INTRODUCTION

Apoptosis is a tightly regulated process of cell death accompanied by clumping of chromatin, nuclear disruption, and formation of cytosol containing apoptotic bodies (1). Genetic and biochemical evidence has suggested that apoptotic death proceeds by one or more ordered pathways. Studies with Caenorhabditis elegans have proved to be particularly insightful for recognizing specific genes in this pathway that govern survival and death (2). Of special interest is ced-3, which is necessary for programmed cell death (2, 3, 4, 5), and ced-9, which acts to block ced-3-promoted apoptosis (4, 6). Mammalian counterparts of these genes have been recognized and extensively characterized; BCL2 and a group of related genes are homologous to ced-9 (7, 8, 9, 10, 11), while ced-3 homologues include the family of interleukin-1 converting enzyme (ICE)¹-like cysteine proteases (7, 8, 10, 12). Although expression of certain ced-9/BCL2 family members may counteract the death promoting actions of ced-3/ICE family proteases (13, 14), the relative positions of these molecules in cell death pathways are unclear.

In addition to ICE family proteases, inducible gene products also appear to be required for apoptotic death in some cell types. For instance, inhibitors of protein or RNA synthesis suppress death of growth factor-deprived neurons (15, 16, 17) as does disruption of the activity of c-Jun, a transcription factor that is induced by growth factor deprivation (18, 19, 20). The relative positions of such gene products in the cell death pathways are also unknown.

Recently, studies have implicated c-Jun kinase (JNK/SAPK) as an obligate component of the cell death pathway in PC12 cells (18). JNK is a member of the mitogen-activated protein kinase superfamily that phosphorylates the N-terminal of c-Jun, thereby activating the transcriptional transactivation potential of this factor (21). JNK is regulated through both Ras-dependent and independent pathways involving MEKK1 and JNKK (MKK4) kinases (22, 23, 24) and is activated in response to multiple stress-inducing stimuli, including interleukin 1, tumor necrosis factor-, metabolic uncouplers, and UV radiation (21, 22). The activity of JNK is also markedly elevated in response to growth factor withdrawal from cultured PC12 cells (18, 25) and sympathetic neurons.² Xia et al. (18) reported that JNK activation is required for apoptosis of neuronally differentiated PC12 cells after NGF deprivation. Transient transfection with constitutively active MEKK1 promoted apoptosis of these cells, while coexpression with dominant-negative forms of c-Jun attenuated this effect (18). Transient expression of dominant-negative forms of MKK4 also diminished death caused by NGF withdrawal (18). With regard to relative position in cell death pathways, the positioning of JNK activation has not been defined.

To understand the mechanisms by which trophic factors govern survival and death of neuronal and other cell types, we and others have exploited the PC12 pheochromocytoma cell line (26). Under serum-containing culture conditions, PC12 cells divide and resemble precursors of adrenal chromaffin cells and sympathetic neurons. Treatment with NGF causes these cells to cease proliferation and induces a neuronal phenotype (26). Withdrawal of trophic support, either by serum deprivation of proliferating neuroblast-like PC12 cells or by NGF/serum removal from neuronally differentiated cells, leads to their apoptotic death (27, 28, 29, 30). NGF withdrawal similarly triggers death of sympathetic neurons both in vivo (31, 32) and in vitro (33, 34).

Apoptotic death of trophic factor-deprived PC12 cells appears to be regulated by pathways that contain many of the elements discussed above. Consistent with apoptosis in other cell systems, overexpression of BCL2 (35, 36) and inhibition of the ICE family proteases (37) attenuates PC12 death caused by withdrawal of trophic support. Moreover, death of neuronal PC12 cells is blocked by inhibition of RNA synthesis (17) and appears to require JNK activation (18).

Examination of cell death in PC12 cells and their sympathetic neuron counterparts has lead to identification of a wide variety of agents that confer protection from loss of trophic support. These include the putative endonuclease inhibitor (38) aurintricarboxylic acid (ATA) (29); cell cycle inhibitors, including ciclopirox, deferoxamine, mimosine, (39), olomoucine, and flavopiridol (25); activators of cyclic AMP-dependent protein kinase (PKA) (28); N-acetylcysteine (NAC) (40, 41), a reducing agent that blocks cell cycle; and agents that elevate or mimic intracellular cGMP, including nitric oxide generators and permeant cGMP analogues (42). Despite the efficacy of these treatments, the relative positions at which they block apoptotic death has again been unclear.

