zVAD-induced autophagic cell death requires c-Src-dependent ERK and JNK activation and reactive oxygen species generation

Abstract

The treatment of L929 fibrosarcoma cells with zVAD has been shown to induce necroptosis. However, whether autophagy is involved or not in this event remains controversial. In this study, we re-examined the role of autophagy in zVAD-induced cell death in L929 cells and further elucidated the signaling pathways triggered by caspase inhibition and contributing to autophagic death. First, we found that zVAD can stimulate LC3-II formation, autophagosome and autolysosome formation and ROS accumulation. Antioxidants, beclin 1 or Atg5 silencing and class III PtdIns3K inhibitors all effectively blocked ROS production and cell death, suggesting ROS accumulation downstream of autophagy contributes to cell necrosis. zVAD also stimulated PARP activation, and the PARP inhibitor DPQ can reduce zVAD-induced cell death, but did not affect ROS production, suggesting the increased ROS leads to PARP activation and cell death. Notably, our data also indicated the involvement of Src-dependent JNK and ERK in zVAD-induced ROS production and autophagic death. We found caspase 8 is associated with c-Src at the resting state, and upon zVAD treatment this association was decreased and accompanied by c-Src activation. In conclusion, we confirmed the autophagic death in zVAD-treated L929 cells, and defined a new molecular pathway in which Src-dependent ERK and JNK activation can link a signal from caspase inhibition to autophagy, which in turn induce ROS production and PARP activation, eventually leading to necroptosis. Thus, in addition to initiating proteolytic activity for cell apoptosis, inactivated caspase 8 also functions as a signaling molecule for autophagic death.

Introduction

Cell death can be largely divided into two classes, programmed cell death and necrosis. Programmed cell death can be further divided into apoptosis and autophagic death. Apoptosis is characterized by caspase activation, cell shrinking, nuclear and cytoplasmic condensation, DNA fragmentation and formation of apoptosomes.¹ The activation of autophagy involves a series of steps: induction, expansion, completion, fusion and degradation.² The induction of autophagy requires Beclin 1 and its interacting partner, class III PtdIns3K, resulting in the generation of PtdIns(3)P on endomembranes. In the induction stage, LC3-I (the homologue of yeast Atg8) is converted to LC3-II by conjugation with phosphatidylethanolamine; the formation of lipidated LC3-II on, or recruitment to, the phagophore is a prerequisite step for phagophore expansion.³ Once the autophagosome is completed, it then fuses with the lysosome, generating an autolysosome. The sequestered cargo together with the LC3-II trapped in the lumen of the autophagosome is degraded within the autolysosome. Autophagy is an important mechanism to maintain cell homeostasis and provides a prosurvival function under starvation conditions and cancer progression.^2–5 However, abnormal autophagy may lead to cell death and pathogenesis of neurodegenerative diseases.^4–7

Because caspase activation is necessary for apoptosis induction, benzyloxycarbonyl-Val-Ala-Asp (zVAD), a cell-permeable pan-caspase peptide inhibitor that irreversibly binds to the catalytic site of caspase proteases and inhibits caspase-mediated apoptosis, is a widely used inhibitor in characterizing apoptotic cell death. However, recent studies have revealed a greater complexity of findings with regard to zVAD, and suggest that caspases might be involved in other types of cell death. In the presence of zVAD to block the apoptotic machinery, many death insults lead to necrotic cell death, which was termed caspase-independent cell death (CICD) or necroptosis.⁸ And in some but not all situations, necroptosis is also characterized by activation of autophagy, which in turn contributes to cell death.^9–12 No matter whether autophagy is involved or not, reactive oxygen species (ROS) and the death domain-containing kinase RIP1 contribute to this alternative form of cell death.^8,13–15

Strikingly, the treatment of murine L929 fibrosarcoma cells with zVAD alone has been shown to induce necroptosis.¹⁴ However, molecular mechanisms underlying zVAD-induced necroptosis in L929 cells, in particular the role of autophagy in cell fate, remain controversial. Lenardo's group showed that autophagosome formation, RIP1 activation and ROS production are involved in this cell death.^16,17 Using RNAi of beclin 1 and Atg7, they find zVAD-induced cell death and autophagic vacuole formation are diminished, suggesting autophagy induced by zVAD contributes to cell death in L929 cells. In that study, the authors also show the necessity of caspase 8 inhibition for autophagosome formation and ROS production induced by zVAD. Moreover, they find the vacuolated cell formation is blocked by c-Jun siRNA, indicating that c-Jun N-terminal kinase (JNK) is an upstream signal of autophagic death induced by zVAD.¹⁶ Later on, the finding of zVAD-induced autophagic death in L929 cells was confirmed by Madden et al. (2007), and they also claim caspase 8 inhibition alone is not sufficient to cause autophagic death. Rather, the activity of a calpain-like protease must also be blocked.¹⁸ On the other hand, Shen et al. reported that suppression of autophagy function via inhibition of lysosomal function contributes to zVADinduced cell death.¹⁹ Moreover, recently they also reported that zVAD-induced necroptosis in L929 cells depends on autocrine production of TNFα, a potent necrotic inducer in L929 cells.²⁰ Based on this controversy, further investigation in zVAD-induced cell death in L929 cells is required. Besides L929 cells, zVAD co-treated with lipopolysaccharide was shown to induce autophagic CICD in macrophages.²¹ In that study, they find that this type of autophagy requires PARP1 activation, and involves ROS-PARP1-autophagy pathway. Moreover, T cells lacking caspase 8 activity are subject to hyperactive autophagic signaling and subvert a cellular survival mechanism into a death process.¹⁰

