Chemotherapy Triggers Apoptosis in a Caspase-8-dependent and Mitochondria-controlled Manner in the Non-Small Cell Lung Cancer Cell Line NCI-H460 Free

NSCLC³ remains a disease with poor prognosis despite recent advances made in early diagnosis and treatment. More than 75% of the patients with NSCLC prove to be potential candidates for chemotherapy at some point during the course of their disease because of the development of metastases(1). However, chemotherapy still shows poor response rates in NSCLC patients, with relatively short duration and rare complete remissions. Identifying the molecular determinants of sensitivity and resistance to chemotherapy in NSCLC as well as in other solid tumor types may help improve its efficacy.

Although the primary intracellular targets and the pharmacological mechanisms of action of the anticancer drugs vary, it has become evident that drug-induced cell kill is, at least partially, mediated by programmed cell death or apoptosis (2). Relevant progress has been made in the understanding of the events underlying drug-induced apoptosis, and several molecules have been implicated in this process. Among them, a family of aspartatic-specific cysteine-proteases named caspases (3) are the central executors of the apoptotic process (4, 5). Thus far, 14 family members have been cloned in mammals, but not all of them seem to be directly involved in apoptosis (6). Caspases are synthesized as inactive zymogens(M_r 30,000–50,000) that possess three domains: an NH₂-terminal domain; a large subunit (M_r ∼20,000); and a small subunit (M_r ∼10,000; Ref.5). In general, the process of caspase activation requires a proteolytic processing between domains, followed by an assembly of the subunits to generate an active tetramer (4, 5, 6).

Caspases are activated in a cascade fashion. It has been proposed that a proapoptotic signal can activate an initiator or upstream caspase,which usually possesses a long prodomain such as caspases-8, -9, and-10. In turn, these initiators can activate the effector or downstream caspases, including caspase-3, -6, and -7, which leads to the biochemical and morphological changes that are characteristic of apoptosis (7). To date, two major caspase pathways or cascades, headed by caspases-8 and -9, have been described and shown to mediate distinct sets of signals (7, 8, 9, 10). The cascade led by caspase-8 is involved in death-receptor-mediated apoptosis such as the one triggered by Fas, TNF, and TRAIL. On activation, these receptors recruit FADD, which in turn binds to procaspase-8, which leads to cleavage into its active form (11, 12). The subsequent activation of the effector caspases occurs either directly or after an amplification step involving mitochondria(13). On the other hand, the caspase pathway headed by caspase-9 is thought to mediate chemical-induced apoptosis. This cascade is triggered by the cytochrome c release from mitochondria, leading to the formation of a complex with Apaf-1 that via its caspase recruitment domain (CARD) binds to procaspase-9(14). This complex, called apoptosome (15) in the presence of dATP can activate procaspase-9, which in turn activates effector caspases (16). In contrast to the caspase-8 cascade, in which the role of mitochondria may vary between different cell types (13), the caspase-9 pathway is clearly mitochondria-dependent.

The Bcl-2 family of proteins plays a crucial role in the function of mitochondria during the apoptotic process (17). Overexpression of the antiapoptotic molecules Bcl-2 or Bcl-x_L has been reported to cause resistance to anticancer drugs (18, 19). These molecules have the capacity to inhibit apoptosis by preventing permeability transition(PT) and/or by stabilizing the barrier function of the outer mitochondrial membrane (20, 21). Stabilization of the mitochondria pores abrogates the release of cytochrome c(22), thereby preventing the activation of procaspase-9.

Besides the action of Bcl-2 family members, other more direct mechanisms of caspase inhibition have been identified. Viruses are known to produce potent caspase inhibitors, such as CrmA made by cowpox virus that preferentially inhibits caspase-1 and caspase-8(23). On the other hand, the baculovirus p35 protein is an example of a broad range inhibitor that blocks multiple caspases(24). On the basis of sequence similarity toward p35,human homologues of IAPs have been identified such as XIAP,cIAP1, cIAP2 and survivin (25, 26, 27). They have been shown to be able, in the case of XIAP, cIAP1, and cIAP2, to inhibit not only effector caspases but also the initiator caspase-9(28).

