Activation of Caspase Pathways during Iron Chelator-mediated Apoptosis*

Iron chelators have traditionally been used in the treatment of iron overload. Recently, chelators have also been explored for their ability to limit oxidant damage in cardiovascular, neurologic, and inflammatory disease as well as to serve as anti-cancer agents. To determine the mechanism of cell death induced by iron chelators, we assessed the time course and pathways of caspase activation during apoptosis induced by iron chelators. We report that the chelator tachpyridine se-quentially activates caspases 9, 3, and 8. These caspases were also activated by the structurally unrelated chelators dipyridyl and desferrioxamine. The critical role of caspase activation in cell death was supported by micro-injection experiments demonstrating that p35, a broad spectrum caspase inhibitor, protected HeLa cells from chelator-induced cell death. Apoptosis mediated by tachpyridine was not prevented by blocking the CD95 death receptor pathway with a Fas-associated death domain protein (FADD) dominant-negative mutant. In contrast, chelator-mediated cell death was blocked in cells microinjected

Iron is a required cofactor in numerous essential enzymes, including respiratory enzymes and ribonucleotide reductase, the enzyme catalyzing the rate-limiting step in DNA synthesis. However, iron can also promote the formation of hydroxyl radicals through Fenton chemistry. Maintaining iron availability for cellular metabolic and growth requirements is critical to the survival of both prokaryotes and eukaryotes.
The availability of extracellular and intracellular iron can be affected dramatically by iron chelators. Studies of iron chelators have largely focused on their ability to mobilize iron from iron-overloaded patients and tissues (1,2). However, the potential uses of chelators extend beyond iron overload. To the extent that chelators prevent the participation of iron in oxygen radical formation, iron chelators may mitigate the oxidative stress damage resulting from ischemia/reperfusion (3), neurodegenerative disease (4,5), inflammation (6), or chemotherapeutic drugs (7). In these settings, chelators are used for their protective cellular and organismal effects.
Iron chelators have also been tested for their potential antiproliferative and cytotoxic effects. Desferrioxamine (DFO), 1 a bacterial siderophore and the drug of choice in the treatment of iron overload, as well as pyridoxal isonicotinoyl hydrazone and related chelators (8), the 8-hydroxyquinoline-based chelator O-Trensox (9), and tachpyridine (10), are actively being explored for their cytotoxic and cytolytic properties.
Iron chelators can induce both cell cycle arrest and programmed cell death, or apoptosis. Thus, treatment of activated T lymphocytes and HL-60 cells with the iron chelators deferiprone or DFO induced apoptosis (11). The human leukemic cell line CCRF-CEM treated with DFO exhibited morphological features of apoptosis after 48 h (12). The iron chelators DFO, dithizone, and hinokitol induced apoptosis in F9 embryonal carcinoma cells (13), and the chelator O-Trensox induced apoptosis in HepG2 cells (9). Treatment of a mouse B cell lymphoma cell line 38C13 with either monoclonal antibodies against the transferrin receptor, DFO, or a defined culture medium without supplemental iron (iron-poor medium), showed that these cells died by apoptosis (14). Finally, iron chelation has been shown to induce apoptosis in Kaposi's sarcoma cells (15) and neuroblastoma cells (16).
Our group has synthesized and described N,NЈ,NЈЈ-tris(2pyridylmethyl)-cis,cis-1,3,5-triaminocyclohexane (tachpyridine) (10,17,18). Tachpyridine is a hexadentate chelator that binds metals through six nitrogen donor ligands. Both tachpyridine and DFO can repress synthesis of the iron-binding protein ferritin, the synthesis of which is critically dependent upon intracellular iron, suggesting that these two chelators share a similar ability to deplete intracellular iron pools (10,19). We have shown previously that tachpyridine is cytotoxic * This work was supported by Grant DK 57781 (to S. V. T.) from the National Institutes of Health. Preclinical development of tachpyridine was supported by a grant from the NCI Rapid Access to Intervention Development Program, National Institutes of Health (to S. V. T.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
§ § To whom correspondence should be addressed: Dept. of Biochemistry, Wake Forest University School of Medicine, Hanes Bldg., Rm because of its ability to bind one or more of the metals iron, zinc, or copper, because complexes of tachpyridine with these metals (but not others) are nontoxic (10). Because N-alkylated derivatives of tachpyridine which are sterically hindered in their ability to react with iron are no longer toxic (10), we have speculated that iron deprivation is an important component of the toxicity of tachpyridine (10). Tachpyridine induces an apoptotic mode of cell death which is p53-independent (20).
