The permeability transition pore as a Ca²⁺ release channel: New answers to an old question

Abstract

Mitochondria possess a sophisticated array of Ca²⁺ transport systems reflecting their key role in physiological Ca²⁺ homeostasis. With the exception of most yeast strains, energized organelles are endowed with a very fast and efficient mechanism for Ca²⁺ uptake, the ruthenium red (RR)-sensitive mitochondrial Ca²⁺ uniporter (MCU); and one main mechanism for Ca²⁺ release, the RR-insensitive 3Na⁺–Ca²⁺ antiporter. An additional mechanism for Ca²⁺ release is provided by a Na⁺ and RR-insensitive release mechanism, the putative 3H⁺–Ca²⁺ antiporter. A potential kinetic imbalance is present, however, because the V_max of the MCU is of the order of 1400 nmol Ca²⁺ mg⁻¹ protein min⁻¹ while the combined V_max of the efflux pathways is about 20 nmol Ca²⁺ mg⁻¹ protein min⁻¹. This arrangement exposes mitochondria to the hazards of Ca²⁺ overload when the rate of Ca²⁺ uptake exceeds that of the combined efflux pathways, e.g. for sharp increases of cytosolic [Ca²⁺]. In this short review we discuss the hypothesis that transient opening of the Ca²⁺-dependent permeability transition pore may provide mitocondria with a fast Ca²⁺ release channel preventing Ca²⁺ overload. We also address the relevance of a mitochondrial Ca²⁺ release channel recently discovered in Drosophila melanogaster, which possesses intermediate features between the permeability transition pore of yeast and mammals.

1. Mitochondria have a large Ca²⁺ problem

In energized mitochondria Ca²⁺ uptake is an electrophoretic process driven by the Ca²⁺ electrochemical gradient, $\Delta \tilde{\mu }\text{Ca}$:

$$\Delta \tilde{\mu }\text{Ca}=zF\Delta \Psi +RT\ln \frac{{[{\text{Ca}}^{2+}]}_i}{{[{\text{Ca}}^{\text{2+}}]}_{\text{o}}}$$ (1)

In respiring, coupled mitochondria the Δψ (negative inside) drives uptake of Ca²⁺, which is transported with a net charge of 2 [1,2] via an inner membrane channel [3], the mitochondrial Ca²⁺ uniporter, MCU [4,5]. Ca²⁺ uptake is charge-compensated by increased H⁺ pumping by the respiratory chain [1,2], resulting in increased matrix pH that prevents the recovery of Δψ, limiting the further ability to accumulate Ca²⁺ [6–8]. Uptake of substantial amounts of Ca²⁺ therefore requires both buffering of matrix pH (to allow regeneration of the Δψ); and buffering of matrix Ca²⁺ (to prevent the buildup of a Ca²⁺ concentration gradient) [6–8]. Buffering of matrix pH is achieved by the simultaneous uptake of protons and anions via diffusion of the undissociated acid through the inner membrane (as in the case of acetate), of CO₂ (which then regenerates bicarbonate and H⁺ in the matrix) or through transport proteins (like the H⁺–Pi symporter) [9]. Buffering of accumulated Ca²⁺ (and therefore the final [Ca²⁺] in the matrix) thus depends in part on the cotransported anion and in part on remarkably ill-characterized matrix constituents. If Pi is the prevailing anion, free matrix [Ca²⁺] becomes invariant with the matrix Ca²⁺ load [10] and the $\Delta \tilde{\mu }\text{Ca}$ favors the accumulation of large loads of both Ca²⁺ and Pi [11], with a predicted Ca²⁺ equilibrium accumulation of 10⁶ if the Δψ is −180 mV [6]. This is never reached because at resting cytosolic Ca²⁺ levels the rate of Ca²⁺ uptake is comparable to that of the efflux pathways, and Ca²⁺ distribution is governed by a kinetic steady state rather than by thermodynamic equilibrium [6,7]. Thus, in energized mitochondria coupling of Ca²⁺ uptake with Ca²⁺ efflux on separate pathways allows regulation of both cytosolic and matrix [Ca²⁺]. Energy is required both for Ca²⁺ uptake and for Ca²⁺ release, owing to the electrophoretic nature of transport on MCU and the 3Na⁺–1Ca²⁺ stoichiometry of NCLX [12], which dissipates the Δψ; the requirement for H⁺ extrusion posed by operation of the Na⁺–H⁺ exchanger (NHE) coupled to NCLX; and the fact that Ca²⁺ efflux on the putative H⁺–Ca²⁺ exchanger is favored by the Δψ [13,14], suggesting a H⁺–Ca²⁺ stoichiometry higher than 2H⁺ per Ca²⁺. A kinetic imbalance is apparent, however, because the V_max of the MCU is of the order of 1400 nmol Ca²⁺ mg⁻¹ protein min⁻¹ while the combined V_max of the efflux pathways is about 20 nmol Ca²⁺ mg⁻¹ protein min⁻¹. This arrangement exposes mitochondria to the hazards of Ca²⁺ overload when the rate of Ca²⁺ uptake exceeds that of the combined efflux pathways, e.g. for sharp increases of cytosolic [Ca²⁺]. Why is then the rate of Ca²⁺ efflux so slow?

