Bisindolylpyrrole triggers transient mitochondrial permeability transitions to cause apoptosis in a VDAC1/2 and cyclophilin D-dependent manner via the ANT-associated pore

Bisindolylpyrrole at 0.1 μM is cytoprotective in 2% FBS that is counteracted by cyclosporin-A (CsA), an inhibitor of cyclophilin-D (CypD). We hypothesized that the cytoprotective effect might be due to transient mitochondrial permeability transition (tPT). This study tested the hypothesis that bisindolylpyrrole can trigger tPT extensively, thereby leading to cell death under certain conditions. Indeed, CsA-sensitive tPT-mediated apoptosis could be induced by bisindolylpyrrole at > 5 μM in HeLa cells cultured in 0.1% FBS, depending on CypD and VDAC1/2, as shown by siRNA knockdown experiments. Rat liver mitochondria also underwent swelling in response to bisindolylpyrrole, which proceeded at a slower rate than Ca2+-induced swelling, and which was blocked by the VDAC inhibitor tubulin and the ANT inhibitor bongkrekate, indicating the involvement of the ANT-associated, smaller pore. We examined why 0.1% FBS is a prerequisite for apoptosis and found that apoptosis is blocked by PKC activation, which is counteracted by the overexpressed defective PKCε. In mitochondrial suspensions, bisindolylpyrrole triggered CsA-sensitive swelling, which was suppressed selectively by pretreatment with PKCε, but not in the co-presence of tubulin. These data suggest that upon PKC inactivation the cytoprotective compound bisindolylpyrrole can induce prolonged tPT causing apoptosis in a CypD-dependent manner through the VDAC1/2-regulated ANT-associated pore.


BP causes apoptosis in a CsA/CypD-dependent manner through tPT under low FBS conditions.
To verify the hypothesis that BP can induce the PT that ultimately causes cell death, we searched for the experimental conditions using HeLa cells stably expressing GFP-tagged cytochrome c. The results are shown in Fig. 1A. BP at 5-10 μM induced cell death after 4-8 h in 0.1% FBS, in contrast with less than 5% of cell death in control cells. The type of cell death was morphologically apoptotic, and this was further supported by a significant delay in the onset of cell death by treatment with Z-VAD-fmk (ZVAD), a caspase inhibitor. When CsA was added either 30 min before the addition of BP or at least 1 h later, it blocked apoptosis, whereas the specific calcineurin inhibitor FK506 did not, suggesting PTP involvement.
To determine the involvement of CypD, we performed knockdown experiments with CypD-targeting small interfering RNA (siRNA). As shown in Fig. 1B, transfection with this siRNA effectively reduced the protein expression and this conferred remarkable resistance to BP-induced apoptosis, whereas treatment with the control siRNA had no effect. These findings justify the conclusion that the presence of CypD is a prerequisite for BP-induced, CsA-sensitive apoptosis.
