Inhibition of Ced-3/ICE-related Proteases Does Not Prevent Cell Death Induced by Oncogenes, DNA Damage, or the Bcl-2 Homologue Bak

Rat-1 fibroblasts expressing the β-oestradiol and the 4-hydroxytamoxifen (4-OHT)–dependent conditional alleles of c-Myc (Rat-1/c-MycER and Rat-1/c-MycER™ , respectively) were isolated as described (Littlewood et al., 1995). Rat-1 fibroblasts expressing Bak from a Gal-4 element under the control of the 4-OHT–activatable chimeric GalER-VP16 transcription factor (Braselmann et al., 1993) have also been described (Rat-1/Bak cells) (Chittenden et al., 1995), as have Rat-1/c-MycER cells constitutively expressing Bcl-2 (Rat-1/c-MycER/Bcl-2) (Fanidi et al., 1992). A further Rat-1 cell line, constitutively expressing the adenovirus E1A (12S) protein (Rat-1/E1A), was isolated by infection of Rat-1 cells with the recombinant retrovirus pBabeNEO containing the entire open reading frame of 12S E1A under the control of the virus LTR and subsequent selection and cloning. Expression of 12S E1A was verified by immunoblotting. Rat-1 cells expressing E1A (12S) or empty vector controls were washed twice in serum-free medium and subsequently cultured in serum-free conditions. Cells were observed for 48 h using time lapse videomicroscopy. Rat-1/E1A cells exhibited marked apoptosis compared to controls demonstrating the ability of E1A to induce apoptosis in the absence of serum. Rat-1/cMycER™ , Rat-1/c-MycER/Bcl-2, and Rat-1/E1A cell lines were maintained in DME supplemented with 10% FCS and 1 mg/ml gentamicin. Rat-1/Bak cells were maintained in phenol-red–free DME supplemented with 10% charcoal-dextran–stripped FCS and 1 mg/ml gentamicin. Cells were selected using puromycin at 5 μg/ml for Rat-1/c-MycER™ , neomycin at 1 mg/ml for Rat-1/E1A cells, and with puromycin at 5 μg/ml and neomycin at 1 mg/ml for Rat-1/Bak cells. Cells were passaged by standard trypsinization and seeded directly onto tissue culture plastic. Cells were serum-depleted by culture in 0.05% FCS for 48 h. Both Myc and Bak expression were induced by addition of 4-OHT to the cell cultures to a final concentration of 100 nM.

4-OHT was obtained from SEMAT (St. Albans, UK). A 1 mM stock solution in 100% ethanol was prepared, stored at −20°C, and diluted in DME as required. ZVAD.fmk and BD.fmk were obtained from Enzyme Systems Products (Dublin, CA). Stock solutions in DMSO were prepared at 50 mM, stored at −80°C, and used as required. Etoposide/VP16, phalloidin, and 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) were obtained from Sigma (Poole, UK). FITC-labeled annexin V was obtained from Bender & Co. GmbH (Vienna, Austria). Secondary antibodies and enhanced chemiluminescence were obtained from Amersham International (Amersham, UK).

Polyclonal antibody to adenovirus E1A protein was the gift of Dr. N. Jones, Imperial Cancer Research Fund, London, UK (ICRF). Polyclonal antibody to PARP (C-2-10) was the gift of Dr. W. Earnshaw (Edinburgh University, Scotland, UK). The anti–lamins A and C antibody (131C3) was the gift of Dr. Y. Raymond (Institut du Cancer de Montreal, Montreal, Canada), together with the control antibody MOPC 21C. The actin antibody was the gift of Dr. Fulvia Verde, ICRF.

Time-lapse videomicroscopic images were acquired on an inverted phase contrast microscope (Diaphot; Nikon, Inc., Mellville, NY) equipped with epifluorescence and collected on either BetaCam or sVHS video tape under the control of an external animation controller (EOS Electronics AV Ltd., Barry, Wales, UK) as described (Evan et al., 1992). Apoptotic cell deaths were scored at the point when they were fully detached and rounded. Cells with the “blebbing” phenotype were deemed to be alive while they remained adherent and actively blebbing and similarly scored as dead once fully detached and rounded.

