In normal pancreatic acinar cells, the oxidant menadione evokes repetitive cytosolic Ca2+ spikes, partial mitochondrial depolarisation,cytochrome c release and apoptosis. The physiological agonists acetylcholine and cholecystokinin also evoke cytosolic Ca2+ spikes but do not depolarise mitochondria and fail to induce apoptosis. Ca2+ spikes induced by low agonist concentrations are confined to the apical secretory pole of the cell by the buffering action of perigranular mitochondria. Menadione prevents mitochondrial Ca2+ uptake, which permits rapid spread of Ca2+ throughout the cell. Menadione-induced mitochondrial depolarisation is due to induction of the permeability transition pore. Blockade of the permeability transition pore with bongkrekic acid prevents activation of caspase 9 and 3. In contrast, the combination of antimycin A and acetylcholine does not cause apoptosis but elicits a global cytosolic Ca2+ rise and mitochondrial depolarisation without induction of the permeability transition pore. Increasing the cytosolic Ca2+buffering power by BAPTA prevents cytosolic Ca2+ spiking, blocks the menadione-elicited mitochondrial depolarisation and blocks menadione-induced apoptosis. These results suggest a twin-track model in which both intracellular release of Ca2+ and induction of the permeability transition pore are required for initiation of apoptosis.
Mitochondrial membrane permeabilization is regarded as an important hallmark of early apoptosis. There is still uncertainty about the mechanism by which the permeabilization of the inner mitochondrial membrane occurs, but one possibility, which has attracted recent interest, is induction of the permeability transition pore (PTP). Elevation of the cytosolic Ca2+concentration may play a role in this process(Ferri and Kroemer, 2001;Kroemer and Reed, 2000).
Elevations of the cytosolic Ca2+ concentration are used as general signalling mechanisms, which activate different processes even in the same cell. Ca2+ can interact with many different molecular targets,and it has therefore become increasingly clear in recent years that signal specificity requires subcellular localization and/or special temporal patterns. A number of different mechanisms allowing such spatiotemporal specificity have been identified (Berridge et al., 2000; Cancella et al., 2000;Parekh, 2000;Petersen et al., 2001).
Pancreatic acinar cells have proved useful models for the study of Ca2+ signalling. Physiological stimulation, with the neurotransmitter acetylcholine (ACh) or the circulating peptide hormone cholecystokinin (CCK), evokes repetitive cytosolic Ca2+ spikes, and these are mostly confined to the apical granular pole(Petersen et al., 2001). This is partly due to the clustering of Ca2+ release channels in the endoplasmic reticulum (ER) extensions in the apical pole(Petersen et al., 2001) and partly due to the firewall of active mitochondria around the apical granular pole, which acts as a Ca2+ buffer barrier(Tinel et al., 1999). Each local Ca2+ spike evokes an exocytotic secretory response(Maruyama and Petersen, 1994). Supramaximal agonist stimulation, eliciting a sustained global cytosolic Ca2+ elevation, has recently been shown to induce intracellular enzyme activation and vacuole formation in the apical granular pole(Parekh, 2000;Raraty et al., 2000).
How do Ca2+ signals that cause apoptosis differ from those that have other effects? Thapsigargin, a specific inhibitor of sarcoendoplasmic reticulum Ca2+-ATPase, can induce apoptosis in a wide variety of epithelial and non-epithelial cell lineages(Baffy et al., 1993;Qi et al., 1997). Recent evidence suggests this may involve activation of caspase 12, which is localised at the ER. This mechanism is stimulus specific in that other inducers of apoptosis not causing stress within the ER do not activate caspase 12 (Nakagawa et al., 2000). Mitochondria also play a critical role in another pathway and orchestrate a wide range of stimuli leading to activation of caspase 9. In response to appropriate intracellular signals, which remain to be clearly defined,cytochrome c and other proteins are released into the cytosol from the intermembrane space of the mitochondria(Liu et al., 1996;Susin et al., 1999). In the cytosol, a complex is formed containing cytochrome c, APAF-1 and procaspase 9,which processes and activates caspase 9(Zou et al., 1997). Caspase 9 then activates caspase 3 and other downstream caspase family members that cleave specific protein targets causing apoptosis(Li et al., 1997). There is also evidence that opening of the PTP in mitochondria is important for the release of cytochrome c. The PTP is a pore formed from a complex of the voltage-dependent anion channel (VDAC), the adenine nucleotide translocase and cyclophilin-D at contact sites between the inner and outer mitochondrial membrane (Crompton, 1999). PTP activators, including Ca2+, induce an open-pore state, possibly causing swelling of the mitochondrial matrix and rupture of the outer membrane with release of apoptogenic proteins from the intermembrane space(Martinou et al., 2000).