Cells were loaded with calcein to directly observe the opening of the PTP with large conductance, and the cytoplsmic calcein signals were quenched with Co 2+ . Calcein signals were much stronger than cytochrome c-GFP signals and were not significantly affected by the latter signal. Calcein signals were lost at 25 min after treatment with 7.5 μM BP; in contrast, tetramethylrhodamine methyl ester (TMRM) signals decreased as early as 15 min after treatment and remained low over 2-3 h while still being significantly higher than those values in the cells treated with carbonyl cyanide-p-(trifluoromethoxy)phenylhydrazone (FCCP). These changes by BP were blocked by CsA pretreatment (Fig. 1C), indicating the induction of tPT. the presence or absence of BP from 5 to 10 μM with or without 3 μM CsA, which was added 30 min before BP addition (pre) or at least 1 h later (post), 20 μM ZVAD or 10 μM FK506. Data are representative of three separate experiments (error bars represent the mean ± SD; *p < 0.005 relative to control and to CsA plus BP; #p < 0.005 relative to 7.5 μM BP at the same time points). (B) Effect of CypD knockdown on apoptosis. (i) Immunoblotting with anti-CypD and actin antibodies of lysates from cells transfected with siRNA targeting CypD (siCypD) or control siRNA. (ii) Cell viability was determined 8 h after treatment with 7.5 μM BP. *p < 0.005. (C) The effects of 7.5 μM BP with or without CsA on mitochondrial entrapped calcein (signals pseudo colored in white) and TMRM signals (red; enhanced images in the insets); the membrane potential upon treatment with 0.5 μM FCCP was considered as a level of mitochondrial depolarization (i). Quantification of calcein and TMRM signals over 3 h (initial fluorescence intensities were normalized for comparative purposes; *p < 0.005 compared with BP treatment at 3 h; + p < 0.001 compared with control at the same time point by Welch's t-test with Bonferroni's correction; n = 100 cells) (ii). (D) Signals for mitochondrial cytochrome c-GFP (Cytc-GFP) at 3 h; different apoptotic stages of the cells are present, with (a) mitochondrial clustering around the nuclei, (b) cytochrome c-GFP release into the cytoplasm, and (c) apoptotic blebs (differential interference contrast micrograph [DIC]). Representative fields were randomly imaged (i). Quantification of cells with cytochorome c-GFP release (b) and apoptotic bodies (c) over 4 h (*p < 0.001 by Welch's t-test with Bonferroni's correction; n = 100 cells) (ii). (E) Time-lapse analysis of the release of cytochrome c-GFP from mitochondria to the cytoplasm and nucleus in response to BP treatment. Scale bars: 10 μm. Images were digitized using FV10-ASW software version 4.2a (https ://www.olymp us-lifes cienc e.com).

Scientific RepoRtS
| (2020) 10:16751 | https://doi.org/10.1038/s41598-020-73667-z www.nature.com/scientificreports/ We evaluated the cytochrome c release from mitochondria leading to apoptosis and observed, different apoptotic stages of the cells were evident at 3 and 4 h, with (a) mitochondrial clustering around the nuclei, (b) cytochrome c-GFP release into the cytoplasm and nucleus, and (c) apoptotic blebs; these changes were blocked by CsA (Fig. 1Di). A quantification analysis showed that the apoptosis was preceded by the release of cytochrome c-GFP, which occurrred after 2-3 h of BP treatment (Fig. 1Dii). A time lapse analysis showed that cytochrome c-GFP was released in the cytoplasm from the mitochondria approximately 2 h after BP treatment, which was almost simultaneously diffused into the nucleus (Fig. 1E). Keeping with the pivotal role of nuclear translocation of cytochrome c for caspase-independent nuclear apoptosis 43 , this nuclear accumulation of cytochrome c-GFP may account for the observation that the inhibition of apoptosis by ZVAD was observed for only about 8 h and virtually all of the cells underwent cell death until 11 h (Fig. 1A). These data suggest that BP treatment may induce prolonged tPT, which transitions to the persistent pore opening associated with cytochrome c-mediated apoptosis.
