The mitochondrial permeability transition pore: Is it formed by the ATP synthase, adenine nucleotide translocators or both?

The mitochondrial permeability transition (MPT) pore is a recognized effector of cell death [1]. While its key features have been described in extensive detail, its molecular components have not been unanimously accepted. For example, the adenine nucleotide translocator (ANT) was initially proposed to constitute the channel-forming unit of the MPT pore but mice lacking two out of the three ANT isoforms in the liver still displayed an MPT pore albeit with increased resistance to Ca²⁺ and loss of MPT regulation by ADP/ATP and other ANT ligands (reviewed in [2]). This sparked an intensive search for its identity leading to subsequent hypotheses placing the ATP synthase as the MPT channel-forming constituent. Most of these studies have relied on gene knockdown approaches to show either decreased MPT-dependent cell death or changes in Ca²⁺ sensitivity for pore opening [2]. In addition, electrophysiology of purified ATP synthase has resulted in the detection of channels with MPT-like conductance properties, which are silenced by adenine nucleotides but remain insensitive to key antagonists including cyclosporin A (CsA) [3].

Recent reports using HAP1 cells with specific deletions of different subunits of the ATP synthase have resulted in an apparently normal MPT [4,5]. Specifically, loss of any of these subunits - which resulted in the inability to form a fully assembled ATP synthase - had no effect on the amount of Ca²⁺ required to trigger MPT-dependent Ca²⁺ release and its sensitivity to key MPT modulators including CsA remained unchanged [5]. HAP1 cells lacking ATP synthase c subunit exhibited decreased swelling kinetics in response to Ca²⁺, which may conceivably be due to a decreased pore size [6]. However, mitochondria with a disassembled ATP synthase still display unchanged polymer-exclusion properties [5]. In light of these studies, the authors concluded that it was ‘very unlikely’ that the ATP synthase forms the channel-unit of the MPT pore [5].

Recent studies have reported that liposome reconstituted ATP synthase dimers can form H⁺-dissipating channels and that Ca²⁺ can increase the channel’s open probability, which exhibits similar electro-physiological signatures when compared with the MPT pore [7]. Furthermore, a recent study by Mnatsakanyan and colleagues suggest ATP synthase monomers can independently form megachannels [8]. In both studies, the amount of Ca²⁺ to activate these channels stands in the millimolar range, which is considerably higher than that used to activate MPT in isolated mitochondria depending on the assay conditions.

A recent study by Karch and colleagues has shown that mouse liver mitochondria lacking all 3 ANT isoforms are extremely resistant to Ca²⁺-induced MPT [9]. When crossed with mice lacking cyclophilin D (CypD), the canonical Ca²⁺-induced MPT was completely abolished. Moreover, mouse embryo fibroblasts lacking ANTs failed to show a canonical ADP-inhibited megachannel. This led the authors to propose that the MPT pore is formed by ANTs alongside a CypD-dependent component [9].

The data at this point indicates on one side that the ATP synthase is most likely not forming the channel unit of the MPT pore. On the other side, highly purified ATP synthase preparations do present a nanosiemen-range conductance resembling the mitochondrial megachannel studied decades ago [11]. Consequently, we decided to test the effects of concomitant MPT and ATP synthase inhibition by measuring the Ca²⁺ retention capacity (CRC) of isolated murine heart mitochondria from either wild type (WT) or Ppif^−/− (CypD KO) mice (Fig. 1). In the presence of the ATP synthase inhibitor oligomycin (Oligo), WT heart mitochondria showed no significant changes on CRC values (Fig. 1A,B). Addition of the CypD inhibitor CsA significantly increased CRC in agreement with previous studies [1]. When combined, CsA plus Oligo significantly decreased the CRC when compared to the values observed with CsA alone, and statistically matched the CRC values obtained under control conditions. To discard potential molecular interactions between CsA and Oligo affecting MPT, we performed a CRC assay on CypD KO mouse heart mitochondria. As shown before [5], the CRC of CypD KO mouse mitochondria duplicates that of WT mitochondria incubated with CsA (Fig. 1B). However, addition of Oligo to CypD KO mitochondria results in significantly lower CRC values, matching that of WT mitochondria in the presence of CsA plus Oligo.

