Reagents, Cell Culture, and Transfection

Mouse embryonic fibroblasts (MEFs) were cultured as described before (Scorrano et al., 2003). Transfection was performed using Transfectin Lipid Reagent (Bio-Rad, Richmond, CA) following the manufacturer's instructions. Cyclosporine H was generously provided by P. Bernardi (University of Padova). All chemicals, unless specified, were from Sigma (St. Louis, MO).

Real-Time Imaging of Mitochondrial Membrane Potential

MEFs, 1 × 10⁵, plated on 24-mm round glass coverslips were transfected and after 24 h were loaded with 20 nM tetramethyl rhodamine methyl ester (TMRM) dissolved in Hanks' balanced salt solution (HBSS) supplemented with 1.5 μM cyclosporine H (a P-glycoprotein inhibitor) or cyclosporine A for 30 min at 37°C. Cells were then placed on the stage of an Olympus IMT-2 inverted microscope (Melville, NY) equipped with a CellR imaging system. After identification of GFP-positive cells, sequential images of TMRM fluorescence were acquired every 60 s using exposure times of 30 ms with a 40×, 1.4 NA Plan Apo objective (Olympus). Cells were excited using a 525/20 BP excitation filter, and emitted light was acquired using a 570/LP filter. Images were stored and analysis of TMRM fluorescence over mitochondrial regions of interest was performed as described in Scorrano et al. (2003).

Confocal Imaging and Morphometric Analysis

MEFs, 4 × 10⁵, plated on 13-mm-diameter glass coverslips (VWR Scientific Products, West Chester, PA) and transfected as indicated in the figure legends were incubated after 24 h in HBSS supplemented with 10 mM HEPES. For confocal images of mitochondrial network, red-channel images were acquired using the detector assembly of a Nikon Eclipse E600FN microscope (Melville, NY) equipped with a Bio-Rad Radiance 2100 Confocal Laser Scanning system. Morphometric analysis was performed as described (Cipolat et al., 2004). Cells were scored with fragmented or coalesced mitochondria when more than 50% of the objects in the image (i.e., mitochondria) displayed aberrant morphology (either fragmented or coalesced). For confocal imaging of mitochondrial movement, mtRFP fluorescence was imaged using a Nikon Eclipse TE300 inverted microscope equipped with a spinning-disk Perkin Elmer-Cetus Ultraview LCI confocal system (Norwalk, CT) and a Orca ER 12-bit CCD camera (Hamamatsu, Bridgewater, NJ).

Polyethylene Glycol Fusion Assay

For polyethylene glycol (PEG) fusion assay, 5 × 10⁵ MEFs were cotransfected with mtYFP or with mtRFP and β-Gal or HA-Fis1. After 24 h cells labeled with different fluorescent proteins were coplated in a 1:1 ratio onto 13-mm round coverslips. Fusion was induced after 24 h by a 60-s treatment with a 50% (wt/vol) solution of PEG1500 in PBS, followed by extensive washes in DMEM supplemented with 10% FCS. To inhibit de novo synthesis of fluorescent proteins, 30 min before PEG treatment cells were incubated with the protein synthesis inhibitor cycloheximide (20 μg/ml), which was subsequently kept in all solutions and tissue culture media until cells were fixed for 30 min with ice-cold 3.7% (vol/vol) formaldehyde in PBS. After two washes with PBS, coverslips were mounted on slides using Anti-Fade Reagent (Molecular Probes, Eugene, OR). Confocal images were acquired and fusion was calculated exactly as described in Cipolat et al. (2004).

Immunofluorescence

MEFs, 6 × 10⁵, plated on 13-mm-diameter glass coverslips (VWR Scientific Products) were transfected as indicated in the figure legends, and after 24 h immunofluorescence was performed as described in Griffiths et al. (1999). Briefly, cells were fixed in 0.25% (wt/vol) paraformaldehyde, incubated overnight with anti Bak (Ab1, Calbiochem, La Jolla, CA), washed, and then incubated with TRITC-conjugated isotype matched antibody. Alternatively, cells were fixed in 4% (wt/vol) paraformaldehyde, incubated overnight with anti-Bax N20 (Santa Cruz Biotechnology, Santa Cruz, CA), and anti-cytochrome c (BD Biosciences, San Jose, CA) washed and then incubated with TRITC- and FITC-conjugated isotype-matched antibodies.