As mentioned above, the role of autophagy in zVAD-induced necroptosis in L929 is still controversial, and the molecular mechanism for ROS induction and JNK activation upon zVAD treatment is ambiguous. Although catalase degradation contributes to ROS accumulation in zVAD-treated L929 cells,¹⁷ we cannot exclude other sources for ROS elevation. Most importantly, how caspase inhibition transduces signal for autophagy formation, if it really exists, has not been fully investigated. PARP1 activation is another issue not solved regarding its contribution in the death caused by zVAD in L929 fibrosarcoma cells. Therefore, in this study, we not only re-examined the role of autophagy in zVAD-induced cell death but also elucidated the signaling pathways triggered by caspase inhibition and contributing to autophagy formation.

Results

Consistent with previous reports, we found that zVAD (20 µM) can induce autophagic features under electron microscopy observation. Autophagosome defined as a double-membraned structure containing cellular contents was induced by zVAD at 5 h and 10 h (Fig. 1A-VI and VII). Furthermore, cells under zVAD treatment exhibited necrosis-like features, which were characterized by mitochondrial swelling (Fig. 1A-VII and VIII) and disruption of the plasma membrane (Fig. 1A-II–IV). We also used tfLC3 construct to confirm zVAD-induced autophagic flux. This double-tagged LC3 construct can monitor the maturation processes of autophagosome and autolysosome with different fluorescent signals.^3,22 As shown in Figure 1B, L929 cells under starvation with treatment of HBSS exhibited both yellow (representing the autophagosome) and red (representing the autolysosome) puncta formation, both of which were not detected in cells of control group. Similarly, we also observed this phenomenon in cells treated with zVAD. zVAD treatment for 6 h can induce yellow puncta formation, and at 8 or 12 h red puncta was obviously detected. In contrast, bafilomycin A₁, which inhibits acidification inside the lysosome and thus impairs autolysosomal maturation, only increased yellow puncta.

Figure 1.

zVAD induces autophagic cell death in L929 fibrosarcoma. (A) L929 cells were treated with zVAD (20 µM) for 5 or 10 h as indicated, then cells were collected and prepared for electron microscopy (TEM) analysis as described in Materials and Methods. The photos shown in (V–VIII) are the amplification of (I–IV), respectively. The arrows indicate the appearance of autophagosomes, and arrowheads indicate swelling mitochondria, which contain the double-membrane structures. (B) Cells were transfected with tfLC3 plasmid for 24 h and treated with zVAD (20 µM) or bafilomycin A₁ (BA, 100 nM), or cultured in HBSS for the indicated time. Cells were prepared as described in the method section then subjected to confocal microscopy analysis. Images are representative from three independent experiments. (C) zVAD (20 µM) and/or bafilomycin A₁ (100 nM) were added for the time periods as indicated, and total cell lysates were prepared and immunoblotted with LC3 and β-actin antibodies. Traces are representative from two independent experiments. The protein levels of LC3-I, LC3-II and β-actin were quantified by densitometry, and LC3-II /I and LC3-II /actin ratios were calculated and shown in the parentheses. (D) L929 cells were transfected with specific beclin 1 and Atg5 siRNA to knock down endogenous expression of Beclin 1 and Atg5 or nontargeting siRNA as a control. After 48 h of transfection, cells were treated with zVAD (20 µM) for 12 h, and then cell viability was measured by the MTT assay. When stimulating with indicated agents, we also collected cell lysates at the same time to determine the silencing efficiency by immunoblotting with anti-Beclin 1 and Atg5 antibodies. Data are mean ± SE M from three independent experiments. *p < 0.05, indicating significant attenuation of zVAD-induced cell death by beclin 1 siRNA. (E) After treatment with zVAD (20 µM) for 6 or 12 h, cathepsin B inhibitor (CBi, E64-d and pepstatin A, 20 µg/ml for each) or calpain inhibitor (CAi, Z-LLY-FMK, at concentration prepared in the commercial kit) for 12 h, cell lysates were prepared for determining cathepsin B and calpain activities. For the in vitro assay, zVAD, CBi or CAi were added to total lysates, and incubated for 30 min. Data are mean ± SE M from three independent experiments. *p < 0.05, indicating significant inhibition of enzyme activity.

Moreover, since LC3 conversion is a prerequisite for phagophore expansion and autophagosome formation, we measured the effect of zVAD on LC3-II expression. Figure 1C showed that upon zVAD treatment for 1–5 h, a gradual decrease of LC3-I and a concomitant increase of LC3-II were induced. Through inhibition of lysosomal activity to block the degradation of LC3-II, bafilomycin A₁ treatment resulted in a time-dependent accumulation of LC3-II. When zVAD was treated in the presence of bafilomycin A₁, we found that the ratio changes of protein level of LC3-II to LC3-I within a 5 h incubation were further increased. Besides the increase of LC3-II/I ratio, the LC3-II/actin ratios under bafilomycin A₁ treatment for 1 and 3 h were also elevated by zVAD. All of the above data derived from electron microscopy, fluorescence microscopy and biochemical assays confirm the previous notion that zVAD can promote autophagy in L929 fibrosarcoma cells.^16,17 Next we also confirmed the previous study and showed the ability of zVAD to induce autophagic cell death in L929 cells. When Beclin 1 and Atg5 protein expression were suppressed using an siRNA approach, zVAD-induced cell death was attenuated as compared to the group of cells receiving control siRNA (Fig. 1D).