Cleavage of poly(ADPribose) polymerase, nuclear lamins A and C, and actin was analyzed by SDS-PAGE and subsequent transfer to nitrocellulose membrane (Immobilon P, Amersham International). Membranes were blocked in TNT buffer (1% TBS, 0.5% Tween 20, and 5% skimmed milk) for 30 min and then incubated in antibodies to PARP (Lamarre et al., 1986), actin, or lamins A and C antibody at 4°C for 12 h in TNT buffer. Membranes were washed twice in TNT and then incubated for 1 h at room temperature (RT) with anti–mouse or anti–rabbit HRP-conjugated secondary antibody as appropriate. Finally, membranes were washed twice in TNT buffer and once in TBS and 0.5% Tween 20, and bound antibodies were visualized by enhanced chemiluminescence.

Standard electron microscopic techniques were used as follows. Cells were pelleted, fixed for 1 h at RT in 2.5% glutaraldehyde in Sorensens buffer, postfixed in osmium tetroxide in the same buffer, and embedded in Araldite resin. Sections were cut on a Reichert ultracut, stained with uranyl acetate and lead citrate, and viewed on an electron microscope (EM 10 CR; Carl Zeiss, Inc., Thornwood, NY). PARP expression and localization was determined immunocytochemically as previously described (Lamarre et al., 1986). Briefly, cells were grown on 8-well chamber slides and fixed with 100% methanol for 10 min at −20°C. Fixed cells were incubated with primary antibody for 60 min at RT in PBS followed by five washes in PBS buffer containing 0.1% Tween 20. After incubation in appropriate FITC-labeled secondary antibodies, cells were viewed using a confocal microscope (model MRC 1000; BioRad Labs, Hercules, CA). Lamins were identified immunocytochemically using anti–lamin A+C antibody (1:100), incubated with cells previously fixed as described above, and preblocked in 10% FCS and PBS. After three washes in PBS, cells were incubated with secondary antibody as described above. For actin staining, cells were fixed in 4% paraformaldehyde in PBS for 10 min, washed in PBS at RT, and then permeabilized with 0.2% Triton in PBS for 5 min. Cells were then rinsed in PBS and incubated with 0.1 μg/ml FITC-phalloidin in PBS for 20 min at RT. Finally, cells were washed several times in PBS, mounted, and viewed as above.

Mitochondrial dehydrogenase activity, used as a marker of cell survival, was assayed by the ability of cells to convert soluble MTT into an insoluble formazan reaction product (Mosmant, 1983). 10 μl of MTT solution (5 mg/ml in PBS) was added to each well of a 96-well plate containing 3,000 cells in 100 μl of medium. The reaction was stopped after 2 h at 37°C by removal of the supernatant and addition of 200 μl of acid propan-2-ol. The plates were left at RT for 15 min in the dark, and absorbance difference at 570–630 nm was measured using a microplate reader and % survival calculated as (Experiment − Blank)/(Control − Blank) × 100%, where the blank is medium + MTT without cells and the Control was the signal from wells with maximum cell survival, i.e., with serum and no 4-OHT.

Cell survival was also assayed by a direct microscopic assessment of the proportion of cells able to exclude the vital dye, trypan blue. Rat-1/cMycER™ cells were grown in 3-cm tissue culture dishes, changed to serum-free medium for 48 h, and then 4-OHT ± ZVAD.fmk/BD.fmk was added (time 0). At the time points indicated, cells were trypsinized and washed in DME, and the number of trypan blue–positive cells was assessed by counting a minimum of 200 cells on a hemocytometer.

Adherent cells were trypsinised, fixed in 70% ethanol, and stained with propidium iodide. Flow cytometric analysis was carried out on a BectonDickinson (Mountain View, CA) FACSstar plus.

Cells undergoing apoptosis in the presence or absence of ZVAD.fmk or BD.fmk were observed by time-lapse videomicroscopy. At the time that a significant proportion of the cells were either apoptoptic (no ZVAD.fmk/ BD.fmk) or blebbing (plus ZVAD.fmk/BD.fmk), the culture was flooded with FITC–annexin V. The same microscopic field as had been observed thus far was then examined by fluorescence and phase microscopy to detect bound annexin V and examine cell morphology, respectively.