Current available data on the molecular mechanism underlying the sequential activation of caspases has led to a model in which caspase-9 is activated on chemotherapy and caspase-8 is activated by death receptor signaling (7, 8, 9, 10). Nevertheless, some reports have described the activation of procaspase-8 after chemotherapy(29, 30, 31, 32, 33). However, this activation seems to be a downstream and secondary event that follows caspase-9 activation(7, 29, 30, 31). In addition, it should be noted that most of these studies were performed in leukemia cell lines or in cell-free extracts, and that relatively little data are available on the activation of caspases in chemotherapy-induced apoptosis in solid tumor cells.

To obtain more insight in the molecular mechanisms underlying drug sensitivity in NSCLC cells, here we examined the caspase cascades involved in triggering apoptosis in the NSCLC cell line H460. Our results—obtained by means of various biochemical assays, including caspase cleavage and activity assays, and by using selective peptide and molecular inhibitors of caspases—clearly point to a central role for caspase-8 in chemotherapy-induced apoptosis in H460 cells.

Drugs were provided as pure substances. Stock solutions of cisplatin(Bristol-Myers Squibb, Woerden, the Netherlands) and gemcitabine (Eli Lilly Research Laboratories, Indianapolis, IN) were made in PBS;topotecan (SmithKline-Beecham Pharmaceuticals, Herts, United Kingdom)was dissolved in water. For each experiment, the stock solutions of the drugs (Table 1) were freshly diluted in culture medium to the indicated final concentration. The caspase-inhibitors Z-VAD-fmk, used at a final concentration ranging from 50 to 100μg/ml and the preferential caspase-8 inhibitor IETD-fmk, added to a final concentration of 20–40 μg/ml, were used 2 h prior to drug exposure. Both of the inhibitors were purchased from Enzyme System Products (Livermore, CA). In some experiments, the caspase-3-like inhibitor DEVD-CHO (Calbiochem, San Diego, CA) was used at a final concentration of 50–100 μg/ml.

The cDNAs encoding CrmA wild type, loss of function mutant CrmA(T291R), Bcl-2, Bcl-x_L, subcloned into the expression vector pEFFLAGpGKpuro, were described previously (32, 33). pcDNA3 vectors encoding caspase-9S, active site mutant caspase-8 (Flice-DN or caspase-8-DN) and FADD-DN have also been described elsewhere (34, 35, 36). The cDNA encoding myc-tagged-XIAP was excised from pCS3MT-XIAP (37) and subcloned into pcDNA3.

The human NSCLC cell lines NCI-H460 (H460), NCI-H322 (H322), and SW1573 and Jurkat-T-leukemia cells were used in the experiments. Cells were cultured in RPMI 1640 supplemented with 10% heat-inactivated FCS (Life Technologies, Inc., Breda, the Netherlands), 2 mml-glutamine, 50 IU/ml penicillin, and 50 μg/ml streptomycin, and grown at 37°C in a humidified atmosphere with 5%CO₂. The cell lines were tested regularly for the absence of Mycoplasma infection. Cells from exponentially growing cultures were used in all of the experiments.

For the generation of stable transfectants, H460 cells were transfected with 10 μg cDNA, using Superfect reagent (Life Technologies, Inc.)according to the manufacturer’s protocol. After 24 h, cells from each transfection were split into five separate culture dishes to ensure that independent lines were established. Selection was made using increasing concentrations of puromycin (Sigma, Zwijndrecht, the Netherlands) ranging from 1 to 2 μg/ml or 200–400 μg/ml Geneticin(Life Technologies, Inc.), depending on the transfected plasmid used. Independent clones were collected and allowed to grow in six-well plates. Clones were tested for the expression of the constructs by Western blotting and were selected for use in the desired experiments.

Cells were plated at a density of 5 × 10⁶ cells in 75-cm² tissue culture flasks (Costar, Cambridge, MA) 24 h before treatment. Cells were incubated for 4–72 h with IC₅₀s or IC₈₀s of cisplatin, topotecan, or gemcitabine(Table 1) in the presence or the absence of peptide inhibitors of caspases when indicated. The analysis of apoptotic cells was performed as described previously(38). Briefly, the extent of cell death was determined by PI staining of hypodiploid DNA or by annexin V-FITC and 7-AAD double staining. For the PI staining, 3 × 10⁵ cells were resuspended in Nicoletti buffer as described previously (39) and analyzed by FACScan (Becton Dickinson, Mount View, CA). The fraction of cells with sub-G1 DNA content was assessed by the Lysis program (Becton Dickinson). Annexin V staining was performed according to the manufacturer’s protocol(Nexins Research, Kattendijk, the Netherlands). After incubation with annexin V, 10 μl of 7-AAD (PharMingen, San Diego, CA) was added and analysis was performed on FACScalibur using CELLQuest software (Becton Dickinson). The percentage of specific apoptosis was calculated by subtracting the percentage of spontaneous apoptosis of the relevant controls from the total percentage of apoptosis.