Even though our results together with those reported by Lenardo and coworkers defined zVAD-treated L929 fibrosarcoma as a good autophagic death model, we also examined the possible action mechanism proposed by Shen's group and contributing to cell death caused by zVAD. zVAD might be able to suppress lysosomal cathepsin B and calpain,¹⁹ thereby leading to interruption of lysosomal function and abrogation of cytoprotective autophagy. To address this point, we conducted enzymatic assays on cathepsin B and calpain in both cell and cell-free systems. The results of Figure 1E showed that at the incubation period where zVAD is capable of inducing autophagic influx (i.e., 6 and 12 h incubation), neither cathepsin B nor calpain activity in L929 was significantly affected by zVAD. When treating zVAD directly in the cell-free enzyme assay mixtures, we still cannot observe inhibition of both enzymes. In contrast, cathepsin B and calpain inhibitors can exert their respective inhibition on cathepsin B and calpain in assays conducted either in L929 cells or cell-free system. Therefore, we rule out the involvement of lysosomal inhibition, at least on that of cathepsin B and calpain, in the action of zVAD and confirm the ability of zVAD to induce autophagic flux as previously reported.^16,17

Previously, it was shown that zVAD-induced autophagic death is JNK- and ROS-dependent.^16,17 Thus, we were interested to understand the causal relationship of both death mediators, as well as the involvement or not of ERK and p38. Using an antioxidant (trolox, which is a water soluble vitamine E analogue), a ROS scavenger (BHA), a NADPH oxidase inhibitor (diphenyleneiodonium, DPI), mitochondrial respiratory chain inhibitors (rotenone, FCCP) and MAPK inhibitors (U0126, SP60015, SB203580), we found that zVAD-induced cell death can be attenuated by antioxidants, mitochondrial respiratory chain inhibitors, JNK and MEK/ERK inhibitors. In contrast, SB203580 (p38 inhibitor) and DPI had no effect (Fig. 2A). We also detected the intracellular ROS level with the fluorescent dye DCFH₂. Figure 2B showed that zVAD can induce intracellular ROS production in a time-dependent manner, and this effect was attenuated by the ROS scavenger BHA and the JNK inhibitor SP600125. Likewise, trolox, U0126, but not SB203580 or DPI, dramatically reduced the ROS production induced by zVAD (Fig. 2C). These results suggest that JNK and ERK signaling pathways are upstream of ROS increase. Next, we suspected that ROS production was derived from mitochondria, because BHA, a ROS scavenger specifically targeting mitochondria,²³ is effective in blocking zVAD-induced ROS production. To verify this point, we used a mitochondria-specific dye MitoSox for ROS measurement. Results showed that zVAD still can induce mitochondrial ROS production in a time-dependent manner (Fig. 2D), indicating mitochondria are the major source for ROS production.

Figure 2.

ROS production from mitochondria mediates zVAD-induced cell death. (A) L929 cells were pretreated with trolox (1 mM), BHA (100 µM), DPI (1 µM), rotenone (Rot, 3 µM), FCC P (10 µM), U0126 (10 µM), SP600125 (SP, 10 µM) or SB203580 (SB, 3 µM) for 30 min, followed by zVAD (20 µM) incubation for 12 h. Cell viability was measured by the MTT assay and expressed as percentages of control. Data are mean ± SEM from three independent experiments. *p < 0.05, indicating significant attenuation of zVAD-induced cytotoxicity by trolox, BHA, rotenone, FCC P, U0126 and SP600125. (B) L929 cells were pretreated with or without BHA or SP600125 for 30 min, and then stimulated with zVAD (20 µM) for the indicated time periods. After treatment, cells were harvested and followed by intracellular ROS measurement. (C) L929 cells were pretreated with trolox (1 mM), BHA (100 µM), DPI (1 µM), U0126 (10 µM), SP600125 (10 µM) or SB203580 (3 µM) for 30 min, and then stimulated with zVAD (20 µM). After 10 h incubation, cells were harvested, followed by measurement of intracellular ROS. Data are mean ± SE M from three independent experiments. *p < 0.05, indicating significant attenuation of zVAD-induced ROS production by trolox, BHA, U0126 and SP600125. (D) zVAD (20 µM) was added for the time periods as indicated. After treatment, cells were harvested followed by measurement of mitochondrial ROS.

After observing that ROS and autophagy are involved in zVAD-induced necrosis, we attempted to find out if autophagy is upstream or downstream of ROS production under zVAD treatment, and if PARP1 is also involved in this event. To this end, we compared the death features of N-methyl-N'-nitro-N'-nitrosoguanidine (MNNG), a PARP1 activator,^24,25 and TNFα with zVAD. In L929 cells, MNNG (250 µM) and TNFα (10 ng/ml) treatment for 12 h could induce cell death. Pre-treating cells with the PARP inhibitor DPQ (10 µM) could effectively diminish cell death in response to zVAD and MNNG, but not to TNFα (Fig. 3A), suggesting the involvement of PARP1 activation in death signals of zVAD and MNNG. Confirming this point, we found zVAD and MNNG could stimulate PARP1 activation as assessed by increasing PAR accumulation (Fig. 3B). In contrast to the rapid increase of PAR formation by MNNG, the onset of the zVAD response is slower. All these results suggest that PARP1 activation is involved in zVAD-induced autophagic death. To clarify the participation of autophagy in MNNG and TNFα-induced cell death, we treated cells with two autophagic inhibitors, which can achieve this action via inhibition of PtdIns3K.²⁶ As expected, wortmannin (WM, 1 µM) and 3-methyladenine (3-MA, 10 mM) can effectively protect cells against zVAD-induced cell death. In contrast, both autophagic inhibitors had no effects on MNNG- or TNFα-induced cell death (Fig. 3A). These results suggest that PARP1 activation is not sufficient to induce cell death by autophagy, and conversely that it might be a downstream event of autophagy.