Apoptosis can be triggered by a wide range of influences, including oncogene deregulation, DNA damage, and expression of killer genes. To determine the general relevance of IRPs to apoptosis, we examined the effect of the cell-permeable IRP inhibitors, ZVAD.fmk and BD.fmk, upon apoptosis induced in Rat-1 fibroblasts by a number of agents. Apoptosis was induced by activation of a deregulated conditional allele of c-Myc (Littlewood et al., 1995), expression of the adenovirus protein E1A (Rao et al., 1992), DNA damage in combination with c-Myc expression using the topoisomerase II inhibitor etoposide (Fanidi et al., 1992), or expression of the killer member of the Bcl-2 family, Bak (Chittenden et al., 1995). Both E1A (Debbas and White, 1993) and etoposide (Lowe et al., 1993) induce apoptosis via a p53-dependent mechanism; c-Myc may act partly via p53 (Littlewood et al., manuscript in preparation) and Bak killing is p53 independent (T. Chittenden, unpublished data).

The Rat-1 fibroblast cell lines were induced, activated, or treated with drug, as appropriate, in the presence or absence of ZVAD.fmk or BD.fmk. Cells were monitored by time-lapse videomicroscopy and scored as dead when they became detached from the substratum. Both ZVAD.fmk and BD.fmk afforded excellent protection from apoptosis in all cases, regardless of the trigger (data for ZVAD.fmk shown in Fig. 1). Thus, all these triggers of apoptosis act via a ZVAD.fmk- or BD.fmk–sensitive agent, presumably one or more of the IRPs.

As shown in Fig. 1, the degree of protection afforded by ZVAD.fmk or BD.fmk is profound. Nonetheless, under the sustained pro-apoptotic influence of c-Myc, E1A, Bak, or continuous DNA damage, all cells were eventually observed to detach from the substratum, undergo partial chromatin condensation, and eventually become permeable, even in the sustained presence of ZVAD.fmk or BD.fmk (see below). Neither higher levels of ZVAD.fmk or BD.fmk nor regular 24 hourly readdition of ZVAD.fmk or BD.fmk to the culture medium was able to delay the eventual demise of cells (data not shown), suggesting ZVAD.fmk and BD.fmk inhibition of cell death is eventually overwhelmed or bypassed. In all cases tested, the effects of ZVAD.fmk and BD.fmk were identical. Accordingly, primarily data for ZVAD.fmk is shown henceforth.

Time-lapse videomicroscopy permits observation both of the onset and the execution of individual apoptotic events within a cell culture. The onset of apoptosis is characterized by the sudden initiation of surface blebbing, together with cytoplasmic fragmentation and exfoliation, from which the name apoptosis derives. In fibroblastic cells, onset of blebbing is rapidly followed by nuclear condensation and collapse and total cell fragmentation, a process that typically takes between 30–60 min (Evan et al., 1992) (see Note Added in Proof for Web information). However, whereas an individual apoptotic event is rapid, individual cells within a culture, undergoing apoptosis induced by activation of c-Myc in low serum, demonstrate asynchronous initiation of apoptosis over a prolonged period. Since such apoptosis is cell cycle independent, the basis of this asynchronicity is unclear, although its incidence is greatly affected by levels of c-Myc protein and the presence of survival factors in the culture medium (Harrington et al., 1994).

Using time-lapse videomicroscopy to observe cells undergoing c-Myc–induced apoptosis in the presence of either ZVAD.fmk or BD.fmk, we were surprised to find that neither inhibitor appears to delay the initiation of apoptosis within the cultures, as judged by the onset of blebbing. Indeed, the kinetics of onset of apoptotic events within ZVAD.fmk-treated cultures were superimposable on those of untreated cell cultures (Fig. 2,A). However, unlike untreated cells, ZVAD.fmk- or BD.fmk-treated cells did not rapidly condense and fragment but continued to bleb and boil for extended periods (Fig. 2,B). During this extended blebbing, often lasting days, cells remain adherent, exclude trypan blue, and exhibit physical activity, all suggestive of sustained metabolism (see Note Added in Proof for Web Information). This latter notion was supported by assaying mitochondrial activity of ZVAD.fmktreated cells using the MTT assay, which showed continued mitochondrial dehydrogenase activity in such cells (Fig. 3 A).

