Temporal Phases in Apoptosis Defined by the Actions of Src Homology 2 Domains, Ceramide, Bcl-2, Interleukin-1β Converting Enzyme Family Proteases, and a Dense Membrane Fraction

The preparation, fractionation, and use of Xenopus egg extracts were as described (28, 29). In experiments with unfractionated extracts, 50 μl of extract was mixed with ATP-regenerating system (consisting of 2.5 μl of a 200 mM solution of phosphocreatine, 1 μl of a 100 mM solution of ATP, pH 7.0, and 1.5 μl of a solution of creatine phosphokinase (5 mg/ml; Sigma Chemical Co., St. Louis). Then, 1–3 μl of a suspension of rat liver nuclei (at ∼10⁶ nuclei per μl) was added, as well as 1 μl of rhodamine-labeled nuclear import substrate (TRITC-HSA–NLS [a TRITC-labeled conjugate of human serum albumin with a synthetic nuclear localization signal peptide]) to monitor nuclear intactness, as described (28). GST–SH2 domain fusions and synthetic peptides were then added from stock solutions at 20–50 times the final concentration. HM-depleted extracts were reconstituted from 45 μl of egg cytosol and 5 μl of light membrane suspension obtained from M-phase extracts, as described (28).

The expression and purification of GST fusion proteins were as described (4). Briefly stated, the GST–SH2 domains of Lck were produced by PCR amplification of the desired region of wild-type (Y192), F192, and E192 Lck constructs (amino acids 121–224), using the proofreading Vent DNA polymerase (New England Biolabs, Beverly, MA). The PCR fragments were inserted in frame into the pGEX-3T prokaryotic expression vector (14), and the resulting GST–SH2 fusion proteins were purified by glutathione–Sepharose affinity chromatography. The fusion proteins were dialyzed (using a microdialysis apparatus [Pierce Chemical Co., Rockford, IL]) against a buffer containing Hepes (20 mM) and KCl (50 mM) before use. Synthetic peptides were obtained from Research Genetics (Huntsville, AL).

Phosphorylation of the synthetic peptide (RNLDNGGFYIAPR) comprising the site surrounding Y192 in Lck was assayed essentially as described previously (27): peptide (4 mg/ml), γ-[³²P]ATP (1 mCi/ml), and Xenopus egg extract (7.4% vol/vol, corresponding to a protein concentration of ∼2.6 mg/ml) were incubated at 22°C for the times indicated, in 25 μl of a buffer containing 50 mM Hepes/K, pH 7.4, 10 mM MgCl₂, 5 mM MnCl₂, and 1.2 mM Na₃VO₄. The reaction was stopped with the addition of 155 μl of 3.2% TCA and 20 μl of a solution of BSA (10 mg/ml). After 30 min on ice, the samples were centrifuged, and 100 μl of each supernatant was applied to a 2.5-cm circle of phosphocellulose membrane (Whatman P-81). The filters were washed seven to eight times in 75 mM o-phosphoric acid in 50-ml conical tubes. After drying, the filters were placed in 4 ml of scintillation fluid and the radioactivity was counted. All the results presented here were qualitatively reproducible. However, the time at which nuclear destruction began to occur varied from extract to extract, making it impossible to average the results from separate experiments in a meaningful way. Each figure is representative of at least four experiments giving similar results.

To investigate whether SH2–PTyr interactions are involved in the cell-free apoptosis system, we prepared GST fusion proteins containing the SH2 domains from several signaling proteins and added them to the extracts. Fig. 1 shows that the SH2 domains from c-Src and c-Abl produced a concentration-dependent inhibition of apoptosis when present at concentrations of ⩾10 nM. A similar effect was seen when fusions of GST with the SH2 domains of Grb-2 and Stat1 were added (not shown).