Mitochondrial autophagy follows tPT and influences the progress of apoptosis. To determine whether the mitochondrial aggregation in BP-treated cells reflects the selective removal of impaired mitochondria by autophagosomes, or mitophagy, lysosomes and mitochondria were labeled with LysoTracker green (whose signals were much stronger than cytochrome c-GFP signals and were not significantly affected by the latter signal), and MitoTracker red, respectively. While control cells had small lysosomal signals with distribution distinct from the mitochondrial signals, BP-treated cells exhibited larger lysosomes in the perinuclear regions after 3 h, which were largely co-localized with mitochondrial signals ( Fig. 2A), indicating autophagolysosomes containing mitochondria. When CsA was added 30 min before BP treatment, it blocked these mitochondrial changes (Fig. 1Ci), indicating that mitophagy follows PTP opening. We inquired about whether or not autophagy plays a role in the course of apoptosis and utilized colchicine (a microtubule polymerization inhibitor) and N-acetyl-Leu-Leu-Norleu-al (ALLN) (a peptide aldehyde inhibitor of lysosomal proteases, including cathepsin-B/L and calpains) 44 . Both reagents delayed significantly apoptosis (Fig. 2B), although all cells eventually died within 12 h. We then examined whether or not colchicine influences the BP-induced release of TMRM, calcein and cytochrome c. Colchicine did not affect the TMRM release and calcein release by BP (Fig. 2Ci), but it delayed the formation of autophagolysosomes and the release of cytochrome c in most cells (Fig. 2Cii), indicating that PTP opening precedes mitophagy and cytochrome c release. These data suggest that mitophagy may play a role in the progression of BP-induced apoptosis.

Involvement of VDAC isoforms in BP-induced apoptosis.
VDAC is a regulator of the PTP. To investigate whether or not BP-induced apoptosis is mediated by VDAC, we knocked down VDAC1 and/or VDAC2 using their specific siRNAs (siVDAC1 and siVDAC2). As shown in Fig. 3A, immunoblotting showed that transfection with each isoform of siVDAC1 and siVDAC2 effectively reduced the protein expression of the target VDAC isoforms without significantly affecting the non-target VDAC isoforms, including VDAC3. We observed that cells doubly transfected with siVDAC1 and siVDAC2 were remarkably resistant to BP-induced apoptosis (Fig. 3B), whereas those treated with siVDAC1 or siVDAC2 had no effect. These findings indicate that both isoforms of VDAC1 and VDAC2 can mediate the BP-induced, CsA-sensitive apoptosis. whether or not BP directly acts on mitochondria, experiments were performed with isolated rat liver mitochondria preloaded with TMRM and calcein. BP was able to trigger PTP opening dose-dependently at 3 and 10 µM in mitochondrial suspensions (Fig. 4A). The quantification of mitochondrial diameters based on differential interference contrast micrographs revealed no significant swelling during BP-induced PTP opening, in contrast to the striking swelling observed when mitochondria were treated with the nonspecific permeabilizing agent alamethicin (Fig. 4A).
Time course experiments were then performed over 20 min with mitochondria that were immobilized in agar 45 As shown in Fig. 4Bi and iii, upon BP treatment, the TMRM signals started to drop within 2-5 min, and calcein release occurred over 15 min; both effects were completely blocked by 3 μM CsA (Fig. 4B). This slow calcein release was in contrast with the rapid release of calcein when mitochondrial swelling was induced by Ca 2+ loading ( Supplementary Fig. S1). These data suggest that the pore size induced by BP may be smaller than that of Ca 2+ .
To determine the involvement of VDAC in BP-induced pore opening, tubulin was used as a VDAC inhibitor 46 . BP was not able to trigger pore opening in the presence of 50 nM tubulin, as indicated by the inhibition of depolarization and calcein release (Fig. 4B), indicating VDAC involvement. Next, we assessed the involvement of ANT with bongkrekate, a membrane-permeant ANT ligand that promotes m-state conformation and inhibits pore opening 35 . Notably, bongkrekate prevented the release of TMRM and calcein signals by BP (Fig. 4B), indicating the involvement of ANT.

