Identification of Caspase-Independent Apoptosis in Epithelial and Cancer Cells

Historically and currently, apoptosis is characterized by a number of morphological criteria including cell shrinkage, alterations in nuclear morphology, the maintenance of membrane integrity, and the formation of apoptotic bodies (Salvesen and Dixit, 1997; Saraste, 1999; Cummings et al., 2000; Saraste and Pulkki, 2000). Generally, these events are believed to be mediated by the activation of a family of cysteine-containing aspartate-directed proteases called caspases (Salvesen and Dixit, 1997; Saraste, 1999; Cummings et al., 2000). However, recent work suggests that these events can occur independently of caspases. For example, cell shrinkage and decreases in mitochondrial membrane potential in Jurkat cells exposed to Ca²⁺ ionophores proceeds in the presence of the pan caspase inhibitor ZVAD-fmk (Bortner and Cidlowski, 1999). Furthermore, toxicant-induced phosphatidylserine externalization occurs in primary T cells (Ferraro-Peyret et al., 2002), renal epithelial cells (Cummings and Schnellmann, 2002), and L1210 cells (Belmokhtar et al., 2001) in the presence of caspase inhibitors and the absence of caspase activity. Other studies suggesting that apoptosis occurs in the absence of caspase activity include those using N-methyl-d-aspartate (Lankiewicz et al., 2000), staurosporine (Lankiewicz et al., 2000; Nutt et al., 2002), and ceramide (Jones et al., 1999; Perry et al., 2000).

We recently demonstrated that 50% of cisplatin-induced renal proximal tubule cell (RPTC) apoptosis proceeded independently of either p53 nuclear translocation or caspase activity (Cummings and Schnellmann, 2002). Apoptosis in RPTC was measured by multiple methods including phosphatidylserine externalization, chromatin condensation, and DNA hypoploidy. Although the above studies support the hypothesis that pathways for caspase-independent apoptosis exist, in many cases insufficient markers were used to assess apoptosis in these studies, or only one model system and one apoptosis-inducing agent was studied. Consequently, it is difficult to link specific apoptotic endpoints to caspase activity. Another issue is the relative role of oncosis during apoptosis in the presence of caspase inhibitors has not been examined in many studies.

It is important to determine the mechanisms of apoptosis in the absence of caspase activity. Such data have clinical implications. For example, the efficacy of caspase inhibitor treatment on acute organ dysfunction following ischemic/reperfusion injury or toxicant exposure is dependent upon a major role of caspases in apoptosis (Iwata et al., 2002; Scarabelli et al., 2002). In cancer, the elucidation of apoptosis in the absence of caspase activity may identify new pharmacological targets for cancer treatment (Johnstone et al., 1999; Kolenko et al., 2000; Jaattela, 2002; Mathiasen and Jaattela, 2002). Furthermore, if noncancerous tissue primarily exhibits caspase-dependent apoptosis whereas cancer cells exhibit caspase-independent apoptosis, it may be possible to kill the cancer cells while protecting normal tissue with caspase inhibitors.

The goal of this work was to determine the prevalence of apoptosis in the absence of caspase activity in normal and cancer cells exposed to multiple apoptosis-inducing agents. In addition, several morphological and biochemical markers of apoptosis were used to determine the ability of caspase inhibitors to alter the endpoints of apoptosis in both normal and cancer cells.

Materials and Methods

Materials. Female New Zealand White rabbits (1.5–2.0 kg) were purchased from Myrtle's Rabbitry (Thompson Station, TN). l-Ascorbic acid-2-phosphate (magnesium salt) was obtained from Wako Bioproducts (Richmond, VA). DEVD-afc (caspase-3 substrate) was purchased from BioVision (Palo Alto, CA). The caspase-3 inhibitor DEVD-fmk, the general pan caspase inhibitor ZVAD-fmk, and annexin V-FITC were obtained from R&D Systems (San Diego, CA). Vincristine and 4′,6-diamidino-2-phenylindole-dihydrochloride (DAPI) were purchased from Calbiochem (La Jolla, CA). Cisplatin, staurosporine, A23187 (calcimycin), propidium iodide (PI), and all other chemicals and materials were obtained from Sigma-Aldrich (St. Louis, MO).