How do cytosolic Ca2+ signals that cause, for example, secretion or intracellular activation of digestive enzymes, act without inducing apoptosis? To answer this question we have investigated how the cytosolic Ca2+ response in intact normal pancreatic acinar cells elicited by ACh and CCK differs from the Ca2+ response caused by the proapoptotic oxidant menadione. Menadione is a quinone that is metabolised by flavoprotein reductase to semiquinone, which can be oxidised back to quinone in the presence of molecular oxygen. In this redox cycle the superoxide anion radical, hydrogen peroxide and other reactive oxygen species are generated(Monks et al., 1992). Menadione can cause elevations in the cytosolic Ca2+ concentration contributing to cell death (Nicotera et al., 1992). The exact in vivo mechanisms remain to be fully defined, although studies of permeabilised cultured hepatocytes suggest that IP3-driven Ca2+ signals can trigger the PTP and induce apoptosis (Chernyak and Bernardi,1996; Costantini et al.,2000; Szalai et al.,1999). A further goal was to determine whether oxidants and Ca2+ could interact to induce the PTP and initiate apoptosis.
We show that menadione induces cytosolic Ca2+ spikes initiated in the secretory pole. However, in contrast to those generated by ACh, there is no tendency to restrict the Ca2+ elevation to this part of the cell. Menadione induces partial mitochondrial depolarisation through induction of the PTP, preventing Ca2+ buffering by mitochondria and permitting spread of Ca2+ throughout the cell. Blockade of either cytosolic Ca2+ spikes or induction of the PTP prevents menadione-induced apoptosis. Mitochondrial depolarisation without induction of the PTP does not induce apoptosis even in the presence of an elevated cytosolic Ca2+ concentration. Together these data indicate that both a rapid rise in the cytosolic Ca2+ concentration and induction of the PTP are essential and act cooperatively to mediate menadione-elicited apoptosis.
Materials and Methods
Fura Red AM, Fluo 4, TMRE, Mito Tracker Green FM, BAPTA AM, Rhod-2 AM,calcein AM were obtained from Molecular Probes. ApoAlert Annexin V FITC Apoptosis Kit and propidium iodide were obtained from Clontech. ZVADfmk was obtained from Calbiochem, menadione and the rest of the chemicals were purchased from Sigma (UK). Bongkrekic acid was obtained from Affinity Research Product Ltd (UK).
Single pancreatic cells or small clusters (two or three cells) were acutely isolated from CD1 mouse pancreas as previously described(Thorn et al., 1993). Briefly,the pancreas was injected with 200 units/ml collagenase solution, incubated for 10-15 minutes at 37°C then agitated manually and by pipette to obtain the isolated cells. Isolation procedure and experiments were performed in a standard buffer containing 140 mM NaCl, 1.13 mM MgCl2, 1 mM CaCl2, 4.7 mM KCl, 10 mM glucose, 10 mM HEPES, pH 7.2 adjusted with NaOH. After the isolation procedure, cells were washed with standard buffer.
Cytosolic Ca2+ was measured by loading pancreatic acinar cells with either Fura Red or Fluo 4 by incubation in standard buffer containing 10μM Fura Red AM or Fluo 4 AM for 45 minutes at 22°C. After incubation,cells were washed with standard solution and used for experiments (excitation 488 nm, emission >515 nm). Fluorescence measurements were done using a Noran confocal microscope with a 60× Nikon objective (1.4 NA). Linescan(4 ms per line) in slow mode (6400 ns) was used with a slit of 25 μm. Images were processed using TwoD Analysis software (Noran): shade correction(dividing by first image) and 3×3(7) low pass filtering were used. Images of Fura-Red-loaded cells were also inverted (using subtraction from saturated image) and shown with linear colour scale. The maximal and minimal values of the Fura Red fluorescence were determined at the end of each experiment by applying 20 μM ionomycin with 1 mM CaCl2 or 10 mM EGTA. Calculations of Ca2+ concentration in single cells were performed in the conventional way for single wavelength indicators(Takahashi et al., 1999). Kd for Fura Red was assumed to be 140 nM and 345 nM for Fluo 4 (Molecular Probes).
Pancreatic acinar cells were loaded with Rhod-2 AM (5 μM) for 30 minutes at 37°C in standard buffer (see cell preparation procedure). After washing with the same buffer, cells were observed by confocal microscopy (excitation 543 nm, emission >560 nm).
Calcein loading procedure
Cells were loaded with 1 μM calcein AM in the presence of 1 mM CoCl2 for 15 minutes at 37°C(Petronilli et al., 1999) in standard buffer solution followed by washing with the same buffer. Mitochondrial calcein fluorescence was measured using confocal microscopy(excitation 488 nm, emission > 515 nm).
Detection of apoptosis by ApoAlert Annexin V FITC apoptosis kit
Isolated cells were divided into several samples: (1) control cells; (2)cells incubated with 20 μM menadione for 15 minutes or (3) for 30 minutes or (4) 5 pM CCK or (5) 20 nM ACh or (6) 100 nM ACh or (7) 1 μM thapsigargin for 30 minutes or (8) for 3 hours. All the cells were incubated at 22°C and subsequently washed using centrifugation at 200 g for 1 minute. After this, ApoAlert Annexin V FITC kit was applied according to manufacturer's instructions. Briefly, the pellet of the cells from each sample was resuspended in the provided binding buffer and centrifuged at 200 g for 1 minute, again resuspended in 200 μl of binding buffer, and 5 μl of Annexin V FITC conjugate was added. The cells plus conjugate were incubated for 15 minutes at 22°C, washed by centrifugation and resuspended in binding buffer and observed by confocal microscopy(excitation 488 nm, emission >515 nm). To distinguish late apoptotic and necrotic cells, we used staining with 1 μM of propidium iodide. Early apoptotic cells and nonapoptotic cells do not show positive staining with this dye (excitation 363 nm, emission >400 nm). Cells whose nuclei stained with propidium iodide were assumed to be necrotic and excluded from analysis. In experiments using apoptosis inhibitors BAPTA or ZVADfmk cells were incubated for 1 hour in a solution containing 25 μM BAPTA AM or 250 μM ZVADfmk followed by application of 20 μM menadione.