Thus, ZVAD.fmk-treated cells appear to initiate apoptosis normally but then become arrested within the apoptotic program. Identical results were obtained when apoptosis was induced by E1A, etoposide, or Bak expression (data not shown).

To investigate further the ZVAD.fmk-mediated block in the death program, we examined a number of biochemical and morphological parameters of apoptosis in the ZVAD.fmktreated blebbing cells. First, we examined the efficacy with which ZVAD.fmk inhibits intracellular IRPs by examining cleavage of known IRP substrates by a combination of Western blotting and confocal immunofluorescence microscopy. Cells were harvested from ZVAD.fmk-treated cultures in which 50–60% of cells exhibited the continuously blebbing phenotype as judged by time-lapse videomicroscopy (typically 24–36 h after induction of apoptosis), and cell lysates fractionated by SDS-PAGE and electroblotted. Blots were probed with antibodies to actin, lamins A+C, and PARP, all known IRP substrates (Lazebnik et al., 1994, 1995b; Gu et al., 1995a; Mashima et al., 1995; Oberhammer et al., 1994), and compared with equivalent lysates from control cultures without ZVAD.fmk. ZVAD.fmk inhibits cleavage of all proteins for up to 48 h (Fig. 4,A). We also examined the immunocytochemical distribution of actin, PARP, and lamin staining in cells with and without ZVAD.fmk treatment by immunofluorescence confocal microscopy (Fig. 4,B). The effect of ZVAD.fmk on actin distribution within cells was examined by staining permeabilized cells with phalloidin-FITC. This showed ZVAD.fmk-treated cells to be morphologically distinct from either log-phase viable or apoptotic cells, exhibiting very obvious cytoplasmic blebbing (Fig. 4 B). ZVAD.fmk treatment prevented the dispersal and reduction of PARP and lamin staining that is typically observed in apoptotic cells. Thus, as reported elsewhere (Fearnhead et al., 1995; Fletcher et al., 1995; Slee et al., 1996), ZVAD.fmk effectively inhibits intracellular cleavage of known IRP substrates.

Next, electron microscopy was performed on Rat-1/cMycER™ cells induced to undergo apoptosis in the presence or absence of ZVAD.fmk (Fig. 5). 24 h after c-Myc activation in low serum, many cells in the non–ZVAD.fmktreated population exhibited features of classical apoptosis, i.e., chromatin condensation, cell shrinkage, cytoplasmic blebbing, and cell fragmentation. In contrast, although cells from ZVAD.fmk-treated populations exhibited gross cytoplasmic blebbing and some dilatation of the endoplasmic reticulum, their organelle structure was otherwise well preserved. More remarkably, nuclei of ZVAD.fmktreated cells resembled those of nonapoptotic cells, with reticular chromatin pattern bounded by intact nuclear lamina and membrane.

We next examined integrity of DNA in ZVAD.fmktreated blebbing cells. Flow cytometric analysis demonstrated that ZVAD.fmk suppresses the appearance of cells with sub-G1 content of DNA characteristic of apoptotic cells (Fig. 3 B). Consistent with this, no degradation of DNA into typical oligonucleosomal fragments (Wyllie, 1980) was detected in DNA extracted from ZVAD.fmktreated cells up to 48 h after c-Myc activation (data not shown).

Apoptosis is classically associated with loss of membrane phospholipid asymmetry: inactivation of a critical “flippase” results in expression of phosphatidylserine on the cell surface, a signal for recognition and ingestion of apoptotic cells by macrophages (Fadok et al., 1992). Surface phosphatidylserine expression was assessed by binding of FITC-labeled annexin V and visualized by fluorescence microscopy (Koopman et al., 1994). Rat-1/c-MycER™ cells undergoing normal apoptosis exhibited annexin V binding, whereas no binding was detected on blebbing cells blocked with ZVAD.fmk or live cells (Fig. 6). Thus, ZVAD.fmk blocks apoptosis before destruction of membranes, mitochondria, or loss of phosphatidylserine asymmetry.