To determine whether the effects of SH2 domains added to the cell-free system were based on specific interactions with phosphorylated ligands, we took advantage of point mutations in the SH2 domain of Lck that disrupt PTyr-dependent binding. Previous studies (4) had shown that the function of the Lck SH2 domain can be regulated by modification of a particular amino acid residue, Y192. Phosphorylation at this site, which is located next to the EF loop of the SH2 domain, markedly lowers the affinity of the SH2 domain for phosphotyrosine-containing protein ligands. Mutants in which this Tyr residue was changed to Phe (F192) and Glu (E192) had been found to mimic the unphosphorylated and phosphorylated states, respectively, of Y192. We added the wild-type and mutant Lck SH2s to the Xenopus cell-free system. As shown in Fig. 2, the SH2 domain from the constitutively inactive Lck-E192 mutant was unable to inhibit apoptosis. On the other hand, the Lck-F192 mutant SH2, which binds constitutively to phosphotyrosine-containing ligands, was as potent an inhibitor as the Src and Abl SH2 domains.

The wild-type Lck SH2 domain, like the nonbinding E192 mutant, was unable to inhibit apoptosis. This result could be explained if we supposed that the wild-type Lck SH2 domain became phosphorylated on Y192 by a kinase present in the extract and consequently had a lowered affinity for the tyrosine phosphorylated ligand, similar to the E192 mutant. This hypothesis predicted that if phosphorylation on Y192 were prevented, the wild-type SH2 domain would function like Lck-F192; i.e., it would inhibit apoptosis in the cell-free system. To test this prediction, we added the tyrosine kinase inhibitor, genistein, either separately or with the wild-type Lck SH2 domain. Genistein, added by itself, had only a slight effect on the kinetics of nuclear destruction in the apoptotic cell-free system (Figs. 3 and 4,B; this suggests that any critical tyrosine phosphorylation events involving proteins endogenous to the extract had already occurred before the addition of genistein). However, the coaddition of genistein and the wild-type Lck SH2 domain produced a marked inhibitory effect, similar to that of the F192 mutant (Fig. 3,A). Therefore, the Lck SH2 domain is likely to be phosphorylated on Y192 by an endogenous protein tyrosine kinase in the Xenopus egg extracts, and thus inactivated. To examine this more directly, we assayed the phosphorylation of a synthetic peptide corresponding to the sequence surrounding Y192. Fig. 3,B shows that this peptide is phosphorylated by the extract in a time-dependent manner; the specific activity was estimated at 7.7 ± 0.2 pmol per min per mg of extract protein. Thus, a kinase is present in the extract that is able to phosphorylate this site in the Lck SH2 domain. We conclude from the experiments shown in Figs. 1–3 that the effects of the exogenous SH2 domains are dependent on specific interactions with endogenous tyrosine-phosphorylated ligands.

To confirm that PTyr–SH2 domain interactions are important for apoptosis in the cell-free system, we examined the effects of adding an exogenous synthetic phosphopeptide, EPQY*EEIPIYLK (abbreviated Y*EEI; Y* represents PTyr), known to interact with Src family SH2 domains (41). Fig. 4,A shows that this peptide was able to inhibit apoptosis even at concentrations as low as 1 nM. In control experiments (Fig. 4,B), we found that a nonphosphorylatable mutant peptide, EPQFEEIPIYLK (FEEI), was ineffective, arguing again that the inhibitory effect is dependent on the specific interaction of SH2 domains with sequences containing PTyr residues. The effect of adding the unphosphorylated peptide (YEEI) was similar to that seen with Y*EEI, suggesting that YEEI can be phosphorylated by kinases endogenous to the extract (Fig. 4,B). In support of this idea, we found that the protein tyrosine kinase inhibitor, genistein, was able to reverse the inhibitory effect of YEEI (Fig. 4,B). To probe the specificity of these effects further, we tested another synthetic phosphopeptide, corresponding to the SH2 domain–binding region surrounding residue Y1021 of phospholipase-Cγ1 (DNDY*IIPLPDPK; 35). This peptide had no significant effect on the apoptotic process in the cell-free system, even at concentrations as high as 1 μM (see Fig. A1). This is additional evidence that interactions involving specific SH2 domains are important in the Xenopus system.

To determine when in the apoptotic process PTyr–SH2 interactions are important, we performed experiments in which the Src SH2 domain–GST fusion was added to the cell-free system at various times after the start of incubation (Fig. 5). As before (Fig. 1), the Src SH2 domain slowed the kinetics of nuclear destruction markedly when added at the start of incubation. However, when the addition of this SH2 domain was delayed by intervals ranging from 30–120 min, its inhibitory activity was gradually reduced. These results demonstrate that the SH2 domain interactions are important throughout the initial portion of the apoptotic process.