Cytotoxicities to cisplatin, topotecan, and gemcitabine were assessed by MTT assay basically as described previously (38, 41). In brief, a suspension of 10,000 cells/100 μl of medium was added to each well of flat- or U-bottomed 96-well plates (Costar, Corning, NY)and allowed to grow. Twenty-four h later, drugs were made up in medium and eight different concentrations were added to the plates at a volume of 100 μl per well, and plates were incubated for 72 h with drugs. Then 20 μl of a solution of 5 mg/ml MTT (Sigma Chemicals, St. Louis, MO) were added to each well and incubated for another 4 h at 37°C. Plates were then centrifuged at 1000 rpm at 4°C for 5 min, and the medium was carefully discarded. The formazan crystals were dissolved in 100 μl of DMSO (ACROS Organics, Geel,Belgium) and absorbance was read at 540 nm using Spectra Fluor (Tecan,Salzburg, Austria). Absorbance values were expressed as a percentage of untreated controls and concentrations resulting in IC₅₀ and IC₈₀ were calculated. The IC₅₀ and IC₈₀ values represent the means of at least three independent experiments.

Clonogenic assays were performed essentially as described previously(40). H460 cells were seeded in triplicate into six-well plates at a concentration of 300 cells per well. After 24 h, drugs(cisplatin, topotecan, or gemcitabine) were added at final concentrations of IC₅₀ or IC₈₀ (Table 1) for 24–72 h. After drug removal,cells were washed twice with PBS and allowed to proliferate in fresh medium. Colonies were counted when they reached the size of 50–100 cells, after staining with 0.1% crystal violet in 0.9% saline for 30 min at room temperature. The number of colony-forming units in treated cultures was expressed as the percentage of untreated controls.

Western blot analysis was performed essentially as described previously(38). In brief, from each sample, 25 μg of protein per lane were separated on 8–15% SDS-PAGE and electroblotted onto polyvinylidene difluoride membranes (Amersham, Braunschweig, Germany). Subsequently, membranes were incubated overnight at 4°C in a solution of PBS supplemented with 5% nonfat dry milk. For immunodetection, the following antibodies were used: anti-caspase-3, anti-caspase-7, and anti-FADD mouse mAbs (Transduction Laboratories, Lexington, KY);anti-caspase-8 mAb (Immunotech, Marseille, France); anti-Flag mAb(Stratagene, La Jolla, CA); anti-cytochrome c mAb and anti-caspase-9 polyclonal antibody (PharMingen, San Diego, CA);anti-Myc mAb and anti-Bcl-x_L polyclonal antibody(Santa Cruz Biotechnologies, Santa Cruz, CA), Bcl-2 mAb (Dako, Santa Barbara, CA), and rabbit Apaf-1 polyclonal antibody (provided by Dr. Xiadong Wang). After 2-h incubation with the primary antibody in dilutions that ranged from 1:500 to 1:2000, membranes were washed in TBST [10 mm Tris-HCl (pH 8.0), 0.15 m NaCl, and 0.05% Tween 20], followed by horseradish peroxidase-conjugated goat-antimouse or goat-antirabbit antibody. ECL (Amersham, Braunschweig, Germany) was used for detection, and protein expression was quantified by densitometry of autoradiographs (Bio-Rad, Model GS-690, Imaging densitometer,Richmond, CA). Protein loading equivalence was assessed by the expression of β-actin.

Cytosolic extracts were prepared essentially as described by Deveraux et al. (28). Briefly, cells were pelleted by centrifugation after being washed once with ice-cold buffer A[20 mm HEPES (pH 7.5), 10 mm KCl, 1.5 mm MgCl₂, 1 mm EDTA, and 1 mm DTT]. Subsequently, cell pellets were resuspended in two volumes of buffer A, incubated for 20 min on ice, and disrupted by 20 passages through a 26-gauge needle. Cell extracts were clarified of mitochondria by centrifugation at 20,000 × gfor 30 min at 4°, and the harvested cytosolic extracts were stored at−80°C.