Electron Microscopy

MEFs (5 × 10⁶ cells) transfected with β-Gal or HA-hFis1 were fixed in a 2.5% (vol/vol) solution of glutaraldehyde in PBS for 30 min. Conventional electron microscopy was then performed as described in Scorrano et al. (2002).

Analysis of Cell Death

MEFs, 6 × 10⁴, of the indicated genotype grown in 12-well plates were cotransfected with pEGFP and the indicated vector. At the indicated times, cells were collected and stained with Annexin-V–Alexa568 according to the manufacturer's protocol. Apoptosis was measured by flow cytometry (FACSCalibur, BD Biosciences) as the percentage of annexin-V–positive events in the GFP-positive population.

Subcellular Fractionation

MEFs transfected as indicated were harvested and resuspended in isolation buffer (IB, 0.2 M sucrose, 10 mM Tris-MOPS, pH 7.4, 0.5 mM EGTA-Tris), and mitochondria were isolated by standard differential centrifugation. Light membranes, which included endoplasmic reticulum (ER) and peroxisomes, were obtained by centrifugation of the postmitochondrial supernatant at 100,000 × g for 30 min. Protein concentration was determined by BCA assay (Pierce, Rockford, IL).

Recombinant Protein Production and Purification

Full-length hFIS1 was cloned in pET15b (Novagen, Madison, WI) and the His-tagged hFIS1 (r-HisFIS1) was expressed in Escherichia coli (BL21DE3). After induction with IPTG, bacteria were lysed, and monomeric FIS1 was recovered in the soluble bacterial fraction and purified by chromatography on nickel–nitriloacetic acid–agarose followed by Q-Sepharose resin. The protein was stored in 25 mM Tris/HCl, 100 mM NaCl, 0.2 mM dithiothreitol, and 30% (vol/vol) glycerol, pH 7.5, at −80°C. p7/p15 recombinant BID was produced, purified, and cleaved with caspase 8 as described in Scorrano et al. (2002).

In Vitro Assays and Cross-Linking

Purified mitochondria (0.5 mg/ml) were incubated at 37°C with p7/p15 BID (3.5 ng) or with r-HisFIS1 (125 ng) in experimental buffer (EB, 125 mM KCl, 10 mM Tris-Mops, pH 7.4, 10 μM EGTA-Tris, 1 mM P_i, 5 mM glutamate, and 2.5 mM malate) in the presence of cytosolic extract (0.5 mg/ml). After 30 min, mitochondria were centrifuged at 8000 × g for 5 min. For BAX activation and insertion, mitochondria were resuspended in 0.1 M Na₂CO₃, pH 11.5, incubated on ice for 30 min, and spun at 100,000 × g for 30 min. Proteins in pellet and supernatant were then separated by SDS-PAGE. BAK oligomerization was assessed as previously described (Scorrano et al., 2002).

Immunoblotting

Protein extracts were separated by SDS-PAGE (NuPAGE, Invitrogen, Carlsbad, CA), transferred onto PVDF membranes (Millipore, Bedford, MA) and probed using the following antibodies: rabbit polyclonal anti-BAK (1:1000 Upstate Biotechnology, Lake Placid, NY), anti-BAX (1:1000, Upstate Biotechnology), anti-hFis1, anti-MnSOD (1:1000, StressGen Biotechnologies, San Diego, CA), anti LDH (1:1000, Fitzgerald Industries International, Chelmsford, MA), anti-Tom20 (1:1000, Santa Cruz), or anti-actin (1:3000, Chemicon International, Temecula, CA). Horseradish peroxidase–conjugated isotype-matched secondary antibodies (Amersham Pharmacia Biotechnology, Piscataway, NJ) were used followed by detection by chemiluminescence (Amersham).

