A preliminary account of this work has been presented at the American Society for Biochemistry and Molecular Biology, Annual Meeting and 8th International Union of Biochemistry and Molecular Biology Conference, Boston, Massachusetts (June 12–16, 2004).
Address correspondence and reprint requests to Salvatore Passarella, Dipartimento di Scienze per la Salute, Università del Molise, Via A. De Sanctis, 86100 Campobasso, Italy, E-mail: email@example.com; firstname.lastname@example.org
We investigated ADP/ATP exchange mediated by the adenine nucleotide translocator and opening of the mitochondrial permeability transition pore in homogenates from cerebellar granule cells en route to apoptosis induced by low potassium. We showed that, in the first 3 h of apoptosis, when maximum cytochrome c release had already occurred, adenine nucleotide translocator function was impaired owing to the action of reactive oxygen species, but no permeability transition pore opening occurred. Over 3–8 h of apoptosis, the permeability transition pore progressively opened, owing to caspase action, and further ADP/ATP translocator impairment occurred. The kinetics of transport and permeability transition pore opening were inversely correlated, both in the absence and presence of inhibitors of antioxidant and proteolytic systems. We conclude that, en route to apoptosis, alteration of the adenine nucleotide translocator occurs, resulting in permeability transition pore opening. This process depends on the action of caspase on pore component(s) other than the ADP/ATP translocator, because no change in either amount or molecular weight of the latter protein was noted during apoptosis, as measured by western blotting. Cell death occurs via apoptosis in the presence of cyclosporin A, the permeability transition pore inhibitor, thus showing that permeability transition pore opening, not needed for cytochrome c release, is also unnecessary for apoptosis to occur.
One of the outstanding problems in the role of mitochondria in apoptosis concerns cellular ATP production and utilization. In apoptosis ATP is mostly produced in the mitochondria by oxidative phosphorylation and reaches the cytosol via the adenine nucleotide translocator (ANT), where it is used in a variety of processes including caspase activation (for references see Bobba et al. 2004). Thus the role of the ANT in apoptosis is crucial.
On the other hand, besides export of ATP, the ANT also participates in mitochondrial permeability transition pore (mPTP) formation, together with other proteins including the voltage-dependent ion channel (VDAC) and Bax–Bcl-2 in the outer membrane and cyclophilin D in the matrix (for references see Halestrap et al. 2002; Le Bras et al. 2005). Although the mPTP is suggested to be an important mediator of apoptosis, the mechanism underlying its opening, particularly in regard to ANT, and its involvement in apoptotic release of mitochondrial proteins, including cytochrome c (cyt c), is the subject of ongoing debate (for references see Tatton and Chalmers-Redman 1998; Gulbins et al. 2003). It has been proposed that pro-apoptotic Bcl-2 family members such as Bax and Bak cause opening of the mPTP, whereas anti-apoptotic members, such as Bcl-2, favour closure of this channel (Reed 1997; Belzacq et al. 2003). However, the role of each component and the interaction between the components of the mPTP remain to be fully established. Recently it was shown that mitochondria from liver of ANT knockout mice still possess mPTP activity, suggesting that ANT is a non-essential component of the mPTP and may not be required for mPTP-associated cell death (Kokoszka et al. 2004).
We have investigated this subject using cerebellar granule cells (CGCs) as a model system. The cells are cultured in a medium containing 25 mm potassium (S-K25 cells) and undergo apoptosis when subjected to a potassium shift to 5 mm (S-K5 cells) (D'Mello et al. 1993). These cells represent a model that, in contrast to others that require use of harmful manipulations (Gorman et al. 1999; De Moliner et al. 2002), mimic a prolonged state of neuronal depolarization in vivo (Borsello et al. 2000). Moreover, CGC homogenates contain coupled mitochondria (Atlante et al. 1998a, 2003b), thus allowing investigation of mitochondrial bioenergetics, and in particular ANT function, in situ.
The time course of apoptosis has already been described in some detail and occurs in two phases. In early apoptosis (0–3 h) (Scheme 1a), production of reactive oxygen species (ROS) occurs, modulation of which is effected by the antioxidant system (AOS), which in turn is modulated by proteasomes (Atlante et al. 2003a). Then cyt c is released from the mitochondria and acts as a bifunctional factor: it can remove ROS as a scavenger and can drive ATP synthesis, perhaps regulating the time course of apoptosis (Atlante et al. 2003b). It has been shown recently (Atlante et al. 2005a) that an increase in ATP and a decrease in ADP occur with 1 : 1 stoichiometry, with the maximum ATP level found at 3 h of apoptosis coinciding with maximum release of cyt c. In late apoptosis (3–8 h) (Scheme 1b), the ROS level reaches a steady state, there is cyt c-dependent caspase activation, and degradation of the AOS due to both proteasome and caspase action occurs, as does caspase-dependent degradation of the released cyt c (Bobba et al. 1999; Atlante et al. 2003a). The level of ATP declines to control values. Interestingly, depending on the level of ROS, CGCs can die via apoptosis, via necrosis or as a result of ‘energy catastrophe’ (Atlante et al. 2003a).
We have measured ADP/ATP exchange and mitochondrial permeability transition (mPT) as a function of time after induction of apoptosis in homogenates from S-K25 and S-K5 cells in the absence or presence of compounds designed to selectively affect ROS production and the AOS and proteolytic system. The aim of this was to determine whether these parameters change en route to apoptosis and, if so, when and how. Moreover, we wished to establish whether ANT transport function and mPTP opening are linked to one another. We show that en route to apoptosis, a double alteration of the ADP/ATP translocator occurs, owing to the action of ROS in the early phase and of caspase in the late phase. The action of caspase alone can trigger the mPT. This process depends on proteolysis by caspases of mPTP protein component(s) other than the ANT, because no change in either the amount or molecular weight of the ANT takes place during apoptosis. Finally, we show that the mPTP is not required for either cyt c release or for apoptosis to occur.