The present studies are aimed at elucidating the relative biochemical positions of molecules that either participate in, or interfere with, the pathways that lead to apoptotic death. In particular, we have investigated whether BCL2, ICE family protease inhibition, inhibition of RNA synthesis, and a variety of additional survival-promoting agents (ATA, NAC, cAMP, cGMP, and deferoxamine) interfere with activation of JNK after withdrawal of trophic support from PC12 cells. By identifying whether these molecules block cell death prior to or subsequent to JNK activation, we can begin to formulate the sequence of events in the cell death pathway.

EXPERIMENTAL PROCEDURES

Human recombinant NGF was kindly provided by Genentech. NAC, ATA, 8-bromo-cGMP, 8-(4-chlorophenylthio)-cAMP, and deferoxamine were purchased from Sigma Diethylenetriamine nitric oxide (DETA·NO) was obtained from Research Biochemical International (Natick, MA). The construct containing the GST-c-Jun fusion protein was a generous gift from Dr. Michael Karin (University of California, San Diego, CA). zVAD-fluoromethylketone (zVAD-fmk) was obtained from Enzyme Systems Products (Dublin, CA). PC12 cells overexpressing BCL2 and an empty neomycin-resistant construct (lines PC12bcl-2.1 and PC12neo.1, respectively) were generated and characterized as described previously (35).

PC12 cells were cultured and passaged as described previously (26, 28) in RPMI 1640 medium supplemented with 10% heat-inactivated horse serum and 5% fetal calf serum and on rat tail collagen-coated 100-mm plates. Neuronally differentiated PC12 cells were obtained by plating 1-2 × 10⁷ naive cells onto 100-mm collagen-coated dishes in the presence of 100 ng/ml NGF for a period of 8-10 days in RPMI 1640 medium supplemented with 1% heat-inactivated horse serum.

Survival experiments were performed as described previously (28). Briefly, naive or neuronally differentiated PC12 cells were extensively washed in serum- and NGF-free RPMI 1640 medium and replated onto 24-well tissue culture dishes with the indicated agent. The final volume of medium in each well was 1 ml. At appropriate times, cells were lysed and numbers of viable cells were determined by counting numbers of intact nuclei. Survival data are expressed as a percentage of cells plated at time 0 ± S.E. (n = 3).

Naive PC12 cells were washed five times on the dish and, after mechanical detachment, subjected to three cycles of centrifugation/resuspension, all with serum-free RPMI 1640 medium. The washed cells were replated onto fresh collagen-coated 100 mm dishes in serum-free RPMI 1640 medium and in the presence/absence of the appropriate survival agent for various times. In the case of deferoxamine experiments, 16-h pretreatment was carried out in the presence of full serum-containing RPMI 1640 medium. Neuronally differentiated PC12 cells were washed on the plate six times with serum- and NGF-free RPMI 1640 medium and incubated in the same medium containing the presence/absence of appropriate agents for various times. After incubation, cells were harvested, extracted, and c-Jun kinase activity was affinity-purified from cell extracts with recombinant GST-c-Jun as described previously (25, 43). The amount of cellular extract used for c-Jun kinase purification was normalized for protein content (400 µg/sample) as quantified by the Bio-Rad protein assay. c-Jun kinase activity recovered from each sample was determined by an in vitro solid-phase kinase radioassay as described previously (43). GST-c-Jun was resolved on 8.5% SDS gels and incorporation of ³²PO₄ into this substrate was quantified by autoradiography and densitometry. To assess the effects of survival agents on JNK activity in vitro, activated c-Jun kinase was isolated from PC12 cells after treatment with 0.5 mM sodium arsenite for 30 min and solid-phase kinase assays were carried out in the presence of constant amounts of activated enzyme and with/without the indicated agents.

