Main

The efficiency of the intrinsic pathway of apoptosis1 depends on the release of mitochondrial pro-apoptotic proteins (e.g., Cytochrome c (Cyt c)) into the cytosol, where they trigger the caspase-dependent execution of the cell.2 Other mitochondria-associated events are: (a) changes in electron transport,3 (b) loss of transmembrane potential (ΔΨm),4 (c) Ca2+ overload5 and (d) re-organization of mitochondrial ultrastructure.6, 7

The regulatory pathways that govern the dynamics of mitochondrial fusion/fission events are the object of general interest for their critical role in both cell physiology and pathology. Their relevance in apoptosis is also widely acknowledged.8, 9 Alterations in ultrastructure occur in numerous diseases and are also a feature of apoptosis. The narrow tubular junctions that connect mitochondrial cristae to the inner boundary membrane support core functions such as: (i) creating gradients of small molecules (adenosine diphosphate),10 (ii) ions’ homeostasis (e.g. Ca2+)11 and (iii) segregation of the respiratory chain complexes.12 Cyt c localizes in the intermembrane space, and disruption of tubular junctions is required to support its release into the cytosol at the onset of apoptotic cell death.13

The endogenous inhibitor of the ATPase, IF1, is a small, basic, heat-stable protein composed of 80–84 amino acids (∼10 kDa) in mammals and predominantly compartmentalized inside the mitochondrial matrix.14 IF1 has the unique capacity to inhibit, through a non-competitive mechanism, the adenosine triphosphate (ATP)-hydrolysing activity of the F1Fo-ATPsynthase without affecting the synthesis of ATP during oxidative phosphorylation. The protein is α-helical along most of its length (∼90 Å15) and at low pH (<6.7) becomes active as a dimer, making this the only condition in which its inhibitory role is exerted.16 The dimerization of IF1—which is critical for the inhibition of the F1Fo-ATPsynthase—involves the C-terminal regions of two monomers (residues 37–8417), which form an antiparallel double-stranded coiled-coil unit stabilized by complementary hydrophobic interactions between the two helices.18 By assessing the role of IF1 on resting mitochondrial physiology and morphology, we recently reported a prominent effect on cristae number and alignment,19 suggesting that IF1 has a role in the regulation of mitochondrial ultrastructure.

Mitochondrial ultrastructure is governed by: (a) the mitofilin/Fcj1 complex, which is essential for cristae development;20 (b) the pro-fusion protein Optic atrophy 1 (Opa1)7 and (c) the dimerization of the F1Fo-ATPsynthase.21 The latter is promoted by IF1, with increased density of cristae, contributing to the morphology of single mitochondria and that of the whole network.22, 23 In addition, dimerization of the F1Fo-ATPsynthase was recently shown to protect cells from starvation,24 highlighting the importance of molecules with a role in this. Beside the outcome on mitochondrial morphology, IF1 is key to the preservation of cellular ATP during hypoxia or ischemia,19 but its outcome on pathways mediating apoptosis remains controversial. IF1 expression is greatly increased in a number of human cancers,25, 26 and recently it has been proposed to interact with the canonical pro-survival pathway, NF-κB,27 priming an adaptive response. Even though this was explained via an IF1-mediated inhibition of mitochondrial oxidative phosphorylation and a resultant increased production of ROS—an aspect which is still controversial23—this implies a mitochondrial downstream effect that would impinge on signaling pathways of cancer cells.

The shaping of the outer mitochondrial membrane is sustained by the pro-fission GTPase Dynamin-related protein 1 (Drp1)28, 29 and, during apoptosis, by the activity of the pro-apoptotic member of the Bcl-2 family Bax,30, 31 whose oligomerization facilitates mitochondrial outer membrane permeabilization.32 Drp1 translocation to mitochondria is a Ca2+-dependent process, as its activity is regulated by phosphorylation and dephosphorylation at several amino-acid residues: a rise in [Ca2+] induces the Ca2+/Calmodulin-dependent activation of Calmodulin kinase I alpha, which phosphorylates Drp1 at Ser600, and Calcineurin, which dephosphorylates Drp1 at Ser637. Both events are crucial to induce the translocation of Drp1 onto the outer mitochondrial membrane.33, 34 Ca2+ per se contributes to apoptosis by facilitating mitochondrial fission when the organelles are overloaded with the ion.35

Moreover, Snyder and co-workers suggested that Cyt c, once released from mitochondria, binds to inositol 1,4,5-triphosphate receptors (IP3Rs), promoting the release of Ca2+ from the endoplasmic reticulum (ER) and increasing Ca2+ accumulation into mitochondria.36, 37 This implies that proteins that limit disassembly of the mitochondrial network, cristae remodeling and release of Cyt c could also affect Ca2+ signaling in apoptosis.

Our data demonstrate that IF1 may modulate apoptosis by regulating mitochondrial morphology, so limiting Cyt c release. They support a model whereby Cyt c promotes ER Ca2+ release, which leads to activation of Drp1 and recruitment of Bax to the outer mitochondrial membrane, where it permeabilises the membrane inducing further Cyt c release; this self-sustaining positive feedback amplification pathway may account for the all-or-none release of mitochondrial Cyt c.38 IF1—via modulation of the mitochondrial architecture—represents an early checkpoint in this pathway, limiting the efficient progression of the apoptotic cascade and protecting therefore the cells from the apoptotic cell death.

