Address correspondence and reprint requests to Susan S. Wigdal, Third Wave Technologies, 502 South Rosa Road, Madison, Wisconsin 53719, USA. E-mail: email@example.com
It has been suggested that release of cytochrome c (Cyt c) from mitochondria during apoptotic death is through opening of the mitochondrial permeability transition pore followed by swelling-induced rupture of the mitochondrial outer membrane. However, this remains controversial and may vary with cell type and model system. We determined that in mouse cerebellar granule neurons, Cyt c redistribution preceded the loss of mitochondrial membrane potential during the apoptotic process, suggesting that the pore did not open prior to release. Furthermore, when mitochondria were morphologically assessed by electron microscopy, they were not obviously swollen during the period of Cyt c release. This indicates that the pore mechanism of action, if any, is not through mitochondrial outer membrane rupture. While bongkrekic acid, an inhibitor of pore opening, modestly delayed apoptotic death, it also caused a significant (p < 0.05) suppression of protein synthesis. An equivalent suppression of protein synthesis by cycloheximide had a similar delaying effect, suggesting that bongkrekic acid was acting non-specifically. These findings suggest that mitochondrial permeability transition pore is not involved in Cyt c release from mitochondria during the apoptotic death of cerebellar granule neurons.
Release of cytochrome c (Cyt c) from mitochondria into the cytoplasm is an important step in the apoptotic death of many cell types (Green and Reed 1998; Desagher and Martinou 2000). Once in the cytoplasm, Cyt c binds to apoptosis protease-activating factor-1 and activates procaspase-9 in an ATP-dependent reaction (Liu et al. 1996; Li et al. 1997). Caspase-9, in turn, activates caspase-3, which then cleaves many important cellular substrates and causes cell death.
The exact mechanism for Cyt c release is unclear. Several studies have shown the importance of various Bcl-2 family members in the regulation of Cyt c release (Kluck et al. 1997; Yang et al. 1997; Eskes et al. 1998). One proposed mechanism for Cyt c release is that the mitochondrial permeability transition pore (MPTP) opens during apoptosis resulting in the loss of mitochondrial membrane potential (Δψm), osmotic swelling of the matrix, expansion of the mitochondrial inner membrane (IM) and bursting of the mitochondrial outer membrane (OM; Vander Heiden et al. 1997). Rupture of the OM then allows intermitochondrial membrane proteins, such as Cyt c, to redistribute into the cytoplasm (Susin et al. 1996; Green and Reed 1998). Consistent with a possible role for MPTP, a loss of Δψm has been reported to precede Cyt c release in several cell types, including hepatocytes (Bradham et al. 1998), PC12 cells (Wadia et al. 1998), and PC6 cells (Heiskanen et al. 1999). In contrast, other studies have shown that Cyt c release precedes Δψm loss in HeLa cells (Yang et al. 1997; Bossy-Wetzel et al. 1998) and glioma cells (Ikemoto et al. 2000). Therefore, the mechanism of Cyt c release may be cell type-specific with MPTP having a role in some cell types but not others.
Cortical (Stefanis et al. 1999), hippocampal (Krohn et al. 1999), and sympathetic (Deshmukh et al. 2000) neurons lose Δψm after Cyt c release. In addition, sympathetic neuron mitochondria do not swell during apoptosis (Martinou et al. 1999). Thus it appears that some neurons may not require MPTP opening for apoptosis. In this study we sought to order Δψm loss and Cyt c release during apoptosis of cerebellar granule neurons (CGNs) and to monitor CGN mitochondria by electron microscopy to determine if mitochondrial swelling could account for the release of Cyt c in these cells. To date, neuronal mitochondrial morphology during apoptosis has only been examined in sympathetic neurons (Martinou et al. 1999). We sought to establish whether mitochondrial morphology in a CNS model system is similar to thatobserved in sympathetic neurons during apoptosis. Our studiesdemonstrate that Cyt c release preceded Δψm loss in CGNs undergoing apoptotic death and that this release was not accompanied by mitochondrial swelling. Thus it seems unlikely that MPTP opening is important for Cyt c release in these cells.