Materials and methods
Tissue culture medium and fetal calf serum were purchased from Gibco (Grand Island, NY, USA) and tissue culture dishes were from NUNC (Taastrup, Denmark). All enzymes and biochemicals were from Sigma Chemical Co. (St Louis, MO, USA). Monoclonal anti-cyt c antibodies against either the denatured protein (7H8-2C12) or against the native protein (6H2.B4) were purchased from Pharmingen (San Diego, CA, USA), monoclonal anti-cytochrome oxidase (COX) antibodies were purchased from Molecular Probes (Eugene, OR, USA) and horseradish peroxidase-conjugated anti-mouse and anti-rabbit antibodies used with enhanced chemiluminescence western blotting reagents were from Amersham (GE Healthcare Europe GmbH, Munich, Germany). Polyclonal ANT antibodies were kindly supplied by Professor A. Halestrap (University of Bristol, Bristol, UK) and glutamate dehydrogenase (GDH) antibodies were kindly supplied by Dr F. Rothe (Institut fur Medizinische Neurobiologie, University of Magdeburg, Magdeburg, Germany). Protein A/G PLUS-agarose was purchased from Santa Cruz Biotechnology (Santa Cruz, CA, USA).
Primary cultures of CGCs were obtained from dissociated cerebella of 7-day-old Wistar rats as described by Levi et al. (1984). Cells were plated in basal medium Eagle (BME) supplemented with 10% fetal calf serum, 25 mm KCl, 2 mm glutamine and100 µg/mL gentamicin on dishes coated with poly-l-lysine. Arabinofuranosylcytosine (10 µm) was added to the culture medium 18–22 h after plating to prevent proliferation of non-neuronal cells.
Induction of apoptosis
Apoptosis was induced at 6–7 days in vitro. Cells were washed and switched to serum-free BME, containing 5 mm KCl and supplemented with 2 mm glutamine and 100 µg/mL gentamicin for the indicated times (D'Mello et al. 1993). Control cells were treated identically but maintained in the same serum-free BME medium containing 25 mm KCl. All the compounds used were added to the cell culture medium simultaneously with induction of apoptosis.
Preparations of cell homogenates and mitochondria
The culture medium was removed and the plated CGCs were washed with phosphate-buffered saline (PBS), containing 138 mm NaCl, 2.7 mm KCl, 8 mm Na2HPO4 and 15 mm KH2PO4, pH 7.4, and then collected. Cell integrity was assessed quantitatively by the inability of cells to oxidize externally added succinate, and by the ability of ouabain to block glucose transport (Sjodin 1989).
Cell homogenate was obtained from a cell suspension by 10 strokes with a Dounce homogeniser at 20°C. Cytosolic lactate dehydrogenase was released and subsequent treatment with Triton X-100 did not cause further release.
Mitochondria were isolated from cell homogenates as described by Almeida and Medina (1998). Mitochondria, incubated in PBS, were checked both for their coupling by measuring the respiratory control index (RCI), i.e. (oxygen uptake rate after ADP addition)/(oxygen uptake rate before ADP addition), which reflects the ability of mitochondria to produce ATP, and for their intactness by measuring the activities of adenylate kinase (ADK; E.C.188.8.131.52) and GDH (E.C.184.108.40.206) in the postmitochondrial supernatant; these are marker enzymes of the mitochondrial intermembrane space and matrix respectively. The ADK reaction was assayed essentially as in Atlante et al. (1998b) at 25°C and pH 7.2 to mimic intracellular pH in a standard coupled spectrophotometric assay, in which the ADK-catalysed synthesis of ATP from ADP was measured by using glucose (2.5 mm), hexokinase [HK; 0.5 enzymatic unit (e.u.)], glucose-6-phosphate dehydrogenase (G6PDH, 0.5 e.u) and NADP+ (0.2 mm). To prevent any ATP production via oxidative phosphorylation, 10 µg oligomycin (OLIGO) and 20 µm atractyloside (ATR) were also present to completely inhibit ATP synthase and the ANT respectively. When determining GDH activity at 25°C the following substrates were used: 10 mm 2-oxoglutarate, 10 mm NH4Cl and 0.2 mm NADH (Ostuni et al. 1993); the NADH oxidation was monitored photometrically at 340 nm as a function of time. Protein content was determined according to (Waddel and Hill 1956) with bovine serum albumin as a standard.
O2 consumption was measured polarographically by means of a Gilson 5/6 oxygraph using a Clark electrode, as described previously (Atlante et al. 1996, 2000). Instrument sensitivity was such as to allow rates of O2 uptake as low as 0.5 natoms/min/mg protein to be followed. The cell homogenate in PBS (about 0.2 mg protein) was incubated in a thermostated (25°C) water-jacketed glass vessel (final volume 1.5 mL).
Safranine O response assay
The safranine O response was monitored as described by Atlante et al. 2003b). Time-dependent absorbance changes were recorded with a Jasco (Gross-Umstadt, Germany) double-beam/double-wavelength spectrophotometer UV-550 with wavelengths of 520 and 554 nm. Measurements were carried out at 25°C in 2 mL standard medium consisting of 200 mm sucrose, 10 mm KCl, 1 mm MgCl2 and 20 mm HEPES-Tris pH 7.2, containing 1 µm safranine O and 0.1 mg protein.
ADP/ATP carrier measurement
Cell homogenate (0.1 mg protein), containing mitochondria, was incubated at 25°C in 2 mL standard medium consisting of 200 mm sucrose, 10 mm KCl, 1 mm MgCl2 and 20 mm HEPES-Tris pH 7.2. Appearance of ATP in the extramitochondrial phase, due to externally added ADP, was revealed as described previously (Passarella et al. 1988; Atlante et al. 2003b), by using an ATP-detecting system consisting of glucose (2.5 mm), HK (0.5 e.u.), G6PDH (0.5 e.u) and NADP+ (0.2 mm) in the presence of P1,P5-di(adenosine-5′)penta-phosphate (AP5A), a specific inhibitor of ADK (Lienhard and Secemski 1973). The rate of NADP+ reduction in the extramitochondrial phase was followed as the absorbance increase at 334 nm, measured as the tangent to the initial part of the progress curve and expressed as nmoles NADP+ reduced per min per mg cell protein. Control experiments were carried out in the presence of ATR to ensure that the ADP/ATP exchange was mediated by the ADP/ATP carrier (see Wanders et al. 1984; Rossignol et al. 2000; Passarella et al. 2003).