Figure 3.

zVAD-mediated ROS production occurs downstream of autophagy, but upstream of PARP activation. (A) L929 cells were pretreated with DPQ (10 µM), 3-MA (10 mM) or wortmannin (WM, 1 µM) for 30 min, followed by zVAD (20 µM), MNNG (250 µM) or TNFα (10 ng/ml) incubation for 12 h. Cell viability was measured by the MTT assay and expressed as percentages of control. (B) Cells were pretreated with BHA (100 µM), trolox (1 mM), SP600125 (SP, 10 µM) or U0126 (10 µM) for 30 min followed by zVAD (20 µM) or with MNNG (250 µM) for the time as indicated. Total cell lysates were prepared and immunoblotted with PAR and β-actin antibodies. (C) L929 cells were pretreated with 3-MA or DPQ for 30 min, and then stimulated with zVAD (20 µM) for 10 h. After treatment, cells were harvested followed by measurement of intracellular ROS. (D) L929 cells were transfected with specific beclin 1 siRNA. After 48 h transfection, cells were treated with zVAD (20 µM) for 10 h, then intracellular ROS was measured. Data represent the mean ± SE M of three independent experiments. *p < 0.05, indicating significant attenuation of zVAD- or MNNG-induced cytotoxicity (A) and ROS production (C and D).

Furthermore, in order to understand the roles of autophagy and PARP1 in zVAD-induced ROS production, PARP1 and autophagic inhibitors were tested in ROS productive response of zVAD. As shown in Figure 3C, L929 cells pretreated with an autophagy inhibitor (3-MA) could inhibit zVAD-induced intracellular ROS production, whereas the PARP1 inhibitor (DPQ) could not. Similar inhibition by 3-MA was observed in zVAD action on mitochondrial ROS production (data not shown). Likewise, silencing beclin 1 also achieved the inhibitory effect on zVAD-induced ROS production (Fig. 3D). These results all together suggest that zVAD-induced ROS production occurs downstream of autophagy, but upstream of PARP1 activation. To further support the previous suggestion, we analyzed the effects of antioxidants on zVAD-induced PAR formation. As shown in Figure 3B, both trolox and BHA treatment abolished PAR induction caused by zVAD. Since ERK and JNK were shown to regulate zVAD-induced ROS production (Fig. 2C), we tested their roles in this respect. Consistent with our scenario, U0126 and SP600125 reduced zVAD-induced PAR expression (Fig. 3B).

The relationship between c-Src and autophagy is still unclear. Previously it has been shown that insulin-induced cell swelling is sensed by integrins and thus transduces a signal for p38 activation via c-Src. This effect leads to the inhibition of autophagic proteolysis in rat liver cells.²⁷ To understand if c-Src plays a crucial role in zVAD-induced autophagic cell death in L929 fibrosarcoma, we examined the effects of the specific c-Src inhibitor PP2. In Figure 4A, we found that PP2 treatment in a concentration-dependent manner confers cell protection against zVAD-induced cytotoxicity. Concomitantly, PP2 markedly reduced zVAD-induced ROS production in the cytosol (Fig. 4B) and in mitochondria (Fig. 4C), suggesting that c-Src activity might mediate ROS-dependent autophagic death induced by zVAD. To further elucidate this event, we knocked down c-Src expression using siRNA. Under efficient silencing of c-Src, we found zVAD-induced cell death and ROS production were attenuated (Fig. 4D). These results highlight a new role played by c-Src in an autophagic cell death model of zVAD.

Figure 4.

c-Src is involved in zVAD-induced autophagic cell death. (A) L929 cells were pretreated with PP2 at the concentrations indicated for 30 min, followed by zVAD (20 µM) stimulation. After 12 h incubation, cell viability was measured by the MTT assay. (B) L929 cells were pretreated with PP2 (10 µM) for 30 min, followed by zVAD (20 µM) stimulation for 10 h, and then cells were harvested for intracellular ROS measurement. (C) L929 cells were pretreated with PP2 (10 µM) for 30 min, followed by zVAD stimulation at the indicated time periods, and then cells were harvested for mitochondrial ROS measurement. The parts on the right hand show the raw data. (D) L929 cells were transfected with Src-targeted siRNA, followed by stimulation with zVAD (20 µM). Cell viability (left part) and intracellular ROS (right part) were determined. Upon stimulating with the indicated agents, cell lysates were collected to determine the silencing efficiency by immunoblotting with anti-Src antibody. Data represent the mean ± SEM of three independent experiments. *p < 0.05, indicating significant attenuation of zVAD-induced cytotoxicity and ROS production.