Results

IF1 counteracts apoptotic remodeling of the mitochondrial structure

The IF1:F1Fo-ATPsynthase expression ratio was manipulated in HeLa cells as in our previous studies,19, 23 using a blank vector as control. Cells expressing different levels of IF1 were exposed to the pro-apoptotic agent Staurosporine (STS), a broad-spectrum protein kinase inhibitor39 and, after 4 h of treatment, prepared for transmission electron microscopy analysis. The same experiment was repeated using etoposide.

As shown in Figure 1Aa, an extensive remodeling of cristae architecture following treatment with both stimuli characterized IF1 knock down cells (−IF1), while the shape of cristae remained intact in cells overexpressing IF1 (+IF1). The number of cristae per mitochondrion was quantified and plotted in Figure 1Ab.

Figure 1
figure 1

Mitochondrial ultrastructure and remodeling during apoptosis depends on IF1 levels. (A) Transmission electron microscope micrographs of HeLa cells before and after treatment with 1 μM STS and 100 μM etoposide (ETO). (a) An increased number of cristae was evident in +IF1 cells, as highlighted by the quantification in panel (b) (n=3, *P<0.05, **P<0.01). (B) Representative confocal images of HeLa cells transfected with mtGFP before and after challenging with 1 μM STS; an overall preservation of the mitochondrial network was seen in cells overexpressing IF1. Quantification of mitochondrial volume in all conditions is reported in panel (b), indicating that IF1 ratio of expression is involved in the morphological adaptation of mitochondria to apoptosis (n=3, **P<0.01)

Full size image

Using the recombinant mitochondrially targeted green fluorescent protein , we then analyzed the mitochondrial network morphology (Figure 1Ba). In control cells, mitochondria were fragmented and disrupted after STS treatment. In −IF1 cells the network was completely disrupted after only 4 h. However, within the same timeframe, no significant alteration was seen in +IF1 cells (Figure 1Bb).

IF1 reduces apoptotic redistribution of Cyt c

Subsequently, we investigated the entity of Cyt c release during apoptosis in control and +/−IF1 cells via confocal imaging analysis. For this purpose, we transfected cells with the recombinant Cyt c-GFP chimera33 and quantified Cyt c redistribution by measuring the ratio between its mean fluorescence values and the relative standard deviation of the fluorescence signal (see Materials and Methods; Figure 2Aa).

Figure 2
figure 2

IF1 limits Cyt c release and apoptotic cell death. (A) Prototypical images of Cyt c release from mitochondria in control, IF1 overexpressing or IF1 knockdown HeLa cells after challenging with (a) 1 μM STS. The redistribution of Cyt c from the mitochondrial compartment into the cytosol was visualized cotransfecting cells with GFP-tagged Cyt c. The graph in panel (b) shows the percentages of control, IF1 overexpressing and IF1 knockdown HeLa cells releasing Cyt c from mitochondria during treatment with 1 μM STS. Data were acquired every hour over a period of 8 h (n=4 cover slips, *P<0.05, **P<0.01). (B) Bar charts showing the percentage of control and +IF1 HeLa cells with (a) redistributed Cyt c-GFP and (b) dissipated ΔΨm at various time points of STS treatment in the absence or presence of the mPTP inhibitor Cyclosporine A (CsA). Notably, in +IF1 cells a CsA-mediated inhibition of Cyt c-GFP release was only evident at 7–8 h, while in control cells CsA had clear effects at 5–6 h, the time point required by the drug to mediate a protective effect on ΔΨm in both conditions

Full size image

After only 2 h of STS treatment, 11% of control cells showed Cyt c redistribution, while in +IF1 cells this was limited to just 3%, and suppression of Cyt c release remained even after 6 h of treatment. Conversely, in −IF1 cells, the release of Cyt c was greater than control (Figure 2Aa). Values of each condition between 0 and 3 h of treatment were quantified and shown in Figure 2Ab. After 7–8 h of STS treatment, Cyt c release from +IF1 cells increased, and there was no longer any significant difference.

ΔΨm was measured simultaneously during Cyt c–GFP redistribution (Supplementary Figure S1a). In tetramethylrhodamine, methyl ester-loaded control and +IF1 cells, this was done at intervals of 30 min over 8 h of STS treatment. ΔΨm values, measured at every time point, were normalized between the initial baseline and the fully depolarized mitochondrial state, which was obtained by adding FCCP (carbonyl cyanide 4-(trifluoromethoxy)phenylhydrazone) at the end of each experiment.

We observed that Cyt c release occurred in tandem with mitochondrial depolarization and that the events took place simultaneously and interdependently in both the control and +IF1 groups (values reported in Figures 2Ba and Bb).

Cyt c release and the loss of membrane potential were both significantly suppressed in +IF1 cells during STS treatment (Figure 2B).

Although it remains contentious,37 it has been suggested that the formation of the mitochondrial permeability transition pore (mPTP) may promote the loss of ΔΨm and, possibly, Cyt c release during apoptosis;40 the diverse incidence of the two events in the two groups of cells could be, therefore, due to differences in mPTP opening.36 Hence, we repeated the analysis in the presence of Cyclosporine A, a pharmacological inhibitor of mPTP.41 In control cells, Cyclosporine A significantly suppressed both redistribution of Cyt c (Figure 2Ba) and dissipation of ΔΨm (Figure 2Bb) after only 5–6 h of STS treatment, while in +IF1 cells the activity of Cyclosporine A was delayed, and a same effect was reached after 7–8 h. These data support a role for the mPTP in the loss of ΔΨm during apoptosis.

Apoptotic cell death is modulated by the IF1:F1Fo-ATPsynthase ratio

Along with the retention of Cyt c and ΔΨm in the initial phases of apoptosis, the exposure of phosphatidylserine on the plasma membrane and the activation of Caspase 3 were also limited by IF1 overexpression and facilitated by its downregulation (Figures 3Aa and Ab).