Spectrofluorimetric assays of proteolytic activity were carried out using synthetic fluorogenic substrates DEVD-AFC to measure caspase-3-like or effector caspase activity and IETD-AFC to assess initiator caspase activity. Both kits were purchased from Clontech Laboratories Inc. (Palo Alto, CA). Assays for caspase-9 activity were performed using the synthetic fluorogenic substrate LEHD-AFC(MBL Co., Nagoya, Japan). Experiments were performed according to the manufacturer’s protocols. Fluorescence was detected using a fluorometer equipped with a 400-nm excitation and a 505-nm emission filter (Spectra Fluor Tecan, Salzburg, Austria). Fold-increase in the protease activity was determined by comparing the levels of the treated cells with untreated controls. The values obtained are depicted in Figs. 1,E, 4,F, 6,C, and 7 B as fluorescence units(F.U. × 10⁶).

Quantitative experiments were analyzed by the Student’s ttest. All of the Ps resulted from the use of two-sided tests and were considered significant when <0.05.

We started the investigation by analyzing the role of caspase-9 in the NSCLC cell line H460. These cells, as well as other NSCLC cell lines have been previously characterized in our laboratory with respect to their sensitivity to drug- and Fas-induced apoptosis (38, 41). H460 cells were selected as representatives for NSCLC cells because of their favorable transfection properties. Stable transfectants were generated using vectors expressing caspase-9S, a spliced variant of caspase-9 with dominant-negative activity(34), or the IAP family member XIAP (Fig. 1,A) and exposed to cisplatin. Interestingly, as shown in Fig. 1,B, caspase-9S did not block drug-induced apoptosis in H460 cells (P = 0.72). In line with this observation, overexpression of the caspase-9 inhibitor XIAP led only to minimal protection against drug-induced apoptosis (Fig. 1,B; P = 0.30). The lack of effect of overexpression of caspase-9S and XIAP on apoptosis was independent of the anticancer agent (cisplatin, topotecan, and gemcitabine), drug concentration (IC₅₀ and IC₈₀), duration of exposure (4–72 h), or clone used in the experiments (data not shown). Moreover, the caspase-3-like or effector activity induced by these drugs was not affected in cells overexpressing caspase-9S (P = 0.27), and only slightly decreased in cells transfected with XIAP (Fig. 1,C; P = 0.20), which suggests that the effector activity triggered by chemotherapy is caspase-9-independent in H460 cells. The small percentage of inhibition provided by the overexpression of XIAP on drug-induced apoptosis and DEVD-AFC cleavage, compared with the absence of blockade achieved by the overexpression of caspase-9S, might be interpreted as an additional effect of XIAP on caspases-3 and -7, also known targets of this IAP member (28). As shown in Fig. 1,D, H460 cells express similar levels of caspase-9 and Apaf-1 as compared with Jurkat-T-leukemia cells, used as a control due to their well-established dependency on caspase-9 activation for chemical-induced apoptosis (28). However, upon exposure to chemotherapy, no substantial cleavage of LEHD-AFC, a preferential substrate for caspase-9 (42), was observed above the control levels in H460 cells (P = 0.2),whereas, in Jurkat-T-leukemia cells, a 6-fold increase was found (Fig. 1,E; P = 0.005). Interestingly, as shown in Fig. 1 A, the overexpression of caspase-9S in H460 cells was accompanied by a reduction in the levels of endogenous caspase-9. Nevertheless, the overexpression of caspase-9S and the reduction on the caspase-9 protein had no effect on the levels of LEHD-AFC cleavage induced by chemotherapy in these cells when compared with H460 cells transfected with an empty vector (data not shown). Taken together, these results indicate that, in H460 cells, caspase-9 is not instrumental in mediating chemotherapy-induced apoptosis.