Cell Sorting

Transfected MEFs, 100 × 10⁶, were analyzed by light forward and side scatter and for YFP fluorescence through a 530-nm bandpass filter as they traversed the beam of an argon ion laser (488 nm, 100 mW) of a FACSAria (Becton Dickinson Biosciences, San Jose, CA). Untransfected control MEF cells were used to set the background fluorescence. Sorted cells were checked for viability by Trypan Blue exclusion.

Respiratory Assays

Intact or digitonin (0.01%, wt/vol) permeabilized MEFs (5 × 10⁶ cells) transfected with either β-Gal or HA-hFis1 were incubated in 0.5 ml of HBSS or EB, respectively. Cellular oxygen consumption was measured using a Clarke-type oxygen electrode (Hansatech Instruments, King's Lynn, Norfolk, United Kingdom).

Calcium Retention Capacity

After sorting, YFP-positive MEFs (5 × 10⁶ cells) were incubated in 1 ml of EB supplemented with digitonin (0.01% wt/vol) for 5 min. After permeabilization of the plasma membrane, 2,5-di-(tert-butyl)-1,4 benzohydroquinone (tBuQH, 100 μM) was added before measurement to inhibit sarcoplasmic ER Ca²⁺ ATPase (SERCA) pumps. The calcium retention capacity (CRC) of mitochondria was assessed using the fluorescent Ca²⁺ indicator Calcium Green-5N (1 μM, Molecular Probes, Eugene, OR) in a Perkin Elmer-Cetus Life Sciences LS50B spectrofluorimeter (Boston, MA) as described previously (Fontaine et al., 1998).

Bax and Bak Are Required for Mitochondrial Dysfunction by hFis1

Apoptosis is characterized by impairment of mitochondrial function, including loss of membrane potential (Δψ_m). We analyzed the relationship between mitochondrial dysfunction and fragmentation induced by hFis1. Real-time imaging of wt MEFs loaded with the potentiometric dye TMRM showed equivalent accumulation of the dye in mock-transfected cells and in cells expressing wt and hFis^K148R (Figure 4A), a result confirmed by a cytofluorimetric analysis of TMRM uptake (unpublished data). Oligomycin, a F₁F₀ ATPase inhibitor, unveils latent mitochondrial dysfunction, by inducing depolarization of damaged organelles that maintain their membrane potential by reversal of the ATPase (Irwin et al., 2003). As expected, Δψ_m was maintained in control transfected wt MEFs upon treatment with oligomycin, whereas mitochondria expressing wt hFis1 but not hFis^K148R underwent depolarization (Figure 4, A and quantification in B). On the other hand, hFis1 did not cause loss of Δψ_m in response to oligomycin in DKO cells (Figure 4, C and quantification in D).

Open in a separate window

Figure 4.

DKO cells are resistant to mitochondrial dysfunction induced by hFis1. (A and C) Pseudocolor-coded, representative images of TMRM fluorescence intensity in wt (A) and DKO (C) cells at 5 (5′) and 40 min (40′) of the acquisition sequence. MEFs cotransfected with GFP and the indicated plasmids (asterisks) after 24 h were loaded with TMRM and imaged as described. Oligomycin (2.5 μg/ml) was added at min 3 of the sequence. (B and D) Quantification of the TMRM fluorescence changes over mitochondrial regions in wt (B) and DKO (D) MEFs. Experiments were carried out as A and C, respectively. Quantification procedure is described in the Materials and Methods. Where indicated (arrows), oligomycin (2.5 μg/ml) and FCCP (4 μM) were added.

