In the present study, we used mitochondrial DNA-depleted Jurkat subclones (ρ0 cells) to demonstrate that Fas agonistic Ab (CH-11), at the concentrations that evoke apoptotic death of the parental Jurkat cells, induced necrosis mainly through generation of excess reactive oxygen species, lysosomal rupture, and sequential activation of cathepsins B and D, and in minor part through activation of receptor-interacting protein (RIP). In the ρ0 cells treated with CH-11, ATP supplementation converted necrosis into apoptosis by the formation of the apoptosome and subsequent activation of procaspase-3. In these ATP-supplemented ρ0 cells (ATP-ρ0), generation of excess ROS and lysosomal rupture were still seen, yet cathepsins B and D were inactivated and RIP was degraded. The conversion of necrosis to apoptosis, RIP degradation, and cathepsin inactivation in ATP- ρ0 cells were blocked by caspase-3 inhibitors. Activities of cathepsins B and D in the lysate of necrotic ρ0 cells were inhibited by the addition of apoptotic parental Jurkat cell lysate. Thus, apoptosis may supercede necrosis.
Apoptosis and necrosis are two major forms of cell death and are generally not related to one another, neither under in vitro experimental conditions nor in in vivo tissues (1). Apoptosis is known to play a critical role in physiological elimination of cells that are no longer needed during morphogenesis or development (2). It is also linked to pathogeneses of some diseases such as T cell depletion in AIDS and premature death of neurons in Alzheimer’s disease and Parkinson’s disease, in addition to therapeutic effects of anticancer drugs.
In contrast, necrosis is a rather pathological type of tissue destruction, such as cardiac (3) or cerebral injury following ischemia-reperfusion (4), cyclosporin A-induced endothelial damage (5), radiation hazard (6), and certain congenital mitochondrial diseases (7). However, recently, investigators claimed that necrotic cell death was implicated in a certain type of physiological cell elimination (8).
The molecular basis of apoptosis has been well elucidated (9). A large amount of active caspase-8 is generated after the stimulation of death receptors, such as Fas, TNF-α receptor type 1, and TRAIL receptors, leading to alterations in downstream mitochondria function and the subsequent caspase cascade.
Independent of the caspase-signaling cascade, reactive oxygen species (ROS)3 has been reported to be involved in apoptosis based on the findings that apoptosis was prevented by various antioxidants, including N-acetyl-l-cysteine (NAC) (10, 11), manganese III tetrakis, a nonpeptidyl mimic of superoxide dismutase (SOD) (12), catalase (13), and overexpression of copper/zinc SOD (14). We have linked the role of ROS to that of caspases by demonstrating that the treatment of Jurkat cells with NAC or SOD blocked the formation of an apoptosome and the subsequent caspase-3 activation pathway induced by Fas ligand (15).
Regarding necrosis, however, the molecular events underlying cell death are relatively less understood. Although involvement of ROS (16, 17), lysosomal enzymes (5), mitochondrial-derived proteins including endonuclease G, Omi/HtrA2, and apoptosis-inducing factor, and, most recently, receptor-interacting protein (RIP) (18) have been reported, the mechanisms of necrotic cell death induced by each of these factors and their relationship to one another are unclear.
Furthermore, despite the fact that apoptosis and necrosis are distinct from each other, curiously, they are interchangeable under certain conditions. Transition from a Fas-induced apoptotic state into a necrotic state was accomplished by depletion of ATP with an ATPase inhibitor, and reversal of necrosis into apoptosis was achieved by ATP supplementation (19). In Jurkat cells treated with a pan caspase inhibitor (18) or in a caspase-8-deficient subline of Jurkat cells (16), strong Fas stimulation that transduces an apoptotic signal in normal Jurkat cells instead induced necrosis.
In murine L929 fibrosarcoma cells modified to overexpress Fas, stimulation with Fas ligand rapidly led to apoptosis. When this apoptotic pathway was blocked by a pan caspase inhibitor, necrosis was evoked with the same ligand (17). Similarly, redirection of cell death toward necrosis by blocking apoptotic machinery with cyclosporin A in arterial endothelial cells was also reported (5). The molecular mechanisms underlying this switching phenomenon between necrosis and apoptosis are unclear.
In the present study, we established a subline (ρ0 cell) of human Jurkat cells in which mitochondrial DNA (mtDNA) was selectively depleted by long-term exposure to ethidium bromide. Using these ρ0 cells, we showed that Fas ligand clearly induced necrosis without using a pan caspase inhibitor. This necrotic cell death was ascribed to the sequential events of release of excess ROS from mitochondria, rupture of lysosomes, and activation of cathepsins B and D in parallel with RIP activation. Finally, we showed that supplementation of ρ0 cells with ATP restored the formation of the apoptosome and the subsequent activation of caspase-3, which in turn inhibited necrotic cell death by blocking the activities of cathepsins B and D and by degrading RIP. Thus, Fas-triggered RIP activation and the parallel cellular events of ROS release from mitochondria/lysosomal rupture/activation of cathepsins to induce necrosis were found to be superceded by apoptosis.
