Intracellular Adenosine Triphosphate (ATP) Concentration: A Switch in the Decision Between Apoptosis and Necrosis

Jurkat cells were grown in RPMI-1640 medium supplemented with 10% FCS. Before the experiments, cells were washed and resuspended in serum-free medium without glucose containing 2 mM pyruvate. After adaptation to this medium, cells were exposed to 0 or 2.5 μM oligomycin for 45 min, and then challenged with STS or an agonist mAb (clone CH-11) against CD95.

Cultures were stained with a mixture of the membrane permeant dye H-33342 (500 ng/ml) and the membrane impermeant dye SYTOX (500 nM) (Molecular Probes, Eugene, OR). Necrotic (damaged plasma membrane; noncondensed nuclei) and apoptotic cells (intact plasma membrane; characteristically condensed or fragmented nuclei) were scored. Alternatively, cells were stained with a combination of 0.5 μM calcein-AM, 1 μM ethidium homodimer (EH-1), and 500 ng/ml H-33342. Stained cells (immobilized on poly-l-lysine–coated cover slips at the bottom of a 35-mm chamber) were examined with a Leica TCS-4D confocal microscope simultaneously for membrane integrity (calcein retention and EH-1 exclusion) and morphological changes of the nucleus and the cell (polarization contrast and H-33342 stain). Results obtained with both methods were similar. Digital images of SYTO-13/EH-1–stained cells were acquired on a Leica DM-IRB microscope (40× lens) coupled to an imaging system (Imaging Corporation, St. Catherins, Canada). Apoptotic nuclei appeared hyperfluorescent and condensed. In necrotic nuclei, the fainter green SYTO-13 stain was displaced by the red EH-1 stain (3). Surface phosphatidylserine expression was analyzed by annexin V–staining and FACS^® analysis as described (13).

ATP was determined luminometrically (14). Addition of glucose (⩽10 mM) to cells in the absence of oligomycin resulted in increases of the ATP concentration ⩽20% of the initial value (data not shown) and did not modify apoptosis. With ATP concentrations between 30% and 50% of the resting control level intermediate forms of cell death (e.g., partially condensed nuclei in ruptured cells), and a slight delay in the kinetics of necrosis were occasionally observed after stimulation with either αCD95 or STS. Addition of oligomycin decreased the intracellular ATP concentration (as compared with initial control levels: 60% within 10 min, 75% within 20 min, and ⩾90% within 30 min) irrespective of the presence of other medium supplements or stimuli. Conversely, after treatment with glucose, ATP rose to ⩾70% within 5 min and reached a new steady state after 15 min.

Field-inverted gel electrophoresis (FIGE) was performed as described previously (3) after embedding 5 × 10⁵ cells into 40-μl agarose blocks. Yeast chromosomes and λDNA concatemers were used as molecular weight markers. Conventional agarose gel electrophoresis (CAGE) was performed on 1.0% agarose gels (15). Gels were stained with SYBR green (Molecular Probes, Eugene, OR). Alternatively, histone-containing DNA fragments were determined quantitatively by ELISA (Boehringer Mannheim, Germany) (16). In this assay we used a lysate obtained from ∼500 cells.

Western blots were performed using an anti-human lamin B antibody (clone 101 B7, Oncogene Research Products, Cambridge, MA). All lanes of 10% polyacrylamide gels were loaded with 10 μg protein, and homogeneous transfer to nitrocellulose membranes was controlled by Ponceau red staining. For immunocytochemistry, cells were fixed with 80% MeOH. Incubation with anti-lamin B antibody (101 B7) was followed by staining with a secondary BODIPY-coupled anti-mouse antibody and 1 μg/ml H-33342. Cells were visualized by threechannel confocal microscopy with a 100× UV lens (polarization, lamin, chromatin).

Initially, we examined the characteristics of cell death caused by the two inducers under conditions of intracellular ATP depletion. Jurkat cells were maintained in a glucose-free RPMI-1640 medium supplemented with 2 mM pyruvate to allow only mitochondrial ATP production. Exposure to either STS or αCD95-induced apoptosis. Intracellular ATP levels were maintained during the first 90 min, but eventually declined after the appearance of the first apoptotic cells (Fig. 1, a and b). When cells were preemptied of ATP using oligomycin, a blocker of mitochondrial ATP synthesis, death occurred by necrosis (Fig. 1, a and c).

We asked ourselves whether maintenance of a certain ATP level was needed to prevent an early breakdown of the plasma membrane. In this case, the development of apoptosis, which would not necessarily require ATP, could have been precluded by a premature demise with apparent necrotic features. However, it became clear that necrosis elicited by either of the two inducers in ATP-depleted cells occurred with a similar or rather longer time course than apoptosis. In this case, cell demise had entirely different morphological and biochemical features (Fig. 1,d ). Nuclei of cells pretreated with oligomycin and then exposed to STS/ αCD95 had normal size and showed no apoptotic chromatin condensation, while the cell membrane was clearly permeable to fluorescent dyes. The typical apoptotic DNA cleavage into 700/300 kbp, 50 kbp, and oligonucleosomal fragments did not occur in Jurkat cells preemptied of ATP (Fig. 2). In addition, exposure of phosphatydylserines on the outer surface of the plasma membrane, a process leading to recognition of apoptotic cells (13), did not occur in ATP-depleted cells, which later underwent necrosis (annexin binding cells after 210 min: STS, 85 ± 9%; oligomycin, 19 ± 4%; STS + oligomycin, 13 ± 7%; control, 11 ± 5%). This suggests that ATP was required for the execution of several distinct processes in the apoptotic program.