RESULTS

To determine the point at which BCL2 prevents cell death of trophic factor-deprived PC12 cells in relation to JNK activation, we examined the levels of JNK activity in serum-withdrawn cultures of wild-type PC12 cells and in PC12 cells stabily transfected with a plasmid containing the neomycin resistance gene with (PC12-BCL2 cells) or without (PC12-2A9) BCL2. As shown in Fig. 1A, serum deprivation resulted in approximately 50% death of wild-type and PC12-2A9 cells by 1 day and 75% death by day 2, whereas NGF treatment completely blocked death. As reported previously (34), PC12-BCL2 cells did not die in the absence of trophic support (Fig. 1A), even after 5 days (data not shown). Consistent with past findings (25), serum deprivation elevated JNK activity in wild-type and control PC12-2A9 cells within 2 h (Fig. 1B). This increase reached a maximum level of approximately 4-20-fold by 4 h of serum withdrawal and was suppressed by NGF treatment (Figs. 1B and 2-5). In contrast to wild-type and control PC12-2A9 cells, there was no increase in JNK activity in serum-deprived PC12 cells overexpressing BCL2, even after 5 days (Fig. 1B and data not shown).

zVAD-fmk, an inhibitor of ICE-like activity, promotes survival of naive and neuronally differentiated PC12 cells but does not attenuate JNK activation after withdrawal of trophic support. A and B, effect of zVAD-fmk on survival of naive (A) or neuronal (B) PC12 cells after withdrawal of serum and NGF/serum, respectively. Each data point is the mean ± S.E. (n = 3) and is relative to the number of cells initially plated. Determinations of survival and JNK activity were performed on the same sets of cells. C and D, effect of zVAD-fmk on JNK activation. Naive (C) or neuronal (D) PC12 cells were deprived of serum and NGF, respectively, for the indicated times. zVAD-fmk and NGF were added immediately following serum/NGF deprivation where indicated. JNK activity was determined using GST-c-Jun protein as substrate as described under ``Experimental Procedures.'' The figure shows autoradiograph depicting levels of ³²PO₄ incorporation into GST-c-Jun during the JNK assay. The densitometric values of the autoradiographic signals indicate phosphate incorporation in cell extracts containing equal amounts of protein at each time point and are normalized such that the zero time point is defined as 1.

Parallel experiments were carried out with BCL2-overexpressing PC12 cultures that had been neuronally differentiated by pre-exposure to NGF in serum-free medium for at least 1 week and then subjected to NGF withdrawal. However, in this case, BCL2 overexpression did not confer protection from death (nor did it block JNK activation). Thus, for neuronal PC12 cells, we were unable to draw conclusions about the relationship between the effects of BCL2 on death and on JNK activation.

Treatment of wild-type PC12 cells (37) as well as of neurons with inhibitors of ICE-family proteases (14, 44) protects them from death caused by withdrawal of trophic support. The permeant tripeptide-based inhibitor zVAD-fmk promotes the survival of PC12 cells (37) after withdrawal of either serum (Fig. 2A) or NGF (Fig. 2B). As with serum deprivation, NGF withdrawal leads to significantly enhanced JNK activity by 4 h (Figs. 2D, 4, and 5; Ref. 18). In contrast to NGF treatment or BCL2 overexpression, zVAD-fmk does not block activation of JNK in response to withdrawal of serum or of NGF. These results indicate that in the PC12 cell system, BCL2 appears to act upstream of JNK activation and that ICE-like proteases, in contrast, appear to be required for cell death at a point either downstream of JNK activation or on an independent pathway required for cell death.

In addition to NGF, BCL2, and ICE family inhibitors, we and others have uncovered a variety of agents that promote survival of naive and neuronally differentiated PC12 cells as well as sympathetic neurons after withdrawal of trophic support. To determine whether these lie upstream of JNK activation, we applied these agents to both serum-deprived naive PC12 cells and NGF/serum-deprived neuronally differentiated PC12 cells under conditions in which they prevent death and then measured JNK activity. In these and other experiments, the magnitude of JNK activation caused by NGF withdrawal from neuronally differentiated cells was consistently severalfold lower than for naive cells deprived of serum. This appears to reflect a somewhat higher basal level of JNK activity in control neuronally differentiated cultures. As shown in Figs. 3 and 4, the survival promoting agents DETA·NO (a nitric oxide generator), ATA, NAC, 8-bromo-cGMP, and 8-(4-chlorophenylthio)-cAMP (CPT-cAMP) all significantly attenuated induction of JNK activity. These observations indicate that each of these agents acts prior to induction of JNK activity in the pathway that leads to apoptotic death.