Materials and Methods
Agonistic anti-human Fas mAb, clone CH-11, and caspase-3-specific inhibitory peptide N-benzyloxycarbonyl-Asp-Glu-Val-Asp-fluoromethyl ketone (Z-DEVD-FMK) were purchased from Medical and Biological Laboratories. NAC, ATP, α1-antichymotrypsin, epoxysuccinyl-l-leucylamido-3-methyl-butane ethyl ester (E-64d), chloroquine, bafilomycin cyclosporin A, and pepstatin A were from Sigma-Aldrich. Bongkrekic acid was from Calbiochem. 3′-O-acetyl-2′,7′-bis(carboxyethyl)-4 or 5-carboxyfluorescein, diacetoxymethyl ester was from DOJINDO.
Jurkat cells (human T cell leukemia) were supplied by American Type Culture Collection and maintained in RPMI 1640 (Life Technologies) with 10% heat-inactivated FBS (Sigma-Aldrich). To deplete mtDNA in Jurkat cells, long-term culture >70 passages was conducted in medium containing 800 ng/ml ethidium bromide (Sigma-Aldrich), as described previously (20, 21). The parental Jurkat cells and a mtDNA-depleted subline were conventionally designated as ρ+ and ρ0 cells, respectively.
Transduction of the manganese superoxide dismutase (MnSOD) gene
MnSOD expression plasmid pRc/CMV-MnSOD was provided by N. Taniguchi (Department of Biochemistry, Osaka University, Osaka, Japan). Briefly, ρ0 cells were transfected with pRc/CMV-MnSOD using cationic lipid DMRIE-C reagent (Life Technologies), following the manufacturer’s instructions. Three days after the transfection, 800 μg/ml G418 sulfate (Calbiochem) was added to the culture medium to eliminate nontransfected cells. Isolated colonies were obtained by the standard limiting dilution method. Transduction of the MnSOD gene was confirmed by the PCR method using the following set of primers: 5′-GCC AGC ACT AGC AGC ATG TTG AGC C-3′ (forward) and 5′-CTT TGG GTT CTC CAC CAC CGT TAG G-3′ (reverse). These primers detect a part of MnSOD cDNA (358-bp DNA fragment). Then the MnSOD transfectants and control clones, which were transfected with pRc/CMV, were designated as ρ0-MnSOD and ρ0-neo, respectively.
Genomic DNA was extracted from 1 × 107 cells by the standard phenol/chloroform method, digested with XbaI (Takara), and loaded onto a 0.8% agarose gel for electrophoresis. After hybridization to GeneScreen membrane (NEN-DuPont), blotting was performed with a mtDNA probe radiolabeled with [α-32P]dCTP (NEN-DuPont) using the BcaBEST labeling kit (Takara). The mtDNA probe mentioned above, which was a full-length cDNA of human cytochrome oxidase subunit (COX) 3, had been amplified beforehand by the PCR method from genomic DNA of normal human monocytes using the following set of primers: 5′-ATG ACC CAC CAA TCA CAT GCC TAT-3′ (forward) and 5′-AGA CCC TCA TCA ATA GAT GGA GAC-3′ (reverse). The PCR product was subjected to agarose gel electrophoresis, and COX3 cDNA was excised and purified using the GeneClean kit (Bio 101).
Cells (5 × 106) were lysed in a buffer containing 1% SDS, 20 mM Tris-HCl (pH 7.4), 5 μg/ml pepstatin A, 10 μg/ml leupeptin, 5 μg/ml aprotinin, and 1 mM PMSF, and then boiled for 5 min. After passage through a 20-gauge needle 10 times and centrifugation at 15,000 rpm for 30 min, the aliquot was boiled in a standard reducing sample buffer for 3 min and subjected to SDS-polyacrylamide gel for electrophoresis. It was followed by transfer to Immobilon-P membrane (Millipore) and hybridization with mouse anti-COX2 mAb (clone 12C4; Molecular Probes), mouse anti-cathepsin B mAb (clone CA10; Calbiochem), rabbit anti-cathepsin D polyclonal Ab (JM-3191; Medical and Biological Laboratories), rabbit anti-β-actin polyclonal Ab (4967; Cell Signaling Technology), or rabbit anti-RIP Ab (ab10472; Abcam). Proteins were visualized by ECL (Amersham Biosciences).
Analysis of cell death
Cells (1 × 106) were cultured with anti-Fas Ab, harvested, centrifuged, resuspended in 100 μl of PBS containing 2.5 μg/ml propidium iodide (PI; Sigma-Aldrich), and incubated for 15 min. These cells were analyzed on a FACScan flow cytometer (BD Biosciences). Cell debris was electronically gated out, based on forward and side light scatter. FL2 fluorescence-positive cells were identified as nonviable.
To distinguish necrotic cell death from apoptotic cell death, the cells stained with PI were inspected using confocal fluorescence microscopy, LSM-GB200 (Olympus). As described previously (22), fragmented nuclei or round ones were identified as apoptotic or necrotic, respectively. At least 300 cells were examined in each experiment. The percentages of apoptotic and necrotic cell death within nonviable cells are presented. For electron microscopic analysis, Hitachi H-7000 (Hitachi) was used after fixation of the cells with glutaraldehyde and OsO4. Fragmented DNA in apoptotic cells was amplified by the PCR method using the ApoAlert LM-PCR ladder assay kit (BD Clontech), according to the manufacturer’s instructions.