Measurement of mPT
To visualize the onset of mPT, either mitochondrial swelling or ADK release from mitochondria was monitored. Mitochondrial swelling was monitored at 25°C by following the absorbance decrease at 546 nm, as described by Bernardi et al. (1999). Cell homogenate was suspended in medium containing 250 mm sucrose, 2 mm HEPES, pH 7.4, 0.5 mm K2HPO4, 1 µm OLIGO, 2 µm rotenone and 5 mm succinate. ADK release was monitored by assaying ADK activity at 25°C, essentially as reported above, in postmitochondrial supernatant of both control and S-K5 cells (about 30 × 106 cells in 1 mL), obtained after homogenization in PBS and centrifugation at 15000 g for 15 min. Controls were carried out to show that none of the compounds used in this paper had any effect on the ADK reaction. In particular, we confirmed that no change in the assayed ADK occurred in postmitochondrial supernatants of cells previously incubated with proteasome and/or caspase inhibitor(s), i.e. MG132 and z-VAD-fmk (benzyloxycarbonyl-Val-Asp (OMe) fluoromethylketone).
Purified mitochondria from either control or apoptotic CGCs were solubilized in 1% Triton X-100, 500 mm NaCl, 8 mm MOPS, 0.04 mm EDTA, pH 7.2, for 30 min on ice. Solubilized mitochondrial proteins (4 µg) were loaded on to a Tricine–sodium dodecyl sulphate polyacrylamide gel (Schagger and von Jagow 1987), separated, and transferred to a polyvinylidene difluoride membrane which was probed with polyclonal anti-ANT antibodies. Monoclonal anti-COX antibodies were used as a loading control. Immunoblot analysis was performed with horseradish peroxidase-conjugated anti-mouse or anti-rabbit antibodies using enhanced chemiluminescence western blotting reagents. Relative optical densities and areas of bands were quantified using a GS-700 Imaging Densitometer implemented with Molecular Analyst Software (Bio-Rad Laboratories, Hercules, CA, USA).
To follow cyt c release, immunoblot analysis on cytosolic fractions was performed as already described, with anti-GDH antibodies used as a loading control (Bobba et al. 1999).
Polarographic detection of cyt c release
In order to detect polarographically the presence of cyt c in the extramitochondrial phase, the ability of the cell homogenate to oxidize ascorbate was checked (Atlante et al. 2001 and references therein). Briefly, because ascorbate cannot permeate the outer mitochondrial membrane per se (Alexandre and Lenhinger 1984), then its oxidation can occur as a result of the release from mitochondria of a component that can oxidize ascorbate and then reduce oxygen via cytochrome c oxidase, i.e. cyt c present in the extramitochondrial phase.
DNA fragmentation analysis
Fragmentation of DNA was performed as in described by Volontèet al. (1994). Briefly, CGCs (6 × 106) were plated in poly-l-lysine-coated 60-mm tissue culture dishes and collected with cold PBS (pH 7.2). After removal of the medium and washing once with cold PBS, CGCs were centrifuged at 3500 g for 5 min. The pellet was lysed in 10 mm Tris-HCl, 10 mm EDTA and 0.2% Triton X-100 (pH 7.5). After 30 min on ice, the lysate was centrifuged at 17 000 g for 10 min at 4°C. The supernatant was digested with proteinase K and then extracted twice with phenol–chloroform/isoamyl alcohol (24 : 1). The aqueous phase, containing soluble DNA, was recovered and nucleic acids were precipitated with sodium acetate and ethanol overnight. The pellet was washed with 70% ethanol, air-dried and dissolved in TE buffer (10 mm Tris-HCl, 1 mm EDTA, pH 7.5). After digestion with Rnase A (50 ng/mL at 37°C for 30 min), the sample was subjected to electrophoresis in a 1.8% agarose gel and visualized by ethidium bromide staining. Soluble DNA from equal numbers of cells was loaded in each lane.
Assessment of neuronal viability
Viable CGCs were quantified by counting the number of intact nuclei after dissolving the cells in detergent-containing solution, as described previously (Volontèet al. 1994).
Statistical analysis and computing
All statistical analyses in this study were performed using SPSS software (SPSS, Chicago, IL, USA). The data are representative of at least three independent neuronal preparations (with comparable results), each with independent measurements (in each figure legend the number of measurements is reported), and are reported as mean ± SD. Statistical significance was evaluated using one-way anova followed by post hoc Bonferroni test. p < 0.05 was considered significant for all analyses. Experimental plots were obtained using Grafit Erithacus software (Horley, Surrey, UK). Numerical analyses were performed using Microcal Origin software (Northampton, MA, USA).
ANT-dependent ATP efflux from mitochondria due to externally added ADP and mPTP opening in either S-K25 or S-K5 homogenates
To determine whether and how ADP/ATP exchange mediated by the ANT, as investigated in cell homogenates, could change in apoptosis, we used a procedure (Passarella et al. 1988) that allows the continuous monitoring of ATP efflux from mitochondria incubated with ADP. Because ATP production via oxidative phosphorylation from externally added ADP requires coupled mitochondria, careful control of the functional integrity of mitochondria in cells en route to apoptosis up to 8 h was needed. To achieve this, cell homogenates obtained at different times after the induction of apoptosis were compared with one another with respect to their RCI, i.e. the ratio (oxygen uptake rate after ADP addition)/(oxygen uptake rate before ADP addition) which reflects the ability of mitochondria to produce ATP, and the activity of ADK and GDH, marker enzymes of the intermembrane space and matrix respectively (Table 1). In both cell homogenates and in isolated mitochondria (not shown) the RCI was found to decrease significantly with time of apoptosis, as judged by anova with Bonferroni post hoc test (see legend for Table 1), with a pronounced increase in uncoupling in the 3–8-h time range. However, the mitochondria remained essentially intact as shown by the fact that release of GDH was less than 10% of the total activity measured in the presence of the detergent Triton X-100. As expected, ADK activity remained constant in cell homogenates. In a parallel experiments, cell homogenates were supplemented with β-hydroxybutyrate (5 mm) as an energy source (see Atlante et al. 2005a) and the generation of membrane potential (Δψ) was checked as a function of time after induction of apoptosis by using safranine O as a photometric probe (Table 1). A decrease in both the rate and extent of formation of membrane potential was found, occurring mainly 3 h after induction of apoptosis.
