Abstract
The mitochondrial permeability transition pore (MPTP) has resisted molecular identification for decades. The original model of the MPTP had the adenine nucleotide translocator (ANT) as the inner membrane pore-forming component. Indeed, reconstitution experiments showed that recombinant or purified ANT generates MPTP-like pores in lipid bilayers. This model was challenged when mitochondria from Ant1/2 double null mouse liver still showed MPTP activity. Because mice contain and express 3 Ant genes, here we reinvestigated the genetic basis for the ANTs as comprising the MPTP. Liver mitochondria from Ant1, Ant2, and Ant4 deficient mice were highly refractory to Ca2+-induced MPT, and when also given cyclosporine A, MPT was completely inhibited. Moreover, liver mitochondria from mice with quadruple deletion of Ant1, Ant2, Ant4 and Ppif (cyclophilin D, target of CsA) lacked Ca2+-induced MPT. Finally, inner membrane patch clamping in mitochondria from Ant1, Ant2 and Ant4 triple null mouse embryonic fibroblasts (MEFs) showed a loss of MPT-like pores. Our findings suggest a new model of MPT consisting of two distinct molecular components, one of which is the ANTs and the other of which is unknown but requires CypD.
One Sentence Summary Genetic deletion of Ant1/2/4 and Ppif in mice fully inhibits the mitochondrial permeability transition pore
Mitochondrial permeability transition pore (MPTP) opening contributes to various pathologies involving necrotic cell death after ischemic injuries or degenerative muscle and brain diseases (1–5). The MPTP is a <1.5 kDa pore within the inner mitochondrial membrane which opens in the presence of high matrix Ca2+ and/or ROS (6, 7). Although numerous regulators of this pore have been identified, the molecular pore forming component of the MPTP remains undefined (8, 9). The adenine nucleotide translocator (ANT) family was originally proposed to be the pore forming component of the MPTP due to the fact that ANT ligands and inhibitors have profound effects on the opening of the MPTP and that reconstituted ANTs in lipid bilayers are able to form channels similar to that of the MPTP (10–15). However, this model was largely rejected when it was reported that mouse liver mitochondria lacking the genes Ant1 and Ant2, were able to undergo permeability transition, albeit this transition required substantially higher levels of Ca2+ to engage (16). Here we investigated whether the third Ant gene in mice, Ant4, compensates for the loss of Ant1 and Ant2 and whether mitochondria completely null for all ANT isoforms would be able to undergo permeability transition (humans contain 4 Ant genes (Ant1-4) while mice lack Ant3) (17). More recently a model in which the mitochondrial F1FO-ATPase serves at the MPTP has been proposed (18–20), although not without controversy as some data dispute the role of the F1F0-ATPase (21–23), thus leaving the MPTP undefined.
We first characterized the tissue distribution of all 3 ANT isoforms in the mouse by Western blotting from mitochondria lysates. In wild-type (WT) mice, ANT1 is the major isoform expressed in the heart and skeletal muscle, while ANT2 is expressed in all tissues examined besides the testis, which predominantly expresses ANT4 (Fig. 1A). Importantly, at baseline mouse liver only detectably expresses ANT2 (Fig. 1A). In mice globally lacking either Ant1 or Ant4, the ANT2 isoform becomes upregulated to compensate for their loss (Fig. 1A). Importantly, we observed that ANT4 expression is induced in liver mitochondria from mice lacking ANT1 and ANT2 (Fig. 1B). This observation of compensation by ANT4 when Ant1 and Ant2 were deleted from the liver likely explains why Kokoszka and colleagues failed to observe a more severe loss of MPTP activity (16). The Ant2 gene was LoxP-targeted to permit tissue specific deletion because full somatic Ant2-/- mice are embryonic lethal (24).