PKC determines the susceptibility to BP-induced apoptosis. Cells under high FBS conditions
(> 2%) were resistant to the induction of apoptosis by BP. What signaling pathways activated by FBS are involved in the resistance to BP? Because PKC is involved in the PTP-mediated pathological process 30−34 we investigated whether PKC could modulate apoptosis induced by BP. In the presence of either staurosporine or BMI, PKC inhibitors, the cells became susceptible to BP even in 2% FBS (Fig. 5A). A second approach was then adopted, namely long-term treatment with 1 ng/ml phorbol 12-myristate 13-acetate (PMA) for 24 h, an established method that downregulates all PKC isozymes except ζ and λ, due to proteolytic degradation. The PKCdownregulated cells also became susceptible to BP (Fig. 5A). These data suggest that the cytoprotection of 2% FBS against BP may depend on the activation of PKC. If this is true, conversely, BP-induced apoptosis in 0.1% FBS will be reversed by the activation of PKC. In fact, PMA treatment reversed the sensitivity to apoptosis in a dose-dependent way, ranging from 0.3 to 1 ng/ml (Fig. 5B). These findings suggest that PKC signaling plays an essential role in the susceptibility to BP.
The possible effect of PKC activation by PMA was validated by confocal microscopic images for cytochrome c-GFP, TMRM and calcein. At 3 h after BP treatment, PMA inhibited the diffusion of cytochrome c-GFP signals in most of the cells and also apparently inhibited the formation of autophagolysosomes (Fig. 5C). However, PMA didn't block the occurrence of tPT (Fig. 5D). Taking a close look, PMA was found to reduce the declining rates of both TMRM and calcein signals in comparison to those observed after treatment with BP alone (compare Fig. 5D with Fig. 1C at 25 min after BP treatment). One plausible mechanism underlying the PMA suppression PKCε involvement via VDAC. We focused on the PKC isozyme, PKCε, and validated its role in PMAinduced cytoprotection against BP using the HeLa cells overexpressing a kinase-deficient mutant of PKCε 47 and control cells simply overexpressing the vector. Immunoblot analysis with PKCε antibody of cell lysates shows expression of an 84 kDa band (Fig. 6A), corresponding to mutant PKCε. In response to BP, both types of cells equally underwent apoptosis; however, upon treatment with 1 ng/ml PMA for PKC activation, cell viability in the inactive PKCε overexpressing cells was reduced to approximately half (p < 0.005) in comparison to control cells (Fig. 6B). This indicates a primary role of the PKCε signaling pathway. We examined the direct effect of PKCε on the PTP and performed a mitochondrial swelling assay with nonenergized rat liver mitochondrial suspensions. Mitochondria were pretreated with recombinant PKCε (rPKCε) or recombinant PKCα (rPKCα) in a phosphorylation buffer containing ATP, Mg 2+ and PMA for 25 min at 25 °C and were subjected to BP treatment. The results are shown in Fig. 7A. While control mitochondria underwent CsA-sensitive swelling in response to three additions of 1 μM BP, the mitochondria pretreated with PMAactivated rPKCε conferred resistance to BP-induced swelling to the comparable levels by CsA treatment; no such resistance was observed with mitochondria pretreated with inactive rPKCε (without PMA) or PMA-activated rPKCα. These findings suggest that the selective phosphorylation of mitochondrial protein(s) by PMA-activated rPKCε may desensitize the PTP to BP.
Because the VDAC is involved in BP-induced PTP opening and apoptosis, we suspected that the inhibition of swelling by rPKCε under a phosphorylation condition was mediated by the VDAC. If so, the presence of the VDAC inhibitor tubulin might eliminate the effect of rPKCε. To test this hypothesis, mitochondria were treated with PMA-activated rPKCε in the presence or absence of tubulin and then were subjected to BP-induced swelling. Figure 7B shows that rPKCε pretreatment with tubulin partially counteracted the rPKCε-induced suppression of BP-induced swelling, whereas tubulin alone had no effect on BP-induced swelling, suggesting that VDAC may involve the effect of rPKCε.

Discussion
BP is cytoprotective at concentrations of < 1 μM under normal cell culture conditions (> 2% FBS). However, the present study showed that BP can induce cytochrome c mediated apoptosis at concentrations of 5−10 μM in low FBS conditions (< 0.1%) and that the apoptosis is regulated by CypD, and mediated by the prolonged induction of tPT (Fig. 1). The tPT may be mediated by the VDAC (Fig. 3) which is linked to the ANT (Fig. 4B) regulated by CypD activity (Fig. 1B). The switch of BP from a cytoprotective effect to an apoptotic effect may be primarily dependent on the absence of serum factors (Fig. 5), which results in the inactivation of PKC, especially PKCε, signaling to mitochondrial proteins (Fig. 6), including the VDAC (Fig. 7).