Table 1. Mitochondrial coupling and intactness in homogenate of cells undergoing apoptosis
Δψ generation rate (ΔA520/min)
Δψ extent (ΔA520)
+ Triton X-100
CGCs were switched from high K+ (S-K25) to low K+ (S-K5) serum-free culture medium for the indicated times, subsequently collected, and suspended in PBS. Cell homogenate (0.1 mg cell protein) was prepared and suspended at 25°C in 1.5 mL PBS.
O2 uptake. The RCI, i.e. (oxygen uptake rate after ADP addition)/(oxygen uptake rate before ADP addition), which reflects the ability of mitochondria to produce ATP, was calculated from oxygen consumption rates obtained after the following additions: β-hydroxybutyrate (5 mm) and ADP (1 mm). Values are mean ± SD from six experiments, carried out by using different cell preparations. *p < 0.05, **p < 0.01, ***p < 0.001 for S-K5 versus S-K25 at the same time; †p < 0.05 versus S-K5 at the preceding time.
Detection of ADK and GDH. Enzymatic activity of ADK and GDH (the latter in the presence or absence of 0.2% Triton X-100) is expressed as nmol NADP+ reduced and NADH oxidized/min/mg cell protein respectively. Mean ± SD values of 10 enzymatic activities obtained from six experiments, carried out by using different cell preparations, are reported. **p < 0.01 for S-K5 versus S-K25 at the same time; †p < 0.05, ††p < 0.01 for S-K5 versus the preceding time in the 3–8-h range, in the absence of Triton X-100.
Δψ generation. Cell homogenates (0.1 mg protein) were suspended at 25°C in 1.5 mL PBS containing safranine O (1 µm). When the absorbance trace remained constant the β-hydroxybutyrate (5 mm) addition was made. Safranine O response was monitored. Both the Δψ generation rate and extent, expressed as ΔA520/min/mg cell protein and ΔA520 [(A520 value before β-hydroxybutyrate addition) – (A520 value after Δψ generation due to β-hydroxybutyrate addition)], are reported. Values are mean ± SD from six experiments, carried out using different cell preparations. *p < 0.05, **p < 0.01, ***p < 0.001 for S-K5 versus S-K25 at the same time; †p < 0.05, ††p < 0.01 for S-K5 versus the preceding time in the 1–8-h range. Statistical analysis was by anova and Bonferroni test.
Having established that mitochondria still remained intact and capable of driving generation of Δψ and ATP synthesis up to 8 h of apoptosis, we investigated whether and how the ADP/ATP exchange mediated by the ANT might change in apoptosis in this time range.
In a typical experiment, homogenates from either S-K25 cells or S-K5 cells, treated with AP5A to inhibit ADK, which itself can provide ATP from externally added ADP (Lienhard and Secemski 1973), were incubated in the presence of the ATP-detecting system (Fig. 1b). The ATP concentration outside the mitochondria was negligible as no increase in the absorbance measured at 334 nm was found in the presence of glucose, HK, G6PDH and NADP+. As a result of addition of ADP (0.04 mm) an increase in the NADPH absorbance was found, indicating the appearance of ATP in the extramitochondrial phase (Fig. 1a). The explanation for this is as follows: ADP enters mitochondria in exchange for endogenous ATP; inside the matrix ATP is synthesized by oxidative phosphorylation, and the newly synthesized ATP exits the mitochondria in exchange with further ADP via the ANT (Fig. 1b).
The rate of NADPH formation for S-K25 cells was about 10 nmol NADP+ reduced/min/mg cell protein, remaining constant up to 8 h. In S-K5 cells this rate was lower, decreasing to 5.5 and 1.6 nmol NADP+ reduced/min/mg cell protein 30 min and 8 h after induction of apoptosis respectively. No significant NADPH formation was found in the presence of OLIGO (10 µm), an inhibitor of ATP synthase, showing that no ATP can be synthesized via substrate-level phosphorylation. Besides OLIGO, ATR, an inhibitor of ANT-mediated transport, decreased the rate of ATP efflux (Fig. 1a, inset). As expected in the light of the Ki value (about 3 µm) (see Fig. 1c and related text) and of the ADP and ATR concentrations, partial inhibition was found. To gain some insight into the residual ability of the mitochondria to drive ATP synthesis by generating a Δψ and to ascertain whether this could limit the rate of ATP production, we energized cell homogenates with β-hydroxybutyrate (Fig. 1a, inset) as before (see Table 1). The rates of ATP production were not significantly increased but, on the other hand, addition of a cocktail of the electron flow inhibitors rotenone, antimycin and mixothiazole (RAM) and cyanide (CN–) prevented ATP production totally (Fig. 1a, inset). Because the ANT activity measured in isolated mitochondria could depend on a number of events, including electron flow across the mitochondrial membrane, electrochemical proton gradient generation, ATP synthase and the adenine nucleotide content of the mitochondria, we investigated the rate-limiting step of the process leading to ATP efflux from the mitochondria. That the rate of NADPH formation mirrored the rate of ADP/ATP exchange was verified by applying the control flux coefficient (control strength) criterion (Wanders et al. 1984; Passarella et al. 2003) for the Dixon plot data when either ATR or OLIGO was used as inhibitor at 0.04 mm ADP. The coincidence of the intercepts on the y axis of the lines fitting the points obtained in the presence of ATR, but not OLIGO, reported in Fig. 1(c), in which only S-K25 cells were investigated, showed that the rate of absorbance increase was that of the ADP/ATP exchange. It was also checked that ATP production in homogenates of cells undergoing apoptosis depended on the ANT rather than on ADK activity. This was done by plotting data for the ANT rate in the absence or presence of ATR at different times of apoptosis as 1/i versus 1/[ATR], where i = 1 − (vI/v0), vI and vo being the rate of ADP/ATP exchange in the presence and absence of ATR respectively (Fig. 1c, inset i). The ordinate intercepts were equal to 1, indicating that no ATP could be produced in a manner insensitive to the ANT inhibitor, for instance via ADK.