To further dissect the effect of hFis1 on mitochondrial function, we turned to measurements of oxygen consumption in intact and digitonin-permeabilized cells. hFis1 did not affect basal or uncoupled respiration of intact wt MEFs (Figure 5A). These results were confirmed in cells permeabilized with digitonin whose mitochondria were energized using substrates for different respiratory chain complexes (Figure 5B). Thus, transient expression of hFis1 does not impair mitochondrial respiration, irrespective of the substrate used to feed the respiratory chain. We next addressed the function of mitochondrial ATPase. A specific in-gel analysis showed that ATPase activity of mitochondria expressing hFis1 was equivalent to that of mock-transfected ones (Figure 5C). Accordingly, cellular ATP content after expression of hFis1 remained unchanged for up to 3 d (3.0 ± 0.5 μM in mock vs. 2.7 ± 0.22 in hFis1-transfected cells at 24 h; 3.7 ± 0.2 vs. 3.9 ± 0.7 at 48 h; 2.8 ± 0.3 vs. 2.9 ± 0.5 at 72 h; n = 3 ± SE). These data indicate 1) that fragmented mitochondria are not necessarily dysfunctional, 2) that apoptosis by hFis1 correlates with induction of organelle dysfunction, and 3) that mitochondrial dysfunction by hFis1 is not caused by an impairment of mitochondrial respiration nor ATPase activity, but requires BAX and BAK.

Open in a separate window

Figure 5.

Mitochondrial respiration, ultrastructure, and ATPase activity in cells expressing hFis1. (A) Oxygen consumption in control and hFis1-expressing intact wt MEFs. Forty-eight hours after transfection, cells were harvested, and 10⁸ cells were incubated in HBSS into an oxygen electrode chamber. Where indicated (arrows), the uncoupler FCCP (2 μM) and the complex IV inhibitor NaN₃ (10 mM) were added. (B) Oxygen consumption in control and hFis1-expressing permeabilized MEFs. Experiments were as in A, except that the cells (10⁸) were incubated in 0.5 ml EB containing 0.01% (wt/vol) digitonin. Where indicated (arrows), glutamate plus malate (G/M, 5/2.5 mM) or succinate (Succ, 5 mM in the presence of 2 μM Rotenone, an inhibitor of complex I), ADP (100 μM), oligomycin (2.5 μg/ml), FCCP (60 nM), and NaN₃ (1 mM) were added. (C) In-gel activity assay of F1-ATPase. Mitochondria from hFis1-transfected and untransfected cells were isolated and solubilized in n-dodecylmaltoside. Solubilized samples were then separated by BN-PAGE as described in experimental procedures. The Blue Native gel was then histochemically stained for ATP hydrolysis activity. Mitochondria isolated from human hearth samples were loaded as a control. (D) Representative ultrastructure of mitochondria in hFis1-expressing cells. Cells transfected with hFis1 were fixed, and standard electron microsopy images were acquired as described. Bar, 100 nm.

The ER Gateway Regulates Apoptosis by hFis1

We measured whether dysfunction was accompanied by cytochrome c release. Cells expressing hFis1 displayed a markedly diffuse cytochrome c immunostaining pattern, a picture similar to that observed in MEFs expressing BAX (Figure 6A). We therefore verified whether hFis1 triggered activation of BAX and BAK. We first used an in vitro assay using purified organelles and recombinant proteins. Incubation of isolated mitochondria with purified cytosol and recombinant active truncated BID (tBID) resulted in the insertion of BAX in the mitochondrial membranes (Figure 6B) and in oligomerization of BAK (Figure 6C). Recombinant HisFIS1 was conversely unable to induce both BAK oligomerization and BAX insertion (Figure 6, B and C). tBID but not hFis1 induced BAX and BAK activation also in situ, as shown by immunostaining with antibodies that specifically recognize the activated forms of these proteins (Figure 6, C and D).

Open in a separate window

Figure 6.