Cell survival was measured following 12 h exposure to Staurosporine, etoposide or C2-ceramide. A significant protection from apoptosis was observed in +IF1 cells and the opposite in those downregulated for the gene (Figure 3B). Perhaps surprisingly, increased IF1:F1Fo-ATPsynthase ratio of expression was also protective from cell death triggered by the agonistic anti-CD95 antibody (α-CD95), which activates the extrinsic pathway of apoptosis (Supplementary Figure S1bi).42 Although not recapitulated by the deletion of the gene, as −IF1 cells undergo cell death in equal measure of control, we asked whether IF1 altered the expression of Caspase 8, which links the extrinsic and intrinsic pathways of apoptosis (Supplementary Figure S1bii). This analysis revealed that Caspase 8 expression was not modulated by changes in IF1 expression but that in +IF1 cells treatment with α-CD95 promotes a more prominent cleavage in the p43/41 fragments, which would account for an increased activation of the pathway.

Figure 3
figure 3

IF1/F1Fo-ATPsynthase ratio of expression regulates cell response to apoptosis modulating the activity of Bax. (A) Bar charts displaying the percentage of control and IF1 up/downregulated HeLa cells with (a) caspase activation and (b) positive Annexin-V staining after 14 h treatment with 1 μM STS. Remarkably, in +IF1 cells caspase activation and exposure of phosphatidylserine to the extracellular environment were lower than in control cells, while an opposite situation was seen in −IF1 cells (*P<0.05, **P<0.01). (B) Chart reporting the percentage of living cells after 12 h treatment with 1 μM STS, 20 μM C2-Cer or 100 μM etoposide (ETO) in control, +IF1 and −IF1 HeLa cells. An augmented resistance to all apoptotic stimuli was evident in +IF1, cells, whereas −IF1 cells showed an opposite effect (n=3, *P<0.05, **P<0.01). (C) (a and b) Western blot analysis of cells modulated for IF1 expression, showing that total cellular levels of Bax were not affected by the expression levels of IF1. Band densities have been quantified and normalized to GAPDH (glyceraldehyde 3-phosphate dehydrogenase). The western blot in panel (c) shows instead that, during STS-induced apoptosis (4 h treatment with 1 μM STS), Bax accumulation was limited in cells overexpressing IF1. After treatment, cells devoid of IF1 exhibited higher levels of Bax in the mitochondrial fraction, indicating an increased mitochondrial recruitment during apoptosis (band densities, quantified and normalized to the β-chain of F1Fo-ATPsynthase, are reported in panel (d); n=6; *P<0.05, **P<0.01). (D) Representative immunocytochemistry images showing Bax translocation to mitochondria in control, +IF1 and −IF1 HeLa cells after 4 h treatment with 1 μM STS

Full size image

IF1 overexpression limits the activity of Bax and Drp1 on mitochondria

During apoptosis, Bax translocates onto the MOM and, after a conformational shift with exposure of the C-terminus, inserts into it, facilitating the release of Cyt c.43 We therefore assessed Bax expression levels in control and +/−IF1 cells. Bax appeared to be evenly expressed in all the cell types at resting physiology (Figure 3Ca).

Fractions of purified mitochondria isolated after 4 h of treatment with STS were probed with an anti-Bax antibody; although almost no staining was detectable in +IF1 cells, the observed band intensity in −IF1 cells was much greater than control cells (n=5) (Figure 3Cc). In addition, this trend was also confirmed by immunocytochemical analysis (Figure 3D).

Mitochondrial outer membrane permeabilisation, effected by Bax, depends on the remodeling of membranes, for which the recruitment of Drp1 to mitochondria is required.30 Mitochondrial fractions tested for Drp1 expression showed that downregulation of IF1 facilitates the relocation of Drp1 onto mitochondria during apoptosis, as the protein was retrieved in greater amount when compared with the cytosolic fraction, whereas the opposite was seen following overexpression of IF1 (Figure 4Aa). As for the analysis of Bax, we used an immunocytochemical analysis to confirm the data (Figure 4Ac).

Figure 4
figure 4

Drp1 activation and relocation on mitochondria depends on IF1. (A) (a and b) Western blot (WB) analysis indicating that Drp1 recruitment to mitochondria during apoptosis (still 4 h treatment with 1 μM STS) was increased after knocking down IF1 (band density quantification for mitochondrial and cytosolic Drp1 is reported below; n=4, *P<0.05, **P<0.01). (c) Immunocytochemical analysis of Drp1 recruitment on mitochondria—still at 4 h treatment with STS. In cells devoid of IF1, Drp1 colocalized on mitochondria at higher levels, as highlighted in the representative images reported and relative intensity. (B) During apoptosis, (a) activated Drp1 (phosphorylated at Ser616) was detectable in WB analysis in the mitochondrial fraction of −IF1 cells (with relative band density quantification reported in panel (b); n=3, *P<0.05, **P<0.01). (C) Chart reporting the percentage of living cells after 12 h treatment with 1 μM STS in control, +IF1 and −IF1 HeLa cells co-transfected with the enzymatically inactive mutant Drp1K38A cDNA. The Drp1-dependent effect of IF1 is revealed by the fact that, when Drp1 was not active, all cell populations were protected from apoptosis, with no detectable differences in the percentage of living cells after STS treatment (n=2, *P<0.05, **P<0.01)

Full size image

In addition, Drp1 activation differed between mitochondria expressing diverse levels of IF1, as extracts of −IF1 cells were more strongly stained with the anti-pospho-Drp1(S616) antibody that recognizes the active form of the protein (Figure 4B).