The same data were re-evaluated in terms of the control coefficient for ANT as a function of time after apoptosis induction. In this case, the regulative role of the ANT in ATP production was estimated by titration with ATR in the presence of a non-rate-limiting ADP-regenerating system, i.e. glucose and HK (Wanders et al. 1984). In this case, as in (Salter 1996), the control coefficient values were very close to unity (Fig. 1c, inset ii).
where J is the oxygen consumption rate, (dJ/d[I]) is the initial slope of the titration curve and [I]max is the inhibitor concentration (20 µm) that inhibits the translocator completely. The ANT titration curves for S-K25 are shown in Fig. 1(c, inset iii).
As ANT control coefficient values were equal to 0.92 ± 0.05 at all times both in controls and in cells undergoing apoptosis (not shown), under our conditions the ANT was controlling the rate of oxidative phosphorylation up to 8 h of apoptosis and consequently ATP appearance in the extramitochondrial phase.
It was also confirmed that the coupled enzymatic system used to detect ATP outside mitochondria is not rate limiting per se, by showing that addition of 5 µm ATP resulted in increase in the rate of absorbance, whereas no rate increase occurred following the addition of both substrate and enzyme components of the ATP detecting system.
In parallel experiments reported in Fig. 2, mPTP opening in S-K25 cells was measured by monitoring both mitochondrial swelling, as a change in absorbance at 546 nm of CGC homogenates (Bernardi et al. 1999; Xia et al. 2002), and the release of ADK, a marker enzyme of the mitochondrial intermembrane space, into the supernatant (not shown) (Single et al. 1998; Atlante et al. 1999). It should be noted that the decrease in light absorbance as well as ADK release is a spontaneous process due to apoptosis and is not calcium dependent (see Kristal and Brown 1999). Moreover, we confirmed that, as in Crouser et al. (2003), swelling and ADK release are strictly correlated with each other (not shown). No mPTP opening occurred at 30 min after apoptosis, whereas it was found at 8 h after induction of apoptosis, but not in S-K25 cells at the same time. The mitochondrial swelling was inhibited if cyclosporin A (CsA), an inhibitor of the mPTP (Bernardi et al. 1999; Basso et al. 2005), was present in the culture. In another experiment, cyt c release, measured by western blotting as in Bobba et al. (1999), was found in S-K5, but not in S-K25, cells 30 min and 8 h after induction of apoptosis either in the absence or presence of CsA (Fig. 2a, inset).
In order to gain further insight into the mechanism by which ADK and cyt c are released, we compared ADK and cyt c release with respect to their sensitivity to CsA (Fig. 2b). No ADK release occurred up to 3 h after induction of apoptosis either in the absence or presence of CsA. Later ADK was released in a manner prevented by CsA (about 90% prevention at 8 h) (p < 0.001). In contrast, cyt c release, detected polarographically (Atlante et al. 2001), was completely insensitive to CsA at the time points investigated. As further confirmation that ADK release is not dependent on mitochondrial damage, but on mPTP opening, Sanglifehrin A, a compound that is not structurally related to CsA but strongly inhibits cyclophilin D and mPTP opening (Clarke et al. 2002), was also used with results undistinguishable from those found with CsA (not shown). Because inhibition by CsA is considered as a marker of the mPTP (Bernardi et al. 1994, 1998; Nicolli et al. 1996; Crompton 1999; He and Lemasters 2002), and as it works as an mPTP inhibitor in the course of apoptosis, this shows that cyt c release does not involve mPTP opening, i.e. mPTP formation occurs downstream of cyt c release.
ANT transport function and mPTP opening in CGCs en route to apoptosis and the role of the cell antioxidant and proteolytic systems
In the experiments reported in Fig. 3, both ADP/ATP exchange (Fig. 3a) and mPTP opening, measured as ADK release into the extramitochondrial phase (Fig. 3b), were investigated as a function of time (0–8 h) of apoptosis. In early apoptosis, the rate of ADP/ATP exchange, at 0.25 mm ADP, was found to decrease with respect to the control, in which it remained constant at all times, with inhibition increasing with progression of apoptosis (about 60% inhibition found at 3 h, p < 0.001); in the same time range, no or negligible mPTP opening was found either in S-K25 or S-K5 cell homogenates.
In late apoptosis, a further decrease in ADP/ATP exchange was found, with 10% residual ANT-dependent transport at 8 h. mPTP opening occurred over the 3–8 h time range in S-K5 cells. Approximately 50% of the maximum opening was found at 5 h. These findings show that during apoptosis both a decrease in ANT transport function and mPTP opening occur.
To obtain some insight into the mechanism by which these processes occur and to determine whether the two processes are linked to one another, S-K5 cells were maintained in the absence or presence of a variety of compounds designed to inhibit certain activities that participate in the processes leading to apoptosis. These included the broad-range caspase inhibitor z-VAD (100 µm) and MG132 (5 µm), a proteasome inhibitor; in addition, captopril (CP, 5 mm) and NH2-triazole (NH2TZ, 10 mm) were used to inhibit superoxide dismutase (SOD) and catalase respectively (for references see Atlante et al. 2003a). Furthermore, the role of mPTP in apoptosis was investigated by incubating cells in the presence of CsA (1 µm). Cells in which antioxidant enzymes were inhibited by CP plus NH2TZ are referred to as No-AOS cells, those with proteasome activity blocked by MG132 are referred to as No-PROTEASOME cells, those in which caspase activity was prevented by z-VAD are referred to as No-CASPASE cells, those treated with the inhibitor pair MG132 and z-VAD are referred to as No-PROTEASOME/CASPASE cells, cells incubated with SOD present are referred to as No-ROS cells, and cells incubated with CsA are referred to as No-mPTP cells. No-AOS/PROTEASOME and No-AOS/CASPASE cells refer to cells undergoing apoptosis in the presence of the respective inhibitors.