Mechanisms of apoptosis mediated by iron chelators are poorly understood. The experiments described here were designed to determine the molecular pathway utilized by iron chelators to induce apoptosis. In particular, we asked whether chelators engage elements of one or both of two well studied apoptotic pathways. The first of these is the death receptor pathway, originally characterized following binding of ligands to extracellular death receptors such as the tumor necrosis factor (TNF) receptor, CD95/Fas and DR5 (for review, see Ref. 21). Binding of ligands to these receptors triggers oligomerization and recruits proteins such as TRADD and FADD/Mort-1 to associate with the complex through their death domains. Activated death receptors in turn recruit and activate caspase 8, a cysteine protease that initiates a cascade of terminal or "executioner" caspases, including caspase 3 and others. Unleashing of executioner caspases results in cleavage of a number of caspase substrates, including protein kinases, cytoskeletonassociated proteins, transcription factors, and proteins involved in DNA repair and chromatin structure such as poly-(ADP-ribose) polymerase (PARP).
A second pathway that leads to apoptosis is the activation of caspases via a mitochondrial mechanism (for review, see Ref. 21). Mitochondrial participation in apoptosis is characterized by the release of cytochrome c from mitochondria into the cytosol. Released cytochrome c forms a complex with procaspase 9 and Apaf-1. In this "apoptosome," procaspase 9 becomes activated, triggers executioner caspases, and initiates terminal events that overlap with those initiated by death receptors. The mitochondrial apoptotic pathway is modulated by members of the Bcl-2 family, proteins that localize to the cytosolic aspect of membranes, including mitochondrial membranes. Members of this family include the anti-apoptotic proteins Bcl-2 and Bcl-XL as well as the pro-apoptotic protein bax (for review, see Ref. 22). Overexpression of the anti-apoptotic Bcl-2 family members Bcl-2 and Bcl-XL prevents cytochrome c release and the mitochondrial permeability transition and promotes cell survival (22).
We report that tachpyridine, DFO, and dipyridyl activate a caspase cascade. Induction of apoptosis by tachpyridine is characterized by an early activation of caspase 9 followed by the sequential activation of caspase 3 and caspase 8. Further, chelator-mediated cell death can be blocked by a dominantnegative caspase 9 and Bcl-XL overexpression. These results indicate that the mitochondrial pathway makes an important contribution to iron chelator-mediated cell death.

MATERIALS AND METHODS
Cell Culture-The human cervical cancer cell line HeLa was obtained from the tissue culture core laboratory (Comprehensive Cancer Center of WFUSM). HeLa cells stably transfected with a pGFP-⌬FADD (dominant-negative mutant) expression vector or pGFP control vector were kindly provided by Dr. Harald Wajant (University of Stutttgart, Stuttgart, Germany). All cell lines were grown in Dulbecco's modified Eagle's medium supplemented with 10% heat-inactivated fetal bovine serum and antibiotics (100 units/ml penicillin, 100 g/ml streptomycin), in a 5% CO 2 incubator at 37°C. For all experiments, cells were seeded at appropriate concentrations to ensure exponential growth.
Iron Chelators and Chemotherapeutics-Tachpyridine was synthesized from cis-1,3,5-triaminocyclohexane as its nitrate salt according to methods published previously (17,18). The identity of tachpyridine was confirmed by 1 H, 13 C NMR as described (17) and was Ͼ99% pure. The compound was prepared as a 1 mM stock in phosphate-buffered saline (PBS), pH 7.4, 0.22 M filter sterilized, and stored at 4°C. Appropriate volumes were added from this stock to cell cultures. DFO was obtained from Sigma, and on the day of the experiment, it was prepared as a 50 mM stock in complete medium, filter sterilized, and added to cultures from this stock. Etoposide and 2Ј,2Ј-dipyridyl were obtained from Sigma, and on the day of the experiment, they were prepared as a 50 mM stock in dimethyl sulfoxide and added to cultures from this stock. The final concentration of dimethyl sulfoxide was 0.4% or less. Dimethyl sulfoxide at these concentrations did not result in caspase activation (not shown). Cycloheximide was obtained from Sigma and was prepared as 1 mg/ml stock in 95% ethanol and stored at Ϫ20°C. TNF-␣ was obtained from R & D Systems (Oxon, UK), prepared as a 10 ng/l stock in PBS, pH 7.4, containing 0.5% bovine serum albumin and stored at Ϫ80°C.