Because the sustained PT leads to mitochondrial energetic failure, it is generally thought to result in necrosis, but it also causes apoptosis 6,28,29,48 . It was previously proposed that apoptosis occurs when the PT affects a subset of mitochondria (since the ATP required for the execution of apoptosis is provided by the remaining intact mitochondria), while necrosis occurs when it affects most mitochondria (since the reverse activation of F-ATP synthase exhausts cellular ATP) 49 . In the present study, BP affected the whole mitochondrial population, but it caused apoptosis. This could be attributed to the BP-induced tPT retaining a low mitochondrial potential for a few hours (Fig. 1B). The low membrane potential prevents the reverse activation of F-ATP synthase and the exhaustion of cellular ATP, allowing the execution of apoptosis. The low membrane potential observed is in contrast with the high membrane potential during tPT in other experimental settings 6,7 . This could be due to a longer opening time or a higher opening frequency of the pore after BP treatment. The tPT triggered by BP may depend on CypD activity, as demonstrated by the inhibition of apoptosis by CsA and siRNA targeting CypD (Fig. 1A,B).
The low mitochondrial membrane potential associated with tPT may also be responsible for the extensive occurrence of mitophagy observed upon BP treatment (Fig. 2), as previously described 50 . Treatment with ALLN Figure 4. Properties of BP-induced PTP opening in isolated rat liver mitochondria energized with succinate, preloaded by TMRM and calcein. (A) The dose dependent effect of BP (3 and 10 μM) in mitochondrial suspensions. Mitochondria were also treated by 10 μg/ml alamethicin. (i) Representative confocal microscopic fields were randomly imaged at 10 min for TMRM and calcein signals and also the same fields were imaged by a differential interference contrast microscope (DIC). (ii) Quantification of fluorescence signals for TMRM (logarithmic scale) and calcein (linear scale) and of the mitochondrial diameters measured from DIC micrograph images (error bars represent the mean ± SD; *p < 0.01; n = 300 mitochondria). Note that BP-induced PTP opening was not accompanied by mitochondrial swelling. (  www.nature.com/scientificreports/ (a lysosomal protease inhibitor) delayed BP-induced apoptosis (Fig. 2B). Colchicine (a microtubule destabilizer) also delayed apoptosis (Fig. 2B), the formation of autophagolysosomes and cytochrome c release (Fig. 2Cii). These effects are probably due to the inhibition of microtubule-based mitochondrial trafficking. Colchicine increases free tubulin and prevents VDAC permeability in cells 51 and Fig. 4Bi shows the tubulin inhibition of BP-induced PTP opening in isolated mitochondria. These data suggest that a potential increase in free tubulin upon colchicine  www.nature.com/scientificreports/ treatment may interfere with BP-triggered tPT; however, this was not the case (Fig. 2Ci), probably because the free tubulin levels were insufficient for the inhibition of the action of BP. Knockdown experiments with siRNA in HeLa cells indicated that VDAC1 and VDAC2 isoforms mediate BP-induced apoptosis (Fig. 3). This is consistent with the inhibition of BP-triggered mitochondrial swelling by tubulin (Fig. 7), whose specificity to the VDAC was demonstrated with the VDAC channel reconstituted into the planar lipid membrane 46,52 and with cancer cells 51,53 . We did not evaluate VDAC3 in these experiments, because the import of NADH in the mitochondrial inter-membrane space is not possible when the porin 1 deficient yeast is complemented by VDAC3, but not by VDAC1 and VDAC2 54 . It has recently been reported that tubulin also blocks VDAC3 55 : the paper showed that even 85 nM of tubulin produced fewer blockage events in VDAC3 in comparison to those by 45 nM tubulin in VDAC1, indicating that VDAC3 is much less sensitive to tubulin than VDAC1. Using 50 nM tubulin we were able to fully inhibit BP-induced swelling, suggesting that there was no involvement of VDAC3. Furthermore, we observed no effect of transfection with VDAC1 siRNA, VDAC2 siRNA or both siRNAs on VDAC3 (Fig. 3A). Thus, it may be concluded that BP-induced tPT is mediated by the VDAC1 and VDAC2 isoforms.