The rate of Ca²⁺ uptake via the MCU is a steep function of extramitochondrial [Ca²⁺] [15]. Increasing rates of Ca²⁺ efflux would increase extramitochondrial Ca²⁺, stimulate Ca²⁺ uptake via MCU and increase overall Ca²⁺ cycling, resulting in energy dissipation [16]. This can be observed by adding the electroneutral 2H⁺–Ca²⁺ ionophore A23187 to respiring mitochondria that have accumulated Ca²⁺, a condition where Ca²⁺ is released and all of the respiratory capacity can be diverted into Ca²⁺ cycling [17]. Thus (and as long as the membrane potential is high) net Ca²⁺ efflux through stimulation of the efflux pathways would have a high energetic cost. The low V_max and early saturation of the efflux pathways by matrix Ca²⁺ are probably designed to pose an upper limit to the energy that can be spent in regulation of matrix and cytosolic [Ca²⁺] through mitochondrial “Ca²⁺ cycling”. As mentioned above, however, this situation exposes mitochondria to the constant threat of Ca²⁺ overload. We have proposed that this event may be prevented by transient openings of the permeability transition pore (PTP), which does mediate mitochondrial depolarization and fast Ca²⁺ release in vitro and possibly in vivo [18].

2. Properties of the permeability transition pore in mammals

The PTP is a high-conductance (≈1.3 nS) inner membrane channel [19–21]. In the fully open state its apparent diameter is about 3 nm [22] and its solute exclusion size ≈ 1500 Da. The PTP open-closed transitions are regulated by physiological effectors, and the consequences of pore opening vary dramatically depending on the open time [16]. Here we shall cover only the basic features of PTP modulation, and we refer the Reader to previous reviews for more information on the rise of the PTP from the status of in vitro artifact to that of effector mechanism of cell death regulated by key signaling cascades [23–25].

Matrix modulators of the PTP include Ca²⁺ through a “permissive” site for opening that can be competitively inhibited by other Me²⁺ ions like Mg²⁺, Sr²⁺ and Mn²⁺; and Pi, which in most species acts as a powerful PTP inducer through a still undefined mechanism. Pore opening is promoted by an oxidized state of pyridine nucleotides and of critical dithiols at discrete sites, both effects being individually reversed by proper reductants [26]. Pore opening can also cause production of reactive oxygen species, as shown by the occurrence of “superoxide flashes” triggered by transient openings of the PTP in cardiomyocytes [27]. The permeability transition is strictly modulated by matrix pH with an optimum at pH 7.4, while the open probability decreases both below pH 7.4 (through reversible protonation of critical histidyl residues [28,29]) and above pH 7.4 (through an unknown mechanism). Opening of the PTP is inhibited by cyclosporin (Cs) A after binding of the latter to cyclophilin (CyP) D, a matrix peptidyl-prolyl cis-trans isomerase encoded by the Ppif gene that facilitates PTP opening [30–32]; indeed, ablation of CyPD approximately doubles the threshold Ca²⁺ load required to open the PTP, which becomes identical to that of CsA-treated, strain-matched wild type mitochondria, while no effect of CsA is observed in CyPD-null mitochondria [33–36]. Major membrane effectors are the inside-negative Δψ_m, which tends to stabilize the PTP in the closed conformation [28]; and electron flux within respiratory chain complex I, with an increased open probability when flux increases [37]. The latter finding led to the discovery that the PT is regulated by quinones [38], possibly through a specific binding site whose occupancy affects the open-closed transitions depending on the bound species [39].