In a previous study, we observed chemotherapy-induced caspase-8 activation that was independent of Fas/FasL signaling in NSCLC cells(38). To further examine the relation between caspase-8 activation and drug-induced apoptosis, we engineered H460 cells to overexpress the caspase-1 and -8 inhibitor CrmA, or a mutated and inactive derivative, CrmA-mut (32), as a control (Fig. 2,A). CrmA has been shown to be a potent inhibitor of apoptosis induced by the TNF family of receptors but a poor inhibitor of apoptosis triggered by other stimuli, such as chemotherapeutic agents (32, 43). In contrast to these reports, an analysis of apoptosis in H460 cells overexpressing CrmA demonstrated at least 90% protection against cisplatin-induced cell death (P = 0.002), whereas no protection was observed in cells transfected with the loss-of-function CrmA-mut(P = 0.65), or with the empty vector (Fig. 2,B). The inhibitory effect of CrmA are known to be mediated by pseudosubstrate sequences in the protein that are preferential for caspase-1 and -8 (32); however, some effects on caspases-9, -10, and -4 cannot be completely discarded(44). Therefore, we examined the effect of expression of DN-acting caspase 8, a more specific inhibitor of caspase-8(35), on chemotherapy-induced apoptosis in H460 cells(Fig. 2,C). Stable transfection with caspase-8-DN blocked 70% of drug-induced apoptosis (Fig. 2 D;P = 0.0031), thus further substantiating the results obtained with CrmA and pointing to a crucial role of caspase-8 in this process. In both CrmA- and caspase-8-DN-overexpressing H460 cells, a similar inhibitory pattern was observed on use of other drugs(topotecan and gemcitabine), drug concentration(IC₅₀), or prolonged times of exposure (data not shown). Furthermore, the results were consistent in independently selected transfectants that overexpressed either CrmA or caspase-8-DN,or when the annexin V and 7-AAD double-staining method was used to assess apoptosis (results not shown).

As mentioned earlier, caspase-8 has been shown to be the apical caspase in death-receptor-induced apoptosis (11, 12). We previously reported the absence of Fas/FasL activation during drug-induced apoptosis in lung cancer cells (38). Nonetheless, caspase-8 can be recruited and activated by other death-receptors such as TNF and DR5 (45). In addition, it has been shown that these death-receptors use FADD as an adaptor molecule in the processing of caspase-8 (46). To exclude the possible involvement of other death receptors such as TNF and DR5 in the activation of caspase-8 on anticancer drug treatment, we generated H460 cells overexpressing a DN form of FADD (Ref.36; Fig. 3,A). As expected, FADD-DN blocked more than 80% of Fas-induced apoptosis in H460 cells; however, it provided no protection against chemotherapy-induced cell death (P = 0.22;Fig. 3,B). Furthermore, as determined by Western blotting,cleavage of caspase-8 was not suppressed in cells overexpressing FADD-DN (Fig. 3 C). These findings demonstrate that caspase-8 activation and concomitant drug-induced apoptosis in H460 cells do not require the activity of FADD-containing receptor signaling complexes.

The results obtained thus far suggest that caspase-8 may function as an apical caspase in the pathway that triggers apoptosis that is induced by chemotherapeutic agents in H460 cells. In analogy with drug-induced caspase 9 activation, we investigated a possible role for mitochondria in the activation of procaspase-8. First, we examined the effect of cisplatin treatment on the release of cytochrome c from mitochondria. As shown in Fig. 4,A, increased levels of cytochrome c are clearly evident in the cytosolic fraction 48 h after treatment. This release preceded caspase activation because exposure of the cells to the preferential caspase-8-inhibitor IETD-fmk or the broad-spectrum caspase inhibitor zVAD-fmk did not block this process (Fig. 4,A). Next, cytochrome crelease was monitored in H460 cells stably expressing CrmA or the antiapoptotic Bcl-2 or Bcl-x_L proteins (Fig. 4,B). CrmA did not affect cytochrome c release;however, this release was almost completely blocked in Bcl-2- and Bcl-x_L-expressing cells in comparison with untreated controls (Fig. 4,C). As predicted from this finding, cisplatin-induced apoptosis was strongly suppressed by both Bcl-2 and Bcl-x_L (Fig. 4,D); in Bcl-2-expressing cells, a complete block of apoptosis was detected(P = 0.001), whereas Bcl-x_L overexpression resulted in an approximately 85% reduction in cell death (P = 0.006). Moreover, Bcl-2 also blocked the processing of caspase-8 and IETD proteolytic activity (Fig. 4, E and F). Similar results were obtained on the use of topotecan or gemcitabine in these experiments (data not shown).