After observing the inhibitory effects of PP2 on ROS production and cell death, we were interested to understand the role of c-Src in zVAD-mediated upstream signaling cascades. Despite some studies that have demonstrated the roles of JNK and ERK in autophagy formation,^28–30 and c-Src in the activation of both kinases, only a paper published recently showed the involvement of Src family kinases in sorafenib-induced autophagic death in gastrointestinal tumor cells.³¹ To clarify how c-Src cross-talks with ERK and JNK, we determined the effects of PP2 on zVAD-elicited ERK and JNK. Figure 5A showed that zVAD can induce a rapid and sustained activation of JNK and ERK within 4 h incubation. Moreover, both effects of zVAD were abolished by PP2, indicating c-Src is functioning upstream to JNK and ERK signaling. Next to verify if c-Src, ERK and JNK activation contribute to autophagy, we used RNAi to knock down c-Src expression for further validation of its role in zVAD-induced autophagy, and JNK and ERK signaling. Figure 5B showed that zVAD-induced LC3-II conversion, and JNK and ERK activation were inhibited after c-Src silencing. JNK and ERK inhibition after SP600125 and U0126 pretreatment, respectively, also blocked zVAD-induced LC3-II conversion (Fig. 5C). These results all together suggest c-Src mediates zVAD-induced JNK and ERK activation, and autophagy.

Figure 5.

c-Src mediates JNK and ERK activation caused by zVAD. (A) L929 cells were treated with PP2 and zVAD for the indicated time periods. Cell lysates were harvested for immunobloting of JNK-p, JNK, ERK-p and ERK. (B) L929 cells were transfected with specific c-Src siRNA to knock down endogenous expression of c-Src or non-targeting siRNA as a control. After 48 h of transfection, cells were treated with zVAD (20 µM) for the indicated time. Cell lysates were harvested for immunobloting of JNK-p, JNK, ERK-p, ERK, LC3, c-Src and β-actin. (C) Cells were treated with SP600125 (SP, 10 µM) or U0126 (10 µM) for 30 min, and then stimulated with zVAD (20 µM) for the indicated time. Cell lysates were harvested for immunobloting of LC3 and β-actin.

Recent studies identified caspase 8 as a c-Src substrate, and demonstrated that such tyrosine phosphorylation by c-Src provides a new mechanism to inhibit caspase 8 activation.^32–34 Moreover, novel enzyme activity-independent actions of caspase 8 in adhesion and epidermal growth factor-induced activation of the ERK pathway were reported.^35,36 In order to link caspase 8 inhibition and c-Src activation upon zVAD treatment, we determined the binding condition of both proteins. Results of co-immunoprecipitation with using c-Src antibody revealed that caspase 8 can bind to c-Src at the resting state. Such constitutive binding was reduced after zVAD incubation for 10 and 30 min (Fig. 6A). zVAD treatment within 4 h does not change the expression levels of caspase 8 or c-Src (Fig. 6B). In addition, cleaved active caspase 8 was not detected following zVAD treatment. Intriguingly, zVAD is able to induce c-Src phosphorylation at Tyr 418, indicating the activation of this kinase (Fig. 6C).

Figure 6.

zVAD induced c-Src dissociation from caspase 8 and activation. (A) After treating with zVAD (20 µM) for different periods, cell lysates were subjected to immunoprecipitation with c-Src antibody, followed by immunobloting with caspase 8 and c-Src antibodies. In (B and C), cell lysates were subjected to immunobloting with caspase 8, Tyr 418 phosphorylated Src, c-Src and β-actin antibodies.

Discussion

Although zVAD, a pan-caspase inhibitor, is widely used to block cell death undergoing apoptosis, it can also drive necroptosis in the presence of extrinsic death receptor ligands, such as TNFα, FasL, TRAIL and LIGHT.^8,13,37–39 In L929 cells, it is intriguing to note that zVAD alone is sufficient to induce cell death, and in particular, unlike other necrosis induced by death receptor activation, it involves autophagosome formation.^16,17 This unusual and cell-specific effect of zVAD in L929 cells becomes a valuable model to elucidate the link between caspase inhibition and autophagy formation. In agreement with previous observations,^6,17,18 our present results support the idea that autophagy contributes to cell death in zVAD-treated L929 cells. zVAD treatment can induce autophagic features in L929 cells at time points earlier than the occurrence of cell necrosis. In this respect, LC3 conversion was detected at 1 h; autophagosome formation was detected at 5∼6 h; autolysosome formation was detected at 8∼12 h; and necrotic cell death characterized morphologically by organelle swelling and disintegration, nuclear degradation and breakdown of the plasma membrane was detected at 10 h. Consistent with the previous notion that autophagy is involved in zVAD-induced cell necrosis, we confirmed that this cell death was blocked by autophagy inhibitors (class III PtdIns3K inhibitors 3-MA and wortmannin) as well as by knocking down beclin 1 and Atg5.

Supporting the ability of zVAD to induce autophagic flux, we found zVAD can increase the formation of autophagosome and autolysosome. In addition, under the treatment with lysosomotropic agent bafilomycin A₁, which inhibits acidification inside the lysosome and thereby the LC3-II degradation, zVAD is still able to further increase the LC3-II/I ratio and LC3-II/actin ratio at a certain time point, suggesting the ability of zVAD to increase autophagic flux. The increase of LC3-II level in parallel to the concomitant downregulation of LC3-I upon bafilomycin A₁ treatment was also observed in mouse embryonic fibroblasts during starvation.⁴⁰ This finding with bafilomycin A₁ suggests the existence of a high rate of autophagic flux and high efficiency of lysosomal processing in L929 cells under resting condition. The reasons that zVAD induces less ratio changes of LC3-II/I compared to bafilomycin A1 might be due to enhancement of LC3-II degradation by zVAD and its ability to regulate LC3-I expression. Moreover, our data rule out the ability of zVAD to inhibit cathepsin B and calpain activities as previously reported.¹⁹

