Apaf-1 and Caspase-9 in p53-Dependent Apoptosis and Tumor Inhibition

See allHide authors and affiliations

Science  02 Apr 1999:
Vol. 284, Issue 5411, pp. 156-159
DOI: 10.1126/science.284.5411.156


The ability of p53 to promote apoptosis in response to mitogenic oncogenes appears to be critical for its tumor suppressor function. Caspase-9 and its cofactor Apaf-1 were found to be essential downstream components of p53 in Myc-induced apoptosis. Like p53 null cells, mouse embryo fibroblast cells deficient in Apaf-1 and caspase-9, and expressing c-Myc, were resistant to apoptotic stimuli that mimic conditions in developing tumors. Inactivation of Apaf-1 or caspase-9 substituted for p53 loss in promoting the oncogenic transformation of Myc-expressing cells. These results imply a role for Apaf-1 and caspase-9 in controlling tumor development.

The p53 tumor suppressor promotes cell cycle arrest or apoptosis in response to several cellular stresses, including mitogenic oncogenes (1). For example, the c-Myc oncogene induces uncontrolled proliferation but also activates p53 to promote apoptosis (2). Although the balance between Myc-induced proliferation and cell death is determined by genotype and external signals, Myc-induced apoptosis inhibits oncogenic transformation (3). Consistent with this view,Myc and other mitogenic oncogenes activate p53 through p19ARF, a tumor suppressor encoded at the INK4a/ARF locus (4). The ARF-p53 pathway is disabled in most human cancers, which implies that an oncogene-activated p53-dependent apoptosis pathway contributes to tumor suppression (4).

How activated p53 promotes apoptosis is unclear, but it may involve Bax (5, 6), a series of p53-inducible genes known asPIGs (7), or signaling through Fas-related pathways (8). Other p53 effectors might include caspases, a family of cysteine proteases that execute apoptotic cell death (9). Signaling procaspases associate with specific adaptor molecules that facilitate caspase activation by induced proximity (10). For example, caspase-9 (Casp9) associates with Apaf-1, and oligomerization of this complex in the presence of cytochrome c can activate a caspase cascade (11). Studies with knockout mice show that the requirement for different death effectors during apoptosis is highly cell-type– and stimulus-specific (12–15). Because p53-dependent apoptosis limits tumor development, the caspase-adaptor complex (or complexes) that acts downstream of p53 may participate in tumor suppression. However, the observation that caspase inhibitors do not prevent cell death induced by Myc or the p53-effector Bax suggests that caspases act too late in these death programs to have a substantial effect on long-term survival (16).

To determine the requirement for the Casp9 and Apaf-1 in apoptosis induced by Myc and p53, early-passage mouse embryo fibroblasts (MEFs) derived from Apaf-1- and Casp9-deficient mice were examined for their response to Myc. c-Myc or a control vector was transduced into wild-type, p53−/−, Casp9−/−, or Apaf-1−/− MEFs with the use of high-titer retroviruses (17), allowing whole populations of genotypically altered cells to be analyzed with minimal expansion in culture. The resulting populations were exposed to apoptotic stimuli that recapitulate tumor microenvironmental stresses (Fig. 1) (18). As expected, Myc sensitized wild-type cells to apoptosis after growth factor depletion (low serum) or hypoxia (19) as well as in suspension (Fig. 1). However, p53−/−, Casp9−/−, and Apaf-1−/− cells expressing Myc were resistant to apoptosis under these conditions. Consistent with previous reports (20), oncogenic Ras(H-rasV12), which cooperates with Myc to transform primary cells to a tumorigenic state, did not prevent Myc-induced apoptosis but slightly enhanced spontaneous cell death (Fig. 1). As in cells expressing Myc alone, disruption of p53, Apaf-1, or Casp9 in cells coexpressing Myc andRas (Myc-Ras MEFs) prevented apoptosis after serum depletion or hypoxia or in suspension (Fig. 1, I to K). In contrast, neither Apaf-1 nor Casp9 was required for tumor necrosis factor (TNF)–induced apoptosis, a program in which p53 is largely dispensable (Fig. 1, H and L) (21). Moreover, p53-dependent apoptosis inMyc-expressing cells did not require the caspase-8–FADD pathway or caspase-2 (22, 23). Of note, the caspase inhibitor zVAD.fmk was three times less efficient at blocking apoptosis of the p53-dependent deaths than genetic deletion of Apaf-1 or Casp9 (24), which implies that zVAD.fmk is an ineffective inhibitor of Casp9.