Primary Culture of RPTC and Cancer Cell Lines. Renal proximal tubules were isolated from New Zealand White rabbits by the iron oxide perfusion method as described previously (Nowak and Schnellmann, 1995; Nowak and Schnellmann, 1996). The human epithelial cancer cell lines Caki-1 (kidney carcinoma), A549 (lung carcinoma), A172 (glioblastoma), and mouse lymphocytic leukemia L1210 cells were purchased from American Type Culture Collection (ATCC; Manassas, VA) and grown under the conditions recommended by ATCC for each cell line. Experiments using RPTC were performed on confluent monolayers (day 6 after isolation). Experiments using cancer cell lines were performed when cells were 80% confluent and at least 24 h after passage.

Treatment of RPTC and Cancer Cell Lines. Apoptosis was studied at times and concentrations of toxicants determined to result in the maximal amount of apoptosis in the absence of oncosis. RPTC were exposed to cisplatin, staurosporine and vincristine, and A23187 for 24 h. Caki-1 and A172 cells were exposed to 50 μM cisplatin for 48 h, whereas A549 and L1210 cells were exposed to 10 μM cisplatin for 36 h. For staurosporine, Caki-1 cells were exposed to 0.3 μM for 48 h, A549 and L1210 cells to 0.3 μM for 36 h, and A172 cells to 1 μM for 24 h. All treatments included diluent controls [dimethyl sulfoxide at <0.1% (v/v)].

Measurement of Annexin V and PI Staining. Annexin V and PI staining were used to assess both apoptosis and oncosis and were determined using flow cytometry as described previously with modifications (Schutte et al., 1998; Goldberg et al., 1999). Briefly, medium was removed, then cells were washed twice with PBS and incubated in binding buffer (10 mM HEPES, 140 mM NaCl, 5 mM KCl, 1 mM MgCl₂, 1.8 mM CaCl₂, pH = 7.4) containing annexin V-FITC (25 μg/ml) and PI (25 μg/ml) for 10 min. Cells were washed three times in binding buffer and released from the monolayers using a rubber policeman, and staining was quantified using a FacsCalibur flow cytometer (BD Biosciences, Franklin Lakes, NJ). For the nonadherent L1210 cells, cell suspensions were isolated by centrifugation at 1000g for 10 min, and the resulting pellet was resuspended in binding buffer and processed. Ten thousand cells were analyzed per measurement.

Measurment of Cell Shrinkage. Cell shrinkage was determined by analysis of forward and side scatter of control and treated RPTC using flow cytometry (Ruppova et al., 1999; Otsuki et al., 2003). Briefly, RPTC were treated with solvent control or ZVAD-fmk for 30 min prior to treatment with toxicants. After 24 h, the medium was removed and stored at 4°C, and the remaining cells were released from the monolayer using Cell Stripper (Mediatech, Inc., Herndon, VA). Cells present in the media were combined with those released from the dish, vortexed, and forward and side scatter was determined using a flow cytometer.

Determination of Caspase Activity. Caspase-3-like activity was used as an indicator of caspase activation and was measured using attached and detached cells and the fluorometric substrate DEVD-afc following the protocols of the Caspase Activity assay kit from BioVision. Activity was monitored as the linear release of the afc side chain and compared with a standard curve generated on the same microplate.