Previous work has shown that the survival-promoting actions of NAC, but not of other survival agents, in serum-deprived PC12 cultures are blocked by actinomycin D (41). Accordingly, we examined whether inhibition of JNK activation by NAC is also transcription-dependent. Naive PC12 cultures were deprived of serum in the presence or absence of NAC, NGF, and CPT-cAMP and of 10 µM actinomycin D. Although actinomycin D did not block the ability of NGF and CPT-cAMP to suppress JNK activation, JNK activity in cultures treated with NAC and actinomycin were even higher than in control cultures (data not shown). These results indicate that inhibition of both death and JNK activation by NAC require transcription.

We previously reported that the Cdk/cell cycle inhibitor, flavopiridol, promotes survival of NGF-deprived neuronal PC12 cells, but does not block elevation of JNK activity under these conditions (25). We therefore tested deferoxamine (DFX), an agent that arrests cell cycle by a mechanism that is not completely understood and that suppresses death of PC12 cells and sympathetic neurons (39). PC12 cells were pretreated overnight with 1 mM DFX and then deprived of serum in presence of the drug. As shown in Fig. 5A, despite treatment with DFX, JNK activation was similar to that in untreated control cultures. Fig. 5C shows survival measured with the same set of cells which verifies the survival promoting activity of DFX in this experiment.

Deferoxamine and actinomycin do not attenuate JNK activation but promote survival of PC12 cells after withdrawal of trophic support. A, effect of deferoxamine on JNK activation in serum-deprived cultures of naive PC12 cells. Those cells treated with deferoxamine after serum withdrawal were also pretreated with this drug overnight. B, effect of actinomycin on JNK activation in NGF/serum-deprived cultures of neuronally differentiated PC12 cells. PC12 cells were deprived of serum or NGF/serum for the indicated times, and the appropriate survival agent was immediately added where indicated. JNK activity was determined using GST-c-Jun protein as substrate as described under ``Experimental Procedures.'' The figure shows autoradiograph depicting levels of ³²PO₄ incorporation into GST-c-Jun during the JNK assay. The densitometric values of the autoradiographic signals indicate phosphate incorporation in cell extracts containing equal amounts of protein at each time point and are normalized such that the zero time point is defined as 1. C, effect of deferoxamine (deferox.) on survival of PC12 cells deprived of serum. D, effect of actinomycin D (actin.) on survival of neuronally differentiated PC12 cells deprived of NGF. Each value is the mean ± S.E. (n = 3) and is expressed relative to the number of cells initially plated. Survival experiments were carried out on same sets of cells used to measure JNK activity.

Actinomycin D suppresses death of neuronally differentiated PC12 cells and sympathetic neurons induced by removal of NGF (17). The data in Fig. 5, B and D, show that in contrast to its survival promoting actions in the same experiment, actinomycin did not prevent the 3-fold induction of JNK activity brought about by NGF withdrawal. In fact, exposure to actinomycin appeared to somewhat increase JNK activity, either in the absence (Fig. 5B) or presence (data not shown) of NGF, indicating that this agent may induce cellular stress. Taken together, our findings suggest that DFX and actinomycin, as zVAD-fmk, but unlike the other molecules discussed above, promote survival at a point either after JNK activation or, alternatively, on a separate death signaling pathway.

Although zVAD, actinomycin, and deferoxamine do not prevent the intracellular activation of JNK activity (as discussed above), it was possible that these agents may act to directly inhibit JNK activity in cells. To evaluate this, JNK was activated in PC12 cells by exposure to arsenite, affinity-purified, and assessed for activity in the presence and absence of each compound at the concentrations used to prevent cell death. As shown in Fig. 6, none of the three agents directly affected JNK activity.

DISCUSSION

Previous studies have shown that multiple molecules prevent the death of naive and neuronal PC12 cells deprived of trophic support (25, 28, 29, 35, 39, 40, 42). Our current study focuses on ordering these relative to JNK activation. The evidence presented here and of others (18) suggests that although the latter maybe necessary, it is not sufficient for PC12 death under conditions of trophic withdrawal.