These data suggest that IF1 overexpression reduces Drp1 translocation to mitochondria, as well as the interdependent translocation and oligomerisation of Bax. We therefore expressed the dominant-negative mutant of Drp1 (Drp1K38A) in +/−IF1 cells, and measured the percentage of living cells after 12 h incubation with STS. Under these conditions, we saw no differences between the three conditions compared with cells expressing naive levels of Drp1 (Figure 4C). The same outcome was obtained in Drp1−/− cells (data not shown).

IF1 limits Ca2+ mobilization during apoptosis activation

According to Boehning et al.,36 Cyt c binds to IP3Rs on the adjacent ER at the beginning of apoptosis, thus enhancing Ca2+ release, mitochondrial Ca2+ loading and opening of the mPTP, so amplifying the release of Cyt c. As the redistribution of Cyt c and the loss of mitochondrial potential were significantly delayed in +IF1 cells, we explored the potential role of Ca2+ signaling in this pathway. Cells transfected with IF1 cDNA and control cells were loaded with Fura-2, AM, treated with STS and imaged over 8 h. Alternative patterns of [Ca2+]c signals were seen with varying frequencies in control and +IF1 cells (Figures 5Aa and Ab). The majority of +IF1 cells (85.60%) showed no change in [Ca2+]c compared with ∼40% of control cells, in which a significant percentage showed either a progressive increase (13.09%) or spikes (46.86) in [Ca2+]c, which were very rarely observed in +IF1 cells. IF1 overexpression reduced STS-induced changes in [Ca2+]c, and this effect correlated with the delay in Cyt c release from mitochondria (as quantified in Figure 2Ab).

Figure 5
figure 5

IF1 overexpression counteracts Ca2+ mobilization and Calcineurin (CaN) activation during apoptosis. (A) (a) Traces of control and +IF1 HeLa cells loaded with 5 μM Fura-2, AM, treated with 1 μM STS and imaged for 8 h, monitoring changes in cytosolic Ca2+ levels. Three types of cytosolic [Ca2+] patterns were observed. Data summarized in panel (b) illustrate that a significantly higher percentage of +IF1 cells showed no changes in cytosolic [Ca2+] levels as compared with control cells. Conversely, a higher percentage of control cells showed rises and spike-like increases in cytosolic [Ca2+] during STS treatment. (n=3 cover slips, 15–29 cells per cover slip, *P<0.05). (B) Representative traces of changes in [Ca2+]c in control and +IF1 HeLa cells exposed to 1 μM STS and then challenged with 500 nM Tg to specifically inhibit the activity of sarco/ER Ca2+-ATPases. The release of Ca2+ in the cytosol was measured to infer ER Ca2+ content. In the absence of (a) STS, ER Ca2+ content was not significantly different in control and +IF1 cells; (b) after 2 h of STS treatment, [Ca2+]ER content was instead greatly depleted in control cells, but better maintained in +IF1 cells. (C) Representative (a) images and (b) kinetics of mitochondrial Ca2+ handling monitored with Rhod-2, AM during treatment with 1 μM STS in control and +IF1 HeLa cells. Although control cells exhibited a rise in [Ca2+]m after challenging with STS, in +IF1 cells mitochondrial Ca2+ levels were stably maintained (normalized signals; 1 h, Control: 1.34±0,39; +IF1: 2.23±0.49, 2 h, Control: 2.21±0.07; +IF1: 1.28±0.13; 3 h, Control: 1.18±0.05; +IF1: 0.92±0.08; n=4, 10 cells per condition, *P<0.05). (D) Histogram reporting the enzymatic activity of CaN in control, +IF1 and −IF1 cells. Cells were treated for 4 h with STS to induce apoptosis or with DMSO as a control. The activity of CaN has been measured and reported as nmol of PO4 released (values are normalized to control). Notably, in −IF1 cells this value was higher at resting conditions and after triggering apoptosis, indicating a higher activation of CaN, while IF1 overexpression inhibits this event (*P<0.05; **P<0.01). (E) Model for IF1 functional role in apoptosis: IF1 preserves the inner mitochondrial membrane structure by increasing the cristae stability,1 reducing the release of Cyt c2 and hence avoiding the Cyt c-induced mobilization of ER-stored Ca2+.3 This limits CaN activity4a,b and hence Drp1 recruitment/activation on mitochondria, also hindering the function of Bax.5 The delayed massive mitochondrial release of Cyt c prevents the formation of the apoptosome into the cytosol and the execution of apoptosis6

Full size image

The differences in [Ca2+]c patterns may be attributed to differences in the initial release of Cyt c, which may modulate ER Ca2+ signaling so amplifying Cyt c release to completion. We therefore measured the ER Ca2+ content using Thapsigargin (Tg, 500 nM) at several time points following exposure to STS. Representative traces of changes in [Ca2+]c following STS treatment and Tg application are shown in Figures 5Ba and Bb. As reported in Supplementary Figure S1ciii, [Ca2+]ER was substantially depleted in control cells compared with +IF1 cells, consistently with the increased release of ER Ca2+ seen in control cells.

Remarkably, as a chance observation, we noted that STS altered the capacitative Ca2+ influx in both control and +IF1 cells; in fact, while in untreated cells [Ca2+]c remained elevated after addition of Tg (Figure 5Ba), it decreased to basal levels in STS-treated cells (Figure 5Bb). A similar pattern occurred up to 6 h of STS treatment, as reported in Supplementary Figures S1ci and cii.