Figure 1

Requirement for Apaf-1 and Casp9 in Myc-induced apoptosis. A control vector (A to D), c-Myc (E to H), or c-Mycand H-rasV12 (I to L) were transduced into early passage MEFs derived from wild-type (open circles), p53−/− (open squares), Casp9−/− (closed circles), and Apaf-1−/− (closed triangles) mice with high-titer recombinant retroviruses (17). Cell populations were incubated in growth factor–poor (low serum) medium (A, E, and I), hypoxic conditions (B, F, and J), suspension (C, G, and K), or murine TNF-α (D, F, and L) for the indicated times, and cell viability was determined by trypan blue exclusion (17, 18). Each point represents the mean ± SD of at least three experiments with two separately transduced populations. Data are normalized to the rate of spontaneous cell death occurring in untreated cells (<10% in all cases, except 25% for Myc-Ras wild-type MEFs).

The similar requirement for p53, Apaf-1, and Casp9 in Myc-induced apoptosis suggests that these genes act in the same pathway. Apaf-1 and Casp9 do not act upstream of p53, because c-Myc or the combination of c-Myc plus Ras efficiently induced p53 protein and activity in Apaf−/− and Casp9−/− MEFs (Fig. 2A). To determine whether Apaf-1 and Casp9 act downstream of p53, a temperature-sensitive p53 (p53val138) fused to green fluorescent protein (GFP) was introduced into wild-type, p53−/−, Apaf-1−/−, and Casp9−/− Myc-Ras MEFs (17), and the apoptotic response of GFP-positive cells was determined at the restrictive (39°C) or permissive (33°C) temperature (Fig. 2B). At 33°C, p53 induced massive apoptosis in wild-type cells and rescued the apoptotic defect in p53−/− cells, but it had no pro-apoptotic effect in Apaf-1−/− or Casp9−/− cells.

Figure 2

Apaf-1 and Casp9 act downstream of p53 to promote apoptosis. (A) Immunoblot analysis of the indicated proteins in vector control (lanes V) or in two independent populations (lanes 1 and 2) of Myc-Ras–transformed MEFs of the indicated genotypes. For wild-type cells, populations 1 and 2 correspond to MEFs extracted from littermates of Apaf-1−/− and Casp9−/− mice, respectively. Thirty micrograms of lysate was prepared from exponentially growing cells and separated onto 8% SDS–polyacrylamide gels for detection of c-Myc (N-262, Santa Cruz), p53 (CM-5, Novocastra), and Mdm2 (mAB 2A10, provided by A. Levine, Rockefeller University) or onto 12% gels for detection of Ras (OP23, Calbiochem) by standard immunoblot procedures. Expression of α-tubulin (clone B-5-1-2, Sigma) is shown as a loading control. (B) Viability of Myc-Ras–transformed cells expressing a temperature-sensitive p53 at the restrictive (39°C, white bars) or permissive (33°C, hatched bars) temperature for wild-type p53 expression. Myc-Ras MEFs were transduced with a retrovirus expressing human p53val138 fused to GFP. Infected cell populations (>80% GFP-positive) were placed in medium containing 1% FBS and retained at 39°C or transferred to 33°C for 48 hours. Cell viability was determined by trypan blue exclusion and by DAPI staining (12). (C) Representative fluorescence micrographs of cells grown at 33°C showing GFP or DAPI (1 μg/ml) fluorescence to monitor p53 expression or chromatin condensation, respectively.