As discussed above, ZVAD.fmk-treated cells induced to undergo apoptosis and exhibiting the blebbing phenotype for several days do eventually detach and become permeable to vital dyes. Analysis of these “late-stage” ZVAD.fmktreated cells revealed a ragged surface membrane, vacuolated cytoplasm, partial chromatin condensation (although less than in “normal” apoptotic cells), and eventual membrane permeability (data not shown). We conclude that these cells appear to have undergone a partial apoptotic cell death, although this will require further characterization.

Both cytokines (Harrington et al., 1994) and Bcl-2 (Fanidi et al., 1992) are effective at suppressing fibroblast apoptosis, which is most evident as a delay in onset of apoptosis, with the extent of the delay depending upon levels expression of c-Myc and the abundance of IGF-1 or Bcl-2 (Evan et al., 1992; Fanidi et al., 1992; Harrington et al., 1994). Intriguingly, time-lapse videomicroscopic analysis reveals that, while IGF-1 and Bcl-2 delay the onset of apoptosis (as judged by the start of membrane blebbing), they do not inhibit completion of the program once initiated. Cells proceed from blebbing to final fragmentation within 20–60 min, essentially the same kinetics as cells cultured in the absence of IGF-1 or overexpression of bcl-2 (Fig. 7, A and B). This is consistent with the notion that Bcl-2, serum and survival factors like IGF-I signaling act upstream of the onset of membrane blebbing, effectively reducing the probability that the apoptotic program will be initiated in any individual cell. In complete contrast, ZVAD.fmk acts downstream of membrane blebbing but before the later stages of apoptosis. To investigate further the temporal relationship between serum survival factors and ZVAD.fmk inhibition of apoptosis, we induced blebbing in ZVAD.fmk-treated Rat-1/c-MycER™ cells by addition of 4-OHT in the absence of serum and then examined whether readdition of serum could reverse the blebbing phenotype and allow cell recovery. Cultures were observed by time-lapse videomicroscopy. At the time serum was added, some cells within the population had just begun to bleb, whereas others had been blebbing for hours. In none of >500 cells examined was there any reversal of blebbing, and all cells that had initiated blebbing eventually went on to die. A study of a representative population is depicted in Fig. 8. From examination of time-lapse video data, it was also apparent that ZVAD.fmk-treated blebbing cells do not proliferate. Of some 1,000 cells examined, no division was ever observed of any cell that had started blebbing. However, this is not due to a cytostatic effect of ZVAD.fmk per se since nonapoptotic cells in ZVAD.fmktreated cultures divide at normal rates, as judged by timelapse videomicroscopy (see Note Added in Proof for Web Information) and DNA profile (Fig. 3 B). Thus, we conclude the onset of blebbing is coincident with loss of clonagenic potential. Interestingly, cells that had not yet initiated blebbing at the time of serum readdition remained viable, never blebbed, and, in the presence of ZVAD.fmk, proliferated to eventually repopulate the culture (not shown). Thus, readdition of survival factors before the onset of membrane blebbing can completely protect both the viability and proliferative potential of cells.

Finally, we asked whether ZVAD.fmk-blocked, blebbing cells might be rescued by removal of 4-OHT from the growth medium, thus inactivating c-Myc protein, which, in the Rat-1/c-MycER™ cells, is dependent upon the continuous presence of 4-OHT (Littlewood et al., 1995). Of some 500 cells examined, we have observed no recovery of any ZVAD.fmk-treated blebbing cell upon removal of 4-OHT (data not shown), affirming the commitment of these cells to eventual death.