Previously, we showed that Bcl-2 can inhibit apoptosis by acting during the early cytoplasmic phase of the apoptotic process (29). To define more precisely the period during which Bcl-2 acts, we performed experiments similar to those described above. Bcl-2 was added at various intervals after the start of incubation, and the time course of nuclear survival was determined. Unlike the SH2 domains, Bcl-2 inhibited apoptosis effectively even if its addition was delayed up to ∼1.5 h (Fig. 6). Thus, it would appear that the events controlled by Bcl-2 occur later in the apoptotic process than those regulated by SH2 domain–PTyr interactions.

Our previous studies also identified a requirement for an HM fraction that was enriched in mitochondria but also contained other membrane material (29). Extracts depleted of heavy membranes were found to lack apoptotic activity, and the addition of HMs restored this activity in a dose-dependent manner. To determine when the HM fraction is required during the time course of the apoptotic process occurring in the cell-free system, we performed an experiment in which the HM fraction was added at various times after the start of incubation. Fig. 7 shows that delaying the addition of the HM fraction by as much as 1.5 h had essentially no effect on the kinetics of nuclear destruction in the apoptotic extract. Only when the HM fraction was added at 2 h after the start of incubation was the time course of nuclear destruction delayed. We conclude that, like Bcl-2, the heavy membranes are required at a relatively late stage in the apoptotic process.

We also showed previously that a soluble ceramide analog could substitute for the HM fraction in the cell-free apoptotic system (24) This finding raised the possibility that the function of HMs was simply to produce ceramide. However, as Fig. 8 shows, ceramide acts earlier in the apoptotic process than the HMs. Like the GST–SH2 domain fusions, ceramide was most active when added at the start of incubation, and its potency declined gradually as the time of its addition was delayed. Thus, it is unlikely that the function of HMs is merely to produce ceramide.

Previous studies from many laboratories have demonstrated the involvement of cysteine proteases from the Ced-3/ICE family in the apoptotic process (see Introduction). To determine whether such proteases are involved in the Xenopus system, we tested the effects of aldehydebased tetrapeptide derivatives that have been shown to be specific inhibitors of ICE family proteases. The compound Ac-DEVD-CHO (DEVD) has been shown to inhibit CPP32-like proteases preferentially, while Ac-YVAD-CHO (YVAD) is more selective for ICE and its close relatives (30). Fig. 9 A shows that micromolar concentrations of DEVD block apoptosis in the Xenopus system. YVAD also inhibited apoptosis in this system, but concentrations ∼500-fold higher were required. These results are similar to those described by Nicholson et al. (1995) (30) for mammalian cells, and argue that a CPP32-like, rather than an ICE-like, protease activity is required for apoptosis in the Xenopus cell-free system. However, because these compounds can inhibit multiple caspases, it is possible that several such enzymes are involved in this system. After this manuscript was submitted, other studies showed directly that CPP32-like caspases are activated in Xenopus egg extracts (3, 18).

Using the same approach as described above, we investigated the effect of delaying the addition of DEVD. The results (Fig. 9 B) showed that the protease activity was required relatively late in the apoptotic process, near the end of the 1.5–2-h latent phase. In this experiment, when AcDEVD-CHO was added as late as 1 h after the start of the incubation, almost no reduction in its inhibitory activity was seen, and only a partial loss of activity was observed when the inhibitor was added after 1.5 h of incubation. In contrast, the inhibitor had no activity when added after 2 h.