In parallel experiments, similar to those shown in Fig. 1(c), we showed that the measured change in absorbance mirrored the rate of ANT-mediated transport. The control coefficients were higher than 0.9, indicating that en route to apoptosis, independent of other effects on mitochondrial bioenergetics, we were monitoring the rate of ANT-dependent transport.
The rates of ATP efflux and mPTP opening were measured at different times of apoptosis. Under No-ROS conditions, almost complete prevention of ANT impairment was found in early apoptosis, showing that the decrease in ANT transport efficiency depends on ROS production. Consistent with this, in a separate experiment not reported here, we confirmed that ROS production increased in this time range (Atlante et al. 2003b). In late apoptosis, in spite of the negligible change in ROS production (Atlante et al. 2003b), progressive impairment in ANT transport was found but to a lesser extent than with S-K5 cells (p < 0.01).
No mPTP opening was found in early apoptosis, showing that ROS cannot themselves cause the mPT, which occurred only in the 3–8-h time range, and to a lesser extent than with S-K5 cells.
Interestingly, in No-CASPASE cells, no opening of the mPTP occurred, either in early or late apoptosis, showing conclusively that mPTP opening requires caspase action. Furthermore, the time-dependent decrease in ANT activity in early apoptosis paralleled that found in S-K5 cells, but in late apoptosis no significant further reduction was found, showing that only the late ANT impairment is caspase dependent.
As expected in the light of the above results, in No-ROS/CASPASE cells, the ANT activity remained largely unaffected with respect to S-K25 cells, and no mPT occurred at all. This, together with the above results, shows that ANT is damaged in two phases as far as ADP/ATP exchange is concerned, with initial damage due to ROS and further additional damage owing to caspase. Conversely, even if damaged by ROS, the ANT can still translocate adenine nucleotides, but cannot participate in mPT formation. Later, as a result of caspase action, the ANT begins to participate in mPTP opening.
In agreement with this conclusion, in No-PROTEASOME cells, in which the cell AOS remains active for longer times and consequently the ROS level is reduced (Atlante et al. 2003b), both the ANT impairment and the mPTP opening were largely prevented (p < 0.001). In No-AOS cells, in which the ROS level increased (Atlante et al. 1998a), ANT impairment increased up to 5 h, but no change in mPTP opening occurred.
In early apoptosis in No-AOS/CASPASE cells, the pattern of initial ANT impairment was found to be similar to that in No-AOS cells, but in late apoptosis the impairment of ANT remained constant albeit at a lower level compared with No-CASPASE cells. This is a further demonstration that in this time range the role of caspase is crucial in determining ANT impairment and mPTP opening en route to apoptosis. In No-AOS/PROTEASOME cells, ANT impairment was greater than that in S-K5 cells, whereas no change in mPTP opening was found, further showing that modification of the ROS level has no effect on mPTP opening. In No-mPTP cells we found no pore opening, without any effect on ANT impairment.
Loss of ANT transport function and mPTP opening are inversely correlated in late, but not in early apoptosis
Another analysis was developed to determine whether and at what stage of the progress towards apoptosis the loss in ANT transport function and mPTP opening depend on one another. Thus, comparison was made between the data mirroring the loss of transport efficiency and the data mirroring pore opening as a function of time. In both cases, the data were expressed as a percentage, the former as (100 −% residual) transport activity and the latter as a percentage of maximum opening. Such an analysis shows the phases of the two processes, which paralleled one another (Fig. 4). The essential features are that loss of ANT transport function and onset of mPTP are independent during early apoptosis, but the situation changes over the period 3–5 h, when mPT sets in and the rate of loss of ANT activity begins to decrease. After that they proceed in parallel and conform to the same mathematical model over the 5–8-h time range when the two processes become essentially linear. That is, in the 5–8-h phase of apoptosis, the progressive decrease in the ANT-dependent transport function occurs simultaneously with mPTP opening.
The same analysis was carried out for the other data in Fig. 3 with essentially the same conclusions: in No-ROS cells in the 5–8-h time range ANT impairment and mPTP opening proceeded similarly. In No-CASPASE cells, the analysis confirmed that the two processes are independent, i.e. ANT damage occurs in the absence of significant mPTP opening.
Thus, the above analysis confirms that, 5–8 h after induction of apoptosis, ANT impairment and mPTP opening proceed simultaneously and conform to the same mathematical model, with the exception of the No-AOS/PROTEASOME cells for which the reciprocal relationship holds only after 6 h of apoptosis.
Mechanism of impairment of ANT-dependent transport function en route to apoptosis
Having established that ANT impairment derives from ROS (in early apoptosis) and from caspase action (in late apoptosis), to elucidate how such an impairment occurs, we investigated the dependence of the rate of ATP efflux on increasing ADP concentrations in S-K25 and S-K5 at different times after induction of apoptosis (Fig. 5). As in earlier experiments, application of control strength criteria showed that the rate of NADP+ reduction mirrored the ADP/ATP exchange (not shown).
Hyperbolic dependence of the exchange rate on ADP concentration was found and the results were analysed using the Michaelis–Menten equation. Loss of ANT-mediated transport activity occurred progressively en route to apoptosis by two different mechanisms: up to 3 h of apoptosis, the Km for ADP, measured as the ADP concentration that gave the half-maximum rate of transport, was found to increase with increasing apoptosis times from 30 ± 1 to 145 ± 12 µm. No change in Vmax was found, characteristic of competitive-like inhibition. This clearly shows that inhibition was due to a decrease in the affinity of the ANT for ADP. As expected in the light of the above results, impairment of ANT transport activity was prevented in No-ROS cells, whereas it remained unchanged in No-CASPASE cells (not shown).