A cell suicide program appears to have arisen early in metazoan evolution and been substantially conserved. Two of the three principal genes directly involved in control and execution of ontogenetic cell death in the developmentally invariant nematode worm C. elegans, ced-9 and ced-3, encode polypeptides that have multiple mammalian homologues: the Bcl-2 (Hengartner and Horvitz, 1994) and ICE-related cysteine protease (Xue and Horvitz, 1995) families, respectively. Both the Bcl-2 and IRP families are implicated in vertebrate cell suicide, a process whose morphological manifestation is apoptosis. In contrast to C. elegans, however, cell suicide in vertebrates is influenced by many diverse factors that include extracellular signals, genotoxic and physical trauma, anoxia, oncogene expression, and immune killing. How all of these differing agents impact upon the underlying cell death mechanism is unclear. Specifically, it is unclear which of the factors that modulate vertebrate cell death are part of the actual execution machinery—the mechanism that physically dismantles the cell—and which are merely regulatory mechanisms that govern the engagement of the execution process.

The best candidates for components of the execution machinery of vertebrate apoptosis are the Ced-3 homologues, the ICE-related proteases. Consistent with this, many diverse studies implicate IRPs in apoptosis induced by insults such as growth factor deprivation (Milligan et al., 1995; Nicholson et al., 1995), loss of contact with extracellular matrix (Boudreau et al., 1995), Fas/TNF (Enari et al., 1995; Los et al., 1995; Duan et al., 1996a; Schlegel et al., 1996), and cytotoxic T cell killing (Darmon et al., 1995). We have therefore investigated the requirement of IRPs in mammalian apoptosis using the IRP inhibitors ZVAD.fmk and BD.fmk. ZVAD.fmk and BD.fmk are particularly useful inhibitors for studying IRP action in intact cells because their intrinsic hydrophobicity permits their entry into cells, whereupon intracellular esterases convert the aspartyl methyl ester to aspartate to generate irreversible inhibitors of IRPs. Both ZVAD.fmk and BD.fmk have been shown to be effective inhibitors of apoptosis in intact cells with broad reactivities amongst tested ICE family proteases (Fearnhead et al., 1995; Slee et al., 1996). In our studies, we observed no differences in effect between ZVAD.fmk and BD.fmk. Consequently, most of the data we show is confined to studies with just one of these inhibitors, ZVAD.fmk.

We investigated the ability of ZVAD.fmk to inhibit apoptosis triggered by diverse promoters of apoptosis: c-Myc and E1A oncogene expression, DNA damage induced by the topoisomerase II inhibitor etoposide, and ectopic expression of the killer member of the Bcl-2 family Bak. In all cases, ZVAD.fmk provided substantial protection from apoptosis that correlated with inhibition of cleavage of the known IRP substrates, actin, lamins, and PARP. Thus, oncogenes, p53 (after DNA damage), and pro-apoptotic members of the Bcl-2 family all induce apoptosis via a ZVAD.fmk-sensitive activity, most probably one or more IRPs. These findings extend the critical role of IRPs in mammalian apoptosis to include stimuli of fundamental importance in the genesis and progression of neoplasia.

However, detailed time-lapse videomicroscopic analysis of cells protected by ZVAD.fmk revealed an unexpected and novel phenotype. An early manifestation of apoptosis in most cells is the onset of vigorous membrane blebbing. Membrane blebbing is a well-described characteristic of apoptosis in many cell types that, in the normal course of apoptosis in fibroblastic cells, is rapidly (20–60 min) followed by cell shrinkage, chromatin condensation, and nuclear and cellular fragmentation. Surprisingly, in cultures of ZVAD.fmk-treated cells induced to initiate apoptosis by the action of c-Myc, E1A, or Bak or the genotoxic agent etoposide, we found that the onset of membrane blebbing was not delayed by the presence of ZVAD.fmk in the culture medium. However, whereas untreated blebbing apoptotic cells rapidly proceed through the latter stages of chromatin condensation and cell fragmentation, ZVAD.fmk-treated cells continue to bleb vigorously for greatly extended periods of time, typically several days. Such blebbing cells exhibit few characteristics of apoptotic cells: cell nuclei remain intact with little evidence of chromatin condensation or nuclear fragmentation, plasma membrane integrity and mitochondrial function are preserved, and phosphatidylserine is not expressed on the cell surface. Thus, ZVAD.fmk appears not to inhibit the onset of membrane blebbing, although it effectively suppresses classical late-stage apoptotic events.