Materials and methods
Cultures were kept primarily neuronal in part by inhibition of the dividing cells using aphidicolin (3.3 µg/mL, Sigma, St Louis, MO, USA). Non-physiological apoptosis was induced by addition of the broad-spectrum protein kinase inhibitor staurosporine (STS, 500 nm, Sigma). The progression of apoptosis was delayed by bongkrekic acid (BA, 30 µm, Calbiochem, San Diego, CA, USA). Protein synthesis was suppressed by cycloheximide (CHX, 5 ng/mL,Sigma). Mitochondrial membrane potential was monitored using the mitochondrial membrane potential dependent dyes JC-1 (5,5′,6,6′-tetrachloro-1,1′,3,3′-tetraethylbenzimidazolylcarbocyanine iodide, 2 µg/mL, Molecular Probes, Eugene, OR, USA) and TMRM (tetramethylrhodamine, methyl ester, perchlorate, 10 nm, Molecular Probes). Mitochondrial membrane potential was abolished by addition of FCCP (carbonyl cyanide p-(trifluoromethoxy) phenyl-hydrazone, 5 µm, Sigma).
The cell culture procedure was taken from Miller and Johnson (1996). Briefly, entire cerebella from post-natal day seven C57BL/6 mice were dissected, the meningeal layer and blood vessels removed and a single cell suspension generated by trypsin digest andtrituration. The cells were plated onto poly-l-lysine-coated (100 µg/mL, Sigma) slides, tissue culture plates, or coverslips at a cell density of 2.5 × 105 cells/cm2 and incubated at 37°C/5% CO2/95% air in K25 + S medium [basal medium Eagle with 25 mm KCl (Life Technologies, Inc., Gibco BRL, Rockville, MD, USA), 10% dialyzed fetal bovine serum (10 000 molecular weight cut-off, Sigma), and 50 units/mL penicillin G sodium and 50 µg/mL streptomycin sulfate (Life Technologies)]. Twenty-four hours after plating, the antimitotic agent aphidicolin (Sigma) was added at a final concentration of 3.3 µg/mL to inhibit the growth of non-neuronal cells. Cells were cultured for 6–9 days prior to experimental procedures.
CGNs maintained in K25 + S undergo apoptosis when deprived of KCl (5 mm) and serum due to induced plasma membrane potential repolarization and withdrawal of trophic support (K5-S medium, D'Mello et al. 1993). The broad-spectrum protein kinase inhibitor, staurosporine (STS), has also been widely used to induce apoptosis (e.g. Krohn et al. 1998; Deshmukh and Johnson 2000). Therefore, apoptotic death was induced by: (i) replacing K25 + S with K5-S medium (basal medium Eagle, Life Technologies); and (ii) addition of 500 nm staurosporine (STS, Sigma), based on a dose–response curve (data not shown). The cultures were rinsed twice in the respective media and incubated for the indicated number of hours. The effects of bongkrekic acid (BA, 30 µm, Calbiochem, San Diego, CA, USA) and cycloheximide (CHX, 5 ng/mL, Sigma), added at the time of apoptosis induction, were also determined.