Intracellular ATP content was estimated using the ATP assay kit (Calbiochem). Briefly, 1 × 106 cells were Fas stimulated, washed, and resuspended in 200 μl of PBS. Then 200 μl of provided releasing reagent was added to facilitate ATP release. The luciferin/luciferase reaction was initiated by adding 100 μl of provided luciferase solution, of which the peak height was recorded by a luminometer, ML2200 (Dynatech Laboratories). Luminescence of ATP content was presented relative to control ρ+ cells, which had not been Fas stimulated.
The production of ROS was quantified fluorometrically using the ROS-sensitive fluorescent probe hydroethidine (HE; Polysciences). Following the stimulation of Fas, cells were harvested, centrifuged, resuspended in 200 μl of PBS with 3 μM HE, and incubated for 30 min on ice. Then the fluorescence was analyzed on a FACScan flow cytometer. Dead cells and debris were excluded by electronic gating of forward and side light scatter. Mean channel fluorescence of ROS production was presented relative to control ρ+ cells, which had not been Fas stimulated.
Lipid peroxidation was assayed using a Bioxytech LPO-586 kit (OXIS). Briefly, 5 × 107 cells were Fas stimulated, washed, and resuspended in 1.0 ml of PBS, and then Dounce homogenized using a type B (loose) pestle. After adding 10 μl of 0.5 M butylated hydroxytoluene (Sigma-Aldrich), homogenate was centrifuged at 3000 rpm for 10 min. Then 200 μl of the aliquot was mixed with 650 μl of provided reagent R1, N-methyl-2-phenylindole, and 150 μl of 12 N HCl (Sigma-Aldrich). Following the incubation at 45°C for 60 min, malonaldehyde, which results from peroxidation of polyunsaturated fatty acids, was detected spectrophotometrically by ImmunonMini NJ-2300 (InterMed) at 586 nm. Absorbance of lipid peroxidation was presented relative to control ρ+ cells, which had not been Fas stimulated.
The activity of caspase-3 and -9 was measured using a colorimetric protease assay kit (Medical and Biological Laboratories). Briefly, after Fas stimulation, 1 × 106 cells were collected and suspended in 50 μl of the provided lysis buffer. After incubation for 10 min on ice, cells were centrifuged at 15,000 rpm for 15 min. The supernatant was mixed with 50 μl of provided 2× reaction buffer and 200 μM p-nitroanilide (pNA)-conjugated colorimetric substrate of caspase-3 or -9. Following incubation for 45 min at 37°C, released pNA was detected spectrophotometrically at 405 nm.
For further confirmation, procaspase-9, procaspase-3, and their cleaved fragments were detected by Western blotting. Briefly, 5 × 106 cells were Fas stimulated, collected, and suspended in ice-cold isosmotic buffer, 200 mM mannitol, 70 mM sucrose, 1 mM EGTA, 10 mM HEPES-KOH (pH 7.4), 1 μg/ml pepstatin A, 1 μg/ml leupeptin, and 1 mM pefabloc SC. Then it was Dounce homogenized with a type B (loose) pestle and centrifuged at 1,000 × g for 10 min to separate nuclei and unbroken cells, of which the supernatant was centrifuged at 8,000 × g for 10 min to pellet heavy membranes (mitochondrial fraction). The aliquot was further centrifuged at 100,000 × g to obtain a cytosolic fraction (S-100). After boiling in a standard reducing sample buffer for 3 min, the S-100 fraction was subjected to SDS-PAGE. It was followed by transfer to nitrocellulose membrane and hybridization with mouse anti-caspase-9 Ab (clone 5B4; Medical and Biological Laboratories) or mouse anti-CPP32 Ab (clone 19; BD Transduction Laboratories), which can detect both proenzyme and their cleaved forms.
Chromatographic detection of apoptosome formation
A HiPrep 16/60 S-300 Sephacryl high-resolution column (Amersham Biosciences) was equilibrated with 5% (w/v) sucrose, 0.1% (w/v) CHAPS, and 20 mM HEPES/NaOH (pH 7.0), and calibrated with a Gel Filtration Protein Standards kit (Amersham Biosciences). Then the cytosolic fraction (S-100) of 2 × 108 cells was applied to and eluted from the column at a flow rate of 0.2 ml/min and fractionated every 20 min. All procedures were conducted at 4°C. Each fraction was concentrated by Centricon YM-3 (Millipore) and analyzed by Western blot with the anti-caspase-9 mouse mAb mentioned above, anti-apaf-1 mouse mAb (clone 94408.11; R&D Systems), or anti-cytochrome c mouse mAb (clone 7H8.2C12; BD Pharmingen).
Analysis of cytochrome c release
A total of 1 × 106 cells was harvested and washed twice with PBS. The cytosolic fraction (S-100) was prepared, as described above. The pellet of mitochondrial fraction was resuspended in 10 mM Tris-HCl (pH 7.4) containing 1% SDS. Western blot analysis was performed with the anti-cytochrome c mouse mAb mentioned above.
Measurement of lysosomal rupture
Lysosomal rupture was analyzed using the lysosomotropic weak base acridine orange (AO; Molecular Probes). This vital dye accumulates in intact lysosomes and exhibits red fluorescence. Thus, the rupture of lysosomes can be monitored as a decrease in red fluorescence. Briefly, following the stimulation of Fas, cells were harvested, centrifuged, suspended in 200 μl of PBS with 5 μg/ml AO, and incubated for 20 min. Then FL2 fluorescence was analyzed on a FACScan flow cytometer. Dead cells and debris were excluded by electronic gating of forward and side light scatter. The percentage of FL3-negative cells was presented as lysosomal rupture relative to control ρ+ cells, which had not been Fas stimulated.