Measurment of Terminal Deoxynucleotidyl Transferase (TdT)-Mediated dUTP-Biotin Nick-End Labeling (TUNEL) Staining. DNA strand breaks during apoptosis were detected by TUNEL of free 3′-OH ends of cleaved DNA. Following toxicant exposure, RPTC were released from the monolayer using Cell Stripper, fixed at 1 × 10⁶ cells/ml in 1% (v/v) paraformaldehyde for 30 min on ice, washed in PBS, and stored in 70% (v/v) ice-cold ethanol at -20°C for at least 48 h. Fixed cells were washed once with PBS and incubated for 60 min at 37°C in 50 μl of TdT-containing solution (Roche Diagnostics, Mannheim, Germany) (20 U of TdT, 10 μlof5× reaction buffer, and 0.25 nmol of fluorescein-12-dUTP). Following TUNEL staining, all samples were washed once in rinsing buffer (0.1% Triton X-100, 5 mg/ml bovine serum albumin) and resuspended in 0.5 ml of PBS containing 10 μg/ml PI and 200 μg/ml DNase free-RNase. TUNEL staining in cells was analyzed within 3 h of staining using a FacsCalibur flow cytometer.

Assessment of Nuclear Morphology. After treatment, RPTC were washed twice with PBS, medium was removed, cells were fixed for 20 min using 10% buffered formalin/4% formaldehyde and washed with PBS. After washing, RPTC were incubated at 4°C with DAPI (16.6 μM final concentration) for 2 h. RPTC were washed three times, covered with mounting media, and cover slips were applied. Visualization of DAPI staining was performed using a Nikon TE300 Eclipse Fluorescence microscope (Nikon, Melville, NY) with excitation and emission filters of 350 and 486 nm, respectively. Apoptotic nuclei were scored based on the appearance of three different morphologies in the absence of PI staining: chromatin condensation = chromatin margination without nuclear condensation; nuclear fragmentation = chromatin margination with nuclear condensation; nuclear condensation = nuclear condensation without chromatin margination.

Protein Determination. Protein determination was performed using the bicinchonic acid assay method as described by Sigma-Aldrich.

Statistical Analysis. RPTC isolated from one rabbit represented one experiment (n = 1). For experiments with cancer cell lines, cells isolated from distinct passages equaled distinct experiments (n = 1). Each experiment represents an n ≥3. The appropriate analysis of variance was performed for each data set using SigmaStat statistical software. Individual means were compared using Fisher's protected least significant difference test with P ≤ 0.05 being considered indicative of a statistically significant difference between mean values.

Results

Effect of Toxicants on Annexin V and PI Binding in RPTC. Previous time- and concentration-dependent studies demonstrated that exposure of RPTC to 50 μM cisplatin results in maximal RPTC apoptosis after 24 h (Cummings and Schnellmann, 2002). To determine the concentration-dependent cytotoxicity of the general protein kinase C inhibitor staurosporine, the anticancer agent and inhibitor of microtubule formation vincristine, and the Ca²⁺-ionophore A23187, RPTC were treated with each compound for 24 h. After treatment, cell death was assessed by determination of annexin V (apoptotic marker) and PI (oncotic marker) binding using flow cytometry (Fig. 1). Staurosporine produced a concentration-dependent increase in annexin V labeling, in the absence of PI staining, with a maximal increase at 2 μM (Fig. 1A). Similar results were seen in RPTC exposed to vincristine or A23187 for 24 h (Fig. 1, B and C). Based on these data, 2 μM staurosporine and vincristine, and 10 μM A23187 were chosen for further study.

Effect of Caspase Inhibitors on Toxicant-Induced Annexin V Binding in RPTC. We demonstrated previously that 50% of cisplatin-induced RPTC apoptosis progressed in the absence of caspase-3, -8, and -9 activity and in the presence of the pan caspase inhibitor ZVAD-fmk (Cummings and Schnellmann, 2002). To determine whether RPTC apoptosis induced by other toxicants proceeds in the presence of caspase inhibitors, RPTC were treated with either solvent control or 50 μM ZVAD-fmk for 30 min prior to exposure for 24 h to cisplatin (50 μM), staurosporine (2 μM), vincristine (2 μM), and A23187 (10 μM). As previously demonstrated (Cummings and Schnellmann, 2002), cisplatin resulted in 25% of RPTC being positive for annexin V and treatment of RPTC with ZVAD-fmk reduced annexin V binding approximately 50% after 24 h (Fig. 2A). Staurosporine, vincristine, and A23187 induced similar increases in annexin V binding. Similar to cisplatin, ZVAD-fmk inhibited staurosporine- and vincristine-induced annexin V binding 50%. However, ZVAD-fmk did not significantly alter A23187-induced annexin V binding. Similar results were seen when the caspase-3 inhibitor DEVD-fmk (50 μM) was used (data not shown).