The only primary consequence of PTP opening is mitochondrial depolarization. Unless single channel events are being recorded openings of short duration may not be detected by potentiometric probes. Since opening events are not synchronized for individual mitochondria [40,41] in population studies they may also be missed due to probe redistribution. We refer the reader to a series of specific studies about the occurrence of PTP openings of different durations in mitochondria in situ, and their consequences on cell viability [40–44]. For openings of longer duration depolarization can be easily measured both in isolated mitochondria and intact cells. As long as the pore is open collapse of the $\Delta \tilde{\mu }\text{H}$ prevents ATP synthesis, and ATP hydrolysis by the mitochondrial ATPase worsens ATP depletion, which together with altered Ca²⁺ homeostasis is a key factor in various paradigms of cell death [45]. Persistent PTP opening is also followed by loss of matrix pyridine nucleotides with respiratory inhibition [46], by equilibration of ion gradients, and by diffusion of solutes with molecular masses lower than about 1500 Da and possible occurrence of swelling, cristae unfolding and outer membrane rupture. This, however, is not an inevitable consequence of PTP opening. Cristae remodeling due to PTP opening can also occur in the absence of outer membrane rupture [47] with mobilization of the large pool of cytochrome c usually compartmentalized in the intracristal spaces [48] allowing increased release of cytochrome c through BAX/BAK channels in an otherwise intact outer membrane [47]. Furthermore, osmotic swelling requires the existence of a colloidosmotic gradient between the matrix and the intermembrane space that may not exist (or may not be large enough) to cause outer membrane rupture in situ.

The possible role of the PTP in physiological Ca²⁺ homeostasis has not been studied thoroughly, in part because the nature of the PTP has remained elusive [23] and therefore modulation of the “channel” itself by genetic methods has not been possible; in part because true PTP blockers have not been developed. Great hopes were raised by the discovery that the PTP is inhibited by CsA [49–51], but it is now clear that CyPD is a key modulator but not an obligatory component of the PTP. It should therefore be borne in mind that Ppif^−/− (CyPD-null) mice are not PTP-null mice, as a permeability transition can still occur, e.g. after an increased Ca²⁺ load [33–36]. This consideration has dramatic implications for the interpretation of results obtained with CsA; similar to the absence of CyPD, CsA can desensitize but not block the PTP, and therefore lack of sensitivity to CsA does not necessarily mean that the PTP is not involved in the event under study. Little attention has also been paid to the fact that expression of CyPD can be modulated (e.g. by muscle denervation [52]), and that only CyPD-expressing mitochondria are expected to respond to CsA. A further element of complexity is that CsA also affects mitochondria in situ by preventing calcineurin-dependent dephosphorylation of the pro-fission protein DRP1, which is essential for its translocation to mitochondria [53]. The resulting inhibition of mitochondrial fission by CsA may thus be a cytoprotective event that is independent of inhibition of CyPD and desensitization of the PTP [53], a finding that should induce some caution in evaluating in situ and in vivo studies based only on CsA rather than on genetic ablation of CyPD or on the use of CyP inhibitors devoid of effects on calcineurin [54–56].

3. The mitochondrial permeability transition pore as a Ca²⁺ release channel: is lack of selectivity a problem or an advantage?