Several lines of evidence shown in this study support the essential role of mitochondria-derived ROS production downstream of autophagy in zVAD-induced cell necrosis. First, there is a correlated attenuation of zVAD-induced cytotoxicity and ROS production by ROS scavengers (trolox, BHA). Second, autophagic inhibitor (3-MA) and si-beclin 1 effectively reduced ROS induction. Third, inhibitors of the mitochondrial respiratory chain (i.e., rotenone and FCCP) remarkably inhibited cell death, suggesting mitochondria are the major source for the zVAD-elicited ROS increase. Fourth, we detected mitochondrial ROS increase under zVAD stimulation and found it could be inhibited by the autophagy inhibitor 3-MA. Fifth, NADPH oxidase, which is responsible for TNFα-induced generation of O₂⁻ in L929 fibrosarcoma,⁴¹ did not participate in the action of zVAD. Therefore, we conclude that under zVAD treatment ROS production in L929 cells from mitochondria is downstream of autophagy, and mediates cell necrosis. In this respect, an amplification loop between autophagy formation and ROS production might exist. It has been demonstrated that starvation induces complex formation between class III PtdIns3K and Beclin 1, which in turn, together with other signals, leads to a local rise in H₂O₂ in the vicinity of mitochondria. This oxidative signal inactivates Atg4, thereby promoting lipidation of LC3, an essential step in the process of autophagy.⁴²

However, unlike the case in zVAD-treated L929 cells, TNFα which also induces ROS-dependent necrosis in L929 cells⁴³ did not trigger autophagy formation based on morphological features and biochemical analyses. MNNG action is another case that can induce ROS production, but not autophagy formation (data not shown). Moreover, 3-MA and wortmannin had no effects on MNNG-, TNFα- and H₂O₂-induced cytotoxicity. Therefore, even though autophagy is induced by ROS in various pathophysiological conditions,^42,44 such as during nutrient starvation, treatment with mitochondrial toxins, or during ischemia and reperfusion, it is not a general phenomenon. At least in L929 fibrosarcoma, it is not the case under treatment with MNNG, TNFα and H₂O₂. Alternatively the autophagy induction by zVAD might rely on an additional specific upstream regulating signal.

It has been reported that MAPK signaling pathways are able to modulate autophagy with different mechanisms. ERK1/2 can phosphorylate G interacting protein (GAIP) and stimulate autophagy.²⁸ Conversely, amino acids stimulate the sustained phosphorylation of GAIP at Ser259, which is involved in the negative regulation of Raf-1, and thus inhibits Ras/Raf-1/ERK1/2-mediated autophagy.⁴⁵ In contrast, it was shown that p38 might limit the constitutive autophagy activity by impeding the fusion of autophagosomes with lysosomes. This blockade, however, could be relieved by transiently activated ERK induced by autophagic stimuli.⁴⁶ The relationship between JNK and autophagy has also been reported recently. In this context, ER stress-induced IRE1-JNK activation is required for autophagosome formation. 29 Subsequent studies further identified the molecular links between JNK and autophagy. Phosphorylation of Bcl-2, which disrupts the Bcl-2-Beclin 1 complex, in turn leading to autophagy, was demonstrated.^47,48 Aside from Bcl-2 phosphorylation, JNK-mediated Beclin 1 expression and p53 phosphorylation were demonstrated to contribute to the autophagic cell death in cancer cells.⁴⁹ A recent study showed that the JNK-dependent accumulation of p62 and AMPK pathway can cooperate to mediate the resveratrol-induced autophagic cell death.⁵⁰ In this study, our results suggest the necessity of ERK and JNK for zVAD-induced LC3-II conversion and autophagic cell death in L929 cells.

Currently, the relationship between Src and autophagy is still unclear. Schliess et al. showed that insulin-induced cell swelling is sensed by integrins, which transduce signaling via Src into p38 activation, and leads to inhibition of autophagic proteolysis in rat liver cells.²⁷ However, the involvement of Src in the induction of autophagy has also been reported recently.³¹ Interestingly, here we found Src is involved in zVAD-induced autophagic cell death. Src inhibitor (PP2) pretreatment or silencing c-Src with siRNA can protect L929 cells against zVAD-induced cytotoxicity and intracellular ROS production. Therefore, it is suggested that c-Src may mediate signaling cascades responsible for autophagy formation.

Previous studies have shown the involvement of caspase 8 inhibition in zVAD-induced autophagic cell death in L929 cells.^16–18 In this study, we further show a novel enzymatic activity-independent action of caspase 8 in Src activation in L929 cells, which strengthens an emerging note that the precursor of caspase-8 also functions as a signaling molecule to regulate cell death, migration and adhesion.³⁴ In this respect, procaspase 8 was shown to form a complex with c-Src in EGF-stimulated neuroblastoma cells.⁵¹ Subsequently c-Src-mediated phosphorylation of caspase 8 at Tyr380 leads to caspase inhibition,^32,33 and then converts caspase 8 from a pro-apoptotic factor to a novel signaling molecule in terms of its ability to regulate cell adhesion and motility.^35,36 Extending the above findings, we surprisingly observed the binding of c-Src and caspase 8 at resting conditions in L929 fibrosarcoma. Moreover, it is interesting to note that such interaction is reduced upon zVAD treatment, leading to the dissociation and activation of c-Src, and then initiation of autophagic cell death. Therefore, our study again strengthens the crucial cross-regulation between caspase 8 and c-Src, and provides an additional mechanism to limit c-Src activation at the basal state. We suggest the binding between c-Src and caspase 8 can be modulated by the caspase 8 conformation; possibly the binding of zVAD to the catalytic groove of procaspase 8 renders conformational changes and hinders the accessibility of c-Src to death effector domains of procaspase 8.