Cells were cultured in LabTek II 8-well glass chamber slides with covers (Nagle Nunc International, Naperville, IL, USA). The cells were then subjected to apoptosis-inducing media for the indicated hours. Standard immunocytochemical procedures were used (Harlow and Lane 1988). The cells were fixed in 4% paraformaldehyde (Sigma), permeabilized in 0.2% Triton X-100 (Sigma) and blocked in 5% normal donkey serum (Jackson ImmunoResearch Laboratories, West Grove, PA, USA). The primary antibodies, in 1% normal donkey serum, and final concentrations were as follows: mouse monoclonal IgG anti-βIII tubulin (1 µg/mL, Promega Corporation, Madison, WI, USA), rabbit polyclonal Anti-ACTIVE® caspase-3 (1 : 250 dilution, Promega Corporation), mouse monoclonal IgG anti-Cyt c (1 µg/mL, clone 6H2.B4, Promega Corporation), rabbit polyclonal anti-GFAP (glial fibrillary acidic protein, 1 : 1000 dilution, Promega Corporation), mouse monoclonal IgM anti-O1 (1 µg/mL, Boehringer Mannheim Biochemica, Indianapolis, IN, USA), mouse monoclonal IgG MOPC-21 (PharMingen, San Diego, CA, USA), mouse monoclonal IgM anti-TNP (PharMingen) and ChromoPure Rabbit IgG, whole molecule (1 µg/mL, Jackson ImmunoResearch Laboratories). The secondary antibodies and final concentrations were as follows: Cy-3 conjugated donkey anti-mouse IgG, Cy-3 conjugated donkey anti-mouse IgM and Cy-3 conjugated donkey anti-rabbit (all 3 µg/mL, all from Jackson ImmunoResearch Laboratories). The slides were then mounted with 120 µL Vectashield Mounting Media with DAPI (Vector Laboratories, Burlingame, CA, USA) and coverslipped (22 mm × 50 mm, Fisher Scientific, Chicago, IL, USA). The anti-MOPC-21, anti-TNP and rabbit IgG whole molecule antibodies were used as non-specific negative controls for the mouse IgG, mouse IgM and rabbit polyclonal antibodies, respectively.
Confocal and fluorescent microscopy
For the JC-1 and TMRM assays, cells were imaged on a MRC 1024laser scanning confocal microscope (Bio-Rad Laboratories, Cambridge, MA,USA) with a mixed gas argon-krypton laser using Laser Sharp software (version 3.0, Bio-Rad Laboratories) at the University of Wisconsin-Madison Keck Neural Imaging Laboratory. The laser emitted three lines in coalignment at 488, 568, and 647 nm. Laser intensity was varied using a motorized wheel with neutral density filters. The voltage to the photomultiplier tubes for each channel, the iris and the black level were kept at the same values. For the JC-1 assays, excitation wavelength was 488 nm and the emissions of the red (680 ± 16 nm) and green (522 ± 16 nm) channels were recorded simultaneously. This ensured that exactly the same field of view was analyzed for both channels. For the TMRM assays, the red channel (680 ± 16 nm) was recorded.
For the immunocytochemical assays, cells were imaged on an Axioplan 2 inverted microscope (Carl Zeiss, Inc., Thornwoood, NY, USA). Color images, taken with a SPOT2 camera (version 2.0, Diagnostic Instruments, Inc., Sterling Heights, MI, USA), were merged using ImagePro Plus software (version 4.0, Media Cybernetics, Silver Spring, MD, USA). For the cell type and condensed nuclei experiments, n ≥ 200 for each treatment group from at leastthree independent cultures. The data presented are the mean percent ± 1 SD. Scoring the number of condensed nuclei by DAPI-staining was performed in a blinded fashion. After all the slides were scored, the labels were unmasked and the scores were matched with the treatment groups.
Cells cultured on coverslips (22 × 22 mm, Fisher Scientific) were loaded with JC-1 (5,5′,6,6′-tetrachloro-1,1′,3,3′-tetraethylbenzimidazolylcarbocyanine iodide, 2 µg/mL, Molecular Probes) for the last 20 min of incubation in the appropriate medium. This dyeconcentrates in mitochondria maintaining a high membrane potential and forms aggregates having a red emission spectrum (Reers et al. 1991). The monomeric form of JC-1 has a green emission spectrum. After rinsing the cultures, the red and green emissions of JC-1 were detected with the TRITC and FITC photomultipliers of the confocal microscope, respectively. JC-1 intensities from individual cells were determined using Metamorph software (version 3.6, Universal Imaging Corporation, West Chester, PA, USA).
For each cell, the red and green intensities of the same, randomly chosen spot within the cytoplasm were determined using Metamorph software (version 3.6, Universal Imaging Corporation). The red intensity value was divided by the green intensity value to give a red to green ratio. This ratio reflects the amount of aggregate JC-1 molecules (red) divided by the amount of monomeric JC-1molecules (green), and thus provides an index of the Δψm for all themitochondria within the spot. Unfortunately, because cerebellar granule neurons are small (5–10 microns in diameter), we were unable to determine the JC-1 red to green ratios for individual mitochondria.