Moreover, although ZVAD.fmk substantially ameliorates the pro-apoptotic effects of c-Myc, DNA damage, and Bak in Rat-1 fibroblasts, most cells nonetheless do eventually detach from the substratum. Such cells exhibit some chromatin condensation (although far less than in normal late-stage apoptotic cells) but preserve intact nuclear lamina; they also have extensively fragmented and vacuolated cytoplasm. Eventually, however, late-stage ZVAD.fmk-treated cells lose their integrity and take up vital dyes—indisputable evidence of their ultimate death. We conclude that late-stage ZVAD.fmk-treated cells undergo a delayed “partial” apoptotic process, although a more detailed characterization of these late-stage cells will be required to establish their precise phenotype. We have investigated whether the delay to cell death afforded by ZVAD.fmk might be extended either by addition of higher initial concentrations of ZVAD.fmk to the growth medium or by its repeated readdition. However, neither alters the kinetics of cell death within Rat-1 cell cultures. It is difficult to assess the significance of this observation because of uncertainty as to the effective intracellular concentration of ZVAD.fmk administered to culture medium, or of the inhibitor's half-life in culture media or within cells. Nonetheless, it is possible that a continued pro-apoptotic stimulus eventually overrides the ZVAD.fmk block or bypasses the ZVAD.fmk-sensitive components of the apoptotic program.

Why should apoptotic cells exhibit prolonged blebbing when treated with ZVAD.fmk? One possibility is that the blebbing we observe in ZVAD.fmk-treated cells is different from the normal blebbing that accompanies apoptosis and has nothing to do with apoptosis. Three factors argue against this. By time-lapse videomicroscopy and light and electron microscopy, the blebbing observed in the presence of ZVAD.fmk appears identical to that seen in control cells undergoing normal apoptosis. In addition, the onset of blebbing in ZVAD.fmk-treated cell populations is identical to that in control populations; it merely continues for far longer as it is not accompanied by cell and nuclear fragmentation. Finally, time-lapse videomicroscopy shows that blebbing ZVAD.fmk-treated cells adhere to live cells in the same way as true apoptotic bodies (Whyte, M., N. McCarthy, and G.I. Evan, unpublished data) (see Note Added in Proof for Web Information), indicating that at least some of the pro-phagocytic processes triggered during normal apoptosis (although not surface display of phosphatidylserine) also occur in ZVAD.fmk-blocked cells. A second possibility is that ZVAD.fmk fails to prevent membrane blebbing because blebbing is not part of the basal apoptotic program but merely a dispensable epiphenomenon that commonly accompanies the process. However, the notion that membrane blebbing is not part of apoptosis is difficult to reconcile with our time-lapse videomicroscopic data, which show that once blebbing is initiated, it is not reversed by the action of antiapoptotic cytokines or by removal of the original pro-apoptotic trigger. Thus, blebbing appears to be concomitant with commitment to death. We also dislike this explanation because it seeks to explain the ZVAD.fmk phenotype by redefining apoptosis so as to exclude a characteristic of the process that is so unique and ubiquitous as to give the phenomenon its name. A third possibility is that membrane blebbing is part of the bona fide apoptotic process but triggered by IRPs that are not inhibited by ZVAD.fmk. This possibility needs to be investigated using a range of IRP inhibitors with differing spectrums of specificity. However, at present the possibility remains that membrane blebbing is triggered by a completely different mechanism that might not involve IRP action at all.