Short interfering RNA (siRNA)
Cathepsin B siRNA (100 pmol; Santa Cruz Biotechnology), cathepsin D siRNA (Santa Cruz Biotechnology), RIP siRNA (Santa Cruz Biotechnology), or control siRNA-A (Santa Cruz Biotechnology) diluted with siRNA transfection medium (Santa Cruz Biotechnology) was introduced into 1 × 106 cells using siRNA transfection reagent (Santa Cruz Biotechnology), according to the manufacturer’s instructions. Six hours posttransfection, transfection medium containing transfection reagent and siRNA was exchanged to normal growth medium with FBS. Then the cells were cultured for another 48 h.
Measurement of cytosolic cathepsin activities
The activity of cathepsin B or cathepsin D was measured using Z-Arg-Arg-pNA (Calbiochem) or Bz-Arg-Gly-Phe-Phe-Pro-4MbNA (Calbiochem), respectively. Briefly, after Fas stimulation, 1 × 106 cells were collected and lysed without disruption of lysosomes in 50 μl of lysis buffer (25 μg/ml digitonin, 250 mM sucrose, 20 mM HEPES, 10 mM KCl, 1.5 mM MgCl2, 1 mM EDTA, 1 mM EGTA, and 1 mM pefabloc (pH 7.5)). After incubation for 10 min on ice, the cells were centrifuged at 15,000 rpm for 15 min. The supernatant was diluted by reaction buffer (50 mM sodium acetate, 4 mM EDTA, 8 mM DTT, and 1 mM pefabloc (pH 6.0)) supplemented with 50 μM substrates of cathepsin B or D mentioned above. The dilution ratio was optimized in each experiment. Following incubation for 45 min at 37°C, released pNA or 4MbNA was detected spectrophotometrically at 405 nm or fluorometrically at emission 410 nm and excitation 335 nm, respectively.
Depletion of mtDNA
To deplete mtDNA selectively, Jurkat cells were exposed to ethidium bromide, as described previously (20, 21). The mtDNA was substantially depleted in ethidium bromide-treated Jurkat cells (ρ0) compared with that of parental Jurkat cells (ρ+), as revealed by both Southern blotting using a probe to COX3 gene and Western blotting using a specific Ab to COX2 (Fig. 1,A). There were no differences in growth rates after 3 days of incubation, in morphologic characters, or in Fas expression, as analyzed by flow cytometry between ρ+ and ρ0 cells (data not shown). To validate that mtDNA-depleted cells were impaired in production of ATP, intracellular ATP content was assessed by a luciferin/luciferase method and results from ρ0 cells were compared with that of ρ+ cells. The ATP concentration in ρ0 cells was almost one-third of ρ+ cells (Fig. 1,B). When the ρ0 and ρ+ cells were stimulated with anti-Fas Ab (CH-11), ATP concentrations gradually decreased, possibly due to the consumption of ATP during the process of cell death (Fig. 1 B).
Fas-induced necrosis of ρ0 cells
After incubation for 12 h with various concentrations of CH-11, ρ+ and ρ0 cells were analyzed by flow cytometry with the fluorescent probe PI, which penetrates nuclei of both apoptotic and necrotic cells and exhibits red fluorescence. The percentage of PI-positive fractions in both cells increased dose dependently and reached a plateau at 100 ng/ml CH-11 (Fig. 1,C). At this concentration, 56% of ρ+ cells stained positively, whereas ρ0 cells were more sensitive to Fas stimulation, exhibiting 92% PI-positive cells (Fig. 1 C).
Using 100 ng/ml CH-11, we then monitored cell death of both ρ0 and ρ+ over a time course (Fig. 1,D). The occurrence of cell death of ρ0 lagged that of ρ+ by ∼1 h, but its magnitude was significantly higher, findings compatible with Fig. 1,C. The nonviable cell number for both ρ0 and ρ+ reached a plateau after 12 h of CH-11 treatment. When DNA fragmentation was examined, it was obvious in ρ+ cells, but not in ρ0 cells (Fig. 1 E).
Furthermore, inverted microscopic analysis revealed that Fas stimulation for 12 h induced morphologically characteristic changes in ρ0 cells, which were readily distinguished from ρ+ cells (Fig. 1,F). Examination by electron transmission microscopy revealed that ρ0 showed typical necrotic death with vacuole formation, maintenance of nuclear membrane integrity, and high electron translucency of the cytoplasm (Fig. 1,G). In contrast, the death of ρ+ was characterized by cytoplasmic shrinkage, condensation of chromatin, and formation of apoptotic bodies (Fig. 1,G), features consistent with apoptotic death. Confocal fluorescence microscopy with PI staining (22) demonstrated that 95% of ρ+ cells exhibited apoptotic nuclei with DNA fragmentation and that 92% of ρ0 cells contained shrunken nuclei characteristic of necrosis (Fig. 1, H and I).