PARP1 hyperactivation-induced necrosis has been implicated in several pathophysiological conditions. Overactivation of PARP1 results in unregulated PAR synthesis and widespread cell death. Previous studies, in most cases using MNNG as a potent PARP1 activator, have revealed that the generation of PAR can trigger intracellular ATP depletion, mitochondrial dysfunction, AIF release, calpain activation and eventually caspase-independent necrotic cell death.²⁵ Even though PARP activation can modulate autophagic cell death,²¹ the signaling pathway between PARP and autophagy is unclear. Currently bidirectional interactions between ROS production and PARP1 activation have been documented. In this respect, some studies show that ROS production can trigger PARP1 activation followed by autophagic cell death.²¹ Conversely, PARP1 activation also was reported to induce ROS generation from mitochondria.²⁵ In this study, we found that the PARP1 inhibitor (DPQ) can attenuate cell death induced by zVAD and MNNG, but not by TNFα, suggesting PARP1 activation is also involved in zVAD-induced autophagic cell death in L929 cells. Conversely, DPQ has no effect on zVAD-induced ROS production, while it can block that induced by MNNG.⁵² These findings suggest that upon zVAD stimulation of L929 cells, PARP1 activation is a downstream event of ROS production. Recent studies also have implicated the close relationship between ER stress and autophagy.²⁹ Based on this, we examined whether ER stress response occurs in zVAD-treated cells. As a result, we did not observe any features of ER stress, for example GRP78 and CHOP induction, in zVAD-treated L929 cells (data not shown).

In summary, we propose the signaling pathways for zVAD-induced cell death in L929 cells. zVAD-induced inhibition of caspase 8 can dissociate and release c-Src for activation, which then transduces signals to ERK and JNK. The activation of both MAPKs in turn sequentially triggers autophagy, ROS accumulation, PARP1 activation and necrotic cell death.

Materials and Methods

Cell culture.

Murine L929 fibrosarcoma cells were cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum (FBS) (v/v), 100 U/ml penicillin, 100 µg/ml streptomycin, and incubated at 37°C in a humidified atmosphere of 5% CO₂ in air.

Reagent.

DMEM (12100-046), FBS (04001-1A), penicillin and streptomycin (03031-1B) were purchased from Gibco BRL. Polyclonal antibodies specific for ERK2 (sc-154), JNK1 (sc-474), c-Src (sc-19), horseradish peroxidase-conjugated anti-mouse (sc-2969), and anti-rabbit (sc-2004) antibodies were purchased from Santa Cruz Biotechnology. Polyclonal antibody specific for β-actin (MAB1501) was obtained from Upstate. PP2 (529573), U0126 (662005), SB203580 (559389), trolox (6-hydroxy-2,5,7,8-tetramethylchroman-2-carboxylicacid) (648471), SP600125 (420119), bafilomycin A1 (196000), and wortmannin (681675) were obtained from Calbiochem. Antibody of Src (2108) and phosphorylated antibodies of ERK (Thr202/Tyr204, 9101) and JNK (Thr183/Tyr185, 9251) were purchased from Cell Signaling Technology. Beclin 1 (612112) and PAR (51811) antibodies were obtained from BD Pharmingen. 3-Methyladenine (3-MA, M9281), carbonyl cyanide p-(trifluoromethoxy) phenylhydrazone (FCCP, C2920), rotenone (R8875), zVAD-fluoromethylketone (zVAD-FMK, H8787), butylated-hydroxyanisol (BHA, B6655), 3-(4,5-dimethylthiazol-2-yl)2,5-dephenyltetrazoliumbromide (MTT, M2128), E-64d (E8640), pepstatin A (P5381), DPI (D2926), and 2′7′-dichloro-dihyrofluorescein diacetate (DCFH₂-DA, D6883) were from Sigma-Aldrich. N-methyl-N'-nitro-N-nitrosoguanidine (MNNG, 2392) was purchased from Chem Service. LC3 antibody (PM036) was purchased from MBL. Rabbit polyclonal anti-Src (pY418) (44660G) and MitoSOX™ (M36008) were purchased from Invitrogen. Cathepsin B and calpain kits (K140-100 and K240-100, respectively) were purchased from Biovision. Atg5 antibody (GTX102361) was purchased from Gentex.

Measurement of cell viability by MTT assay.

Cells (10⁴/ml) grown in 96-well plates were incubated with indicated drugs. For MTT assay, MTT (5 mg/ml) was added for 1 h, then the culture medium was removed; the formazan granules generated by live cells were dissolved in 100% DMSO and shaken for 10 min. The OD at 550 nm and 630 nm was measured using a microplate reader. The net absorbance (OD₅₅₀-OD₆₃₀) indicates the enzymatic activity of mitochondria and implicates the cell viability.

RNA interference.