Here we generated mice null for all ANT isoforms in the liver by crossing total somatic Ant1-/-, Ant4-/- with Ant2-loxP (fl) mice with an albumin promoter driven Cre transgenic mouse line (Ant2fl/fl-Alb-Cre) (16, 25). The loss of all ANT isoform expression in the liver was confirmed by Western blot analysis on isolated liver mitochondria (Fig. 1C). In all liver-based experiments Ant1-/- Ant4-/- Ant2fl/fl mice were used as controls because ANT2 is the overwhelming isoform expressed in liver (Fig. 1A and C). More importantly, this strategy allowed us to directly compare littermates (Ant1-/- Ant4-/- Ant2fl/fl-Alb-Cre versus Ant1-/- Ant4-/- Ant2fl/fl). TMRE staining showed comparable membrane potential that was sensitive to FCCP treatment in mitochondria from triple null livers and the ANT2 only expressing controls (Ant1-/- Ant4-/- Ant2fl/fl) (Fig. 1D). H&E-stained histological liver sections and transmission electron microscopy also showed no overt pathology in Ant triple nulls versus ANT2 only expressing mice (Fig. S1A). Surprisingly, Ant triple null mitochondria from liver also showed no difference in stimulated and maximum oxygen consumption rate compared with ANT2 only, suggesting ANT function is somehow compensated in these mitochondria (Fig. S1B).
To investigate the status of MPTP activity we used the 2 most commonly employed MPTP assays in isolated mitochondria: that of Ca2+ retention capacity (CRC) using step-wise 40 μM Ca2+ additions over time, and that of absorbance-based assessment of mitochondrial swelling. Mitochondria from WT mouse liver only took up 3 stepwise Ca2+ additions before opening and dumping their Ca2+ into the solution, which contains a fluorescence indictor (Fig. 2A). This Ca2+-induced MPTP opening caused immediate mitochondria swelling indicated by a decrease in light absorbance of these organelles in solution (Fig. 2B). Consistent with previous results (16), loss of ANT1 and ANT2 desensitized Ca2+-induced MPTP opening similarly to the level of WT mitochondria treated with cyclosporine A (CsA) (26, 27), which is an inhibitor of CypD that regulates the MPTP (Fig. 2A and B). Mitochondria lacking all isoforms of the ANT family are substantially further desensitized to Ca2+-induced MPTP opening, however, the MPTP can still open with extreme levels of Ca2+ (Fig. 2A and B), confirming that the ANT family is still dispensable for MPTP opening in liver. However, triple null mitochondria also treated with CsA showed Ca2+ uptake to the physical limits of the assay at 4.8 mM before mitochondria stopped taking up Ca2+, although mitochondria swelling never occurred (Fig. 2A and B), unless the non-specific permeabilizing agent alamethicin was added (Fig. 2B, arrowhead).
We observed a similar synergistic effect in the reciprocal experiment when Ppif-/- mitochondria lacking CypD were treated with ADP, a known desensitizer of the MPTP and ligand for the ANTs (Fig. 2C and D). Together these data reveal that MPTP requires at least one ANT family member or CypD. These studies further suggest that there may be at least two separate channels which comprise the total MPTP activity in liver: one channel consisting of the ANT family that can escape CypD regulation and another that is absolutely dependent on CypD to open.
To further address the data obtained in purified mitochondria with CsA in Ant triple null mice, we independently generated quadruple null mice for Ant1, Ant2, Ant4 and Ppif. We compared the CRC and swelling measurements in liver mitochondria from these quadruple null mice versus Ant1-/- Ant4-/- Ant2fl/fl control (ANT2 only) mice (Fig. 3A-C). While ANT2 only mitochondria took up a few boluses of 40 μM Ca2+ before opening and swelling, the quadruple null never underwent swelling and took up Ca2+ boluses until they assay was saturated (Fig. 3A and B). Electron microscopy showed that liver mitochondria from Ant1-/- Ant4-/- Ant2fl/fl control mice, Ant1-/- Ant4-/- Ant2fl/fl-Alb-Cre mice, and Ppif-/- mice had normal mitochondria at baseline in Ca2+ free buffer, but at 800 μM Ca2+ all showed profound swelling. Conversely, mitochondria from Ant1-/- Ant4-/- Ppif-/- Ant2fl/fl-Alb-Cre quadruple null livers failed to show swelling (Fig. 3C) despite our also using half the amount of mitochondria (1 mg versus 2 mg) to correct for baseline absorbance (Fig. S2A). Similar to the Ant triple null mitochondria, Ant1-/- Ant4-/- Ppif-/- Ant2fl/fl-Alb-Cre quadruple null mitochondria displayed no alteration in morphology or stimulated oxygen consumption under energized conditions (Fig. S2B and C). This is the first time that MPT has ever been genetically inhibited, thus demonstrating its molecular existence.