Besides, neither wortmanin nor 3-methyladenine, which were known to suppress autophagy, inhibited Fas-induced cell death of ρ0 (data not shown). Also, the knockdown of Beclin1 or Atg5, which were both essential signal transducers of autophagy, did not inhibit Fas-induced cell death of ρ0 (data not shown). LC3-II, into which LC3-I was converted in autophagic process, was not increased when ρ0 cells were stimulated by Fas (data not shown). These results indicated that Fas-induced cell death of ρ0 was not autophagy.
Activities of caspase-3 and -9 in ρ0 treated with Fas
To ensure that the apoptotic caspase cascade was not playing a crucial role in necrotic death of ρ0, we measured the activities of both caspase-3 and -9 induced by Fas stimulation using a colorimetric assay. The activities of these enzymes in ρ+ cells started to increase at 3 h and peaked at 6 h after Fas stimulation, whereas there was little activation in ρ0 cells (Fig. 2,A). Likewise, when cleavage of procaspases-9 and -3 were examined by Western blotting, cleaved forms (active forms) of these proenzymes were clearly detected in ρ+ cells, but were undetectable in ρ0 cells (Fig. 2 B). Otherwise, caspase-8 was activated in ρ0 cells as well as in ρ+ cells by the Fas stimulation (data not shown).
Impaired formation of apoptosome in Fas-stimulated ρ0
To confirm the above observation, we next examined by the chromatographic fractionation method (15) the formation of the apoptosome, the ATP/dATP-dependent process (23) of assembling apaf-1, procaspase-9, and cytochrome c into a complex. As shown in Fig. 3, A and C, without Fas stimulation, apaf-1, procaspase-9, and cytochrome c in both ρ+ and ρ0 cells were detected in accordance with their respective molecular weights in fractions 4–9, 12–15, and 17–19, respectively. After stimulation of Fas, each component in ρ+ cells was detected in the high m.w. fractions 1 and 2, indicating the formation of the apoptosome (Fig. 3,B). In contrast, the components in ρ0 cells all remained in the low m.w. fractions regardless of Fas stimulation (Fig. 3 D), indicating no apoptosome formation in these cells even after Fas stimulation.
Dependency of Fas-induced necrosis on ROS in ρ0
We next explored the implication of ROS in necrotic cell death by using flow cytometric analysis with the ROS-sensitive fluorescent probe HE. In ρ0 cells, intracellular ROS increased soon after Fas stimulation (1 h) and peaked after 6 h of stimulation (5-fold elevation). Generation of ROS in ρ+ cells was only 2-fold higher after 6 h (Fig. 4 A).
When ρ0 cells were treated with NAC, which did not suppress the activation of caspase-8 (data not shown), but clearly scavenged the ROS generation induced by the Fas stimulation (Fig. 4,B), the induction of cell death observed after Fas stimulation dose dependently inhibited (Fig. 4,C). This observation was supported by additional experiments using MnSOD transfectants of ρ0, which generated much less ROS than mock transfectants (Fig. 4,D). The susceptibility of the MnSOD transfectants to Fas stimulation was largely impaired when compared with mock transfectants (Fig. 4 E). These results indicate the dependency of necrotic death on ROS in ρ0 cells. Besides, other antioxidants such as butylated hydroxyanisole and 2,2,6,6-tetramethyl-4-piperidinol-N-oxyl also effectively inhibited Fas-induced necrosis as well as ROS production of ρ0 cells (data not shown).
As for the mechanisms by which large excess of ROS was induced in ρ0 cells by Fas stimulation (Fig. 4,A), we focused on the possible involvement of the induction of mitochondrial permeability transition (MPT) and resultant cytochrome c release. By using cyclosporin A or bongkrekic acid, which suppressed the induction of MPT and resultant cytochrome c release in ρ0 cells induced by Fas stimulation (Fig. 5,A), Fas-induced ROS production in ρ0 cells was effectively inhibited (Fig. 5,B). Furthermore, the basal level of ROS in ρ0 cells was not higher than that in parental Jurkat, ρ+ cells (Fig. 5 C). We surmise that by the trigger of Fas stimulation, the excessive ROS production is induced in ρ0 cells due to the insufficiency of respiratory chain, which usually consumes oxygen.
Increased lipid peroxidation in Fas-stimulated ρ0
Because ROS is known to damage subcellular structure by lipid peroxidation of membranes, we measured the relative levels of lipid peroxidation by the malonaldehyde method in ρ0 cells. One hour after Fas stimulation, lipid peroxidation became evident, although at a lower level (Fig. 4,F) relative to the induction of ROS (Fig. 4 A). Lipid peroxidation levels gradually increased to >3-fold after 6 h of stimulation, reaching a plateau at 12 h. In ρ+ cells, no appreciable amount of lipid peroxidation was detected after 1 and 3 h of Fas stimulation, but a 1.5-fold increase was observed after 6 h. These results indicate that lipid peroxidation occurred more frequently in ρ0 cells than in ρ+ cells, although its onset appeared to be late compared with that of ROS generation.