Fig. 1.

ATP synthase activity regulates CypD-dependent MPT in isolated mouse heart mitochondria. (A) Representative traces showing the Ca²⁺-retention capacity (CRC) profile of isolated cardiac mitochondria under control conditions and in the presence of 5 μM oligomycin, 1 μM of CsA or both inhibitors. (B) CRC profile under each condition for isolated heart mitochondria from wild type (WT) and CypD knockout (CypD KO) mice of the Sv129 strain [13]. Data are presented as mean + SEM from 5 independent biological replicates for each group. Group differences were assessed using a two-way ANOVA with Scheffe’s post-hoc test. *P < 0.05 vs. WT control, † P < 0.05 vs. WT CsA, ‡ P < 0.05 vs. CypD KO Control. (C) Working model proposing coordinated activities of ANT and the ATP synthase. Under this model, primary activation of an ANT-driven MPT with micromolar Ca²⁺ levels (1) would then trigger the ATP synthase megachannel (2). (D) In the absence of ANT, MPT-like activity would only be evident at non-physiological millimolar Ca²⁺ loads [9]. (E) In the absence of an assembled ATP synthase, CsA-sensitive MPT-like activity (i.e the ~0.3nS channel reported in [6]) would be monitored at micromolar Ca²⁺ levels.

These results suggest that inhibition of c subunit rotation with oligomycin can indeed affect the MPT’s Ca²⁺ threshold and that an un-inhibited ATP synthase is required for a CypD-dependent MPT. However, if we consider results showing a ‘canonical’ CsA-sensitive MPT upon chronic ATP synthase deletion [5], plus the additive effects of CypD deletion in an ANT triple knockout context, which require substantial amounts of Ca²⁺ for MPT onset [9], a reasonable working hypothesis would be that the MPT pore can be formed by ANT alongside the ATP synthase.

Under this working hypothesis, micromolar Ca²⁺ would initially open a CsA-sensitive pore at the level of ANT with peak conductances ~0.3 nS [6]. This in turn would open the megachannel formed by the ATP synthase comprising the canonical MPT pore (Fig. 1C). This model can explain why there is no MPT in the absence of ANT at micromolar Ca²⁺ levels [9], but only present at submillimolar Ca²⁺ levels (i.e. Ca²⁺ threshold for the ATP synthase megachannel) [7,8,11] in a CsA-sensitive fashion due to known interactions between the ATP synthase and CypD (Fig. 1D) [12]. Finally, in the absence of the ATP synthase, the CsA- and BKA-sensitive ~0.3nS channel likely formed by ANT can induce MPT when triggered with micromolar Ca²⁺ levels (Fig. 1E) [5]. The model presented herein is consistent with the known multiple conductance and ‘dimeric’ nature of the MPT pore [10] as well as the known ‘promiscuous’ biding nature of CypD [2] and would explain the BKA- and CsA-sensitive channel in ATP synthase defective cells detected by Walker’s and Pavlov’s groups respectively [5,6]. Nevertheless, we would like to caution the reader that our approach still presents limitations and further experiments are warranted to either reconcile apparently irreconcilable studies or to fully discard a role for some of the MPT pore components addressed herein. Thus, if this model is right then the Ca²⁺-induced Ca²⁺ release monitored in the absence of all ANT isoforms [9] should be suppressed with oligomycin, as this macrolide can decrease the channel’s open probability of purified ATP synthase preparations [8].

Taken together, the findings suggesting that either ANT or the ATP synthase can form mitochondrial pores per se are compatible with the model proposing that the MPT pore is formed through the concerted activities of both candidates.