BCL2 has been shown to protect assorted cell types from death evoked by various stimuli (45, 46, 47). In particular, this protein suppresses death of PC12 cells (35, 36) and sympathetic neurons (45) induced by withdrawal of trophic support. We report here that BCL2, in serum-deprived PC12 cell cultures, prevents, and therefore acts upstream of, JNK activation.

The mechanism(s) by which BCL2 prevents cell death is not well understood. It has been suggested that it may act by regulating levels of reactive oxygen species. Hockenberry et al. (48) and Kane et al. (46) showed that BCL2 blocks lipid peroxidation in dexamethasone-treated 2B4 cells and glutathione-depleted GT1-7 neural lines, respectively. More recent reports, however, demonstrate that BCL2 blocks apoptotic death of PC12 cells and several lymphoid cell lines under hypoxic conditions in which generation of reactive oxygen species would be minimal (47, 49). This argues against the likelihood that the sole mechanism of BCL2 action is interference with reactive oxygen species mechanisms. An alternative is that BCL2 acts at a relatively proximal position in one or more apoptotic signaling pathways. Xia et al. (18) have hypothesized that at least in PC12 cells, the balance between ERK and JNK signaling may determine whether a cell undergoes apoptosis and that trophic factor withdrawal triggers death by activating JNK and reducing ERK activity. In this context, it is intriguing to speculate that BCL2 may influence upstream signaling cascades (either by blocking an activating pathway or stimulating an inhibitory pathway), thereby preventing JNK activation and apoptosis. In support of this, BCL2 protects against apoptotic responses in cells without nuclei (50) and has been reported to associate with R-Ras (51) and RAF-1 (52).

Our study also addressed the relative positions of JNK activation and the ICE family proteases. To do so, we employed zVAD-fmk, an inhibitor of ICE family proteases that blocks PC12 cell death. To date, at least six members of the ICE family of cysteine proteases have been described (5, 12, 53, 54, 55, 56). Although the identity of the specific family member(s) responsible for the death of trophic factor-deprived PC12 cells is presently undefined, the poor efficacy of the more specific ICE inhibitor, acYVAD-cmk, in rescuing the cells, and the absence of enhanced interleukin- 1 release in response to trophic factor withdrawal (37) indicates that ICE itself is unlikely to be a candidate. This appears to rule out the possible scenario that JNK activation in trophic factor-deprived PC12 cells is a consequence of ICE activation and interleukin-1 production.

Our experiments demonstrated that zVAD-fmk does not affect JNK activation. This indicates that stimulation of the JNK signaling cascade in response to serum/NGF withdrawal does not require ICE family proteases. This in turn implies either that the ICE family protease actions lie downstream of the JNK signaling cascade or that the two lie on independent pathways, both of which are required for death.

The contrasting actions of BCL2 and zVAD-fmk on JNK activation also indicate potential orders of these in apoptotic death pathways. It has been reported that BCL2 counteracts the apoptotic effects of ICE over-expression (13). This, coupled with our present findings, indicate the following: 1) if BCL2, JNK, and ICE proteases lie in the same pathway, then BCL2 is upstream of ICE-like proteases, and the two are separated by at least the JNK signaling cascade (Fig. 7A); 2) if JNK and ICE proteases lie in separate pathways, then each pathway is subject to regulation by BCL2 (Fig. 7B).

In addition to BCL2 and an ICE family inhibitor, a number of other agents promote PC12 cell survival by what appears to be diverse mechanisms. To further understand the roles and relative positions of these agents in the pathways that govern cell death and survival, we also examined their effects of JNK activation.

ATA inhibits death of a variety of cell types,including trophic actor-deprived PC12 cells and sympathetic neurons (29). This agent is an endonuclease inhibitor and has been speculated to act at a relatively late step in apoptotic death (i.e. to prevent DNA fragmentation by endonucleases) (29, 57). Surprisingly, ATA was found to act prior to JNK activation. Recent reports, however, demonstrate that ATA stimulates phosphatidylinositol 3-kinase, phospholipase C, and ERK phosphorylation in PC12 cells (58). This suggests that ATA acts upstream of JNK by evoking signaling events that repress its activation.