The siRNA duplexes specific for mouse c-Src, Beclin 1 and Atg5 were obtained from Dharmacon RNA Technologies. The siRNA contained four RNA sequences in a pool SmartPool selected from the NCBI RefSeq Database by a proprietary algorithm. The control nontargeting pool is a pool of four functional nontargeting siRNAs with guanine cytosine content comparable to that of the functional siRNA but lacking specificity for known gene targets. The 100 nM RNAi was transfected by means of MicroPorator (Promega, USA), and the condition was PulseVoltage (V): 1,600; Pulse Width (ms): 10; PulseNumber: 4. To achieve gene silencing, we transfected the cells with the siRNA for 48 h and followed by the drug treatment, then the gene silencing effects were evaluated by western blot analysis.

Immunoblot analysis and immunoprecipitation.

After stimulation, cells were lysed in 1% Triton X-100 lysis buffer (20 mM Tris-HCl, pH 7.5, 125 mM NaCl, 1% Triton X-100, 1 mM MgCl₂, 25 mM β-glycerophosphate, 50 mM NaF, 100 µM Na₃VO₄, 1 mM PMSF, 10 µg/ml leupeptin and 10 µg/ml aprotinin). The cell lysates were resolved by SDS-PAGE and analyzed by western blotting with specific antibodies. To determine the binding between caspase 8 and c-Src, cells treated with zVAD were washed twice with PBS, lysed in 500 µl RIPA lysis buffer (contained 150 mM NaCl) and centrifuged at 14,000 rpm, 4°C for 30 min. The supernatant was collected and pre-cleaned by normal IgG and 10 µl protein A/G-agarose beads for 1 h at 4°C. After centrifugation, supernatant incubated with c-Src antibody at 4°C rocking overnight and then added 10 µl protein A/G-agarose beads rotated at 4°C for another 30 min. The immunocomplexes were washed three times with cold lysis buffer (contained 150 mM NaCl) and twice with 300 mM NaCl containing lysis buffer. The precipitated complexes were added 50 µl 2X Laemmli sample buffer and heat the samples to 95°C for 5 min. The samples were then pelleted down by centrifuge, and the supernatant was fractionated on 10% SDS-PAGE, transferred to nitrocellulose membrane, and probed with specific anibody.

Cytosolic and mitochondrial ROS detection.

For measuring cytosolic and mitochondrial ROS, we used DCFH₂DA and MitoSOX™, respectively. DCFH₂DA can readily enter cells and be cleaved by esterase to yield a polar, nonfluorescent product, DCFH. ROS in the cells promotes the oxidation of DCFH to yield the fluorescent product DCF. MitoSOX™ is live-cell permeant and is rapidly and selectively targeted to the mitochondria. Once in the mitochondria, MitoSOX™ Red reagent is oxidized by superoxide and exhibits red fluorescence (Ex/Em: 510/580). After treating with indicated time periods, cells were collected, then incubated in PBS containing fluorescent reagent (5 µM) for 30 min at 37°C. After incubation, cells were washed by PBS twice, trypsinized, re-suspended in 0.5 ml PBS, and immediately submitted to flow analysis by using FACScan flow cytometer (Becton Dickinson).

Electron microscopic detection of autophagosome.

Cells were collected and fixed with 4% paraformaldehyde in 0.1 M cacodylate buffer at 4°C overnight. Cells were then rinsed with cacodylate buffer and post-fixed with 1% osmium tetroxide in 0.1 M cacodylate buffer and subsequently dehydrated in a graded series of ethanol (75%, 85%, 95% and 100%), and embedded in an Epon-araldite mixture. After the resin had solidified, the plastic culture dish was broken up to release the resin containing the embedded cells. Specimens were sectioned with a Diatome diamond knife on a Rechert Ultracut E ultramicrotome (Leica, Wetzlar, Germany). Ultrathin sections were stained with uranyl acetate and lead citrate and viewed with an H-7100 electron microscope (Hitachi, Tokyo, Japan). All images were taken by an ORCA-ER CCD digital camera system (Advanced Microscopy Techniques, Danvers, MA USA).

Measurement of cathepsin B and calpain activity.

Cells were treated with zVAD, cathepsin B inhibitor (CBi, E64-d and pepstatin A, 20 µg/ml for each) or calpain inhibitor (CAi, Z-LLY-FMK, at concentration prepared in the commercial kit) for the indicated time. Then total cell extracts were prepared with manual instructions. For cell-free enzyme assay, cell lysates of cells without drug treatment were prepared with manual instructions then were treated with zVAD, cathepsin B inhibitors or calpain inhibitor for 30 min. Cathepsin B substrate sequence RR labeled with AFC (amino-4-trifluoromethyl coumarin) and calpain substrate Ac-LLY-AFC were used. The release of free AFC was detected by fluorometer equipped with 400 nm excitation filter and 505 nm emission filter. The changes of cathepsin B and calpain activity were determined and expressed as percentages of the control group.

Plasmid transfection and confocal microscopy analysis.

The tandem fluorescent-tagged LC3 construct (tfLC3), which was generated by Dr. T. Yoshimori's group,²² was purchased from Addgene (Cambridge, MA USA). The transient transfection was performed by means of MicroPorator (Promega, USA), and the condition was PulseVoltage (V): 1,600; Pulse Width (ms): 10; PulseNumber: 4. We transfected the plasmid to cell for 24 h and followed with the drug treatment. The treated cells were fixed with 4% paraformaldehyde and examined by a confocal microscopy (ZEISS, LSM 510 META Confocal Microscope).

Statistical evaluation.

Values are expressed as the mean ± SEM of at least three experiments. Analysis of variance was used to assess the statistical significance of the differences, with a p value of <0.05 being considered statistically significant.