Importantly, bongkrekate inhibited BP-induced pore opening in isolated liver mitochondria (Fig. 4B). This strongly suggests that ANT, and not ATP synthase, may form a pore for the BP-induced tPT, in line with two recent studies demonstrating the involvement of ANT in the PT 38,39 . Moreover, BP-induced mitochondrial swelling was observed to proceed at a very slow rate (Fig. 7) as compared with Ca 2+ -induced swelling (Fig. S1), and be of limited extent (Figs. 4Ai and 7), suggesting that the pore size may be smaller than that of the classic PTP channel, which is in good agreement with the properties of the CsA-sensitive ANT channel in ATP synthase c-subunit knockout mitochondria 38 .
BP-induced apoptosis was inhibited by 2% FBS, but not inhibited by 2% FBS in the co-presence of PKC inhibitors (Fig. 5A); it was also inhibited by PMA treatment (Fig. 5B). These data indicate the involvement of PKC signaling. PMA treatment did not block the tPT but prevented the complete loss of the mitochondrial membrane potential even at 3 h; the mitochondrial membrane potential remained high (19 ± 8%, representing the mean ± SD of the initial TMRM level) (Fig. 5C), compared with the complete loss at 3 h in control (Fig. 1Cii). This retention may be the primary mechanism for the apoptosis inhibition with PMA treatment, as it can block the shift to persistent PTP opening from tPT by preventing the loss of the mitochondrial matrix molecules essential for respiration via the opened pores 56 . The apoptosis inhibition by 2% FBS was accompanied by the inhibition of BP-induced tPT (unpublished data). This fact suggests that other signaling pathways, in addition to the PKC pathway, are needed for the complete inhibition of the tPT of BP. In PKC signaling, PKCε may be involved, as the overexpression of kinase-inactive PKCε counteracted the PMA-inhibition of BP-induced apoptosis by approximately half (Fig. 6). PKCε appears to directly phosphorylate mitochondrial proteins, since PMA-activated rPKCε (but not rPKCα) was able to inhibit BP-induced mitochondrial swelling to the same extent as CsA in isolated rat liver mitochondria (Fig. 7), which was attenuated by the VDAC inhibitor tubulin (Fig. 7B). These data imply a VDAC-mediated mechanism; however, whether PKCε directly phosphorylates VDAC or has an indirect effect on the VDAC through phosphorylating other mitochondrial proteins is unclear. It was reported that PKCε was able to phosphorylate VDAC directly and inhibit Ca 2+ -induced swelling of isolated cardiac mitochondria 33 , although these are challenged 57 . It should be noted that in contrast to the occurrence of Ca 2+ -dependent PTP www.nature.com/scientificreports/ opening in ischemia-reperfusion damage, the BP-induced tPT is apparently independent of the mitochondrial Ca 2+ accumulation. In fact, Ca 2+ preloading was not needed for the BP-induced tPT in isolated liver mitochondria (Figs. 4 and 7), and neither an increase in the cellular Ca 2+ levels at 30 min after BP treatment nor any effect of BAPTA-preloading was observed in Hela cells (Supplementary Fig. S2). Furthermore, even though cellular Ca 2+ might be increased by BP treatment, the rapid decrease in the mitochondrial membrane potential upon BP treatment is expected to prevent the mitochondria from taking up Ca 2+ .