If a Ca²⁺ concentration gradient exists between the matrix and the external medium (or the cytosol) PTP opening leads to Ca²⁺ release. In 1996 we have proposed that the PTP may serve as a mitochondrial Ca²⁺ release channel [18], and a specific point that needs to be discussed is whether the lack of selectivity is a real problem, or rather an essential feature that allows fast and effective release of matrix Ca²⁺. Ca²⁺ efflux down its concentration gradient via a Ca²⁺-selective channel would be opposed by the buildup of a Ca²⁺ diffusion potential [57]. According to the Nernst equation, the magnitude of the Ca²⁺ diffusion potential when no charge-compensating species are present is −30 mV per decade of the matrix/cytosol Ca²⁺ concentration ratio. This diffusion potential is reduced to less negative values (with corresponding increase of the Ca²⁺ efflux rate) by any charge-compensating current (influx of positive charges, efflux of negative charges, or both). Since the inner membrane has a very low permeability to charged species, the rate of Ca²⁺ efflux would be extremely slow and essentially limited by the H⁺ permeability. In other words, to obtain a significant rate of Ca²⁺ efflux via a Ca²⁺-selective channel the inner membrane permeability should be increased as well. An unselective pore of large size like the PTP confers the advantage of providing charge compensation within the channel itself, thus allowing maximal Ca²⁺ flux (i.e. Ca²⁺ release would occur at zero potential). This would make Ca²⁺ release possible even for vanishingly small [Ca²⁺] gradients, regulation being achieved through modulation of the pore open time. It should be noted that no K⁺ and Na⁺ concentration gradients exist across the inner membrane because the slow electrophoretic uptake of K⁺ and Na⁺ is compensated by the K⁺–H⁺ exchanger and the NHE, respectively. Thus, PTP opening can lead to selective Ca²⁺ release without perturbation of K⁺ and Na⁺ homeostasis, and no evolutionary pressure may have existed for the development of cation selectivity. In this respect the PTP is strikingly similar to the Ca²⁺ release channel of the sarcoplasmic reticulum, which operates as a Ca²⁺-selective channel despite its large size (>38 Å), high conductance for monovalent cations (≈1 nS at saturating K⁺), permeability to solutes like glucose and low permeability ratio when both K⁺ and Ca²⁺ are present (P_Ca/P_K ≈ 6) [58].

4. A Ca²⁺ release channel in Drosophila mitochondria: the missing link between the PTP of yeast and mammals?

We have recently argued that, in spite of clear differences between species, the PTP has been conserved in evolution from yeast to mammals [59]. Yeast mitochondria do possess a permeability pathway that resembles the mammalian PTP, also called the Yeast Mitochondrial Unselective Channel (YMUC) [60,61]. Based on the features of the PTP of yeast and mammals as of 1998, Manon et al. concluded that regulation of YMUC is too different from that of the mammalian PTP for yeast to be a good model of the latter [60]; in the light of new results we believe that the differences are not as fundamental as suspected earlier, and we refer the reader to a recent review specifically devoted to this aspect [59]. Other than in mammals and yeast, occurrence of a permeability transition has been established in plants, amphibians and fish including zebrafish (Danio rerio) [62] (see [59] for references). A mitochondrial Ca²⁺ release channel (mCRC) that we recently identified in cells from Drosophila melanogaster [63] deserves a special mention, as it may represent an evolutionary variant of the PTP displaying remarkable selectivity toward Ca²⁺ and H⁺.