ROS-induced lysosomal rupture in ρ0 cell
Based on our experiments and those using an ischemia-reperfusion model (24, 25, 26) to examine the involvement of lysosomal proteases in necrotic cell death, we were prompted to examine whether ROS generated by Fas stimulation caused lysosomal rupture in ρ0 cells. Lysosomal rupture in ρ0 cells as monitored by FACS using an AO indicator occurred rather slowly after Fas stimulation (Fig. 6,A) compared with the rapid rise of ROS (Fig. 4,A) and caspases (Fig. 2), but showed a quite similar curve to that of lipid peroxidation (Fig. 4,F). However, the percentage of lysosomal rupture in ρ0 after 6 and 12 h of stimulation was significantly higher (60%) than in ρ+ cells (10%). Treatment with NAC and MnSOD transfection clearly inhibited lysosomal rupture in ρ0 cells (Fig. 6, B and C), suggesting the involvement of ROS in lysosomal rupture. When ρ0 cells were treated with CH-11 in the presence of lysosomal stabilizers chloroquine (Fig. 6,D) and bafilomycin (Fig. 6 E), the nonviable cell number significantly decreased, indicating lysosomal rupture in necrosis.
Role of cathepsins B and D in necrotic cell death of ρ0
To identify the molecules whose activations are associated with lysosomal rupture, we examined the effect of various protease inhibitors on necrosis of ρ0. A combination of inhibitors for lysosomal proteases, cysteine protease inhibitor E-64d and aspartic protease inhibitor pepstatin A, but not serine protease inhibitor α1-antichymotrypsin, substantially suppressed necrotic death of ρ0 cells (Fig. 7 A). This indicates the role of proteases after release from ruptured lysosomes in necrotic death of ρ0.
Because E-64d and pepstatin A effectively suppressed death of ρ0, we examined the possible involvement of cathepsins released from lysosomes in necrosis. As shown in Fig. 7, B and C, activities of cathepsins B and D markedly increased in ρ0 cells upon CH-11 treatment as compared with ρ+ cells. To confirm the role of these cathepsins in necrotic death of ρ0, we examined necrosis of ρ0 in which expression of cathepsin B and/or D was suppressed by siRNA. As shown in Fig. 7, D and E, susceptibility of ρ0 treated with cathepsin B siRNA and cathepsin D siRNA in combination with CH-11 to necrosis was substantially reduced.
Role of RIP in necrotic cell death of ρ0 induced by CH-11
RIP has been previously shown to be involved in necrosis of Jurkat cells, which were treated with high doses of Fas ligand in the presence of a pan caspase inhibitor (18). We, therefore, knocked down RIP using specific siRNA (Fig. 10,A) and examined the sensitivity of RIP knocked down (RIPKO) ρ0 cells to CH-11. As shown in Fig. 10 B, although the difference was small, the number of nonviable RIPKOρ0 cells was significantly lower than the number of nonviable ρ0 cells. When RIPKOρ0 cells were treated with NAC, almost complete inhibition of CH-11-induced cell death was observed. These results indicate that, in addition to lysosomal cathepsins, RIP is also involved in necrosis of ρ0, although the latter involvement is a rather minor event.
Conversion of necrosis to apoptosis in ρ0 by ATP supplementation
To elucidate whether supplementation of ATP to ρ0 cells converts necrotic death to apoptosis, we added ATP to culture medium of ρ0 cells and stimulated them with CH-11. As shown in Fig. 8,A, in ATP-supplemented ρ0 cells, cell death occurred immediately after CH-11 stimulation with no lag time in a similar manner as ρ+ cells that underwent apoptosis upon CH-11 stimulation (Fig. 1,D). The ATP-supplemented ρ0 cells showed dose-dependent morphological conversion from necrosis to apoptosis as ATP concentrations increased. At 10 mM ATP, almost complete conversion to apoptosis was achieved (Fig. 8,B). This dose-dependent morphological conversion was consistent with the observation of DNA laddering, increasing the appearance of DNA fragmentation as the ATP concentration increased (Fig. 8,C). The activation of both caspase-3 and -9 detected by colorimetric analysis (Fig. 8,D) as well as by Western blotting (Fig. 8,E) was also almost completely recovered by ATP treatment. Furthermore, apoptosome formation in ρ0 cells by addition of ATP was clearly confirmed (Fig. 3 E).
We then explored cellular events relevant to necrosis in ATP-supplemented ρ0 cells. As shown in Figs. 4 and 6, ROS production, lipid peroxidation, and lysosomal rupture observed in ATP-supplemented ρ0 cells occurred at almost the same magnitude as ATP-nonsupplemented ρ0 cells, whereas activities of cathepsins B and D were dose dependently suppressed by the addition of ATP (Fig. 9, A and B). These results suggest that the caspase cascade signal evoked by ATP supplementation somehow inhibited cathepsins that were released from ruptured lysosomes.
Caspase-3 as a necrosis-inhibitory factor
To examine whether the caspase cascade negatively regulated necrosis through inhibition of cathepsins, we treated ATP-supplemented ρ0 cells with an inhibitor of caspase-3, an apoptosis promoter. As shown in Fig. 9, C–E, the caspase-3 inhibitor Z-DEVD-FMK induced necrosis in ATP-supplemented ρ0 cells with full recovery of cathepsin B and D activities.