Uptake of Ca2+ by mitochondria was then measured using Rhod-2, AM (Figure 5Ca). After 2 h of STS treatment, mitochondria of +IF1 cells showed no significant increase in Ca2+ content, while this was elevated in control cells. Quantified data are plotted in Figure 5Cb.

In cells overexpressing IF1, the limited mobilization of Ca2+ also reduced the activity of the serine/threonine protein phosphatase Calcineurin. Activity of Calcineurin, the Ca2+-dependent activator of Drp1, was reduced after treatment with STS in +IF1 cells, with an opposite result in −IF1 cells (Figure 5D).

The hierarchy of the events we attribute to IF1 is presented as a cartoon in Figure 5E.

Discussion

Mitochondrial fragmentation and cristae remodeling augment Cyt c release to completion within minutes during apoptosis.2, 38 Proteins involved in the regulation of mitochondrial dynamics have an active part in this process, opposing or facilitating the structural changes of the organelle.44 For example, Opa1 maintains the shape of cristae, reduces the mobilization of Cyt c and delays cell death,45 while Drp1, which is essential for mitochondrial fission, has an opposite effect, by sustaining the oligomerization of Bax.30 Here, we demonstrate that IF1 also contributes to the structural re-arrangement of mitochondria during apoptosis, modifying the mobilization of Cyt c and so altering the downstream cascade of events.

Our data suggest that IF1 overexpression limits mitochondrial release of Cyt c, so that the downstream events that amplify Cyt c release and complete the apoptotic program are delayed. Overexpression of IF1 preserved cristae morphology (Figure 1) and reduced Cyt c release (Figure 2), Ca2+ signaling (Figure 5), Drp1 activation (Figure 4) and Bax oligomerisation (Figure 3) for 7 h after STS treatment (Figures 1, 2, and 4), also delaying: (i) caspase activation, (ii) phosphatidylserine translocation and (iii) cell death (Figure 3). The opposite was seen in cells in which IF1 expression was suppressed, as just 2 h of treatment with STS were sufficient to produce an almost complete re-distribution of Cyt c (Figure 2Aa).

Interestingly, IF1 was also protective when apoptosis was activated via the CD95/APO-1/Fas signaling pathway (Supplementary Figure S1bi). Because the Caspase 8 cascade, via cleavage of BID,46 merges on the mitochondrial pathway. The increased activation retrievable in +IF1 cells in which apoptosis is delayed, does imply a prominent mitochondria mediated effect.

Mitochondria were not immune from the apoptotic-associated events when IF1 was overexpressed, as both migration of Cyt c and dissipation of ΔΨm did take place, even if at a later stage—after 7 h of incubation with STS. Even though at this time point the events on which protection by IF1 is primed in the initial phases are statistically normalized, it has to be noted that (a) the trend toward delay of Cyt c release and ΔΨm dissipation remains similar to those statistically proven at 3–4 h and (b) modulation of both events occurring earlier in time are associated with a significant protection from cell death when IF1 is overexpressed (Figure 3).

Notably, such a regulation of apoptosis efficiency by IF1 appears independent of the canonical action of the protein as an inhibitor of the reverse mode of the F1Fo-ATPsynthase.47 One might predict that apoptosis itself should predispose to a similar scenario, with reversal of the ATP synthase and accelerated ATP consumption so that ΔΨm dissipation would occur faster in +IF1 cells. This was not the case: even though IF1 dimers were still present after STS treatment with a greater degree (Supplementary Figure S1d), these did not influence ΔΨm that dissipated concomitantly with the release of Cyt c and, consequently, with the permeabilisation of mitochondria.

The regulation of apoptosis by IF1 embeds other elements of the process, such as (a) the multidomain protein Bax and (b) Drp1. Cuezva and co-workers have proposed that IF1 overexpression influences proteins of the Bcl-2 family by increasing the expression of Bcl-XL.27 For this reason, we explored whether expression of Bax was also altered, but we found no changes in its expression despite alternative degrees of IF1 expression. We only observed changes during active apoptosis and just in the mitochondrial localization of the pro-apoptotic molecule, an event that was almost absent in IF1-overexpressing cells (Figures 3Cc and Cd). Even though such an effect may be due to the increment of the anti-apoptotic counterparts of the same family—hence in line with the findings of an increased presence of BcL-XL48, 49—other events may still modify the anchoring capacity of Bax, and the translocation/activation of the pro-fission protein Drp1 is one of these (Figures 4A and B).30 The effects of IF1 on mitochondrial morphology (Figure 1) primed the analysis based on the recently proposed model of Martinou and colleagues, in which the degree of mitochondrial Drp1 is instrumental for promoting Bax-contributed remodeling of the membranes.30 This explains our findings whereby in −IF1 cells a dramatic increase in mitochondrial translocation of Drp1 is followed by activation (Figure 4B).

Thus IF1, although intramitochondrially localized, can reach out and affect mediators of apoptosis localized in the cytosol (e.g., Drp1), as confirmed by the fact that modulation of IF1 expression was ineffective in altering the apoptotic response when cells expressed the dominant-negative mutant of Drp1 (Figure 4D). A similar scenario—with different actors—was also proposed in a previous relevant work, with a role for NF-κB as the underlying proliferative pathway via which IF1 exploited its anti-apoptotic and tumorigenic role.27 As this was linked to concomitant changes in mitochondrial respiration that we have not observed under any condition, we remain unsure how to reconcile these different data sets.

It is possible to speculate that reduced translocation of Drp1 to mitochondria depends on alterations in the ratio of hFis-150 per mitochondrion—considering the organelle’s elongation after Atpif1 upregulation—however, because at resting physiology we find the same amount of mitochondrial Drp1 regardless the IF1:F1Fo-ATPsynthase ratio, this may be mediated by upstream Ca2+-dependent events.