For each time point, the red to green ratios of at least 50 cells from five fields of view were determined (n = 250 cells per time point). This experiment was performed from three independent cultures (n = 750 total cells per time point). To account for slight variations in loading time, a control culture, equivalent to ‘time zero’, was loaded with JC-1 in parallel with each time point. Thus the ratios could be directly normalized to an untreated culture in the exact same experimental conditions. The ratios for each cell in each sample were averaged and the average percent change from the zero time point was calculated (e.g. average of the 9-h time point ratios divided by the average of the zero time point ratios). For simplicity, the average ratio value at the zero time point was designated 100%. For each subsequent time point, the average percent change from the corresponding control cultures were averaged and plotted as the mean percent of untreated cells ± 1 SD. For the positive controls, 5 µm FCCP (carbonyl cyanide p-(trifluoromethoxy) phenyl-hydrazone, Sigma) was added to abolish Δψm.
To verify the JC-1 assay results, mitochondrial membrane potential was also monitored using TMRM (tetramethylrhodamine, methyl ester, perchlorate, 10 nm, Molecular Probes). The accumulation of the methyl esters of TMRM in mitochondria is driven by membrane potential. After 11.5 h in K5-S, the treated and control cultures were incubated for 30 min in K25-S (for control condition) or K5-S (for apoptotic condition) containing 10 nm TMRM. The dye was washed out twice and observed in the third wash of media. The 12 h time point was compared to the zero time point. For each time point, 35 cells were examined.
Protein synthesis assay
Cells were plated in 24-well tissue culture plates (Corning Costar). Twenty-four hours prior to the end of the experiment, fresh K25 + S + 3.3 µg/mL aphidicolin media with 30 µm BA or CHX (1, 5, 10, 15, 100 ng/mL, and 1 µg/mL final concentrations) was added. Four hours prior to the end of the experiment, fresh K25 + S + 3.3 µg/mL aphidicolin media + BA or CHX + 10 µCi/mL TRAN 35S-label (70%l-methionine, 15%l-cysteine) was added. A ‘no protein synthesis’ control, to determine the amount of free 35S binding to the filters, was performed by adding media plus TRAN solution to untreated cultures at the end of the 4 h and then immediately performing the next steps. The media were then removed and the cells were lysed in 1% Triton X-100/PBS (phosphate-buffered saline), precipitated in 10% chilled trichloroacetic acid (Sigma) for 1.5 h at 4°C and filtered in a manifold onto a nitrocellulose square (Schliecher and Schuell, Keene, NH, USA). The manifold well and nitrocellulose filter was rinsed twice with 5 mL 10% chilled trichloroacetic acid. Then the nitrocellulose was dissolved in scintillation fluid (Fisher Scientific) for 1 h at room temperature and the incorporation of radiolabeled amino acids was determined with a LS3801 liquid scintillation counter (Beckman Coulter, Inc., Fullerton, CA, USA). Each experiment was performed with n = 4–6 per treatment from three independent cultures. The data presented is the mean percent change in protein synthesis ±1SD.
Cells were grown on coverslips (22 × 22 mm, Fisher Scientific) placed in 35-mm tissue culture dishes (Fisher Scientific) and covered with media. The cultures were subjected to apoptosis inducing media (K5-S and 500 nm STS) for 0, 12 and 24 h with or without 30 µm BA. The slips were carefully rinsed in 0.1 m phosphate buffer for 5 min at room temperature and fixed in 2% paraformaldehyde/2% glutaraldehyde/0.1 m phosphate buffer for 20 min at room temperature. The samples were then rinsed twice in 0.1 m phosphate buffer for 5 min at room temperature. Next, the samples were post-fixed in 0.1 m phosphate buffer plus 2% osmium tetroxide for 1 h at room temperature and dehydrated in a graded series of alcohol solutions. After embedding in Durcupan (Fluka, Switzerland), the glass slips were removed from the cells by etching with hydrofluoric acid. Seventy-nanometer sections were then cut parallel to the slips, post-stained with uranyl acetate and lead citrate and examined with a Philips CM120 transmission electron microscope at the University of Wisconsin-Madison Medical School Electron Microscope Facility. The morphology of 30–40 cells per K5-S and STS treatment group and 10–15 cells per BA-treated group was determined.