The mechanisms underlying the inhibitory action of caspase-3 on cathepsins were then investigated. The possibility that caspase-3 directly cleaved these cathepsins is unlikely because there is no target sequence, DEVD, for caspase-3 in these enzymes. Furthermore, no apparent cleavage bands of cathepsins were observed in ATP-supplemented ρ0 cells treated with CH-11 (data not shown). Another possibility was that some unknown cofactor(s) that binds to cathepsins to augment their activity was cleaved by caspase-3. However, there was no apparent difference in silver staining patterns of precipitates with anti-cathepsin B and D Abs between ρ0 and ATP-supplemented ρ0 cells (data not shown). Interference of cathepsin activity by shifting the cytosolic pH away from the optimal range as the result of digestion of various target molecules in the cytosol by caspase-3 was disproved by the observation that there was no significant difference in cytosolic pH of ρ0 cells and ATP-supplemented ρ0 cells (data not shown). When we mixed cytosol of CH-11-treated ρ0 cells with CH-11-treated ρ+ cells in different proportions, as indicated in Fig. 10, C and D, the actual activities of cathepsins B and D were significantly lower than those theoretically predicted, indicating inhibition of cathepsins in ρ0 cells by cytosolic components of apoptotic cells. Besides, the activities of caspase-9 and -3 in Fas-stimulated apoptotic ρ+ cell lysate were decreased by the adding of Fas-stimulated necrotic ρ0 cell lysate with just a dilution effect of mixing (data not shown).
Degradation of RIP by caspase-3
To examine the possible effect of caspase on RIP, we performed Western blotting of RIP from ATP-supplemented ρ0 cells in the presence or absence of Z-DEVD-FMK. As shown in Fig. 10,E, RIP completely disappeared in ATP-supplemented ρ0 cells concomitantly with the appearance of lower m.w. RIP species. Treatment of ATP-supplemented ρ0 with Z-DEVD-FMK restored the original RIP band, compatible with the notion that RIP, which has the caspase-3-sensitive motif DEID at aa 314–317, is being degraded by caspases. A postulated schematic overview of necrotic signaling in ρ0 cells was shown in Fig. 10 F.
In the present investigation, we established a mtDNA-depleted subline of Jurkat cells (ρ0) to study the molecular mechanism of necrosis induced by Fas signal stimulation on the basis of previous reports that various mitochondrial diseases are associated with tissue damage characteristic of necrotic cell death (7). Furthermore, in mtDNA-depleted syndromes, necrosis of hepatocytes is generally associated (27) and cyclosporin A caused deep mitochondrial damage, followed by necrosis of endothelial cells (5).
Results obtained indeed indicate that ρ0 underwent necrosis upon stimulation with agonistic Fas Ab (CH-11), suggesting that in mitochondrial diseases (7) or mtDNA-depleted syndromes (27), chronic exposure of tissues (cells) to endogenous death triggers (such as Fas ligand and possibly TNF) by circulation might be the cause of tissue damage by necrosis. The molecules responsible for necrosis in ρ0 were found to be ROS, lysosomal enzymes (particularly cathepsins B and D), and RIP.
The role of ROS in necrosis has been noted in various experimental models and cell lines (1, 16, 17). However, the mechanisms inducing necrosis by ROS are not fully understood at present (1), and, with the exception of cyclosporin A, evoked necrotic death of endothelial cells in which ROS caused lysosomal rupture and release of cathepsins, promoters of necrosis.
In the present study, we describe a similar necrosis pathway involving ROS/lipid peroxidation/lysosomal rupture/cathepsin activation in Fas-triggered ρ0 cells, suggesting that these sequential events might be a general scheme of ROS-related necrosis, irrespective of cell types or triggering agents. Incidentally, we have previously shown that ROS is also involved in apoptotic cell death of CH-11-treated Jurkat cells (ρ+) by facilitating apoptosome formation (15). Thus, ROS is considered to be a common and key factor for both apoptosis and necrosis, and, therefore, scavenging ROS may be a reasonable means of preventing tissue damage, irrespective of the type of cell death.
Although ROS-related cellular events were primarily responsible for necrotic death in ρ0 cells, RIP was also found to participate in necrosis, but as a minor player in the present study. The RIP pathway has been identified as a major causative component for necrosis in Jurkat cells treated with soluble Fas ligand with cross-linking Abs, an extremely strong stimulator of Fas, in the presence of a pan caspase inhibitor (18). The reason for the apparently less pivotal role of RIP in necrosis of ρ0 cells than in Jurkat cells may be that we did not use a pan caspase inhibitor and used relatively weak Fas stimulation (CH-11 at a concentration of 100 ng/ml compared with 1000 ng/ml in a previous report (18) with no pan caspase inhibitor).
Reversal of necrosis by addition of ATP was clearly seen in ρ0 cells in accordance with the previously reported fact that Jurkat cells, which underwent necrosis by depletion of ATP from culture medium, showed an apoptotic type of death when ATP was supplemented into the medium (19). This indicates that in ρ0 cells, ATP deficiency, not generation of excess ROS, determines necrotic cell fate. In other words, ρ0 cells never undergo necrosis even in the presence of excess ROS if ATP is sufficiently supplied, suggesting that apoptosis may be the primary type of cell death, whereas necrosis occurs only when the apoptosis cascade is not functioning.
This dominance of apoptosis over necrosis was proven by the subsequent observation that treatment of ATP-supplemented ρ0 cells with caspase-3 inhibitor resulted in necrosis of the cells (Fig. 9 C).