It has been reported that caspase-8 can be activated by caspases-9 in a mitochondria-controlled manner (29). Although we showed that caspase-9 is not involved in the apoptotic process in H460 cell(Fig. 1), we wanted to further exclude this alternative. In H460 cells that overexpressed caspase-9S, caspase-8 cleavage remained clearly detectable after cisplatin treatment (Fig. 5). Taken together, these data provide strong evidence for the existence of a mitochondria-dependent pathway that is able to activate procaspase-8, independently from caspase-9,during anticancer drug-induced apoptosis.

It has been suggested that caspase-8 can be activated downstream of caspase-3 (29). To rule out this possibility and to further order the sequence of events in drug-induced apoptosis in H460 cells, we exposed the cells to the caspase-3-like inhibitor DEVD-CHO during cisplatin treatment. As shown in Fig. 6,A, this inhibitor had no effect on caspase-8 cleavage, although we could observe a 30–50%decrease in apoptosis (result not shown). In addition, a more downstream event, such as PARP cleavage, was prevented by overexpression of CrmA and caspase-8-DN (Fig. 6,B), likely attributable to the suppression of DEVD-like activity triggered by cisplatin exposure in these stable transfectants (Fig. 6 C). These results show that effector caspases are activated after the cleavage of procaspase-8 in drug-induced apoptosis in H460 cells.

We further investigated whether the findings that indicated a central role for caspase-8 in chemotherapy-induced apoptosis in H460 cells would apply also to other NSCLC cell lines. Therefore, H460 cells,together with the NSCLC cell lines H322 and SW1573, were exposed to both cisplatin and the preferential caspase-8 inhibitor IETD-fmk. The extent of blockade of drug-induced apoptosis was analyzed in comparison with the effect obtained with the broad-spectrum caspase inhibitor zVAD-fmk. Jurkat-T-leukemia cells were selected for use in these experiments as a control, because the pattern of inhibition of chemotherapy-induced apoptosis provided by IETD-fmk and zVAD-fmk has been reported in these cells (31, 45). As shown in Fig. 7,A, a similar pattern of suppression of chemotherapy-induced apoptosis (50–60%) was provided by IETD-fmk in all of the three NSCLC lines analyzed(P = 0.025). Interestingly, the protection provided by IETD-fmk was similar to the blockade obtained with the pancaspase inhibitor zVAD-fmk, which again suggested a crucial role of caspase-8 (Fig. 7,A). In contrast, in Jurkat cells, IETD-fmk provided no protection (P = 0.2), whereas zVAD-fmk blocked at least 85% of drug-induced apoptosis(P = 0.006). These results were reproducible on using other drugs (topotecan or gemcitabine), concentrations(IC₅₀ or IC₈₀), or time point for analysis (data not shown). We also analyzed the cleavage of IETD-AFC as a marker of caspase-8 activity in the NSCLC cell lines H460, H322, and SW1573 during cisplatin- and Fas-induced apoptosis. The findings were then compared with the results observed in Jurkat cells under the same experimental conditions. Fas-induced death was used as a marker for caspase-8 activity during apoptosis. As shown in Fig. 7,B, the amount of IETD-AFC cleavage generated during cisplatin and Fas-induced apoptosis was similarly high in all of the three NSCLC cell lines analyzed. In contrast, in Jurkat cells, Fas exposure led to a substantial increase in IETD-AFC cleavage, whereas exposure to cisplatin had hardly any effect on caspase-8 activity (Fig. 7 B). These results are consistent with the findings observed with the caspase-inhibitor IETD-fmk. We thus conclude that our findings pointing to an important role for caspase-8 in chemotherapy-induced apoptosis in H460 cells can be likely extrapolated to other NSCLC cells.

Finally, to further examine the relevance of caspase-8 in drug-induced cytotoxicity in H460 cells, we questioned whether the inhibition of this caspase would simply delay cell death or whether it would be able to rescue cells, favoring clonogenic survival. Clonogenic growth assays were performed in which control vector-transfected cells,together with CrmA and caspase-8-DN overexpressing cells, were exposed to IC₅₀ and IC₈₀concentrations of anticancer drugs. CrmA- and caspase-8-DN-expressing cells showed a significant increase in clonogenic potential when compared with cells transfected with the empty vector or the parental line (Table 2; P = 0.015 and 0.025, respectively). Similar observations were made in H460 cells overexpressing Bcl-2 or Bcl-x_L(P = 0.006). In contrast, stable expression of the caspase-9 inhibitors, caspase-9S and XIAP, did not increase clonogenic survival (Table 2), being consistent with their inability to suppress drug-induced apoptosis.