Subcellular fractionation of immunoblotting for cytochrome c
Cytosolic and mitochondrial fractions from control and menadione-treated cells (treated for 2 minutes) were prepared according to Schuler et al.(Schuler et al., 2000). Cells were isolated from the pancreas of two mice in ice-cold buffer containing 150 mM KCl, 10 mM Hepes (pH 7.2), 1 mM EDTA, 1 mM DTT and protease inhibitor cocktail tablets (Roche, USA). The protein content of the fractions was determined by Bradford assay (Sigma, UK). All subcellular fractions were analysed by dot-blot and western blot analysis. Samples were boiled in Laemmli buffer for 5 minutes and subjected to electrophoresis in 10-20%SDS-polyacrylamide gels using Bio-Rad Mini-PROTEAN 3 Cell, followed by transfer on nitrocellulose membranes (0.2 μm, Sigma). After blocking with tris buffered saline containing 5% nonfat dry milk and 0.05% Tween 20,nitrocellulose membranes were exposed to the primary antibodies (cytochrome-c H-104, rabbit polyclonal, Santa Cruz Biotechnology, Inc.) and horseradish peroxidase-conjugated anti-rabbit secondary antibodies (Sigma, UK). Antibody binding was detected using SuperSignal West Pico Chemiluminescent Substrate according the protocol (SuperSignal, Pierce).
Measurement of caspase 3 and caspase-3-like activity using fluorescent caspase 3 substrate
A pellet of pancreatic acinar cells (treated with menadione alone or pre-treated with BAPTA AM before menadione application or untreated (control cells)) was resuspended in small volume of media containing 10 μM fluorogenic caspase 3 substrate PhiPhiLux (OncoImmunin). Following incubation for 1 hour at 37°C and 5% CO2, the pellet was washed twice with standard buffer and observed under confocal microscopy. The excitation of PhiPhiLux substrate was 488 nm, emission >515 nm.
Measurement of caspase 9 activity
Caspase 9 activity was measured using a carboxyfluorescein derivative of benzyloxycarbonyl leucylglutamylhistidylaspartic acid fluoromethyl ketone(zLEHD-FMK), which is a potent inhibitor of caspase 9 (FAM-LEHD-FMK). FAM-LEHD-FMK enters the cells and covalently binds to caspase 9 to activate it as well as to caspase 4, 5 and 6. The caspase 9 detection procedure was performed according the protocol of CaspaTag caspase 9 (LEHD) Activity Kit(Intergen). Briefly, 300 μl aliquots of induced and control cells were transferred to fresh tubes and 10 μ1 of 30× Working Dilution FAM-LEHD-FMK was added directly to the cell suspension followed by mixing and incubation for 1 hour at 37°C. After incubation, cells were washed with 1× Working Dilution Wash Buffer and resuspend in standard buffer. Live cells (propidium iodide negative) containing bound caspase 9 inhibitor were analysed using confocal microscopy (excitation 488 nm, emission > 515 nm).
Measurements of mitochondrial membrane potential changes
Changes of mitochondrial membrane potential were detected with tetramethylrhodamine ethylester (TMRE) using Leica SP-2 confocal microscope. TMRE accumulates in mitochondria according to the Nernst equation and has been used previously to assess relative changes in mitochondrial potential. No quenching of cellular fluorescence with TMRE was detected in our experiments. Decrease in fluorescence corresponds to depolarisation of mitochondrial membrane potential. Cells were loaded with TMRE at a concentration of 0.5μM in standard solution for 30 minutes at 37°C (excitation 488 nm,emission >550 nm). To determine the subcellular localisation of mitochondria, cells were coloaded with 0.1 μM of Mito Tracker Green FM for 15 minutes (excitation 488 nm, emission <550 nm) and then washed in standard buffer.
We considered that a cell was positively stained with Annexin V FITC if the fluorescence of this cell was higher than the averaged value for the control group of cells plus 3 standard deviations of fluorescence of this control group. The same criterion was applied in experiments with FAM-LEHD-FMK and PhiPhiLux. Significance was determined using an analysis of variance using SPSS software. A value of P<0.05 was considered to be significant.