hFis1 does not activate BAX and BAK. (A) hFis1 induces release of cytochrome c. wt MEFs were cotransfected with mtRFP and the indicated plasmid. After 24 h, cells were fixed and immunostained for cytochrome c, and confocal images of mtRFP and cytochrome c were acquired. Images are representative of 80 different cells in three independent experiments. Bar, 15 μm. (B) hFis1 does not trigger BAX insertion into mitochondrial membranes. Purified mitochondria (50 μg) were incubated with p7/p15 BID or with r-HisFis1. Mitochondria were then treated with 0.1 M Na₂CO₃, and alkali-resistant (pellet) and -sensitive fractions (supernatant) were separated by centrifugation. Proteins were separated by SDS-PAGE and immunoblotted using the indicated antibodies. (C) hFis1 does not induce BAK oligomerization. Purified mitochondria (50 μg) were incubated as in A, and proteins were cross-linked where indicated by incubating with bismaleimidohexane (BMH, 10 mM) for 15 min. Samples were separated by SDS-PAGE and immunoblotted with an anti-BAK antibody. Asterisk, BAK multimers; arrow, BAK monomer. (D) hFis1 does not trigger BAX activation in vivo. MEFs were transfected with indicated plasmids, and 48 h after transfection cells were stained for Tom20 and for activated BAX using a BAX-NT antibody and counterstained with FITC- and TRITC-conjugated isotype-matched secondary antibodies. (E) hFis1 does not trigger BAK activation in vivo. wt MEFs were cotransfected with the indicated plasmid and mtYFP. After 48 h cells were fixed and immunostained with a monoclonal anti-BAK antibody and counterstained with a TRITC-conjugated secondary antibody. Bar, 15 μM.

The lack of direct activation of BAX and/or BAK by hFis1 prompted a more detailed evaluation of its killing mechanism. DKO cells also display lower ER steady state Ca²⁺ concentration ([Ca²⁺]_ER), a gateway for Ca²⁺-dependent apoptotic stimuli (Scorrano et al., 2003). Selective correction of the ER and the mitochondrial defect of DKO cells by expression of SERCA or by mitochondrially targeted Bax (mtBAX) enabled us to test which pathway was recruited by hFis1 (Scorrano et al., 2003). As expected, apoptosis by tBID totally depended on the correction of the mitochondrial pathway. On the other hand, hFis1 induced apoptosis only in the DKO-SERCA–corrected MEFs (Figure 7A). In line with this, apoptosis by tBID was not affected when we depleted ER Ca²⁺ stores by culturing wt MEFs in low Ca²⁺ (Pinton et al., 2000), and this maneuver fully prevented death by hFis1 (Figure 7B). Finally, mitochondrial dysfunction by hFis1 in DKO cells depended on ER Ca²⁺ levels, as shown by correction of the ER defect by SERCA (Figure 8, A and quantification in B). Taken together, these experiments show that the apoptotic program elicited by hFis1 is controlled by the ER death gateway requiring adequate [Ca²⁺]_ER.

Open in a separate window

Figure 7.

The BAX, BAK ER gateway controls death by hFis1. (A) hFis- induced apoptosis is restored in DKO-SERCA but not in DKO-mtBAX MEFs. MEFs of the indicated genotype were cotransfected with GFP and the indicated plasmids. Apoptosis was determined as the percentage of GFP-positive, Annexin-V–positive cells by flow cytometry. Data represent mean ± SE of six different experiments. (B) hFis1 induced cell death is inhibited by low extracellular [Ca²⁺] and by NAC. wt MEFs were transfected as in A, and after 4 h NAC (2.5 mM) was added to the media. Apoptosis was monitored 48 h after transfection. Where indicated (0.1 mM Ca²⁺), wt MEFs were incubated in KRB supplemented with EGTA for 3 h and then maintained in complete DMEM containing 0.1 mM Ca² before transfection.

Open in a separate window

Figure 8.