It has been proposed that Cyt c, once released from the mitochondria into the cytosol, binds to IP3Rs, inducing ER Ca2+ release37 and mitochondrial Ca2+ overload, which promotes swelling and opening of the mPTP.51 In addition, Drp1 dephosphorylation by Calcineurin, and its subsequent translocation to mitochondria, is also mediated by Ca2+. As IF1 delayed the release of Cyt c and suppressed ER Ca2+ mobilization in both cytosolic and mitochondrial compartments, we explored the activity of Calcineurin as the logical link between Ca2+ and Drp1.33, 52 Consistent with this model, in +IF1 cells activation of the protein phosphatase was much reduced.

Doubts remain about whether the elongated shape of mitochondria affects the contact sites between ER and mitochondria and, ultimately, the Ca2+-buffering capacity of mitochondria; however, the kinetics of cytosolic [Ca2+] during STS treatment being equal in both the control and +IF1 cells after Tg challenge, we ruled out this option.

Thus, we propose that, besides its role in ischemia,19 IF1 also protects from apoptotic cell death, primarily by influencing mitochondrial ultrastructure and membrane remodeling.19, 23

Our data support a model of apoptosis that involves a cascade of actions, which will serve as an amplification mechanism to ensure the rapid, complete and efficient release of all mitochondrial Cyt c within a short time frame. Some elements of the scheme, outlined in Figure 5, have been controversial, but it turns out that increased expression of IF1 proves an invaluable tool to interrupt this cascade, revealing the importance of downstream processes.It is therefore tempting to propose that the increased expression of IF1 described in human cancers may point to a tumorigenic role for IF1, given its antiapoptotic action; assaying IF1 levels in cancer biopsies may thus provide a useful prognostic indicator for chemotherapy responsivity.

Materials and Methods

Experiments were carried out in Human Cervical Adenocarcinoma cells (HeLa), wild-type and Drp1−/− mouse embryonic fibroblasts. Cells were transfected as explained previously.19

Investigating mitochondrial morphology

Confocal images of mitochondrially targeted GFP-positive cells were acquired as previously described.23 For electron micrograph analysis, transfected cells were grown on gridded cover slips and 36 h after transfection identified by fluorescence microscopy and their localization marked. Cells were processed as described in precedence19 and images acquired with a 2 JEOL 1010 transmission electron microscope (Jeol Ltd., Tokio, Japan). Acquired images were analyzed using Image J (National Institutes of Health, Bethesda, MD, USA) and ‘Lucida’ (Kinetic Imaging, Andor Technology plc., Belfast, UK) to define the number of cristae per mitochondrion.

Cell death analysis

At a confluence of ∼50%, cells cultured on Ø 22 mm borosilicate cover slips were transfected with IF1 cDNA, IF1 siRNA or yellow fluorescent protein (YFP) cDNA (as a control). After 36 h of transfection, cells were incubated for 12 h at 37 °C, 5% CO2 with one of the following apoptosis inducers: 1 μM STS (Calbiochem 569 397; Merck Millipore, Billerica, MA, USA), 100 μM etoposide (Sigma-Aldrich Co, St. Louis, MO, USA), 20 μM C2-ceramide or 1 μg/ml anti-CD95 (Clone CH11) Ab (No. 05-201, Millipore, Merck Millipore). After that time, a period established to achieve significant death but leaving a reasonable number of surviving cells for adequate data collection, cells were washed with recording medium and cover slips were assembled into a chamber for confocal imaging. Annexin V and poly-caspases analysis were carried out according to the manufacturers’ instructions (BD Pharmingen TM 550 911; BD Biosciences, San Jose, CA, USA), Poly-Caspases FLICA kit, category No. 92, IT) with a Zeiss LSM confocal laser scanning microscope (Carl Zeiss Group, Oberkochen, Germany) (inverted configuration on an Axiovert 200), using a ‘Fluar’ × 40/1.30 oil immersion objective. Images were analyzed using the LSM Image Browser software (Carl Zeiss Group).