Effect of menadione on the cytosolic Ca2+concentration
We investigated the effect of menadione on the cytosolic Ca2+concentration in isolated pancreatic acinar cells and compared this effect with that of ACh. Menadione (20 μM) induced repetitive cytosolic Ca2+ spikes (n=27 cells)(Fig. 1A, upper trace)(Han et al., 2000). Lower concentrations of menadione did not induce consistent Ca2+responses. As similar responses were evoked in the absence of external Ca2+ (n=17) (Fig. 1A, lower trace), the primary and main source of Ca2+must be intracellular stores. As has been demonstrated previously, similar Ca2+ signals could be obtained in response to stimulation with 100 nM ACh (n=10) (Fig. 1B). It is well known that this secretagogue mobilizes Ca2+ from the endoplasmic reticulum (ER)(Petersen et al., 2001). Application of 100 μM ACh in the absence of extracellular Ca2+substantially reduced the subsequent release of Ca2+ evoked by menadione, suggesting that these two agents release Ca2+ from the same store (n=6 for each trace)(Fig. 1C).
Menadione induces apoptosis
We then investigated whether menadione induces apoptosis in normal pancreatic acinar cells. Six hours after isolation, 64±8%(n=128) of the menadione-treated cells were propidium-iodide positive compared with 15±6% (n=109) of the untreated cells(Fig. 2C). All subsequent studies were then undertaken less than 1 hour after isolation, when less than 5% of the cells treated with menadione were propidium-iodide positive. In order to identify individual cells undergoing apoptosis, we employed two techniques based on fluorescence confocal microscopy. First, phosphatidyl serine becomes redistributed to the outer leaflet of the plasma membrane in cells undergoing apoptosis and can be detected with fluorescently labelled Annexin V (Martin et al.,1995). Cells were incubated for 30 minutes with menadione at 22°C with subsequent application of Annexin V FITC to menadione-treated and control cells to detect apoptosis. All propidium-iodide negative cells in a microscope field were analysed in a double blind manner. At least five microscope fields of view were evaluated. The result for each cell is presented separately as a single bar (Fig. 2A). No cells were Annexin V FITC positive in the control group. The average fluorescence intensity of the Annexin V FITC stained control(non-apoptotic) cells was 71.5±12.3s.d. (arbitrary units), n=69. Cells with fluorescence higher than 108.5 (average value plus 3×s.d. value) were considered apoptotic. After incubation with menadione, the fluorescence intensity of Annexin V FITC increased markedly in comparison with control cells (86.3% of Annexin V FITC positive cells in a population of cells treated with menadione for 30 minutes and 66.7% of Annexin V FITC positive cells in population of cells treated with menadione for 15 minutes). The intensity of fluorescence of control group cells and cells incubated with menadione for 30 minutes (n=96) or 15 minutes(n=13) were significantly different (P<0.001). Cells incubated for 30 minutes with either CCK at its physiological concentration (5 pM) or 25 nM ACh or 100 nM ACh did not display statistically significant changes in Annexin V FITC fluorescence compared with control cells(Fig. 2A) (n>7 for each group). It has been reported previously that the specific inhibitor of the sarcoendoplasmic reticulum Ca2+-ATPase, thapsigargin, can induce apoptosis in pancreatic islet and haemopoietic cells(Baffy et al., 1993;Qi et al., 1997;Zhou et al., 1998). Treatment of pancreatic acinar cells with thapsigargin (1 μM) for 30 minutes did not induce apoptosis within the timeframe of the experiment(Fig. 2A) (n=15). The thapsigargin concentration used is maximal and effectively releases all Ca2+ stored in the endoplasmic reticulum in pancreatic acinar cells within about 15 minutes (Mogami et al.,1998; Toescu et al.,1992). However, after treatment of cells with thapsigargin (1μM) for 3 hours, 37.9% of the cells in the population became apoptotic(n=30) (Fig. 2A). Hoechst staining of the nuclei of cells treated with 20 μM menadione for 3 hours demonstrated chromatin condensation characteristic of apoptosis(Fig. 2D,E)(Gukovskaya et al., 1997). No nuclear condensation could be detected at earlier time points.
To check that the increase in Annexin binding to menadione-treated cells was due to apoptosis, the effect of Z-VADfmk, a cell membrane permeable inhibitor of caspases 1-10, was studied(Garcia-Calvo et al., 1998). Treatment of cells with 250 μM Z-VADfmk for 1 hour reduced the percentage of apoptotic cells in the menadione-treated population from 86.3% to 21.4%(Fig. 2A). Furthermore,Z-VADfmk significantly reduced absolute Annexin fluorescence compared with cells treated with menadione alone (n=15, P<0.001).
We then determined whether a rise in the cytosolic Ca2+concentration is required for menadione-induced apoptosis. We have demonstrated previously that cytosolic Ca2+ spikes induced by ACh can be abolished by buffering cytosolic Ca2+ with BAPTA, in spite of the fact that the rate of Ca2+ release from the ER is actually enhanced under this condition (Mogami et al., 1998). BAPTA prevented any significant change in Annexin V FITC fluorescence of cells exposed to menadione (no Annexin V FITC positive cells in population, n=12) (Fig. 2A). Similar results have been noted in the pancreatic cell line AR4-2J (Sata et al., 1997). BAPTA loading abolished the menadione-elicited cytosolic Ca2+ spike but did not prevent a very slow rise in the cytosolic Ca2+concentration (Fig. 2B). These data, together with data showing that neither CCK nor ACh induce apoptosis,suggest that a rise in the cytosolic Ca2+ concentration is essential, but insufficient, for the induction of apoptosis by menadione.