At prolonged times of apoptosis, a very large decrease in Vmax for transport was observed, but without further substantial change in the Km value (125 ± 15 µm), resembling the onset of a non-competitive type of inhibition and actually reflecting a marked reduction in transport capacity. In this case, in No-CASPASE cells, but not in No-ROS cells, prevention of ANT inhibition was found (not shown).
Impairment of ANT function does not depend on proteolysis of the protein
To determine whether the decrease in ANT transport function could be a result of a reduction in the number of carrier molecules, perhaps resulting from proteolysis following activation of caspases, both the amount and the size of the ANT were monitored by western blotting en route to apoptosis in isolated mitochondria. No change in either the molecular weight or the amount of the monomeric (32 kDa) or dimeric (64 kDa) isoforms of ANT was found in S-K25 or S-K5 cells taken at 1, 5 and 8 h (Fig. 6). None of the amino acid sequences of the three ANT isoforms retrieved from the SwissProt Data Bank was found to contain a consensus sequence for proteolysis by caspase. In this regard, it is worth noting that, among other putative proteins involved in the mPTP, only Bcl-2 protein and HK contain the caspase consensus cleavage site DXXD (where X represents an unspecified amino acid). In particular, Bcl-2 has been found to be a specific target for caspase 3 and is cleaved at Asp-34 in the course of apoptosis (Grandgirard et al. 1998).
The mPTP is not required for apoptosis to occur in S-K5 cells
The results described above, together with those previously published (Atlante et al. 2003a,b), show that mPTP opening takes place only in the late phase of apoptosis, when maximum cyt c release has already occurred (Bobba et al. 1999) (see Fig. 2).
This is at variance with previous reports based on other experimental systems, in which occurrence of the mPT was suggested as being responsible for the release of mitochondrial proteins needed for apoptosis to occur, including cyt c (Bradham et al. 1998; Wadia et al. 1998; Heiskanen et al. 1999; Borutaite et al. 2003). This poses the question of whether the mPTP is needed at all for apoptosis in CGC cells. Confirmation that the mPTP is, in fact, not needed for apoptosis to occur is provided by the results of an experiment in which we investigated the effect of CsA on the time course of apoptosis monitored as described in Atlante et al. (2003a) (Fig. 7). Cell viability was measured as a function of time after apoptotic stimulus in the absence or presence of CsA. In the same samples, we checked the ability of actinomycin D (Act D, 1 µg/mL) to prevent cell death (D'Mello et al. 1993). We found that the decrease in viability occurred as reported previously (D'Mello et al. 1993; Atlante et al. 2003a) and that CsA had no effect on the process (Fig. 7). In the presence of Act D, complete prevention of death was found (p < 0.01), as expected if cell death occurs via apoptosis (D'Mello et al. 1993), To confirm this, DNA laddering, a specific hallmark of apoptosis, was checked at 24 h after the induction of apoptosis. DNA laddering was found only in S-K5 cells, both in the absence and presence of CsA. It was completely absent in the same cell samples in the presence of the transcriptional inhibitor Act D as well as in S-K25 cells (Fig. 7, inset). As expected, a decrease in the rate of death was found for No-ROS and No-CASPASE cells (p < 0.05).
In this paper, we have addressed a major topic in the bioenergetics of apoptosis, namely the role of ANT which is involved in ATP export from mitochondria in exchange for cytosolic ADP and/or in mPTP formation. We approached this issue by using CGC homogenates as a model system. This model is unique as mitochondria are intact and coupled (Table 1) and thus provide a ‘physiological’ environment for the investigation of ANT in situ. The function of the ANT en route to apoptosis that emerges from the results reported here is as follows (Scheme 1a′ and b′). In early apoptosis (Scheme 1a′) a decrease in the transport efficiency of the ANT occurs, caused by a ROS-mediated post-translational modification as shown by its prevention by SOD (Fig. 3) and resulting in reduced affinity for ADP (Fig. 5). The application of experimental conditions described by Atlante et al. (2003b), in which either antioxidant or proteolytic systems are impaired with modulation of ROS production, helps to substantiate the above conclusion in that the activity of the AOS, which is higher in the absence of digestion by proteasomes, results in reduced impairment of the ANT; in contrast, a deficit in the AOS system and consequent increase in ROS production leads to further ANT damage. It should be noted that impairment of the ANT has been suggested to be due to ROS-dependent cardiolipin oxidation (Nakagawa 2004). However, in rat liver mitochondria we have already found that ROS can impair the ANT carrier as investigated here, in a manner prevented by the thiol reagent mersalyl. This raises the possibility that thiol(s) in ANT molecules may be the target of ROS (see Atlante et al. 1989; Le Bras et al. 2005). This might suggest a chemical modification in the carrier molecule as responsible for the change in ANT activity; unfortunately a similar experiment cannot be carried out in CGCs undergoing apoptosis.
During early apoptosis in S-K5 cells, mPT does not occur, but cyt c is already released (Fig. 2; Bobba et al. 1999, 2004). This confirms previous work (Atlante et al. 2000), in which the time course of cyt c release and pore opening as monitored by release of ADK were compared and the former was shown to occur first (Bobba et al. 1999). Such a conclusion, which was confirmed by (Wigdal et al. 2002), who showed that the Δψ is intact, is in contrast to a variety of reports in which the mPTP has been proposed to determine cyt c release (Bradham et al. 1998; Wadia et al. 1998; Heiskanen et al. 1999; Le Bras et al. 2005). This discrepancy might explained by the different experimental models used; in particular, we caution against the use of complex systems in which harmful manipulations are used that might cause unpredictable cellular damage.