In this report, we have shown that apoptosis in the Xenopus cell-free system is inhibited by the addition of SH2 domains from several signaling proteins (Figs. 1–3 and 5). Our results with the Lck SH2 domain (Figs. 2 and 3) demonstrate that the effect is specifically due to interaction with tyrosine-phosphorylated ligands, in that it is abrogated by phosphorylation on residue Y192 of Lck; phosphorylation at this site was shown elsewhere to disrupt binding of the SH2 domain to tyrosine-phosphorylated ligands (4). Similarly, the inhibition produced by the ligand peptide, YEEI, was dependent on phosphorylation of the tyrosine residue required for interaction with SH2 domains. Thus, our results show specificity in the sense that they reflect the PTyr-dependent binding of SH2 domains and their ligands. On the other hand, the exogenous SH2 domains appear to be acting in a generic manner, inasmuch as the SH2 domains from a number of proteins can block apoptosis in this system when present at concentrations of at least 10 nM. Our results therefore do not identify which particular SH2-containing protein(s) and corresponding tyrosinephosphorylated ligand(s) are involved in the cell-free apoptosis system.

The synthetic phosphopeptide Y*EEI inhibited apoptosis in the extract at lower concentrations than those required for the GST–SH2 domain fusions. The reason for this is unclear. Possibly these reagents act in different ways, e.g., by interacting in a regulatory manner with specific molecules. However, we consider it likely that these molecules act by disrupting PTyr-dependent interactions between proteins endogenous to the extract. What could be the function of these endogenous proteins? SH2 domains are frequently found in signaling molecules that regulate cell proliferation. Such molecules include tyrosine kinases, tyrosine phosphatases, and the adapter molecules (e.g., Grb2) that link these enzymes to other members of signaling cascades. Grb3-3, a naturally occurring variant of Grb2 that contains a truncated, nonfunctional SH2 domain, was found to promote apoptosis, presumably by acting as a dominant negative inhibitor of a cell survival pathway (10). Our results (Figs. 1–5) are just the opposite: i.e., they suggest that some SH2 domains can be involved in promoting cell death, not survival. We suggest that the fate of the cell may rest in the balance between apoptotic and survival pathways, both controlled in part by SH2 domain interactions. After this manuscript was submitted, it was reported that c-Crk, an adapter protein containing an SH2 domain, is required for apoptosis in Xenopus egg extracts, and that the SH2 domain of Crk inhibits apoptosis potently in this system (8). The activity of Crk could account for our results; however, an involvement of other proteins containing SH2 domains cannot be excluded at present.

In view of our findings that suggest a role for PTyr–SH2 domain interactions in apoptosis, it may seem surprising that genistein, an inhibitor of multiple tyrosine kinases, has little or no effect on apoptosis in the Xenopus system. This paradox can be resolved if we suppose either that the critical phosphorylation events have already taken place before the preparation of the extracts, or that the kinases responsible for these critical phosphorylation events are insensitive to genistein. We have also observed that pervanadate (1–10 mM), an inhibitor of protein tyrosine phosphatases, inhibits apoptosis effectively in the Xenopus system (unpublished data; this inhibition is not merely due to the presence of peroxide in the pervanadate preparation, as control experiments with added H₂O₂ [10 mM] showed no effect on the kinetics of apoptosis). This result may suggest that tyrosine dephosphorylation of specific proteins is required for the apoptotic process in the Xenopus system. However, it is also possible that pervanadate blocks apoptosis by inhibiting activities other than phosphatases.

The role of the heavy membrane fraction is still under investigation. Previously we reported that a soluble ceramide analog can induce apoptosis in HM-depleted extracts (24). Further experiments have shown that this effect is apparently due to the synergistic effect of ceramide and a small amount of contaminating heavy membranes still remaining in the HM-depleted extracts (Farschon, D.M., and D.D. Newmeyer, unpublished results). We note that both the production of ceramide and the molecules that respond to ceramide are reported to reside in the plasma membrane (e.g., 22). If so, to explain the activity of ceramide in HMdepleted extracts, we would need to postulate that some amount of contaminating plasma membrane–derived material is present. The HM fraction could also contain a certain amount of such membranes. Thus, we considered the possibility that the function of our HM preparation was simply to increase the amount of plasma membrane–producing ceramide. However, as Figs. 7 and 8 show, time- dependent changes occur in the cell-free system that at first diminish its responsiveness to ceramide and later make it responsive to the HM fraction. Thus, it is unlikely that the function of HMs is merely to produce ceramide. Recent studies have shown that, during apoptosis, mitochondria release cytochrome c into the cytosol, leading to the activation of CPP32-like proteases. Bcl-2, through its interactions with the outer mitochondrial membrane, blocks the efflux of cytochrome c (18a, 18, 48). Whether ceramide can directly or indirectly modulate the release of cytochrome c from mitochondria is still unknown.