Statistical analyses of the Cyt c and JC-1 experimental results were performed using Dunnett's procedure for multiple comparisons with a standard (α = 0.05; JMP version 3.2.6 statistical analysis package, SAS Institute, Inc., Cary, NC, USA) and assigning the zero hour time point as the standard. Statistical analyses comparing and the zero time point and the 12 h time point for the TMRM experiment were performed. Statistical analyses comparing and the time points of the percent of cells with condensed nuclei were also performed for the ± BA, ± CHX, and BA versus CHX graphs. The data were analyzed by the Student's t-test, two-sample assuming unequal variance (α = 0.05) using Microsoft Office 2000 Excel Data Analysis ToolPak (Microsoft Corporation, Redmond, WA, USA).
CGN cultures are predominately neuronal
CGN cultures were stained with antibodies against three cell specific markers: βIII tubulin (Fig. 1a, neurons), GFAP (Fig. 1b, astrocytes) and O1 (Fig. 1c, oligodendrocytes). Sister cultures were stained with non-specific primary antibodies of the same immunoglobulin type or secondary antibody only, the latter to confirm the specificity of all antibodies used (data not shown). The percent of each cell type was plotted (Fig. 1d) and determined to be 96.5 ± 1.44% neurons, 3 ± 1.33% astrocytes and 0.5 ± 0.20% oligodendrocytes.
CGN undergoing K5-S- and STS-induced apoptosis lose Cyt c staining
A time course of Cyt c staining loss was performed (Fig. 2) which showed a significant (p < 0.05) Cyt c loss by 5 h in K5-S medium and 4 h in 500 nm STS. Interestingly, the majority of cells with condensed nuclei lacked Cyt c staining altogether (Fig. 2, insert). Our findings with CGNareconsistent with the reports in sympathetic neurons undergoing apoptosis induced by neurotrophic factor-deprivation(Neame et al. 1998; Deshmukh et al. 2000) and HeLa cells undergoing STS-induced apoptotic death (Goldstein et al. 2000) where Cyt c, once released into the cytoplasm, is rapidly degraded.
Because the majority of cells with condensed nuclei also lacked Cyt c staining (Fig. 2, insert), the rate of Cyt c was very similar to the inverse of the survival curve (Fig. 3e). Cultures induced to undergo apoptotic death by K5-S or STS exhibited condensed nuclei in approx. 50% of cells by 10 h post-induction and peaked at approx. 90% condensed nuclei at 20–24 h, consistent with previous studies examining K5-S-induced CGN apoptotic death (D'Mello et al. 1993; Miller and Johnson 1996). Theincrease in the number of condensed nuclei is evident both by phase microscopy and DAPI staining and is shown in Fig. 3(a–d) at 0 and 8 h post-K5-S induction. Like Armstrong et al. (1997), we found the increase in the number of cells with condensed nuclei corresponded with an increase in the number of cells with activated caspase-3, a hallmark indicator of apoptosis. Very few cells had active caspase-3 at 0 h (Fig. 4a) while numerous cells had active capsase-3 at 8 h post-K5-S induction (Fig. 4b).
Δψm is lost during apoptosis
The JC-1 red to green ratio was used to monitor mitochondrial membrane potential. Δψm decreased over time during apoptosis (Fig. 5) and dramatically decreased by 88.15% with acute addition of FCCP (5 µm), a potent uncoupler of oxidative phosphorylation that rapidly depolarizes the Δψm. Δψm was significantly lower in K5-S- and STS-induced apoptosis at 18–20 h post-induction (p < 0.05, Fig. 5,graph). The JC-1 results were confirmed using a different mitochondrial membrane potential dye, TMRM. We examined TMRM accumulation from cells at zero (control cultures) and 12 h post-apoptosis-induction with K5-Smedia. TMRM intensity at the 12 h time point was 102 ± 8 SEM ofthe control intensity. Like the JC-1 experimental data, there was no statistical difference at 12 h post-K5-S apoptosis-induction from the control by t-test (p > 0.1; n = 35) as measured by the TMRM dye.