The temporal sequence of cyt c release en route to apoptosis is consistent with the results of Atlante et al. 2003a, who showed that at least in early apoptosis the released cyt c can contribute to mitochondrial energization, probably resulting in increased ATP levels (Atlante et al. 2005b).
In late apoptosis (Scheme 1b′), together with CsA-sensitive mPTP opening (Figs 2 and 3), a further progressive decrease in ANT transport function, i.e. of ADP/ATP exchange, which can be prevented (Figs 3 and 4) by the caspase inhibitor z-VAD, occurs. However ANT itself was proved not to be the caspase target because the ANT protein did not change in either amount or molecular weight during apoptosis as measured by western blotting (Fig. 6).
We consistently found a decreased Vmax which was due to a change in the transport efficiency (Kcat) rather than to a decrease in the number of carrier molecules (Fig. 5). It might be argued that the changes in Km and Vmax values could depend on alteration of one of the processes leading to ATP production via oxidative phosphorylation, including the electrochemical proton gradient and the ATP synthase. Such a possibility is ruled out by the evidence that en route to apoptosis the rate-limiting step is the ATR-sensitive step, i.e. transport across the mitochondrial membrane. However, to further substantiate that the decrease in ANT activity does not depend on the failure of mitochondria to generate Δψ, which is used as driving force for ATP efflux, we confirmed that addition to S-K5 CGC homogenates of β-hydroxybutyrate, a respiratory substrate that enters mitochondria via diffusion with an increase in Δψ (Table 1), is not accompanied by recovery of ANT transport activity (not shown). Moreover, we have found in preliminary experiments that the activity of ATP synthase, assayed as described by Atlante et al. (1989), did not decrease up to 3 h of apoptosis (not shown) and that mitochondrial oxidation can still result in Δψ generation (Table 1).
On the other hand, increasing mPTP opening occurs in a strictly caspase-dependent manner. Nonetheless, as opening in mitochondria previously subjected to ROS is faster (Figs 2–4) and because no further mPTP opening occurs when the ROS level increases, as in No-AOS cells, we propose that a threshold level of ROS is required to enforce caspase action, even if ROS by themselves cannot cause mPTP opening.
In the light of the inverse overlapping sensitivity of ANT and mPTP both to the caspase inhibitor and to other inhibitors used to block cell systems participating in apoptosis (Fig. 4), and because no change occurs in either the amount or the size of ANT (Fig. 6), we conclude that in CGCs undergoing apoptosis a functional alteration of ANT takes place resulting in ANT becoming a functional component of the pore responsible for the mPT. Such a conclusion is supported by the inverse correlation between the kinetics of loss of transport efficiency by the ANT and the increase in mPTP opening, as calculated both in S-K5 cells and in cells treated with inhibitors at 5–8 h after induction of apoptosis (Fig. 4).
We suggest that a possible target of caspase-dependent proteolysis is Bcl-2 and/or HK, the only mPTP proteins endowed with a caspase consensus sequence (Scheme 1b′). In this regard, it has been shown that Bcl-2 protein is cleaved at Asp-34 by caspases during apoptosis (Kirsch et al. 1999). The possibility that in mPTP formation caspase action on Bcl-2/HK and the consequent effect on carrier function depend on Bax insertion into the mitochondria (Belzacq et al. 2003; Sharpe et al. 2004; Verrier et al. 2004) cannot be ruled out. Interestingly, ANT residues that can bind Bax have already been identified (Halestrap et al. 2002). Unfortunately investigation of the possible change of Bcl-2 and HK structure cannot be carried out under our experimental conditions.
It seems clear that caspases, but not the CsA-sensitive mPTP, are responsible for the occurrence of apoptosis. This was confirmed by the finding that the dependence of the viability of S-K5 cells on time up to 24 h was sensitive to Act D, but not to CsA, i.e. cells die via apoptosis in the absence of mPT (Fig. 7). Here, in contrast to Kokoszka et al. (2004), we confirm that mPT occurs as a result of modification of the ANT and that no mPT can take place without this alteration in the ANT that leads to its becoming a pore component. We show that in CGC apoptosis, mPTP is not needed for apoptosis to occur or for the release of cyt c, consistent with previous findings (Baines et al. 2005; Nakagawa et al. 2005).
Our findings are partially consistent with two recent papers by Kroemer's group (Belzacq et al. 2002, 2003), in which either the channel or the translocase activity of the purified ANT was investigated in artificial systems. The proposal of the ‘ménage a trois’ in Kroemer terms, in which the ANT interacts with Bax or Bcl-2, is somewhat clarified here as we show that first the caspase-dependent modification of a mPTP protein must occur, perhaps Bcl-2 which is the only putative target of caspase activity (along with HK). On the other hand, we show that mPTP opening requires ANT damage with failure in its transport properties. Thus, in contrast to Kokoszka et al. (2004), we conclude that the ANT is an essential component of the CsA-sensitive mPTP.
Although our results show that the mPTP is not required for the early stages of apoptosis, we cannot rule out that it may be involved in the final stages (up to 24 h), when leaky uncoupled mitochondria finally release their own contents.
The authors thank Professor Shawn Doonan for his critical reading of the manuscript, Professor Andrew Halestrap of the University of Bristol, Bristol, UK for his kind gift of polyclonal anti-ANT antibodies, Dr Fritz Rothe of the University of Magdeburg, Magdeburg, Germany for his kind gift of anti-GDH antibodies, and Mr Vito Petragallo for skilful technical assistance with tissue culture.
This work was partially financed by Fondo per gli Investimenti della Ricerca di Base ‘Malattie Neurodegenerative come conseguenza di un alterato processamento di proteine neuronali. Modelli animali e colture cellulari in vitro’ and Fondo Intergravito Speciale per la Ricerca ‘Stress ossidativo e bioenergetica mitocondriale nella patogenesi delle malattie neurodegenerative’ (to AA), by Ministero dell’ Istruzione e della Ricera – Contributi straordinari di ricerca/aree obiettivo 1 (to EM) and by Fondi di Ricerca di Ateneo del Molise and FIRB RBNE03B8KK_003 (to SP).