Studies to unravel the molecular basis of chemoresistance in tumor cells may help to design new strategies for cancer treatment. In this respect, great progress has been made in the understanding of the pathways that underlie chemotherapy-induced apoptosis, in which the sequential activation of caspases has been found to be required for the execution phase of the apoptotic process. A model has been proposed in which two different caspases, caspase-8 and -9, mediate distinct types of apoptotic stimuli: caspase-8 being the apical caspase in death-receptor-induced apoptosis, and caspase-9 being activated on anticancer drug treatment (7, 8, 9, 10). Although this model may explain the situation found in leukemia cells, it is not clear whether it can be applied to apoptotic responses triggered in solid tumors cells. In this study, we have investigated the involvement of caspase-8- and -9-dependent pathways in chemotherapy-induced apoptosis in NSCLC cell line H460.

Different lines of experimentation clearly indicate the importance and apical role of caspase-8 and the lack of function of caspase-9 in chemotherapy-induced apoptosis in H460 cells. Firstly, stable overexpression of the caspase-9 inhibitors caspase-9S and XIAP in the NSCLC cell line H460 failed to suppress apoptosis that was induced by various chemotherapeutic drugs, which correlated with the observed lack of caspase-9 activity in these cells (Fig. 1). Secondly, drug-induced apoptosis was blocked in H460 transfectants that expressed known inhibitors of caspase-8, e.g., CrmA and caspase-8 DN(Fig. 2). These results are in line with the observation that the preferential caspase-8 peptide-inhibitor IETD-fmk blocked druginduced apoptosis in NSCLC cells (Fig. 7,A). The similar blockade of drug-induced apoptosis induced by IETD-fmk and the substantial amount of cleavage of IETD-AFC generated by exposure to chemotherapy in three distinct NSCLC cell lines suggest that our results may be characteristic of NSCLC cells, rather than being cell type-specific phenomena related to H460 cells. In addition, these findings are in contrast to the results obtained with drug-induced apoptosis in Jurkat cells, which we included as control in our studies(Figs. 1 and 7). These findings in Jurkat cells are in line with a previous report in HL-60 cells (31). Our findings in H460 cells expressing CrmA are at odds with the reported lack of suppression of drug-dependent apoptosis by CrmA in lymphoid cells(32, 47) but are in line with another report showing a blockage of drug-induced by CrmA in leukemia cells (48). A possible explanation for this discrepancy may be an effect of CrmA on other caspases. In fact, when analyzing our results, we have to take into consideration that both CrmA and chemical caspase inhibitors like IETD-fmk have been shown to inhibit caspases other than caspase-8 in cell-free systems (44). However, the confirmation of our findings in H460 cells overexpressing caspase-8-DN gives weight to our data and point to a crucial role of caspase-8 in the process of drug-induced apoptosis in the NSCLC cell line H460.

Some investigators have reported the activation of procaspase 8 in drug-induced apoptosis. However, in these studies, the activation of caspase-8 by chemotherapy appeared to be a mere bystander effect and not a key event in the execution of the apoptotic program(29, 30, 31, 49). An alternative pathway for chemotherapy-dependent activation of caspase-8 has been proposed in leukemia, neuroblastoma, and colon cancer cells and would be based on Fas/FasL signaling (50, 51, 52). However, we and others have been unable to substantiate this model (30, 38, 45, 53, 54, 55); and, moreover, our current finding that drug-induced apoptosis in H460 cells is not inhibited by FADD-DN overexpression further demonstrates that death receptors are not involved in mediating this response (Fig. 3). In H460 cells, caspase-8 activation appears to be an apical event in the execution phase, because we conclude from our finding that this activation precedes the chemotherapy-mediated increase in caspase-3-like activity and the PARP cleavage that take place independently from caspase-9 activation (Figs. 1 and 6).