p53 activation by oncogenes and tumor microenvironmental stresses may provide a natural brake to tumor development (1, 4). If Apaf-1 and Casp9 are required for p53 action under these circumstances, then inactivation of either one should facilitate oncogenic transformation. To test this hypothesis, we assayed wild-type, p53−/−, Apaf-1−/−, and Casp9−/− Myc-Ras MEFs for their ability to produce colonies in soft agar (Fig. 3) and form tumors in immunocompromised mice. Consistent with the ability of p53 to suppress transformation, p53-deficient cells formed >100 times the number of colonies in soft agar as wild-type cells (Fig. 3A). Cells without Apaf-1 or Casp9 formed a similar amount of colonies as p53-deficient cells. Thus, inactivation of Apaf-1 or Casp9 can enhance the long-term survival of oncogenically transformed cells.

Figure 3

Effect of Apaf-1 and Casp9 on colony formation in soft agar; 2 × 104 cells of the indicated genotypes were resuspended in 0.3% Noble agar (in DMEM supplemented with 10% FBS) and plated on 6-cm plates containing a solidified bottom layer (0.5% Noble agar in growth medium). Giemsa staining (A) and phase-contrast microscopy (B) of a representative experiment 2 weeks after plating. The plating efficiencies determined by counting colonies containing ≥40 cells in random fields were <1%, 65%, 75%, and 80% for wild-type (wt), Apaf-1−/−, Casp9−/−, and p53−/− MEFs, respectively.

To determine the potential impact of Apaf-1 and Casp9 inactivation on tumor development, we injected wild-type, p53−/−, Apaf-1−/−, and Casp9−/− Myc-Ras MEFs subcutaneously into immunocompromised mice and monitored for tumors at the sites of injection. As expected, p53 reduced the tumorigenicity of Myc-Ras–transformed cells (Table 1 and Fig. 4). Hence, at low numbers of injected cells, wild-type cells produced tumors at a lower frequency than p53−/− cells, and the tumors that formed grew at dramatically reduced rates (Table 1 and Fig. 4). At higher cell numbers, wild-type cells produced tumors more readily (Table 1), perhaps because higher concentrations of autocrine survival factors protected cells from apoptosis in vivo. Still, tumors derived from p53-deficient cells displayed more rapid growth.

Figure 4

Effect of Apaf-1 and Casp9 on tumorigenicity ofMyc-Ras–transformed cells. Myc plusRas transformed populations derived from wild type (white circles), p53−/− (white squares), Apaf-1−/−(black triangles), and Casp9−/− (black triangles) were injected subcutaneously into each rear flank of female NSW athymic nude mice (Tatonic Farms) within two passages of gene transfer. Tumor volume (V) was estimated from caliper measurements of the length (L) and width (W) as V = (L ×W 2)/2. For each genotype, eight sites were injected; each data point represents the mean ± SD of measurements derived from only those sites that formed tumors (seeTable 1). Bromodeoxyuridine (BrdU) incorporation–DNA content analysis indicated that the cell-cycle distribution of the injected populations was similar; for example, the amounts of cells in S phase after a 4-hour BrdU pulse were 41%, 46%, 45%, and 49% for wild-type, Apaf-1−/−, Casp9−/−, and p53−/− Myc-Ras cells, respectively.

Table 1

Tumorigenicity of Myc-Ras transformed cells. Indicated numbers of Myc-Ras–expressing MEFs were injected into athymic nude mice as described in Fig. 4. Ratio of injections producing tumors and volume (mean ± SD) of those tumors that formed at 2 weeks after injection are shown. ND, not determined; NA, not applicable.