To estimate the ATP concentration sufficient to allow the progression of apoptosis, we clamped the ATP concentration of oligomycin-treated cells at different levels by supplementing the medium with glucose (Fig. 3,a). Cells were then treated with the inducers. In STS-treated cells, an ATP depletion >50% (i.e., a residual ATP concentration <0.8 nmol × mg prot⁻¹ 45 min after the addition of the stimulus, when the apoptotic features were not yet apparent) was sufficient to change the mode of demise from apoptotic to necrotic. Higher ATP concentrations instead favored the ordered continuation of the apoptotic program. With αCD95, an ATP loss ⩾70% (i.e., a residual ATP concentration of ⩽0.5 nmol × mg prot⁻¹) was invariably followed by necrosis (Fig. 3, b and c). Similar results were obtained in experiments with fructose replenishment. In cells pretreated with oligomycin and then challenged with αCD95, a fructose concentration (40 mM) that limited the ATP loss (i.e., residual ATP above 50%) changed the mode of cell death from necrosis to apoptosis. Addition of similar metabolic substrates, e.g., glycerol (50 mM), dihydroxyacetone (50 mM), or lower fructose concentrations (10 mM), which were not able to restore intracellular ATP levels, did not modify the shape of cell death. Thus, it appears that the availability of the ATP generated by glycolysis is sufficient to allow the execution of cell death by apoptosis. In the absence of glucose and oligomycin, mitochondrial ATP generation stimulated by the addition of pyruvate would naturally replenish also the extramitochondrial pool.

To determine at which phase of the death program ATP was required, we added oligomycin at different times after triggering cell death with STS or αCD95. When ATP depletion was inflicted up to 90 min after the challenge with either of the two stimuli, cell death still shifted from apoptosis to necrosis (Fig. 3,d ). In a complementary experimental setup, we found that glucose addition to replenish ATP in cells treated with either STS or αCD95 plus oligomycin directed the death process towards apoptosis up to 2 h after the challenge (Fig. 3,d ). Thus, it seemed that ATP was required immediately before the characteristic nuclear changes, perhaps to ensure that nuclear collapse, the ordered packaging of chromatin, and DNA degradation occur before cell lysis. Accordingly, DNA fragmentation was strongly reduced in necrotic cells (i.e., challenged with STS or αCD95 in the presence of oligomycin) as compared with those undergoing apoptosis, although the total percentage of cell death was approximately the same (Fig. 4, a and b). Differences in the fate of the nucleus became evident when we examined lamin degradation, a process pivotal for the nuclear collapse that is effected by the activation of certain caspases during apoptosis (17, 18). Cleavage of lamin B was significantly reduced under conditions of ATP depletion leading to necrosis (Fig. 4, c and d ), which suggests that activation of the proteolytic system responsible for lamin cleavage is specific for apoptosis.

Because activation of caspases is also an integral component of the CD95 signaling system that triggers apoptosis (19, 20) we used caspase inhibitors to test whether the same signaling system would be active in necrosis and apoptosis. In this case, caspase inhibitors prevented αCD95-induced cell death regardless whether apoptotic or necrotic (Fig 4, a and b), in agreement with other findings that both types of cell death share similar signals and initial mechanisms (21–23).

A cytoplasmic controller (24) seems to direct downstream events resulting in the typical apoptotic features including proteolysis of nuclear elements and DNA degradation. Furthermore, according to a series of elegant experiments with reconstituted cell-free systems, an apoptogenic protease (25) released from mitochondria upon permeability transition (26) can promote the nuclear changes. In cells, mitochondrial permeability transition can be triggered by a variety of stimuli acting on the mitochondria directly, or by a yet unknown upstream controller (25, 26). Thus, either the upstream controller or the execution system(s) may be modulated by the availability of ATP. The latter may be provided by gycolysis or, in tissues with high metabolic demand, primarily by the mitochondria (3, 27). In the model described here, restoration of glycolytic ATP generation was sufficient to allow the ordered execution of apoptosis. Thus, apoptosis and necrosis would be two extremes of a continuum of possible types of cell demise, whose shape and implications for the neighboring tissue would be decided by the availability of ATP, and likely by additional factors in the dying cell and in scavenger cells. This would explain the frequent coexistence of both types of cell death in pathological situations, e.g., cerebral ischemia (28) or inflammatory liver failure (5, 6), where individual cell death within the tissue would be decided by the energy supply.

We are grateful to Ms. H. Naumann and to Mr. T. Schmitz for the excellent technical assistance and to Dr. E. May for stimulating discussions.

This work has been supported by grants from the European Community (EV5V CT94 0508) (ENV4CT96-0169), by the European Science Foundation to A.F. Castoldi.