Mitochondria were studied in digitonin-permeabilized S₂R⁺ cells [63], which are derived from Drosophila late embryonic stages and represent a variety of tissue precursors [64]. Of interest, mitochondria of S₂R⁺ cells possess all the classical Ca²⁺ transport pathways found in mammalian mitochondria, i.e. (i) the RR-sensitive Ca²⁺ uniporter (which is consistent with the existence of orthologs of MCU [4,5] and of MICU1 [65] in the Drosophila genome); (ii) the Na⁺–Ca²⁺ antiporter recently identified as NCLX [12], whose ortholog also exists in Drosophila; (iii) the RR-insensitive putative H⁺–Ca²⁺ antiporter mediating Ca²⁺ release at high membrane potential; (iv) a tetracaine-sensitive, RR-insensitive Ca²⁺ release pathway that opens in response to matrix Ca²⁺ loading or to depolarization, and mediates Ca²⁺ release [63]. The properties of the Drosophila mCRC appear to be intermediate between those of the PTP of mammals and yeast. Like the mammalian PTP, the Drosophila Ca²⁺ release pathway is inhibited by tetracaine [66] and opens in response to matrix Ca²⁺ loading, inner membrane depolarization, thiol oxidation and treatment with relatively high concentrations of NEM [63]. Like the yeast PTP (and at striking variance from the mammalian pore) the Drosophila mCRC is inhibited rather than stimulated by Pi; and is insensitive to ADP, quinones and to CsA [63], a finding that matches the absence of a mitochondrial CyP in Drosophila. Together with the inhibitory effect of Pi, the most striking difference between the Drosophila mCRC and the PTP is the selectivity for the transported species. At the onset of Ca²⁺-dependent Ca²⁺ release Drosophila mitochondria undergo depolarization, suggesting that the putative channel is also permeable to H⁺; yet no matrix swelling is observed even in KCl-based medium, indicating that the channel is not permeable to K⁺ (and to Cl⁻) in spite of the fact that the hydrated radius of Ca²⁺ is larger than that of K⁺ [63]. Lack of swelling was confirmed by lack of cytochrome c release and by ultrastructural analysis, and was not due to peculiar structural features of Drosophila mitochondria because matrix swelling and cytochrome c release readily followed the addition of the pore-forming peptide alamethicin [63].

The key features of the yeast and mammalian PTP, and of the Drosophila mCRC are summarized in Table 1. The most remarkable differences are the permeability to solutes, the selectivity for Ca²⁺ and the effects of Pi. Since nearly all yeast strains lack an MCU, the Ca²⁺-dependence of the yeast PTP has not been easy to assess although it was known that the yeast PT is favored by added Ca²⁺ [67]. Recent work using the Ca²⁺ ionophore ETH129, which mediates electrophoretic Ca²⁺ transport, has demonstrated that the PTP is favored by Ca²⁺ uptake in mitochondria from Saccharomyces cerevisiae [68], and based on this finding we suggest that matrix Ca²⁺ may be required for onset of the PT in yeast as well (Table 1). Lack of inhibition of Drosophila mCRC by CsA can be easily explained by the lack of a mitochondrial CyP. On the other hand, since a matrix CyP is found in yeast and its enzymatic activity is inhibited by CsA [69] we suspect that the CyP ability to interact with the PTP occurred later in evolution.

Table 1.

Properties of mammalian and yeast PTP, and of the Drosophila mCRC.

	Mammals	Yeast	Drosophila
Permeability to solutes up to ≈1500 Da	Yes	Yes	No
Selective Ca2+ release	No	No	Yes
Matrix Ca2+	Required	May be required	Required
Matrix Pi	Inducer	Inhibitor	Inhibitor
mt Cyp	Yes	Yes	No
CsA	Inhibitor	No effect	No effect
Tetracaine	Inhibitor	Not tested	Inhibitor
Redox sites	Yes	Yes	Yes

5. The PTP as a mitochondrial Ca²⁺ release channel: only a hypothesis?

In 1992, Altschuld et al. demonstrated that CsA significantly increases net Ca²⁺ uptake and decreases Ca²⁺ efflux in isolated cardiomyocytes, as measured by radiolabeled ⁴⁵Ca²⁺, without having any impact on cell morphology or viability [70]. The effect of CsA was concentration-dependent and specific to mitochondria, as ATP-dependent Ca²⁺ uptake by the sarcoplasmatic reticulum was not affected. The selectivity of the effects for mitochondrial Ca²⁺ and the very short incubation time with CsA (15 min) suggest that the PTP was being affected, and these data may represent the first piece of evidence that the pore contributes to Ca²⁺ cycling in mitochondria of living cardiomyocytes, and that reversible pore opening may be a physiological process in heart cells [70]. On the other hand, Eriksson et al. found that fluxes of Ca²⁺, Mg²⁺ and adenine nucleotides in perfused rat livers following hormonal stimulation were unaffected by previous administration of CsA, even if the PTP was demonstrably desensitized by CsA in mitochondria isolated from the same CsA-perfused livers [71]. The Authors concluded that regulation of mitochondrial ion and metabolite homeostasis is independent of the PTP [71], yet as discussed above a negative result is not as informative.