View this table:

Apaf-1 and Casp9 had a profound effect on suppressing the tumorigenicity of Myc-Ras MEFs (Table 1 and Fig. 4). Inactivation of Apaf-1 and Casp9 substantially reduced the number of cells required to form tumors (Table 1), which grew at rates similar to tumors derived from p53-deficient cells (Fig. 4). This enhanced tumorigenicity was most likely due to inefficient apoptosis, because Apaf-1−/− and Casp9−/− Myc-Ras MEFs displayed marked apoptotic defects (see Fig. 1) and showed no differences in proliferation (Fig. 4). Moreover, tumors derived from wild-type cells displayed two- to fourfold more apoptotic bodies than tumors derived from Apaf-1−/− or Casp9−/−cells (24). Importantly, all injected Myc-Ras cells were nonclonal populations that were minimally passaged in culture, yet Apaf-1−/− and Casp9−/− Myc-Ras MEFs formed tumors rapidly (tumors were visible 4 days after injection). Therefore, it is unlikely that rare secondary mutations contribute to tumor acceleration. These results demonstrate that defects in the effector mechanism of p53-dependent apoptosis can facilitate tumor development.

Cytochrome c is required for activation of pro-Casp9, and cytochrome c release from mitochondria appears to be essential for some apoptotic programs (25). Of note, both pro-apoptotic oncogenes and p53 can affect mitochondrial function. For example, the adenovirus E1A oncogene [which also activates p53 (26)] and the p53 effector Bax facilitate cytochrome c release (27). Furthermore, the PIGs may act to influence redox metabolism and the mitochondrial permeability transition (7). Although it is possible that p53-mediated activation of the Fas pathway might indirectly target the mitochondria through caspase-8 (8, 28), we see no evidence that Apaf-1 and Casp9 are required for Fas or TNF-induced apoptosis (13) (Fig. 1). Therefore, our data support the view that immediate effectors of p53 in apoptosis target the mitochondria, thereby releasing cytochrome c to activate Casp9.

It has been unclear whether inactivation of apoptotic components acting downstream of cytochrome c should enhance long-term survival or only delay cell death (25). We show that disruption of Apaf-1 or Casp9 allows efficient colony formation in soft agar and rapid tumor development in nude mice. These data imply that Apaf-1 or Casp9 actively limits tumor development and suggest that cytochrome c release is not necessarily a lethal event. Perhaps the ability of Apaf-1 or Casp9 deficiency to promote long-term cell survival reflects the high proliferation rate of Myc-expressing cells, which may allow substantial cell accumulation as a result of inefficient cell death. Alternatively, the ability of cytochrome c release to commit a cell to death may depend on cell type, genetic background, or apoptotic stimulus.

Essential components of p53 apoptotic pathways act as tumor suppressors. For example, p19ARF and Bax act upstream or downstream of p53 in oncogene-induced apoptosis, respectively, and mutations in either gene facilitate tumorigenesis in mice and humans (4, 6). Although mutations in Apaf-1 or Casp9 genes have not been described, each maps to loci disrupted in human tumors. Apaf-1 is located on chromosome 12q22-12q23 (15), a region that is altered in some pancreatic and ovarian carcinomas (29). Casp9 is located at chromosome 1p34-1p36.1 (30), a hotspot for loss of heterozygosity in several cancers, including some neuroblastomas (31). Interestingly, many neuroblastomas tolerate N-Myc amplification but rarely harbor p53 mutations (31–32), and neuronal hyperplasia is one of the hallmarks of Casp9-deficient mice (18). Irrespective of the frequency at which Apaf-1 and Casp9 are mutated in human cancers, our data clearly demonstrate that disruption of the effector mechanisms of apoptosis can facilitate oncogenic transformation and tumor development.

  • * Present address: Department of Immunology, Medical Institute of Bioregulation, Kyushu University, Fukoka, Japan.

  • To whom correspondence should be addressed. E-mail: lowe{at}


View Abstract

Stay Connected to Science

Navigate This Article