Evaluation of Cyt c redistribution

Cyt c release was followed using the Cyt c-GFP fusion protein. Cells were seeded on Ø 22 mm borosilicate cover slips and, when ∼50% confluent, cotransfected with either IF1 and Cyt c-GFP cDNAs (+IF1 cells), IF1 siRNA (Hs_ATPIF1_2; SI00308112, Qiagen, Hilden, Germany) and Cyt c-GFP cDNA (−IF1 cells) or an empty vector and Cyt c-GFP cDNA (control cells). Thirty-six hours after transfection, each cover slip was mounted into a chamber for live-cell imaging and inserted into the thermostatted chamber of a Zeiss LSM 510 confocal laser scanning microscope (inverted configuration on an Axiovert 200 frame), with temperature constantly set at 37 °C. Images were acquired using a ‘Fluar’ × 40/1.30 oil immersion objective and an Argon laser (488 nm line; 1% laser power) to excite GFP, which fluorescence was captured through the HFT 488 dichroic mirror with BP 500/530 band-pass filter. At the start of each experiment, one image was captured for baseline measurements; then, 1 μM STS was added to recoding medium and images were hourly acquired until end (8 h after STS addition for +IF1 cells, 4 h for −IF1 cells). Analysis was carried out using AQM Advance 6 software (Andor, Belfast, UK) to obtain ratios of the S.D. of the green-fluorescent signal relative to the mean (S.D./Mean) for the Cyt c–GFP fluorescence emission; this gives a quantifiable measure of Cyt c redistribution in each region of interest analyzed. In brief, when the Cyt c-GFP chimeric protein is localized into mitochondria, the S.D. is high because there are bright pixels (corresponding to mitochondria) interspersed with black pixels (corresponding to cytoplasm). Once Cyt c–GFP redistributes through the cytosol, the S.D. falls as the signal is evenly distributed throughout the cell; this needs to be normalized with respect to the mean signal simply because S.D. is also a function of the overall intensity. The S.D./Mean values of Cyt c–GFP signal in all the analyzed cells were binned to explore the distribution. Changes in distribution of the ratio following treatment with STS from time 0–4 h were plotted for control and +IF1 cells. The distribution for control cells displayed a leftward shift from time 0–4 h after STS treatment compared with that of +IF1 cells, showing that more control than +IF1 cells had undergone Cyt c redistribution. Control cells analyzed at 4 h from STS addition were expected to consist of two populations, one showing Cyt c release and the other one without. Indeed, the distribution plot for control cells at 4 h was fitted well by the sum of two Gaussian distributions representing the two sub-populations of cells. An intersection point between the two Gaussian populations was demarcated and threshold S.D./Mean values of the Cyt c-GFP signal were set to define the two populations. Cells with S.D./Mean ratios of <0.6 were defined as those that showed Cyt c redistribution, and cells with S.D./Mean >0.7 were considered as those that did not show Cyt c release from mitochondria.

Ca2+ measurements

The red-fluorescent ratiometric Ca2+ indicator Fura-2, AM was used to measure cytosolic-free Ca2+ levels in live cells. At low concentrations of the indicator, 340/380 nm excitation ratio allows accurate measurements of the intracellular concentration of the divalent cation. Once ∼50% confluent, cells seeded on Ø 22 mm borosilicate cover slips were transfected with either IF1-YFP cDNA or YFP cDNA. After 36 h of transfection, cells were loaded with 5 μM Fura-2, AM and in presence of 0.005% (w/v) pluronic acid (30 min incubation at RT—∼21 °C). The loading solution was then removed, and cells were washed twice with modified Hanks’ Buffered Salt Solution (HBSS 138 mM NaCl, 5.4 mM KCl, 0.25 mM Na2HPO4, 0.44 mM KH2PO4, 1.3 mM CaCl2, 1.0 mM MgSO4, 4.2 mM NaHCO3, pH 7.4). Each cover slip of cells was assembled into a purpose-built chamber, a fresh aliquot of modified HBSS with or without 1 μM STS was added to the cells and placed on the stage of a Leica TCS SP5 II confocal laser scanning microscope (Leica Microsystems GmbH, Wetzlar, Germany). Temperature inside the environmental chamber of the microscope was constantly maintained at 37 °C. Images were acquired for 8 h using a HCX PL APO 40 × /1.25-0.75 oil immersion CS objective and collecting Fura-2 emission at 505 nm in the ratio mode with excitation of the dye at 340 and 380 nm light (using an UV laser, a 400 long pass dichroic mirror and a 510/540 band pass filter). Settings were adjusted to minimize phototoxicity. Acquired images were analyzed with the AQM Advance 6 software making a 340/380 nm.

For experiments measuring ER Ca2+ content, cells loaded with Fura-2, AM and treated with 1 μM STS for the required number of hours were challenged with Tg, a sarcoplasmic reticulum Ca2+-ATPase inhibitor, to prevent Ca2+ reuptake into the ER.

Mitochondrial Ca2+ levels were analyzed with the red-fluorescent Ca2+ indicator Rhod-2, AM. Cells were seeded on Ø 22 mm borosilicate cover slips and transfected with either IF1-YFP cDNA or YFP cDNA when ∼50% confluent. After 36 h, cells were loaded with 5 μM Rhod-2 in the presence of 0.05% (w/v) pluronic acid in modified HBSS (45 min incubation at 37 °C, 5% CO2). After four washes with modified HBSS to remove excess dye, cover slips were assembled into a live-cell imaging chamber and an aliquot of modified HBSS with 1 μM STS (or DMSO (dimethyl sulfoxide) vehicle as a control) was added. Cover slips were then seeded into the thermostatted chamber of a Zeiss LSM 510 Confocal Laser Scanning Microscope (inverted configuration on an Axiovert 200 frame) and cells were imaged using a ‘Fluar’ 40 × /1.30 oil immersion objective and a green Helium-Neon laser (543 nm line; 1% laser power) with a BP 565/615 band-pass filter. Settings were chosen to reduce photobleaching of the fluorophore and the pinhole set to give an optical slice of ∼2 μm. Acquisitions were taken every 30 min to obtain a time course of mitochondrial Ca2+. Images were analyzed with the LSM Image Browser.