Apoptosis could be envisioned as occurring in three stages. The first of these, “initiation,” would refer to the events that cause entry into the common death pathway. The second stage, which might be termed “sentencing,” encompasses the intracellular events that commit the cell irreversibly to the death process. Finally, the cell enters the “execution” stage, in which effector molecules, such as particular nucleases and proteases, accomplish the overt changes associated with apoptotic cell death. Each of these hypothetical stages could itself involve a series of events. To understand the apoptosis machinery, it will be necessary to know both the identities of the molecular participants in the process and whether these molecules act in sequence or in parallel.

In this report we have investigated the function of several components of the apoptotic machinery: SH2 domains, the HM fraction, ceramide, ICE family proteases, and Bcl-2, particularly with regard to the time intervals during which these reagents could affect the kinetics of nuclear destruction in the cell-free system. The events we examined fell into three main time windows, as depicted in Fig. 10. We propose that these experimentally defined time periods correspond to the three conceptual phases of apoptosis listed above.

The earliest distinguishable phase in the cell-free system, which we call “initiation,” occurs in the first ∼60–90 min of incubation. (Events occurring before lysis of the eggs may of course also be important for apoptosis initiation, but are outside the scope of our investigations.) The importance of SH2 domains and ceramide in this period (Figs. 5 and 8) is consistent with the expectation that these molecules would function in signal transduction. The second period, which occurs ∼90–120 min after the start of incubation, is the time when Bcl-2 protein, the heavy membrane fraction, and the CPP32-related protease(s) act (Figs. 6, 7, and 9). We propose that this represents a “sentencing” phase, during which the system becomes committed irreversibly to the apoptotic pathway. Finally comes the third, or “execution” phase, beginning after ∼1.5–2 h of incubation. This is the period when overt changes occur in the cell nuclei, culminating in extensive DNA fragmentation and nuclear destruction. During this phase, we also observed the cleavage of two polypeptides whose proteolysis is characteristic of apoptotic cell death (25, 33): fodrin, a cytoskeletal protein endogenous to the extract, and lamin B, present in the rat liver nuclei added to the system (18a). Our kinetics data suggest that the cleavage of fodrin and lamin B occur later than the proteolytic events inhibited by DEVD. Thus, this system appears to involve at least two distinct protease activities, which may belong to an enzymatic cascade.

It is important to note that the timing of apoptotic events in the Xenopus system is somewhat variable from extract to extract. This may be due, for example, to variation between animals in the proportion of eggs that are undergoing apoptosis. In extreme cases, the onset of nuclear destruction in a given extract can be as early as 45 min or as late as 4 h after incubation has begun. However, in the majority of experiments, the various events occurred within ± 30 min of the times indicated in Fig. 10. This means that, while the data in this report show that Bcl-2, the HM fraction, and DEVD-inhibitable protease activity all act in roughly the same time period, other approaches have been required to determine the sequence in which they act. Studies have now shown that mitochondria release cytochrome c into the cytosol, and that this event is blocked by Bcl-2 (18, 48); cytochrome c, when present in the cytosol, causes the activation of CPP32-related caspases (23, 18a).

The occurrence of kinetically distinguishable phases in apoptotic extracts now serves to underscore the value of this cell-free system for dissecting the crucial intracellular events involved in apoptosis. In particular, we expect this system to be useful in uncovering both the activation mechanisms and the critical substrates of the CPP32/Ced-3 family proteases.

We are indebted to Rajasekaran Baskaran and Jean Wang for helpful advice and gifts of Abl, Src, and Stat1 SH2 domain GST fusions, as well as to David Jones for a GST fusion of the SH2 domain of Grb-2. Gratitude is expressed to our colleagues Ruth Kluck and Doug Green for advice and comments on the manuscript.

This work is supported by grants to D.D. Newmeyer from the American Cancer Society (DB-97) and National Institutes of Health (NIH) (GM50284) and to T. Mustelin from NIH (GM48960 and AI35603).