We showed recently that NAC promotes PC12 cell survival by a mechanism that is independent of its capacity to up-regulate intracellular GSH or to act as an anti-oxidant (41). This compound increases ERK activity and phosphorylation in PC12 cells, raising the possibility that NAC blocks JNK activation by regulating an upstream signaling pathway (59).

CPT-cAMP prevents death of PC12 cells and neurons by a mechanism dependent on activation of protein kinase A (PKA) (28, 60), while NO generators and 8-bromo-cGMP appear to promote survival by a mechanism involving elevation of intracellular cGMP (42). It thus appears that the signaling cascades regulated by PKA and cGMP interfere with the pathway responsible for activating JNK in response to loss of trophic support. Although ATA, NAC, PKA, and cGMP all regulate early signaling pathways and suppress JNK activation, it is not clear whether any or all intersects with the JNK signaling pathway at the same or different points.

Unlike the above molecules, the cell cycle blocker deferoxamine appears to act either downstream of JNK or on an independent pathway required for death. We recently observed that flavopiridol and olomoucine, two CDK inhibitors that block cell cycle progression and promote survival of neuronal PC12 cells, also fail to prevent JNK activation after NGF withdrawal (25). We and others have hypothesized that trophic factor-deprived neurons undergo apoptosis, because they inappropriately attempt to re-enter the cell cycle (25, 61, 62). In accordance with this, agents such as deferoxamine (39), flavopiridol (25), and olomoucine (25) that block cells in G₁/S, promote neuronal survival. However, it may not be that cell cycle arrest per se promotes survival by a mechanism independent or downstream of JNK. Both NAC (40) and CPT-cAMP (39) have been suggested to block death by a cell cycle mechanism since each inhibits PC12 cells from undergoing DNA synthesis at the same concentrations at which it prevents death. If this is correct, then it is the mechanism by which any given agent arrests cell cycle that correlates with its position relative to JNK activation rather than simply its ability to block cell proliferation.

The mechanisms by which JNK activation participates in cell death are unknown. Previous findings indicate that death of trophic factor-deprived neuronally differentiated PC12 cells, as in the cases of neurons (15, 16), appears to require transcription (17). We currently report that activation of JNK in response to trophic factor deprivation occurs even in the presence of an actinomycin D concentration that blocks greater than 90% of transcriptional activity in PC12 cells (63). This implies that JNK activation occurs by a post-translational mechanism and is not a consequence of genes regulated in response to trophic factor withdrawal. In addition, our studies indicate that the JNK signaling cascade lies either upstream of the transcriptional activity required for death or on an independent pathway. One mechanism by which JNK may contribute to neuronal cell death is by activating the transcription of apoptotic signaling genes. Consistent with this, JNK phosphorylates and enhances the transcriptional activity of c-Jun (21), and blockade of c-Jun activity in NGF-deprived sympathetic neurons by anti-c-Jun antibodies (20) or by ectopic expression of dominant-negative forms of c-Jun (18, 19) inhibits their apoptotic death.

This model, however, appears to be at odds with the observed characteristics of certain other cell death systems. As an example, death of serum-deprived naive PC12 cells does not require new synthesis (28). This suggests that if JNK activation is required for death of such cells, then it must act by a mechanism that does not include new gene transcription. Similarly, Verheij et al. (64) reported that death of U937 cells caused by a variety of stresses requires JNK activation, but is not blocked by inhibition of protein synthesis. An alternative mechanism raised by these authors is that activated JNK phosphorylates c-Jun, which in turn sequesters other transcription factors whose activities are required for survival. We have observed, however, that promotion of survival of serum-deprived naive PC12 cells by a number of other agents, including NGF and CPT-cAMP, does not require transcription (28). Therefore, a third alternative mechanism is that once activated, JNK phosphorylates and modulates the activities of other required components of a cell death pathway. Among such targets may be molecules that regulate the cell cycle and/or ICE family members.

In summary, we have identified a number of molecules, including BCL2 that appear to block death of trophic factor-deprived PC12 cells by acting at points upstream of JNK activation. In contrast, several other survival-promoting agents, including an inhibitor of ICE family proteases and actinomycin D act either downstream of JNK (Fig. 7A) or on independent pathways (Fig. 7B). These observations permit an initial ordering of elements involved in cell death.