In conclusion, our study demonstrated the pivotal role of tPT in BP-induced apoptosis, which is mediated by the ANT-associated pore, regulated by CypD. It is likely that whether BP-induced tPT is apoptotic or not depends on the PKC signaling, presumably, to the VDAC. These data will help define the cytoprotective mechanism by which BP protects against Ca 2+ -mediated oxidative cell death, which can be counteracted by CsA.

Methods
Animal studies. Wistar female rats were purchased from Japan SLC. Animal experiments were approved (#1800011) by the President of Josai International University after the review by Animal Care and Use Ethics Committee, and were carried out according to relevant guidelines and regulations.
Agar-embedded mitochondria were prepared 45 : aliquots (5 μl) of mitochondrial suspensions, which were prepared by diluting calcein-loaded mitochondria 4 times in assay buffer containing DMSO, 3 μM CsA, 50 μM bongkrekate (Calbiochem) or 50 nM tubulin (Cytoskeleton), were mixed with 5 μl of 2% (w/v) type VII agarose in assay buffer (37 °C), with half of these mixtures placed in circular areas with a diameter of 15 mm circumscribed with a PAP pen on a 24 mm × 32 mm cover glass. After solidification, agar was covered with 0.3 ml of assay buffer. Tubulin was prepared according to the manufacturer's instruction.
Immunoblotting. Commercially available primary antibodies were used: rabbit polyclonal antibodies against CypD (239784, Calbiochem), PKCε (sc-214, Santa Cruz), actin (sc-1615, Santa Cruz), VDAC2 (ab47104, Abcam) and VDAC3 (ab130561, Abcam), and a mouse monoclonal antibody against VDAC1 (ab14734, Abcam). Cells were washed 3 times in PBS in 48-well plates and each well on a heat block was treated with 10 μl of boiled sample buffer (2% SDS, 1% NP40, 5% sucrose in 62 mM Tris-HCl, pH 6.8) containing protease inhibitors (50 μM PMSF, 50 μM leupeptin, 1 μg/ml aprotinin, and 0.5 μM pepstatin). After scraping, the recovered cells were further disrupted by sonication at output 10 for 30 s with a TOMY ultrasonic disruptor UD-200 and centrifuged at 20,000 × g for 3 min at 4 °C. The supernatants, supplemented with 2-mercaptoethanol and bromophenol blue, were again boiled and were loaded into a 12% (for CypD), 10% (for VDAC) or 8% (for PKCε) sodium dodecyl sulfate-polyacrylamide gel electrophoresis. The proteins were transferred to a PVDF membrane. Membranes were blocked with 5% non-fat milk for 2 h in PBS, and then incubated for 1.5−2 days at room temperature with specific primary antibodies: After washing with 0.1% tween 20 in PBS, blots were incubated overnight with biotinylated secondary antibodies and detected using the ABC method (PK-6100, Vector Labs). Confocal microscopy. Fluorescent signals were monitored using Olympus FV-300 and FV-1000 laser scanning confocal microscopes equipped with a × 40 and × 60 oil immersion objective for 35-mm glass-bottomed dishes and cover glasses, respectively. GFP, LTG and calcein were then excited by 488 nm and fluorescence was detected with 505−530 nm band-pass filter; TMRM and MTR were excited at 543 nm and fluorescence was detected with a 560−620 nm filter. For the acquisition and quantification of confocal signals, calcein, cytochrome c-GFP, and TMRM fluorescence signals were calculated by subtracting the background signals from signals in regions of interest, manually applied using the Olympus FV10-ASW software program. The background signals for TMRM were obtained after treatment with 0.5 μM FCCP.

Statistics.
All results are representative of at least three different mitochondrial isolations and at least three cell death experiments. Data were statistically analyzed using a 1-way or 2-way ANOVA, followed by a multiple comparison analysis with a t-test with Bonferroni's correction.