Mitochondrial dysfunction by hFis1 is mediated by permeability transition. (A and B) hFis1 induces mitochondrial dysfunction in DKO-SERCA MEFs. DKO-SERCA MEFs cotransfected with GFP, and the indicated plasmid (asterisks in A) were loaded with TMRM, and fluorescence imaging was performed as described in Figure 3. Data represent mean ± SE of three independent experiments. (C and D) Mitochondrial dysfunction induced by hFis1 is inhibited by CsA and NAC. wt MEFs cotransfected with GFP and Myc-hFis1 (asterisks in C) were loaded with TMRM in the presence of 2.5 mM NAC or 1.5 μM CsA when indicated. TMRM fluorescence over mitochondrial regions was imaged, stored, and analyzed as described in Figure 3. Where indicated (arrows in B and D), oligomycin (2.5 μg/ml) and FCCP (4 μM) were added. Data represent mean ± SE of five different experiments. (E) Cell sorting of YFP-expressing wt MEF. Twenty-four hours after transfection with YFP-hFis1 (YFPFis), mtYFP or mtYFP plus hFis1^K148R (Fis^K148R), wt MEFs (10⁸) were washed, harvested, and sorted as described in the Materials and Methods. Dot plot histograms of YFP fluorescence in control (top panel) and transfected (bottom panel) cells are shown. R2 indicates sorted population. Sorted cells were lysed in RIPA buffer, and equal amounts of protein (40 μg) were separated by SDS-PAGE and immunoblotted with the indicated antibodies. (F) Representative traces of mitochondrial calcium retention capacity (CRC). Sorted (5 × 10⁶) cells were permeabilized with digitonin (0.001% wt/vol) in experimental buffer, and Ca²⁺ uptake was measured after the fluorescence changes of the Ca²⁺ indicator Ca-Green. Where indicated (arrows), 5 μM Ca²⁺ were added. Final volume, 1 ml, pH 7.4, 37°C. (G) Quantitative analysis of CRC of mitochondria from digitoninpermeabilized cells. CRC of mitochondria of sorted wt MEFs transfected with the indicated plasmids was measured exactly as in F. Data represent mean ± SE of five different experiments.

Death by hFis1 Involves Mitochondrial Permeability Transition and Production of Reactive Oxygen Species

Apoptotic stimuli passing trough the ER gateway can impinge on the mitochondrial permeability transition pore (PTP), a Ca²⁺-dependent inner mitochondrial membrane channel whose dysregulated opening participates in death after selected stimuli (Bernardi, 1999). We wanted to test whether mitochondrial dysfunction induced by hFis1 could be ascribed to dysregulation of PTP. To this end, we tested the effect of cyclosporine A (CsA), an immunosuppressive drug that is also a well-characterized inhibitor of the PTP (Bernardi, 1999). CsA prevented depolarization induced by oligomycin in MEFs expressing hFis1 (Figure 8, C and quantification in D). Prevention of production of reactive oxygen species (ROS) can block opening of the PTP by a variety of inducers, including Ca²⁺ (Kowaltowski et al., 1996). The glutathione precursor N-acetylcysteine (NAC) prevents PTP opening (Reid et al., 2005) and inhibits apoptosis by several agents (Hockenbery et al., 1993; Zamzami et al., 1995). NAC blocked apoptosis (Figure 7B) and mitochondrial dysfunction (Figure 8, C and D) induced by hFis1 in wt MEFs, further substantiating a role for the PTP in this.

We next decided to measure CRC, an indicator of PTP open probability, in mitochondria expressing different levels of hFis1. After transfection of wt MEFs with a YFP-tagged version of hFis1 (that induced death as efficiently as wt hFis1; unpublished data), cells were sorted. Expression of high levels of hFis1 was confirmed by specific immunoblotting of sorted samples (Figure 8E). Increased hFis1 resulted in markedly reduced mitochondrial CRC in these digitonin-permeabilized, sorted cells (Figure 8, F and quantification in G). Conversely, hFis1^K148R did not reduce CRC (Figure 8, F and quantification in D). These results show that hFis1 regulates PTP. Moreover, fission does not necessarily correlate with increased PTP opening probability.

Transient openings of the PTP have been associated with mitochondrial remodeling, which ensures complete release of cytochrome c. We therefore verified whether mitochondria of cells expressing hFis1 displayed an altered ultrastructure. Electron microscopy showed that in ~40% of the mitochondria the expression of hFis1 resulted in a more positive curvature of the cristae and in a widening of the narrow tubular junction (Figure 5).