K5-S- and STS-induced apoptosis does not cause mitochondrial swelling
Mitochondrial morphology during K5-S- and STS-induced apoptosis was examined at 0, 12 and 24 h post-treatment by electron microscopy (EM). As shown in Fig. 6(b; 12 h in K5-S medium) and Table 1, the majority of mitochondria were intact. Very few mitochondria were swollen and most appeared similar to mitochondria from control cells (Fig. 6a; 0 h in K5-S medium). Figure 6(c) shows a rare example of a swollen mitochondrion at 12 h in K5-S medium.
Table 1. Mitochondrial morphology during K5-S- and STS-induced apoptotic death
Normal/total profiles (% normal mitochondrial profiles)
Number of cells examined
CGN cultures were induced to undergo apoptotic death by incubation in K5-S or 500 nm STS for 0, 12 and 24 h. Electron micrographs were taken and the numbers of normal and swollen mitochondrial profiles were counted.
0 h K5-S, no BA
12 h K5-S, no BA
24 h K5-S, no BA
0 h STS, no BA
12 h STS, no BA
24 h STS, no BA
0 h, 30 µm BA
12 h K5-S, 30 µm BA
24 h K5-S, 30 µm BA
12 h STS, 30 µm BA
24 h STS, 30 µm BA
Bongkrekic acid modestly delays K5-S- and STS-induced apoptosis by suppression of protein synthesis
Bongkrekic acid (BA) is an inhibitor of the MPTP (Green and Reed 1998). Marzo et al. (1998) showed that 50 µm BA prevented the pro-apoptotic effects of Bax in Rat-1 fibroblast cells. We titrated BA from 300 µm to 30 µm and found that, in mouse cerebellar granule neurons in culture, 30 µm BA delayed nuclear condensation at the same rate as 50 µm (data not shown). Therefore, we chose to use 30 µm BA for this study. When the effect of BA on our apoptosis model system was examined, we found that it significantly (p < 0.05), albeit modestly, decreased the number of condensed nuclei from 8 to 16 h and 20 h post-induction of K5-S apoptosis, but not at 24 h (Fig. 7a). BA also significantly (p < 0.05), but modestly, decreased the number of condensed nuclei from 10 to 14 h in STS-induced apoptosis, but not thereafter (Fig. 7b). To ensure that BA did not affect mitochondrial morphology, we examined cultures treated with BA by EM and observed similar mitochondrial profiles with or without BA (Table 1).
De novo protein synthesis is a requirement for apoptosis ofmany neurons including CGNs (D'Mello et al. 1993; Armstrong et al. 1997). Recently, cyclosporin A, which inhibits protein synthesis, was shown to inhibit MPTP in sympathetic neuronal cultures (Kirkland and Franklin 2001), so we examined the effect of BA on CGN protein synthesis and found synthesis was inhibited 28.3 ± 5.9%(Fig. 8). To determine whether the modest delay in K5-S-induced apoptosis was due to inhibition of MPTP opening or protein synthesis, we used 5 ng/mL CHX, a known inhibitor proteinsynthesis, to induce approx. 28% (27.3 ± 1.7%) protein synthesis inhibition in our model system (Fig. 8). When the survival curves of CHX and BA were compared, they were nearly identical, differing significantly (p < 0.05) only at 12 h post-induction (Fig. 9).
There are two main hypotheses of how proteins, including Cyt c, are released from the mitochondria during apoptotic death. Several investigators suggest that a pore is formed that allows Cyt c and other mitochondrial proteins to redistribute into the cytoplasm. Alternatively, others suggest that mitochondrial proteins are released non-specifically due to MPTP opening which allows the influx of water and solutes into the matrix resulting in IM swelling and OM rupture.