In analogy with the mechanisms that can activate caspase-9, we studied the role of mitochondria in drug-dependent caspase-8 activation in H460 cells. On the basis of the following observations, we conclude that this activation requires a mitochondrial step (see Fig. 4):(a) overexpression of Bcl-2 and Bcl-x_L blocks caspase-8 cleavage as well as IETD-AFC proteolytic activity; (b) overexpression of the caspase-8 inhibitors CrmA and caspase-8-DN fails to block cytochrome crelease from mitochondria after exposure to chemotherapy; and(c) IETD-fmk and zVAD-fmk do not block cytochrome c release.

We are aware of one earlier report (56) describing mitochondria-controlled caspase-8 cleavage in neuroblastoma cells in which, on exposure to betulinic acid, caspase-8 was activated by apoptosis-inducing factor. Furthermore, in analogy with the mechanism of drug-dependent caspase-9 activation, a possibility would be that Apaf-1, together with cytochrome c and caspase-8 could form an alternative apoptosomal complex. This possibility is not unprecedented, inasmuch as caspase-8 has been found to be able to interact with Apaf-1, although these findings are under dispute(29, 57). In addition, caspase-8 has been shown to be able to cleave and, thus, activate all known caspases in vitro(58), which may be taken as an additional evidence for its candidacy as an initiator caspase in death-receptor-independent types of apoptosis (7). We are presently further exploring this potential mechanism of caspase-8 activation in NSCLC cells. Another question to be addressed is why NSCLC cells use a caspase-8-dependent pathway instead of the more commonly used caspase-9 route. When we take into consideration that caspase-9 and Apaf-1 are expressed at normal levels in H460 cells (Fig. 1 D), our findings suggest that caspase-9 itself or some component of the complex that is responsible for its activation may be constitutionally inhibited or nonfunctional in H460 cells. To the best of our knowledge, the present report provides the first indication for the existence of a constitutional mechanism of inhibition at this level in cancer cells. Possible means that render caspase-9 nonfunctional in these cells are currently under investigation in our laboratory.

An important issue that is disputed in cancer research concerns the correlation between chemotherapy-induced apoptosis and clonogenic survival (59). Controversial data have been generated not only about upstream members of the death machinery—such as p53, p21,and c-myc (9, 60, 61, 62)—but also about caspases. The use of peptide inhibitors of caspases was shown to abrogate the morphological hallmarks of apoptosis; however, peptide inhibitors failed to interfere with the final commitment of cells to die (63). In contrast, a correlation between apoptosis and clonogenic survival has been reported in cells overexpressing CrmA (32). In this study, we also found a close correlation between the suppressive effect of overexpression of the caspase-8 inhibitors, caspase-8-DN, and CrmA, on drug-induced cell death and the clonogenic outgrowth of H460 cells. Taken into a broader perspective, our results suggest that constitutional mechanisms of caspase inhibition may have a considerable impact on the resistance of cancer cells to chemotherapy, because, upon caspase blockade, cells were allowed to proliferate and were not doomed to die by an alternative or delayed mechanism.

In conclusion, our findings in H460 cells provide a novel picture of the events involved in the apoptosis induced by anticancer agents that challenges the assumption that chemotherapy-induced apoptosis is preferentially mediated in a caspase-9-dependent manner. We demonstrate that after mitochondria activation by anticancer drugs, caspase-8 is activated in a caspase-9-independent manner, and, in turn, mediates the activation of effector caspases leading to apoptosis. In fact, because of differences from the model reported in other cell types, our data suggest that the caspase pathway activated might vary not only according to the sort of stimuli, but also depending on the cellular context, as proposed previously (9). These findings may help to explain the difference in chemosensitivity between NSCLC cells and leukemia cell lines cells, and it would be particularly interesting if these observations can be confirmed in other solid tumor lines.

We thank Drs. David L. Vaux and Andreas Strasser (The Walter and Eliza Hall Institute of Medical Research, Victoria, Australia)for CrmA WT, loss-of-function CrmA, and Bcl-2/Bcl-x_L expression plasmids, respectively;Dr. Claudio Vincenz (The University of Michigan Medical School, Ann Arbor, MI) for FADD-DN and Flice-DN expression vectors; Dr. Kuni Matsumoto (Graduate School of Science, Nagoya University, Nagoya,Japan) for plasmid containing XIAP. We are also grateful to Dr. Xiadong Wang (University of Texas Southwestern Medical Center, Dallas,TX) for Apaf-1 antibody.