Similar results have been repeatedly reported in various cell types, such as Fas-transfected murine fibroblast L929 cells (17), murine thymocytes (28), caspase-8-deficient Jurkat cells (16), ATP-depleted Jurkat cells (19), Jurkat cells stimulated by soluble Fas ligand with cross-linking Abs (18), and cyclosporin A-treated endothelial cells (5), of which the caspase cascade was either blocked by inhibitors or genetically impaired. However, in these previous reports, the molecular mechanism for the regulation of necrotic cell death by the caspase cascade was not elucidated. The biological implication of apoptosis dominance over necrosis is merely a matter of speculation, but could be a reflection of ontogeny, meaning that necrosis is simply an older phenomenon in ontogenical sequence than apoptosis.
In the present study, we determined that the activity of cathepsins, which appeared to be the main executor of necrosis, was inhibited by the addition of cytosol from Jurkat cells (ρ+ cells) stimulated with CH-11 in the absence of a caspase inhibitor. The most plausible explanation for this inhibitory effect of ρ+ apoptotic cytosol on cathepsins from ρ0 necrotic cytosol is that the small peptides produced by cleavage of intracellular caspase-3 target proteins work as the cathepsin inhibitors. In fact, there are several proteins, such as Fodrin, DNA-dependent protein kinase catalytic subunit, protein kinase Cθ, and plasma membrane calcium-transporting ATPase 4, which have both target sequences for caspase-3 cleavage (DEVD/DEID) and inhibitory sequences for cathepsin B (LVK) (29). However, identification of every one of these peptides is extremely laborious and remains as a future task. Incidentally, other possibilities, such as that caspase-3 altered intracellular pH from the optimal range for cathepsins and that caspase-3 cleaved cathepsins B and D, are not likely to occur (data not shown).
We also describe an inhibitory effect of caspase-3 on RIP activity, a minor executor of necrosis in ρ0 cells, as evidenced by the fact that degradation of RIP in ATP-supplemented ρ0 cells upon CH-11 stimulation was inhibited by caspase-3 (Fig. 10 E). Because there is a caspase-3 target sequence (DEID) at the C-terminal end of RIP (30) and the C-terminal cleaved RIP protein is calculated to have a molecular mass of 11 kDa, our observation of a 65-kDa band in lysates of ATP-supplemented ρ0 cells is reasonable. Thus, both major and minor pathways of necrosis in ρ0 cells were considered to be under the control of caspase-3. Incidentally, it has been surmised that caspase-8, not caspase-3, is responsible for the cleavage of RIP upon Fas stimulation (18), based on the fact that caspase-8-deficient Jurkat cells underwent necrosis (16, 18) and that caspase-8 is a potent proteolytic enzyme for RIP (31). However, the present result suggesting that caspase-3 is the key controller of necrosis-apoptosis conversion does not contradict these findings because caspase-3 is downstream of caspase-8 and was shown to be capable of similarly cleaving RIP in vitro (30).
In the previous literature (17), L929 cells died Fas-induced necrotic cell death in the complete absence of all caspase activities with the induction of excessive ROS production, whereas caspase-8 is actually activated in the absence of ATP and required for the downstream events in ρ0 cells. Caspase-8 can be activated and evoke MPT in the ATP-depleted conditions of rho0 cells because ATP-dependent step of caspase cascade is downstream of MPT, that is to say apoptosome formation (32). Taking together these findings, the essential event of necrotic cell death is an excessive ROS production, irrespective of how ROS production is induced. Additionally, it is therefore not surprising that RIP is not much involved in this process, because it has been reported for RIP to lead directly to the production of ROS through activation of NADPH oxidase in a caspase-independent manner, and apparently independent of mitochondria (33).
In conclusion, we used ρ0 cells stimulated with agonistic anti-Fas Ab to demonstrate that excess ROS, which caused lysosomal rupture and release of cathepsins, is a major pathway mediating necrosis, whereas activation of RIP was a minor contributor. Reactivation of the caspase cascade by ATP supplementation almost completely shut down these necrotic events through inactivation of cathepsins and RIP by caspase-3, suggesting dominance of caspase-induced apoptosis over necrosis. Based on these results and the fact that ROS plays an essential role in both apoptosis (15) and necrosis, we suggest using antioxidative agents rather than caspase inhibitors, ATP supplementation, or cathepsin inhibitors to treat tissue damage, irrespective of necrotic or apoptotic cell death.
We thank Dr. Naoyuki Taniguchi (Department of Biochemistry, Osaka University, Osaka, Japan) for the gift of the MnSOD expression plasmid pRc/CMV-MnSOD.
The authors have no financial conflict of interest.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This study was supported in part by the Japan Society for the Promotion of Science.
Abbreviations used in this paper: ROS, reactive oxygen species; AO, acridine orange; COX, cytochrome oxidase subunit; E-64d, epoxysuccinyl-l-leucylamido-3-methyl-butane ethyl ester; HE, hydroethidine; MnSOD, manganese superoxide dismutase; MPT, mitochondrial permeability transition; mtDNA, mitochondrial DNA; NAC, N-acetyl-l-cysteine; PI, propidium iodide; pNA, p-nitroanilide; RIP, receptor-interacting protein; RIPKO, RIP knocked down; siRNA, short interfering RNA; SOD, superoxide dismutase; Z-DEVD-FMK, N-benzyloxycarbonyl-Asp-Glu-Val-Asp-fluoromethyl ketone.