Western blotting

At a confluence of ∼50%, cells cultured on 100 mm Petri dishes were transfected with either IF1 cDNA, IF1 siRNA or YFP cDNA (as a control). After 36 h of transfection, cells were incubated with 1 μM STS supplemented recording medium (4 h incubation at 37 °C, 5% CO2) to induce apoptosis. Subcellular fractionation was then executed. Cells were mechanically harvested in 5 ml of ice-cold PBS with a cell scraper and centrifuged to pellet cells. Pellets were resuspended in 1 ml of isotonic buffer supplemented with protease inhibitors, and cells were lysed by homogenization with 60 strokes using a Dounce homogenizer (Sigma-Aldrich Co.). Homogenates were centrifuged for 10 min at 600 × g at 4 °C; pellets (nuclei and debris) were discarded and supernatants were centrifuged for 10 min at 13 000 × g at 4 °C. Pellets (mitochondria) were dissolved in 1 ml isotonic buffer and centrifuged as above; while supernatants were centrifuged at 100 000 × g, and the resulting supernatants were designated as cytosolic fractions. Mitochondria were then lysed by incubating mitochondrial pellets with ad hoc lysis buffer (50 mM Tris-HCl pH 7.4, 50 mM NaCl, 1% Triton X-100, 1 mM EDTA, 1 mM EGTA, 5 mM MgCl2) supplemented with protease inhibitors for 30 min at 4 °C on rotating platform, followed by 30 min centrifugation at 14 000r.p.m. in a refrigerated benchtop centrifuge; supernatants were designated as mitochondrial fractions. Proteins’ concentration was analyzed with Pierce BCA Protein Assay Kit (Thermo Scientific Inc., Rockford, IL, USA).

Western blotting was carried out as previously described, using antibodies against IF1 (1 : 1000, a generous gift from Prof. Sir John Walker), Bax (N-20) (1 : 100; Sc-493, Santa Cruz Biotechnology Inc., Santa Cruz, CA, USA), Drp1 (C-5) (1 : 1000; Sc-271583, Santa Cruz Biotechnology), phospho-Drp1(Ser616) (1 : 1000; #3455, Cell Signaling Technology Inc., Beverly, MA, USA), Caspase-8 (1C12) (1 : 750; #9746, Cell Signaling Technology), ATPase β-subunit (1 : 10 000; ab14 730, Abcam plc., Cambridge, UK), GAPDH (glyceraldehyde 3-phosphate dehydrogenase; 1 : 20 000; ab9482, Abcam) and TSPO (1 : 10 000; ab109497, Abcam).

Immunocytochemistry

For investigating on Bax recruitment to mitochondria in cells modulated for IF1 expression, at a confluence of ∼50%, cells growth on Ø 22 mm borosilicate cover slips were transfected with either IF1-YFP cDNA, IF1 siRNA and YFP cDNA or YFP cDNA (as a control). After 36 h of transfection, apoptosis was induced by treating cells with 1 μM STS in recording medium (4 h incubation at 37 °C, 5% CO2); as a control, cells were treated with DMSO vehicle. Cells were then washed with warm PBS and fixed with 4% paraformaldehyde in PBS (15 min incubation at RT). Following washes with PBS at RT, cell permeabilization was achieved through incubation with 0.1% Triton X-100 in PBS (30 min at RT). Cells were then washed three times with PBS for 5 min at RT. Background or unspecific staining were avoided using normal goat serum blocking solution (10% normal goat serum, 3% bovine serum albumin in 0.01% Triton X-100 in PBS; 1 h incubation at RT). Cells were incubated in the dark overnight at 4 °C with the primary antibody to Bax (6A7) (sc-23959) at dilution 1 : 200 (in 0.01% Triton X-100 in PBS). After five washes with PBS to remove unbound primary antibodies, Alexa Fluor 555-conjugated secondary antibody to mouse IgG was added at dilution 1 : 1000 (in 0.01% Triton X-100 in PBS; 1 h incubation in the dark at RT). Unbound secondary antibodies were removed by washing cells with PBS five times. Nuclei were stained with 10 μM TO-PRO®-3 iodide in PBS (20 min incubation in the dark at RT). After five washes with PBS, cover slips were mounted with AF1 mountant solution (Citifluor Ltd., London, UK). Negative controls were performed by omission of primary or secondary antibody. Cells were imaged on a Zeiss LSM 510 Confocal Laser Scanning Microscope (inverted configuration on an Axiovert 200 frame) using a ‘Fluar’ × 40/1.30 oil immersion objective. To avoid contamination between fluorescent dyes emission, images were acquired in three consecutive scans, using an Argon laser (488 nm line) for YFP, a green Helium-Neon laser (543 nm line) for Alexa Fluor 555 and a red Helium-Neon laser (633 nm line) for TO-PRO-3 iodide. Illumination intensity was kept to a minimum (0.1–2%) to avoid bleaching of the fluorescent signals. Settings were chosen to minimize bleaching. The fluorescence emission of YFP was captured through the HTF 488/514 dichroic mirror with BP 500/550 band pass filter, Alexa Fluor 555 through the HFT 488/543 dichroic mirror with BP 535/590 and TO-PRO-3 through the HTF UV/488/543/633 dichroic mirror with LP 650 high-pass filter. Acquired images were processed with the LSM Image Browser software.

The same procedure was followed for immunocytochemistry analysis of mitochondrial translocation of Drp1. Cells were cotransfected with the mitochondrially targeted red fluorescent protein and a blank vector (control cells), IF1 cDNA (+IF1 cells) or an siRNA against IF1 (−IF1 cells). Cells were then treated with STS for 4 h and stained with an anti-Drp1 antibody and an Alexa Fluor 488-conjugated anti-Mouse IgG antibody (Invitrogen; Life Technologies Ltd., Paisley, UK). DAPI was used for nuclear staining.

Calcineurin activity assay

At a confluence of ∼50%, cells cultured on 100 mm Petri dishes were transfected with IF1 cDNA, IF1 siRNA or an empty vector (as a control). After 36 h of transfection, cells were incubated with 1 μM STS supplemented medium (4 h incubation at 37 °C, 5% CO2) to induce apoptosis; DMSO vehicle was used as a control. After treatment, Calcineurin activity was measured using a Calcineurin cellular activity assay kit (Enzo Life Sciences Inc., Farmingdale, NY, USA), following manufacturer’s instructions.