DISCUSSION

Changes in mitochondrial shape have been recently identified as a common subroutine of apoptosis. Genetic disruption of profusion mitochondria-shaping proteins impairs function of the organelle, whereas excess fission results in cytochrome c release and apoptosis. An emerging view predicts a linear relation between fragmentation of mitochondria and their permeabilization. Whether and how fission necessarily results in permeabilization of mitochondria is however largely unknown. A current model implies that components of the mitochondrial fission machinery can cross-talk with members of the Bcl-2 family to regulate egress of cytochrome c from mitochondria. Multidomain proapoptotic members seem likely candidate to participate in the release of proapoptotic factors after fragmentation of the organelles, but their role in this process remains unclear. Here we genetically dissected the mechanism of death by the profission protein hFis1 and found that fragmentation of mitochondria and apoptosis can occur independently.

hFis1 Does Not Require BAX and BAK To Fragment Mitochondria

Proapoptotic multidomain Bcl-2 family members are multifunctional gatekeepers of organelles that participate in the amplification of apoptotic signals (Scorrano and Korsmeyer, 2003). It has been suggested that they participate in apoptotic fragmentation of mitochondria, based on the colocalization of BAX with sites of fission and on the ability of overexpressed BAX to promote mitochondrial shape changes (Frank et al., 2001; Karbowski et al., 2002). hFis1, a component of the mitochondrial fission machinery, and the intrinsic death stimulus staurosporine, equally fragment mitochondria in wt and Bax, Bak doubly deficient cells. Fragmentation by hFis1 was not associated with a decrease in mitochondrial fusion or with a decreased or disordered movement of the fragmented organelles. Thus, both BAX and BAK are dispensable for fission induced by hFis1 or by a widely used intrinsic apoptotic stimulus. In accordance with this, expression of Bcl-X_L blocked death, but not fission induced by hFis1 (James et al., 2003). Moreover, because DKO cells do not undergo mitochondrial permeabilization, our results contribute to place fission temporally in the course of apoptosis. At least in MEFs, fragmentation occurs before the release of mitochondrial IMS proteins, such as DP/TIMM8a, which has been reported to trigger apoptotic translocation of Drp1 to mitochondria (Arnoult et al., 2005). On the other hand, we cannot exclude that mitochondrial fission by staurosporine in DKO cells is the morphological epiphenomenon of complete inhibition of fusion occurring in the course of apoptosis (Karbowski et al., 2004). This possibility was reinforced by silencing of hFis1 expression that only partially inhibited fission induced by staurosporine. In any case, this would imply that during cell death by staurosporine fusion is inhibited independently from BAX and BAK as well as from the activation of the postmitochondrial program of apoptosis.

DKO cells display multiple defects, including a reduction in the steady state ER Ca²⁺ levels (Scorrano et al., 2003). DKO cells proved therefore informative to address the role of Ca²⁺ in controlling mitochondrial shape. DKO cells displayed normal steady state morphology and underwent equivalent fission when hFis1 was expressed. This implies that adequate [Ca²⁺]_ER, crucial for several mitochondrial functions, is not required to control morphology of the organelle. Likely, dynamic changes in cytosolic and/or mitochondrial Ca²⁺ can conversely modulate mitochondrial shape by yet unidentified mechanisms, such as regulation of kinase/phosphatase couples (Yi et al., 2004).

The ER Gateway Controls Apoptosis by hFis1

The resistance of DKO cells to death induced by hFis1 suggested that hFis1 would directly or indirectly activate these multidomain proapoptotics to provide an efflux pathway for cytochrome c. A set of in vitro experiments using purified organelles and recombinant proteins excluded a direct effect of hFis1 on BAX and BAK activation, as well as on cytochrome c release (J.-C. Martinou, unpublished results). This did not depend on additional factors missing in this reconstituted assay, because activation of multidomain proapoptotics was also excluded in situ. These surprising results suggest that in the case of hFis1 induced fission, insertion and oligomerization of BAX at sites of fragmentation is not favored as in other paradigms of apoptosis (Frank et al., 2001).