The mechanism of action of Cyt c release may vary with the model system. For example, hepatocytes (Bradham et al. 1998), PC12 cells (Wadia et al. 1998), and PC6 cells (Heiskanen et al. 1999) have been shown to lose Δψm priorto Cyt c release. In contrast, cortical (Stefanis et al. 1999), hippocampal (Krohn et al. 1999) and sympathetic (Deshmukh et al. 2000) neurons lose Δψm after Cyt c release. We sought to determine the order of events during apoptosis in cerebellar granule neurons. In addition, since mitochondria from CNS neurons have not been studied during apoptosis, we examined whether CNS mitochondria swell during apoptosis.
For this study, we first demonstrated that our cell culture was primarily neuronal by immunocytochemistry (Fig. 1) and determined the apoptotic time course of two induction methods (Fig. 3), K5-S (physiological) and using the broad-spectrum protein kinase inhibitor, staurosporine (STS). Forboth induction methods, apoptosis was monitored by counting the number of condensed nuclei. We observed significant cell death by 4–5 h (Fig. 3; p < 0.05). Apoptosis in these cultures was accompanied by the activation of caspase-3 (Fig. 4), a hallmark indicator of apoptosis. That we observed active caspase-3 was consistent with earlier findings in this model system (Armstrong et al. 1997).
Our study demonstrates that OM rupture does not occur during mouse cerebellar granule neuronal apoptosis in cell culture and that Δψm is intact during the time at which Cyt c is released, indicating that MPTP is most likely not involved in CGN apoptotic death. Cyt c loss was significant (Fig. 2; p < 0.05) by 4–5 h post-induction by both K5-S and STS. However, mitochondrial membrane potential loss, indicated by JC-1 staining (Fig. 5), and confirmed by TMRM staining (data not shown), was not lost until 18–20 h post-K5-S- and STS-induction. Furthermore, examination of the mitochondria by EM demonstrated that the majority of mitochondria were not obviously swollen as late as 24 h post-induction (Fig. 6 and Table 1).
This is consistent with the findings of Martinou et al. (1999) who showed that rat sympathetic neuron mitochondria do not swell during NGF-withdrawal-induced apoptosis. Our data are also consistent with that of Al-Abdulla and Martin (1998) who showed that mitochondria from dorsal lateral geniculate nucleus neurons induced to undergo apoptosis by unilateral occipital cortex ablation in adult rats also maintained their morphological integrity until late end-stage apoptosis. Similarly, von Ahsen et al. (2000) have shown that Xenopus egg and human cell line HL-60 mitochondria do not swell in response to various apoptotic inductions. In contrast, Bradham et al. (1998) have shown that isolated mitochondria from hepatocytes do swell during apoptosis, as do human Jurkat T cells and mouse FL5.12 pro-B cells (Vander Heiden et al. 1997).
How mitochondria maintain Δψm after releasing Cyt c is not entirely clear. Simbula et al. (1997) have shown that in liver cells undergoing apoptosis, Δψm maintenance is through an ATP-dependent reversal of ATP synthase. Alternatively, Waterhouse et al. (2001) have demonstrated in single HeLa cells expressing green fluorescent protein tagged-Cyt c that a residual level of Cyt c in the cytoplasm is sufficient to allow the continuation of the electron transport chain and ATP production.
It is possible that our study missed a transient loss of mitochondrial membrane potential, which may only be detectable for minutes or seconds; however, this seems unlikely. We determined the red to green JC-1 ratio for 750 individual cells at each time point. If a cell were to transiently lose its mitochondrial membrane potential, there would be no aggregated JC-1 molecules (red signal, indicative of intact mitochondrial membrane potential) and the red to green ratio would be zero. No such cell was ever observed. Instead, we detected a gradual loss of red signal as compared to green signal over the course of hours (significant loss was seen at 18 h post apoptosis induction, Fig. 5).