Menadione-induced Ca2+ signals are not restricted to the apical secretory pole
We have previously shown that at the peak of the ACh-induced cytosolic Ca2+ response, there can be a very significant cytosolic Ca2+ concentration gradient along a line connecting the secretory granular pole with the nucleus (up to 400 nM/μm)(Gerasimenko et al., 1996). We therefore examined the spatial distribution of the menadione-induced Ca2+ signals using confocal linescan measurements. The cytosolic Ca2+ concentrations were determined along a line connecting the secretory granule region and the nucleus(Fig. 3e-h). The location of the granular region can be seen in the transmitted light images(Fig. 3a-d) although the nucleus is the bright area in the Fura-red images on the right side of the cell (Fig. 3e-h). A low ACh concentration (25 nM) induced localised Ca2+ spikes. These Ca2+ signals remained confined to the apical granule-containing area, owing to the perigranular mitochondrial belt(Cancela et al., 2000;Tinel et al., 1999), and therefore did not enter the nucleus (Fig. 3i). Higher doses of ACh (100 nM) usually induced large responses,which also started in the secretory granule area but then spread into the basal area. Spreading of the Ca2+ signal into the nucleus was delayed by 1-7 seconds (n=10) compared to entry into the non-nuclear region (Fig. 3j). The ratio of Ca2+ wave speed propagation into the nuclear region (angle β)compared to speed into the non-nuclear region (angle α) was 0.4±0.1. This delayed entry into the nucleus is most likely due to the recently discovered mitochondrial ring surrounding the nucleus(Park et al., 2001). Menadione-induced Ca2+ signals were also initiated in the secretory granular zone, but the response spread into the nuclear area without delay with nearly the same speed as into the non-nuclear region (ratio:0.8±0.1, n=7). The ratio was significantly different when compared to ACh (P<0.01) (Fig. 3k,m-q).
We have recently shown that active mitochondria, localized in a ring surrounding the granular region, serve as a buffer barrier for Ca2+signal progression from the granular region to the basolateral areas(Tinel et al., 1999). To test the proposition that the Ca2+ wave elicited by menadione can spread fast to the nuclear region because of inhibition of mitochondrial function,antimycin A, a well characterized inhibitor of mitochondrial electron transport (Singer, 1979), was applied together with ACh. In this case the cytosolic Ca2+ response was very similar to its response to menadione(Fig. 3l) (speed ratio:0.9±0.1, n=5, P>0.3), confirming the mitochondrial involvement in the delayed appearance of the Ca2+ signal in the nucleus in response to ACh stimulation under normal conditions. Mitochondrial Ca2+ uptake (Park et al.,2000) could result in a slower rate of Ca2+ elevation in the nucleus (Fig. 3), which could be important by preventing activation of Ca2+-dependent nuclear enzymes.
Menadione causes initial Ca2+ loading into mitochondria followed by unloading
The results above suggest that ACh causes Ca2+ loading into mitochondria, thereby delaying the spread of Ca2+ away from the secretory pole, whereas menadione prevents buffering of Ca2+ by mitochondria. To test this hypothesis, we compared the effects of ACh and menadione on mitochondrial Ca2+ loading with Rhod-2. This fluorescent dye has been well characterised as a reporter of changes in the Ca2+ concentration inside mitochondria in living cells(Duchen, 1999). We therefore measured changes in rhod-2 fluorescence from the perigranular mitochondrial belt (Tinel et al., 1999). ACh(100 nM) caused an initial rapid increase in the mitochondrial Ca2+concentration followed by a slower decrease back towards basal Ca2+levels (Fig. 4A). In contrast,menadione only caused a very small and transient increase in the mitochondrial Ca2+ concentration followed by a prolonged and marked reduction(Fig. 4A, n=19). These results indicate that although menadione causes a small initial uptake of Ca2+ into the mitochondria, this is countered by induction of a Ca2+ efflux pathway. This is likely to account for the rapid menadione-induced spread of Ca2+ through the cell(Fig. 3k).
Menadione induces depolarisation of mitochondria
To gain further insight into potential Ca2+ efflux pathways from mitochondria, we measured changes in the potential across the inner mitochondrial membrane in isolated pancreatic acinar cells using the membrane-potential-sensitive fluorescent dye TMRE. Cells were loaded with TMRE and MitoTracker Green to check the localisation of the mitochondria. The TMRE distribution was identical to the distribution of Mito Tracker Green(Fig. 4Ba-c).Figs. 4Ba and 4Be show the TMRE fluorescence intensity from the mitochondria (green box) and the nucleus(yellow box) of the same cell measured simultaneously. The mitochondrial TMRE fluorescence intensity decreased after application of menadione, indicating a reduction in the mitochondrial membrane potential, whereas the nuclear TMRE fluorescence, which is close to background, remained constant(Fig. 4Be) (n=20). The depolarisation was transient and the mitochondria regained their full electrical potential some 500 seconds after the menadione application. Furthermore, the depolarisation was only partial as the protonophore mCCCP (10μM) had a much greater effect (Fig. 4Be). The menadione-induced depolarisation was markedly reduced,but not abolished, by loading the cells with BAPTA(Fig. 4C, n=20). It has been suggested that the mitochondrial depolarisation may be the result of caspase activity (Marzo et al.,1998b), but this is not the case in isolated pancreatic acinar cells, as Z-VADfmk did not block the mitochondrial depolarisation(Fig. 4C, n=20).