To extend these findings to another model system we also generated mouse embryonic fibroblasts (MEFs) lacking all ANT isoforms. An adenovirus expressing Cre recombinase was used to delete the Ant2 gene, shown by Western blotting (Fig. 4A). In contrast to the liver mitochondria, mitochondria isolated from Ant1-/- Ant4-/- Ant2fl/f-(AdCre) MEFs (triple null) did not show compensation in their ability to transport ATP/ADP so that these mitochondria had almost no respiratory capacity (Fig. 4B). However, these MEFs lacking all 3 Ant genes still showed mitochondrial membrane potential (ΔΨ) in culture with TMRE fluorescence (Fig. 4C), although this is likely due to support from glycolysis and reverse mitochondrial ATPase activity to maintain ΔΨ. To further explore this concept that Ant triple null MEFs depend on glycolysis we grew them in glucose free, galactose-containing culture media making glycolysis an ATP futile process. Under these conditions the Ant null MEFs died within 24 hours whereas the control MEFs were unaffected (Fig. 4D).
A final assay that is used to assess MPTP channel activity is patch clamping of the inner mitochondrial membrane to identify the characteristic conductance properties of this inducible pore. Mitoplasts isolated from WT and ANT2 only expressing MEFs (Ant1-/- Ant4-/- Ant2fl/fl) showed channel opening and conductance levels consistent with the MPTP and importantly ADP, an MPTP suppressor, inhibited all such pore opening events (Fig. 4E and 4F). However, patching of mitoplasts from Ant triple null MEFs showed a much lower prevalence of large pores, and all such rare pores were unresponsive to ADP, suggesting that no MPTP like channels were present without ANT (Fig. 4E and G). These results suggest that the ANT family is the primary inner membrane channel forming component of the MPTP in MEFs.
In summary, we have genetically abolished MPT by deletion of the ANT family and CypD in liver mitochondria. This represents the first complete inhibition of this activity and it provides molecular proof that the MPTP exists and is not an artefactual ex vivo phenomenon. Our data support a model of the MPTP consisting of only ANT in MEFs, yet in liver mitochondria it appears to exist as two distinct molecular components, with one being comprised of the ANTs (Fig. S3). In liver mitochondria this ANT containing MPT activity can occur in the absence of CypD, yet a second alternative activity, completely requires CypD isomerase activity. The possibility that the MPTP is composed of at least two pore-forming components may explain why identifying its gene products have been so challenging. In order to identify the CypD-dependent alternative MPTP channel, the ANTs must be accounted for either by genetic deletion or pharmacological inhibition. With this new understanding previously speculated MPTP-forming components, including various other slc25a family members and the mitochondrial F1FO-ATPase (18–20, 28, 29), should be re-evaluated.
Funding
This work was supported by a grant from NIH to J.D.M. (R01HL132831), funding from the Howard Hughes Medical Institutes to J.D.M., funding from the Fondation Leducq to J.D.M, and the American Heart Association to J.K. (17SDG33661152)
Author Contributions
J.K. and J.D.M designed the study and wrote the manuscript. J.K., M.J.B., H.K., M.A.S., N. L., and P.M.P performed experiments and analyzed data. N.T. and D.C.W. shared previously generated mutant mouse lines.
Competing interests
The authors have no financial or other competing interests related to the current study
Data and materials availability
All data are contained within the figures of the paper or supplemental figures, or available upon reasonable request.
Acknowledgments
We thank Dr Douglas C. Wallace (Children’s Hospital of Philadelphia) for supplying Ant1-/- and Ant2-loxP targeted mice. We thank Gyorgy Hajnoczky for evaluation of the manuscript.