Effect of Caspase Inhibitors on Toxicant-Induced RPTC Shrinkage. All toxicants produced RPTC shrinkage to varying degrees (Fig. 2B). Staurosporine induced the highest degree of cell shrinkage, followed by cisplatin, A23187, and vincristine. Treatment of RPTC with ZVAD-fmk inhibited cisplatin- and staurosporine-induced cell shrinkage by approximately 50% but had no effect in RPTC exposed to vincristine or A23187. Similar results were seen using DEVD-fmk (data not shown).

Toxicant-Induced Caspase-3 Activation in RPTC. To ensure that the toxicants used were activating caspases, caspase-3 activity was determined in RPTC after toxicant exposure (Fig. 3). As previously reported (Cummings and Schnellmann, 2002), cisplatin exposure resulted in a 50-fold increase in caspase-3 activity as measured by cleavage of the fluorometric caspase-3 substrate DEVD-afc after 24 h. Staurosporine, vincristine, and A23187 also induced significant but smaller increases (10-fold) in caspase-3 activity after 24 h. Pretreatment of RPTC with ZVAD-fmk completely inhibited caspase-3 activation induced by all toxicants tested. Similar results were seen with DEVD-fmk (data not shown).

Toxicant-Induced RPTC TUNEL Staining. Cisplatin increased RPTC TUNEL staining to 18 ± 3% compared with 7 ± 4% in controls, whereas vincristine increased TUNEL staining to 15 ± 3%. Staurosporine and A23187 induced higher levels of TUNEL staining than either cisplatin or vincristine (Fig. 4). Treatment of RPTC with ZVAD-fmk prior to toxicant treatment decreased staurosporine-induced TUNEL staining approximately 50%, but had no effect on cisplatin-, vincristine-, and A23187-induced TUNEL staining.

Alterations in Nuclear Morphology during Toxicant-Induced RPTC Apoptosis. Cisplatin exposure alone resulted in 32% of RPTC being positive for chromatin condensation, a small increase in nuclear condensation, and no increase in nuclear fragmentation compared with controls (Figs. 5 and 6). In contrast, the most prominent type of nuclear morphology present in staurosporine-treated cells was nuclear fragmentation, followed by smaller amounts of chromatin condensation and nuclear condensation (Figs. 5 and 6). Vincristine exposure did not significantly increase chromatin condensation or nuclear fragmentation, but did increase nuclear condensation compared with controls (Figs. 5 and 6). Finally, treatment of RPTC with A23187 increased nuclear condensation but had no effect on chromatin condensation or nuclear fragmentation compared with controls (Figs. 5 and 6).

ZVAD-fmk did not alter toxicant-induced chromatin condensation (Fig. 6A) or nuclear fragmentation (Fig. 6B). In contrast, ZVAD-fmk blocked cisplatin-, staurosporine-, and vincristine-induced nuclear condensation but had no affect on A23187-induced nuclear condensation (Fig. 6C). Similar results were seen with DEVD-fmk (data not shown).

Toxicant-Induced Apoptosis in Cancer Cell Lines. The above data demonstrate that toxicants induce apoptosis in RPTC in the presence of caspase inhibitors. To determine whether apoptosis proceeds in the presence of caspase inhibitors in other cell models, we tested the ability of ZVAD-fmk to inhibit apoptosis in four different cancer cell lines exposed to cisplatin and staurosporine. Cisplatin and staurosporine exposure resulted in a concentration-dependent increase in annexin V binding, in the absence of PI staining, in all cancer cell lines tested (Figs. 7 and 8). However, at higher concentrations and in some cases, increases in cells staining positive for both annexin V and PI were detected. Based on these data, one concentration was chosen for the study of the effect of caspase inhibitors on toxicant-induced apoptosis in cancer cells.