Two recent publications based on Ppif^−/− cells and mice do provide clear support for a role of the PTP in Ca²⁺ homeostasis [72,73]. Adult cortical neurons from wild type and Ppif^−/− mice were treated with either ATP (to activate P2Y purinergic receptors) or with depolarizing concentrations of KCl (to open plasma membrane voltage-dependent Ca²⁺ channels), both stimuli causing a robust increase of both cytosolic and mitochondrial [Ca²⁺] that was indistinguishable in neurons of the two genotypes [73]. Application of the two stimuli together, however, resulted in much higher levels of mitochondrial [Ca²⁺] in the Ppif^−/− neurons, suggesting that the threshold for PTP activation had been reached in the wild type but not in the CyPD-null mitochondria in situ. Thus, it appears that the regulatory role of CyPD (and of PTP opening) becomes crucial only for relatively large mitochondrial Ca²⁺ loads exceeding the capacity of NCLX and of the Na⁺-insensitive release pathway [73]. In other words, the PTP could be silent unless high Ca²⁺ loads saturate the NCLX and the H⁺–Ca²⁺ exchanger, allowing matrix [Ca²⁺] to rise enough to trigger pore opening. It should be noted that the transient stimulation of these Ca²⁺-mobilizing pathways did not induce cell death either in wild type or in CyPD-null neurons, suggesting that the mitochondrial PTP-activating response was part of a physiological process that is consistent with occurrence of reversible PTP opening [42].

In adult mice ablation of CyPD increases resistance to acute ischemia-reperfusion injury both in the heart and brain [33,35,36,74], while in neonatal mice, where the membrane-permeabilizing effects of BAX predominate, Ppif^−/− mice were instead remarkably sensitized suggesting age-related changes in the mitochondrial response to injury [74]. An age-related phenotype was also discovered in the hearts of Ppif^−/− mice, which displayed an intriguing decrease of maximum contractile reserve matched by increased shortening and relaxation times with longer decay of cytosolic Ca²⁺ transients [72]. Ppif^−/− mice were also unable to compensate for the increase in afterload caused by transaortic constriction, displaying a larger reduction in fractional shortening, decompensation, ventricular dilation, fibrosis and congestive heart failure; all consequences of CyPD ablation were cured by heart-selective reexpression of CyPD, strongly indicating that the maladaptive phenotype of Ppif^−/− mice depends on a primary disturbance of myocyte mitochondria rather than on an underlying systemic response [72]. Metabolic in vivo analysis demonstrated a significant increase in the glucose to palmitate ratio, suggesting a metabolic shift from fatty acid oxidation to glycolysis associated with increased activity of pyruvate dehydrogenase and α-ketoglutarate dehydrogenase. Direct measurement of total mitochondrial Ca²⁺ content of Ppif^−/− hearts showed a 2.6-fold increase, which was matched by greater mitochondrial Ca²⁺ transients in myocytes treated with CsA. Very importantly, under continuous pacing PTP desensitization with CsA decreased the rise time in Ca²⁺ accumulation and prolonged the recovery time after pacing, findings that are entirely consistent with the PTP acting as a Ca²⁺ release channel to prevent Ca²⁺ overload [72].

6. Summary and conclusions

The hypothesis that transient opening of the PTP may serve the physiological function of regulating matrix Ca²⁺ by preventing Ca²⁺ overload appears theoretically justified, and supported by a limited but extremely solid set of experimental results. Confusion may have arisen from the interpretation of results based on the use of CsA as well as of Ppif^−/− mice, which have too often been erroneously considered null for the PTP as well. Much more work is obviously needed, and we hope that this short review will help rekindle interest and experimental testing of this subject.