In order to explore the relationship between changes in the cytosolic Ca2+ concentration and the reduction in mitochondrial membrane potential further, these two parameters were measured simultaneously in mitochondrial regions of cells loaded with Fluo 4 and TMRE. Application of 100 nM ACh induced a substantial elevation in the cytosolic Ca2+concentration but no change in the mitochondrial membrane potential. Subsequent application of menadione induced both a renewed Ca2+rise and a clear decrease in the mitochondrial membrane potential(Fig. 4E-G) (n=7). In agreement with these data, CCK in concentrations from 5 pM to 10 nM also failed to elicit mitochondrial depolarization, although the protonophore FCCP,in the same experiments, induced collapse of the potential across the inner mitochondrial membrane (Raraty et al.,2000). Taking all the data presented inFig. 4 together, it is apparent that menadione causes both a rise in the cytosolic Ca2+concentration and a reduction in the mitochondrial membrane potential. Since ACh fails to elicit mitochondrial depolarization, it is clear that this phenomenon cannot solely be a consequence of the cytosolic Ca2+rise.
Menadione induces opening of the permeability transition pore
Mitochondrial depolarisation during apoptosis has been reported to be the result of induction of the PTP (Marzo et al., 1998a; Szalai et al.,1999). To test whether this is true in pancreatic cells, we preincubated cells for 45 minutes with the PTP inhibitor bongkrekic acid at a concentration of 50 μM before exposing the cells to menadione. Treatment with bongkrekic acid either completely prevented mitochondrial depolarisation or reduced it dramatically (Fig. 4C, n=23) suggesting that the mitochondrial depolarisation is caused by PTP induction. Furthermore, antimycin A causes mitochondrial depolarization, and this is not blocked by bongkrekic acid,suggesting that menadione and antimycin A are causing mitochondrial depolarization through different mechanisms (n=11)(Fig. 4D). To test the PTP induction hypothesis further, cells were loaded with calcein-AM in the presence of cobalt chloride to quench fluorescence from all cellular domains except from within mitochondria(Petronilli et al., 1999). Using this protocol there was punctate calcein fluorescence around the secretory pole and nuclei, consistent with mitochondrial staining(Fig. 5A,B)(Tinel et al., 1999). Menadione caused a loss of mitochondrial calcein fluorescence, which was completely blocked by 50 μM bongkrekic acid(Fig. 5C,D). Although this protocol does not distinguish between calcein efflux and Co2+influx, it is consistent with induction of the PTP. Loss of mitochondrial calcein fluorescence was not simply a result of mitochondrial depolarisation,as neither antimycin A (Fig. 5E) nor FCCP (data not shown) caused significant loss of calcein fluorescence.
Menadione causes efflux of cytochrome c into the cytoplasm and activates caspase 9 and 3
To determine whether mitochondrial cytochrome c translocates from the mitochondria to the cytoplasm, mitochondrial and cytoplasmic fractions were collected following exposure to menadione or vehicle control for 2 minutes. Cytochrome c was then detected and quantified by immunoblotting. Following exposure to menadione for 2 minutes only, cytochrome c was released into the cytosolic fraction (Fig. 6A,B). BAPTA significantly reduced cytochrome c release from the mitochondria(Fig. 6A,B) (n=5, P<0.005).
A family of cysteine proteases called caspases are known to be activated in apoptosis and are among the principal effector systems of apoptotic cell death. We therefore determined the role of caspase 9, a caspase at the apex of the mitochondrial caspase cascade(Thornberry and Lazebnik,1998). Caspase 9 activity was determined using the fluorescent probe FAM-LEHD-FMK (see Materials and Methods). The averaged value of fluorescence of FAM-LEHD-FMK-stained control cells was 55.4±9.1.s.d. arbitrary units (n=20). Cells with fluorescence higher than 82.7 arbitrary units (average value plus 3×s.d. values) were considered to be FAM-LEHD-FMK positive. A significant increase in FAM-LEHD-FMK fluorescence was detected in all cells after incubation with menadione for 30 minutes(n=15, P<0.001) (Fig. 6C). Intracellular caspase 3 activity was determined using PhiPhiLux-G2D2 - a fluorogenic substrate for caspase-3-like proteases(Zapata et al., 1998). The averaged value of fluorescence of PhiPhiLux-stained control cells was 54.7±7.4s.d. arbitrary units (n=32). Cells with fluorescence higher than 76.8 arbitrary units (average value plus 3×s.d. values) were considered PhiPhiLux positive. A rise in the fluorescence intensity was observed owing to an increase in caspase-3-like activity (n=44, P<0.001) (95.5% of PhiPhiLux positive cells in population)(Fig. 6D). Buffering of cytosolic Ca2+ with BAPTA completely prevented activation of both caspase 9 (n=15) and 3 (n=20)(Fig. 6C,D), demonstrating that Ca2+ is acting at a point upstream of caspase activation. Blockade of caspase activation with Z-VADfmk did not prevent Ca2+ spike formation (data not shown, n=13). These data exclude the possibility that the menadione-elicited rise in the cytosolic Ca2+concentration is the result of caspase 3 activation. Activation of caspase 9 was not simply the result of inhibition of the electron transport associated with oxidative phosphorylation, as a combination of ACh and antimycin A did not activate caspase 9 significantly (n=16)(Fig. 3l,Fig. 6C). Applications of 25μM ACh or 5 pM CCK did not activate caspase 3 significantly(Fig. 6D) (n=10 for each group).