Effect of Caspase Inhibitors on Apoptosis in Cancer Cell Lines. Cisplatin induced significant increases in annexin V binding, in the absence of PI staining, in all cancer cell lines tested (Fig. 9, data for RPTC are from Fig. 2). Treatment of RPTC with ZVAD-fmk prior to cisplatin exposure had no affect on annexin V binding in L1210 cells but reduced annexin V binding to control levels in A172, A549, and Caki-1 cells (Fig. 9A). In contrast, ZVAD-fmk did not reduce staurosporine-induced annexin binding in any cancer cell line tested (Fig. 9B). Cisplatin and staurosporine did not induce oncosis, either in the presence or absence of ZVAD-fmk, as determined by PI staining.

The activation of caspases in these cancer cell lines was verified by measurement of caspase-3 activity (Fig. 10, data for RPTC are from Fig. 3). Similar to RPTC, cisplatin exposure increased caspase-3 activity in all cell lines tested (Fig. 10A). The highest levels of activity were present in A172 and Caki-1 cells with a smaller degree of activation present in L1210 and A549 cells. As in RPTC, treatment of all cancer cell lines with ZVAD-fmk completely inhibited cisplatin-induced caspase-3 activation (Fig. 10A). Staurosporine significantly increased caspase-3 activity in all cancer cell lines tested, and treatment of cells with ZVAD-fmk completely inhibited staurosporine-induced caspase-3 activity (Fig. 10B).

Discussion

Studies in primary cultures of RPTC demonstrated that 50% of cisplatin-induced apoptosis proceeds in the presence of multiple caspase inhibitors, including the general pan caspase inhibitor ZVAD-fmk, and in the absence of caspase-3, -8, and -9 activation (Cummings and Schnellmann, 2002). Our previous study, like those mentioned in the Introduction, was limited by the fact that only one agent was used to induce apoptosis, only one cell model was studied, and a limited number of apoptotic endpoints was examined. Thus, the goal of this study was to determine the occurrence and pathways of apoptosis in the presence of caspase inhibitors in multiple models using both biochemical and morphological markers of apoptosis and oncosis.

Using a variety of markers, caspase-independent apoptosis was widely observed in renal epithelial cells exposed to a diverse group of drugs/toxicants with different cellular effects. These results suggest that caspase-independent apoptosis is a common occurrence that must be considered when studying apoptosis. Caspase-independent apoptosis also was identified in numerous cancer cells. However, the advent of caspase-independent apoptosis was dependent on both the drug/toxicant and cell line studied. Thus, caspase-independent apoptosis needs further examination in all models.

Although all of the toxicants used in this study caused renal cell apoptosis, the diverse toxicants produced marked differences among the different endpoints. For example, cisplatin, staurosporine, vincristine, and A23187 produced equivalent levels of annexin V binding. However, the levels of cell shrinkage, TUNEL staining, caspase-3 activity, and nuclear morphology varied greatly and did not correlate to annexin V binding or to each other (Table 1). The two agents more commonly used in the literature to produce toxicant-induced apoptosis (cisplatin, staurosporine) exhibited the endpoints commonly used to identify and characterize apoptosis (annexin V binding and caspase activity and cell shrinkage). In contrast, vincristine and A23187 produced apoptosis with only a small amount of, or no, cell shrinkage, chromatin condensation, and nuclear fragmentation (Table 1). These results clearly point out the diversity in cellular responses to toxicants that initiate apoptosis.

There also were marked differences in RPTC responses to the caspase inhibitors ZVAD-fmk and DEVD-fmk. Caspase inhibition had no effect on cisplatin-, vincristine-, and A23187-induced TUNEL staining, whereas caspase inhibition decreased staurosporine-induced TUNEL staining approximately 50%. Yet caspase activity was completely inhibited. With the exception of vincristine-induced nuclear condensation, caspase inhibition had no effect on toxicant-induced chromatin condensation, nuclear fragmentation, and nuclear condensation. The marker most sensitive to caspase inhibition was annexin V, whereas cell shrinkage, TUNEL staining, chromatin condensation, nuclear fragmentation, and nuclear condensation were mostly insensitive.