The phenotype of continuous membrane blebbing that we observe in ZVAD.fmk-treated apoptotic cells appears to contradict several reports that indicate that chemical and viral IRP inhibitors can afford long term protection against apoptosis. For example, peptide and viral inhibitors of IRPs protect motor neurons against apoptosis after factor withdrawal (Gagliardini et al., 1994; Martinou et al., 1995; Tewari et al., 1995), protect cells from Fas and TNFinduced killing (Beidler et al., 1995; Enari et al., 1995, 1996; Los et al., 1995), and block developmental apoptosis in Drosophila (Hay et al., 1995) and C. elegans (Xue and Horvitz, 1995). In some of these instances, it is possible that the protected cells have nonetheless initiated a membrane blebbing program that is not easily discernible using static imaging techniques. Furthermore, it is possible that blebbing cells remain capable of sustaining some of the functions of normal living cells within the soma, at least temporarily. In other situations, cell death may be triggered by directly activating or recruiting IRPs, in which case inhibiting IRP activity might be sufficient to prevent completely all manifestations of apoptosis. For example, the DEVD-specific IRP CPP32β is directly cleaved and activated by the cytotoxic T cell granule serine esterase Granzyme B and both the CD95 and TNF-R1 cytotoxic signaling pathways directly recruit (and presumably activate) the IRP FLICE/MACH via their “Death Domains” (Boldin et al., 1996; Muzio et al., 1996). By inference, the Drosophila Reaper, another Death-Domain protein, may act similarly to recruit directly downstream IRPs during developmental apoptosis in the fly. In the cases of oncogene deregulation, DNA damage/p53, and Bak, our data indicate that all these triggers of apoptosis act on some upstream regulator that can independently activate both IRPs and membrane blebbing.

By time-lapse videomicroscopic analysis, we observe a clear difference in the anti-apoptotic actions of the IRP inhibitor ZVAD.fmk, on the one hand, and the well-characterized antiapoptotic factors Bcl-2 and IGF-I, on the other. The actions of both Bcl-2 and IGF-I are to lower the probability of initiation of the apoptotic program, but neither has any effect on the kinetics of each apoptotic event (as defined as the time from onset of blebbing to cell fragmentation) once initiated. In complete contrast, ZVAD.fmk does not inhibit initiation of the apoptotic program, as judged by the onset of membrane blebbing, but prolongs each individual apoptotic event by inhibiting the cellular fragmentation that normally rapidly ensues. Importantly, cells that have initiated membrane blebbing but are blocked in completion of the apoptotic program by ZVAD.fmk cannot be rescued by the action of antiapoptotic survival factors and thus appear to be beyond a point of commitment to die. In addition, ZVAD.fmk-blocked blebbing cells do not divide; they have no clonagenic potential and from the stance of their neoplastic potential may therefore be considered as “genetically dead.” The clear implication from this is that, unlike more upstream inhibitors of apoptosis like Bcl-2 and survival factors, IRP inhibitors (whether viral or chemical) will not act as generic carcinogens because they inhibit the apoptotic program downstream of an irreversible point where replicative potential is lost.

In conclusion, we have shown that oncogene deregulation, DNA damage, and expression of the Bcl-2 family member Bak all induce apoptosis in mammalian fibroblasts via a ZVAD.fmk-inhibitable mechanism, probably one or more IRP. Nonetheless, time-lapse videomicroscopic analysis clearly distinguishes between the antiapoptotic actions of Bcl-2 or the survival factor IGF-I on the one hand and IRP inhibition on the other. Both Bcl-2 and IGF-I signaling act to suppress initiation of the apoptotic program, yet have no effect on the kinetics of each apoptotic event. In contrast, ZVAD.fmk has no effect on initiation of apoptosis, as determined by membrane blebbing, but acts to arrest each apoptotic program before completion. This observation raises the intriguing possibility that membrane blebbing is a discrete subprogram operating during mammalian apoptosis that can lead to cell death via an IRP-independent mechanism. If true, this will constrain the potential therapeutic use of IRP inhibitors. However, such speculations can only be validated by the development of specific inhibitors for each IRP and a more detailed understanding of the intracellular targets and kinetics of these enzymes.

We thank Derek Davies for flow cytometric analysis, Dr. Andrew Edwards and Peter Jordan for help with confocal microscopy, Dr. Nasser Hajibagheri for the electron micrographs, and our colleagues at the ICRF for advice, comment, and support.

4-OHT

4-hydroxytamoxifen

BD.fmk

t-butoxy carbonyl-Asp.fluoromethylketone

FLICE

FADD-like ICE

ICE

interleukin-1β-converting enzyme

IGF

insulin-like growth factor

IRP

ICE-related protease

MACH

MORT1-associated CED-3 homologue

MTT

3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide

PARP

poly-(ADP-ribose) polymerase

RT

room temperature

ZVAD.fmk

benzyloxycarbonyl-Val-Ala-Asp(O-methyl)-fluoromethylketone