Bongkrekic acid inhibits menadione-induced apoptosis
Previous studies have suggested that induction of PTP is essential for the full induction of apoptosis, though controversy surrounds this point(Martinou et al., 2000). We therefore determined whether bongkrekic acid prevents apoptosis induced by menadione. Preincubation with bongkrekic acid before exposure to menadione prevented apoptosis in the majority of cells as measured by Annexin V FITC staining (n=29) (Fig. 2A). Only 14.3% of cells were apoptotic. Bongkrekic acid completely prevented activation of caspase 9 as measured by FAM-LEDH-FMK fluorescence (no FAM-LEDH-FMK positive cells, n=16)(Fig. 6C) and markedly reduced caspase 3 activity (19% of PhiPhiLux positive cells) as measured by Phiphilux-G2D2 fluorescence (n=31)(Fig. 6D). Together these data suggest that activation of the PTP is an essential element in the process of apoptosis induction elicited by menadione.
Our experiments on normal pancreatic acinar cells provide new insights into the role of Ca2+ in the induction of apoptosis. By comparing the actions of the oxidant menadione and the physiological secretagogues ACh and CCK, important similarities and differences have emerged, which shed new light on the control of apoptosis. Two general problems are of special interest. Is Ca2+ involved in the induction of apoptosis? And, if so, how do normal Ca2+-generating stimuli that are required for the initiation of secretion (exocytosis) avoid eliciting apoptosis? The data presented here show that the oxidant menadione, at a concentration inducing apoptosis, evoked repetitive cytosolic Ca2+ spikes owing to the release of Ca2+ from the same internal stores as ACh(Fig. 1). The data also show that blockade of Ca2+ spiking by loading the cytosol with the Ca2+ chelator BAPTA (Fig. 2B) prevented the induction of apoptosis(Fig. 2A,Fig. 6).
Menadione induces apoptosis as defined by the release of cytochrome c from mitochondria, activation of caspase 9 and 3, Annexin V staining and, at later time points, condensation of nuclear chromatin. Although nuclear condensation often occurs simultaneously with Annexin staining, this depends on the cell type and stimulus (Boersma et al.,1996). Menadione in concentrations above 50 μM can induce necrosis, but in our experiments all cells that were stained positive with propidium iodide (and therefore were potentially necrotic) were excluded from analysis. In some cell types, including β-cells from pancreatic islets,depletion of Ca2+ in the endoplasmic reticulum can induce apoptosis after 6 to 48 hours of exposure (Baffy et al., 1993; Qi et al.,1997; Zhou et al.,1998). However, in the pancreatic acinar cells, loss of Ca2+ from the ER was, by itself, insufficient to induce apoptosis. This was seen in the experiments with specific blockade of the ER Ca2+ ATPase, in which 30 minutes of exposure to the ER Ca2+ pump inhibitor thapsigargin failed to induce apoptosis(Fig. 2A). Even after 3 hours of thapsigargin treatment, only 37.9% of the cells were Annexin V FITC positive as compared with 86.3% of the cells treated by menadione for 30 minutes. Although a longer exposure to thapsigargin can induce apoptosis in pancreatic acinar cells, the time course and the mechanism are probably different from the apoptosis induced by menadione. Buffering cytosolic Ca2+ with BAPTA does not prevent stimulant-evoked emptying of the Ca2+ stores in the ER but actually enhances Ca2+ release owing to a lack of Ca2+-mediated negative feedback on the Ca2+ release channels (Mogami et al., 1998). This indicates that it is the menadione-elicited rise in the cytosolic Ca2+ concentration, rather than the reduction in the Ca2+ concentration within the ER, that is needed for the induction of apoptosis.
Our studies show that ACh induces a significant uptake of Ca2+into the mitochondria (Fig. 4A). In contrast, menadione only stimulates a very transient uptake of Ca2+ into the mitochondria followed by a sustained efflux(Fig. 4A). The spatially unrestricted cytosolic Ca2+ concentration rise elicited by menadione could play a causal role in the induction of apoptosis, but this is unlikely, as ACh combined with antimycin A, which also elicits a spatially unrestricted Ca2+ rise (Fig. 3l), does not activate caspase 9 or apoptosis(Fig. 4D,Fig. 6B). Furthermore bongkrekic acid does not prevent the antimycin-A-induced mitochondrial depolarization, although it blocks the menadione-elicited reduction of the mitochondrial membrane potential. This demonstrates that antimycin A does not induce the PTP. These data indicate that it is induction of the PTP that is important for apoptosis and not mitochondrial depolarisation. Furthermore,loss of ATP is excluded as a potential trigger for apoptosis within the time frame of these experiments. It is interesting to note that high concentrations of ACh and antimycin A cause an increase in the nuclear Ca2+concentration, but this rise would appear not to be sufficient on its own to activate nuclear endonucleases, as these agents do not evoke apoptosis.