Selective correction of ER versus mitochondrial defect of DKO cells has been used to define the gateway through which a given apoptotic stimulus passes (Scorrano et al., 2003). hFis1 induced death in SERCA-, but not in mtBAX-corrected DKO MEFs. Further evidences showed that hFis1 required adequate ER Ca²⁺ to kill, placing it genetically downstream of the ER gateway of apoptosis. Our results provide a molecular explanation for the previous finding that death by hFis1 is blocked by Bcl-X_L (James et al., 2003), which also regulates release of Ca²⁺ from the ER (Li et al., 2002). The small reduction in capacitative Ca²⁺ entry (CCE) promoted by hFis1 (Frieden et al., 2004) can therefore be viewed as a compensatory mechanism to protect cells with high levels of this protein, because CCE dictates Ca²⁺ refilling of the ER (Jaconi et al., 1997). In line with this, direct measurement of free [Ca²⁺]_ER using targeted aequorin showed that expression of hFis1 reduced in rate and extent of ER refilling by ~40% (unpublished data). Whether other conditions of excessive fragmentation kill via the ER gateway still needs to be elucidated, but this remains a viable possibility that extends ER-mitochondria cross-talk to a previously unexpected process.

Latent Mitochondrial Dysfunction Precedes Death by hFis1

Mitochondria lacking essential components of mitochondrial fusion display multiple functional defects, including greatly reduced respiratory rates (Chen et al., 2005). This holds true also in yeast, where most deletion mutants of genes involved in mitochondrial fusion are petite, as a consequence of their respiratory incompetence (Dimmer et al., 2002). Moreover, the importance of profusion proteins for cellular physiology is confirmed by their ablation in the mouse, which results in early embryonic lethality (Chen et al., 2003). We therefore reasoned that increasing mitochondrial fission by expressing hFis1 could have similar results. On the other hand, hFis1 did not impair mitochondrial fusion nor did it affect a variety of bioenergetic parameters, including total ATP content, mitochondrial respiration, or membrane potential. This raised the question of the mechanism of killing by hFis1, which apparently did not involve a gross destabilization of mitochondrial function.

A detailed analysis of mitochondrial membrane potential in situ in cells expressing hFis1 showed conversely a latent organelle dysfunction, unveiled only after treatment with oligomycin. This drug is expected to induce slight hyperpolarization, because the inhibition of the F₁F₀ ATPase blocks diffusion of protons back into the matrix of phosphorylating mitochondria. On the other hand, oligomycin induced depolarization of wt mitochondria expressing hFis1, which most probably maintain their membrane potential by reversal of the ATPase. Such a depolarization was not observed in hFis1-transfected DKO cells and was restored by genetic correction of DKO with SERCA. A more detailed analysis of respiration of intact and digitonin-permeabilized cells ruled out gross impairment of the respiratory chain. Moreover, assembly of the ATPase and total cellular ATP content were not affected by high levels of hFis1. When hFis1 is overexpressed, basic bioenergetics appears therefore to be conserved. On the other hand, hFis1 reduces the mitochondrial threshold to stresses, as highlighted by the oligomycin-induced depolarization. It should be noted that similar experiments have been performed in HeLa cells, where oligomycin had apparently no effect (Frieden et al., 2004). One explanation can be found in the short experimental window used by Demaurex and colleagues that precluded highlighting of the slow depolarization induced by oligomycin (Gugliucci et al., 2002; Irwin et al., 2003). Latent mitochondrial dysfunction unveiled by oligomycin characterizes cells from mouse models of collagen VI deficiency (Irwin et al., 2003), as well as prostate cancer cells where accumulation of arachidonic acid was induced by pharmacological inhibition of its metabolism (Gugliucci et al., 2002). In both these models, dysfunction was prevented by the PTP inhibitor CsA, suggesting a role for this inner mitochondrial membrane channel.

We thank A. Cabrelle (Venetian Institute of Molecular Medicine Flow Cytometry Facility) for the skillful FACS analysis and C. Frezza for help with figures. L.S. is an Assistant Telethon Scientist of the Dulbecco-Telethon Institute. This research was supported by Telethon Italy; Associazione Italiana Ricerca sul Cancro, Italy; Compagnia di San Paolo; and Human Frontier Science Program Organization. E.A. was a recipient of a fellowship from the Société Académique de Genève and of the Swiss National Science Foundation (subsidy 3100A0-109419/1).