In addition to RPTC, cancer cells also demonstrated diverse responses to caspase inhibitors. Unlike its effect in RPTC, ZVAD-fmk totally inhibited cisplatin-induced increases in annexin V binding in almost all cancer cell models studied. These results demonstrate that cisplatin-induced apoptosis in several cancer cell lines, including a renal cell carcinoma, proceeds through caspase activity. However, staurosporine-induced annexin V binding in these same cancer cell lines was unaffected by ZVAD-fmk. These results illustrate that staurosporine-induced apoptosis in several cancer cell lines does not require caspase activity, a distinct difference from cisplatin. The inability of caspase inhibitors to decrease staurosporine-induced phosphatidylserine externalization in the cancer cell lines studied is not a novel finding, as previous studies have demonstrated that caspase inhibition has no effect of phosphatidylserine externalization including those performed in L1210 cells (Belmokhtar et al., 2001). Taken together with the RPTC data, these results demonstrate diversity in cellular responses to caspase inhibition and demonstrate that caspase-independent apoptosis is not a mechanism specific to one cell type.

The mechanisms mediating caspase-independent apoptosis in RPTC or cancer cells are not fully elucidated. However, in the absence of caspase activity, a number of cellular events still occurred in drug/toxicant-exposed RPTC and cancer cells including phosphatidylserine externalization (as measured by annexin V binding), alterations in nuclear structure, and cell shrinkage. The mechanisms controlling these events must be intrinsic to the mechanism of caspase-independent apoptosis. In regards to this hypothesis, phosphatidylserine externalization can occur independently and dependently of caspases (Vanags et al., 1996; Kagan et al., 2000). Noncaspase-mediated increases in phosphatidylserine externalization can occur in response to increases in intracellular Ca²⁺, which alters scramblase and translocase activity, two mediators of phosphatidylserine externalization (Vanags et al., 1996; Kagan et al., 2000). Additionally, noncaspase proteases may be activated that cleave the cytoskeleton proteins attached to phospholipids, including focal adhesion kinase and the actin-capping protein α-adducin (van de Water et al., 1999; van de Water et al., 2000). However, the identity of these proteases is not known. Similarly, alterations in nuclear structure (chromatin condensation, nuclear fragmentation, DNA strand breaks) may be controlled also by noncaspase-mediated proteolytic activation of DNases. These would be the same DNases that are activated by caspases during normal apoptosis (Paddenberg et al., 1996; Enari et al., 1998; Samejima et al., 1998). Finally, the mechanisms controlling caspase-independent cell shrinkage most likely involve alterations in ion channels/transporters such as those involved in K⁺ transport (Yu et al., 1997; Yu and Choi, 2000). In support of this hypothesis, K⁺ efflux does mediate cell shrinkage in a caspase-independent manner in lymphocytes undergoing chemotherapeutic-induced apoptosis (Bortner and Cidlowski, 1999; Zhang et al., 1999). Preliminary studies to investigate caspase-independent apoptosis demonstrate that inhibition of poly(ADP-ribose) polymerase-polymerase, mitogen activated protein kinases, or phospholipase A₂ have no affect on caspase-independent RPTC apoptosis (data not shown).

In conclusion, we demonstrate that apoptosis induced by diverse stimuli occurs in the presence of caspase inhibitors in several different models of apoptosis. A pharmacological technique was employed to inhibit caspases in this study, because advances in molecular techniques not withstanding, no current molecular method exists allowing for the inhibition of known caspase family members in a single cell model. Even if such a model existed, it may not account for undiscovered caspases. The pharmacological methods employed herein allow for an initial screening of cell models for the possible existence of caspase-independent apoptosis. Such studies are critical in determining whether inhibition of caspase-independent apoptosis will decrease toxicant-induced cell death in normal epithelial cells or alter cell death in tumor cells.