2+ antiporter and the Ca2+ ATPase of the plasma membrane, the ER Ca2+ ATPase (Satrustegui et al., 1996; Pottorf et al., 2000, and references therein). It is perhaps no surprise, then, that the capacity of aged neurons to maintain low cytosolic [Ca2+] and/or to restore low cytosolic [Ca2+] following depolarization-induced Ca2+ transients is frequently impaired (e.g. Villalba et al., 1995; Buchholz et al., 1996). As argued by Buchholz et al. (1996), this will place greater reliance on mitochondria in shaping the Ca2+ transients. In broad support, when the uncoupler DNP was used to limit mitochondrial Ca2+ uptake, the transient amplitude was increased in senescent cells (superior cervical ganglion), but not in adults.

In spite of the tendency towards higher than normal cytosolic free [Ca2+] in aged neuronal cells, there is evidence for diminished Ca2+ compartmentation in mitochondria, at least in certain types, e.g. hippocampal neurons (Villalba et al., 1995; Satrustegui et al., 1996). Because increased cytosolic Ca2+ would normally result in increased mitochondrial Ca2+, this inverse behaviour points to impaired Ca2+ uptake in aged neuronal cells. Satrustegui et al. (1996) report a decline in the Ca2+ uniporter activity in aged neuronal mitochondria. However, Ca2+ conductance measurements are needed to discriminate between decreased amounts of uniporter (protein) and decreased driving force for Ca2+ accumulation (ΔψM). Indeed, an age-related decline in ΔψM has also been reported in neurons, as in other cells. Thus, Xiong et al. (2002) observed low ΔψM in resting cerebellar granule neurons and, in particular, following Ca2+ transients. Inadequate Ca2+ activation of mitochondrial respiration was suggested, as might indeed result, for example, from a reduced amount of uniporter. It would be interesting to know whether PT pore opening makes any contribution to low ΔψM as, evidence suggests, it does in other aging cells. Again experiments with CSA may help to resolve this question.