The menadione-induced mitochondrial depolarisation was partial, transient and somewhat dependent on the cytosolic Ca2+ concentration rise,but not influenced by caspase inhibition(Fig. 2B,Fig. 5,Fig. 6). As the normal Ca2+ spikes evoked by ACh did not induce mitochondrial depolarization (Fig. 6), it is clear that the ability of menadione to elicit a reduction in the potential across the inner mitochondrial membrane, even under conditions of cytosolic BAPTA loading, could not be explained simply as a consequence of the cytosolic Ca2+ signal generation. These results contrast with those of Szalai et al. (Szalai et al., 1999),who found that Ins(1,4,5)P3-driven Ca2+ signals were sufficient to induce mitochondrial depolarisation in a permeabilised hepatocyte cell line.
The mitochondrial depolarization caused by menadione is probably due to induction of the PTP, as bongkrekic acid, a specific inhibitor of the PTP(Crompton, 1999;Marzo et al., 1998b), markedly reduced or abolished the menadione-elicited mitochondrial depolarization(Fig. 4C). Further evidence for induction of the PTP was obtained with the fluorescent probe calcein(Petronilli et al., 1999). Menadione, but neither FCCP nor antimycin A, caused loss of calcein fluorescence, which indicated that a large enough pore had been induced to allow either calcein efflux or Co2+ influx. Either way, this pore was blocked by bongkrekic acid, providing further evidence for PTP induction. It has been reported that mitochondrial depolarisation could be induced by the action of caspases (Marzo et al.,1998b), but this was not the case in our experiments, as ZVADfmk did not prevent depolarisation (Fig. 4C). We hypothesise that the transient uptake of Ca2+into mitochondria is sufficient to induce the PTP in the presence of oxidants.
The role of the PTP in apoptosis is controversial. Evidence has been reported that induction of the PTP is tightly coupled to cytochrome c release(Fulda et al., 1998;Heiskanen et al., 1999). Furthermore, it has been shown that the proapoptotic protein Bax binds to the adenine nucleotide translocator, a component of the PTP, to induce apoptosis(Marzo et al., 1998a). However, apparently conflicting results have also been reported, where loss of mitochondrial potential occurs after cytochrome c release, and it is prevented by inhibition of caspases (Bossy-Wetzel and Green, 1999; Bossy-Wetzel et al., 1998; Finucane et al.,1999; Nomura et al.,1999). In our studies we found that bongkrekic acid blocked the induction of apoptosis by menadione in the majority of cells studied(Fig. 6). Within the resolution of our analysis, mitochondrial depolarisation occurred simultaneously with the initiation of Ca2+ oscillations induced by menadione(Fig. 4G).
A remarkable feature of menadione-induced apoptosis is that it is very rapid. Cytochrome c release from mitochondria can be detected within 2 minutes of drug application (Fig. 6A). Similar results have been reported in experiments on isolated mitochondria where substantial cytochrome c release was found 5 minutes after exposure to Ca2+ (Gogvadze et al.,2001). Caspase 9 and 3 activations can be detected 30 minutes after exposure to menadione. This time course of activation is faster than many other examples of apoptosis but is not without precedence. For example,in apoptosis induced by Ins(1,4,5)P3-linked mitochondrial calcium signals, similar short periods of caspase induction have also been described (Szalai et al.,1999). Furthermore activation of caspase 9 and 3 is consistent with the kinetics for assembly of the apoptosome, which can occur within 5 minutes (Cain et al., 2000). However it should be noted that chromatin condensation could not be detected until 3 hours after exposure to menadione. Thus our data are consistent with previous studies of the induction of apoptosis in pancreatic cell lines(Sata et al., 1997).
Our data are consistent with a model in which the Ca2+ released from the ER by menadione is transiently taken up by the mitochondria. In the presence of oxidants, this causes induction of the PTP(Chernyak and Bernardi, 1996)and therefore releases Ca2+ from the mitochondria, allowing a wave of Ca2+ to traverse the rest of the cell(Fig. 7). Induction of the PTP allows release of cytochrome c, which leads to the processing and activation of caspase 9 and caspase 3, externalisation of phosphatidylserine and apoptosis (Fig. 7). We cannot exclude the possibility that elevations in the cytosolic Ca2+concentration have additional actions such as regulation of the phosphorylation of proteins critical for apoptosis, for instance the pro-apoptotic protein BAD (Wang et al.,1999). Neither mitochondrial depolarization per se or loss of ATP is responsible for the triggering of apoptosis, as treatment with ACh and antimycin A do not cause apoptosis, because the PTP is not induced under these non-oxidising conditions. This contrasts with the action of submaximal ACh concentrations. In these cases, Ca2+ released from the ER into the granular area is trapped by perigranular mitochondria, which, by means of Ca2+ uptake, act as a firewall and thereby prevent spread to the rest of the cell (Fig. 7).
This work was supported by an MRC programme grant to O.H.P. and A.V.T. O.H.P. is an MRC Research Professor.