Unfortunately, we were unable to directly examine the role of MPTP in CGN apoptotic death. When MPTP was inhibited by BA, we observed that BA modestly delayed apoptosis in mouse CGN in culture (Fig. 7). However, we have previously shown in sympathetic neurons (Kirkland and Franklin 2001) that several of the drugs that inhibit specific pathways of apoptosis also inhibit protein synthesis when examined in cell culture, and protein synthesis is an absolute requirement for progression of apoptosis. In CGN cell culture, BA inhibited protein synthesis by approx. 28% (Fig. 8). To determine whether the delay in apoptotic death was due to the specific inhibition of MPTP or the general suppression of protein synthesis, the protein synthesis inhibitor cycloheximide (CHX) was used. The effect of CHX protein synthesis suppression, at an equivalent level to that of BA (Fig. 8), on the time course of K5-S-induced apoptosis in CGN was determined. The survival curves of the CHX- and BA-treated cultures were indistinguishable (Fig. 9), differing at only one time point (12 h, p < 0.05). Therefore, our data suggest that the slight anti-apoptotic effect of BA was likely mediated by protein synthesis suppression. This also raises the general concern that drugs should first be tested for their inhibition on protein synthesis prior to the assumption that they are specific only for their intended targets within the apoptotic pathway.
There is increasing evidence to suggest that Bcl-2 family members act to regulate Cyt c release during apoptosis (Desagher and Martinou 2000). Regulation occurs at many levels including phosphorylation state, intracellular localization and intermember homo- and hetero-dimerization. For example in Cos-7, HeLa and Hep G2 cell lines, Bax and Bak have been shown by confocal and electron microscopy to cluster adjacent to mitochondria during apoptosis (Nechushtan et al. 2001). Furthermore, the anti-apoptotic protein Bcl-xL completely and specifically inhibits these clusters. In contrast, other pro-apoptotic proteins Bid and Bad circumscribe the OM throughout the apoptotic process.Bax homodimers have been shown to insert into the OMfollowed by the rapid release of Cyt c (Goping et al. 1998; Gross et al. 1998). The idea that Bcl-2 family members may form pores that release Cyt c is based upon the structural similarity of Bcl-xL and Bid to the pore-forming domains of bacterial colicins and diphtheria toxin (Muchmore et al. 1996).
Tsujimoto and colleagues have suggested that Cyt c release involves the formation of pores by the Bcl-2 family members and the voltage-dependent anion channel (VDAC). A role for VDAC in apoptosis was shown in vitro in proteoliposomes (Shimizu et al. 1999) and in vivo in yeast cells (Shimizu et al. 2000b). In these studies, both VDAC and Bax were required for Cyt c release and Δψm loss. This effect was not observed with VDAC and Bid in proteoliposomes or Bax and another MPTP protein, the adenine nucleotide transport (ANT) in yeast cells. The Bcl-2 family members may interact with VDAC through the BH domains. The BH4 domain of Bcl-xL was shown to be sufficient for inhibition of VDAC activity, Cyt c release and Δψm loss in HeLa cells, isolated rat liver mitochondria as well as VDAC proteoliposomes (Shimizu et al. 2000a). Recently, a biological role for VDAC in apoptosis was demonstrated by showing that anti-VDAC antibodies could prevent Bax-induced Cyt c release and Δψm loss when added to isolated rat liver mitochondria and when microinjected into HeLa cells (Shimizu et al. 2001). These effects were not seen in Bid-induced apoptosis, again suggesting that there may be several different mechanisms by which Cyt c is released.
The exact roles of the Bcl-2 family members and MPTP remain elusive and may depend on the cell type and model system. However, in mouse CGN apoptosis in cell culture, MPTP does not appear to be involved in Cyt c release, consistent with other studies of neuronal cultures. Δψm was maintained during the time in which Cyt c was released. In addition, EM analysis demonstrated that the OM was not ruptured during apoptosis. Thus, the more likely candidates for facilitating Cyt c release during CGN apoptotic death may be the Bcl-2 family members and/or VDAC.
This study was supported in part by the National Institute of Health (no. NS73110 to JLF). We thank Angela Okragly and Julia Kasprzak for their technical expertise and Drs Michael Betlach and Kim Knoche for their advice on the statistical analyses. We also thank the members of the Promega Corporation Immunology and Neurobiology R & D group for their input throughout the course of this work.