Time Lapse Video Microscopy-HeLa cells were plated onto T-25 flasks (Corning, Cambridge, MA) and allowed to attach overnight. Chelator was added, and cells were placed in dedicated incubator of an inverted phase contrast video microscope (Axiovert, Zeiss) for observation. A cell was considered to have entered apoptosis at the first appearance of cell shrinkage and membrane blebs, morphological hallmarks of apoptosis (24).
Western Blot Analysis-Treated or control cells were harvested by scraping into medium, washed in ice-cold PBS, and pellets were frozen until analysis. For caspase blots, pellets were suspended in ice-cold Nonidet P-40 lysis buffer (50 mM Tris, pH 8.0, 5.0 mM EDTA, 150 mM NaCl, 0.5% Nonidet P-40) containing a mixture of protease inhibitors (Complete; Roche Molecular Biochemicals), including phenylmethylsulfonyl fluoride (0.1 mg/ml), and sonicated for 15 s. Aliquots representing equal amounts of protein from each lysate were separated on a 12 or 16% SDS-PAGE and analyzed by Western blotting using a rabbit antihuman caspase 3 polyclonal antibody (Stressgene, San Diego, CA), a mouse anti-human caspase 9 monoclonal antibody (Stressgene), or a rabbit anti-human caspase 8 polyclonal antibody (PharMingen, San Diego). For Western blot analysis of PARP cleavage, pellets were suspended in ice-cold lysis buffer (62.5 mM Tris⅐Cl, pH 6.8, 6 M urea, 10% glycerol, 2% SDS, 0.00125% bromphenol blue, and 5% ␤-mercaptoethanol), sonicated for 15 s, and then incubated at 65°C for 15 min. Lysates representing equal numbers of cells were then loaded onto 12% SDS-PAGE and analyzed by Western blotting using mouse anti-PARP monoclonal antibody (Oncogene Research Products, Cambridge, MA). Primary antibodies were detected using a peroxidase-conjugated secondary antibody and enhanced chemoluminescence according to the manufacturer's instructions (Amersham Biosciences). Equivalent loading and protein transfer were confirmed by Western blot using a mouse anti-human ␤-actin antibody (Sigma) and by staining membranes with Ponceau S (Sigma). For quantitation, Western blots were prepared with an internal standard curve consisting of three dilutions of a sample of cleaved caspase; densitometric analyses of these standards were linear with dilution (correlation coefficient r ϭ 1.0, 0.96, and 0.99 for caspase 9, 3, and 8, respectively). Signal intensities were quantitated using the ChemiImager 5500 and AlphaEase FC software (Alpha Innotech Corp., San Leandro, CA). Signal intensities for experimental samples fell within the range of the standard curve.
Caspase Enzymatic Assays-Enzymatic assays were carried out in 96-well microtiter plates (Costar, Cambridge, MA). Treated or control cells were harvested by scraping into medium, washed in ice-cold PBS, and pellets were frozen at Ϫ80°C until analysis. Pellets were suspended in ice-cold lysis buffer obtained from Alexis (San Diego), placed on ice for 10 min, and then centrifuged at 13,000 ϫ g for 15 min at 4°C. 100 g of protein was diluted in 50 l of lysis buffer and added to 40 l of reaction buffer (20 mM PIPES, pH 7.2, 100 mM NaCl, 10% sucrose, 0.1% CHAPS, 20 mM ␤-mercaptoethanol). 10 l of the appropriate caspase substrate was then added to give a final concentration of 100 M. The plates were incubated for 3 h in a 5% CO 2 incubator at 37°C. The substrates for each caspase were as follows: Ac-DEVD-pNA for caspase 3-like activity (Alexis), Ac-LEHD-AFC (Calbiochem) for caspase 9 activity, and Z-IETD-AFC (Calbiochem) for caspase 8 activity. These substrates are preferentially but not exclusively cleaved by the indicated caspases (25). The substrates were prepared as 1 mM stocks according to the manufacturer's instructions and stored at Ϫ20°C. Levels of released p-nitroanilide (pNA) were measured by measuring absorbance at 405 nm. Alternatively, levels of released 7-amido-4-methylcoumarin (AFC) were measured using excitation at 420 nm and emission at 517 nm.
Cell Viability Assay-500,000 HeLa pGFP-⌬FADD or pGFP control cells were seeded in 60-mm dishes (Becton-Dickinson Labware, Franklin Lakes, NJ) and allowed to attach overnight. The next day, plates were treated with 25 M tachpyridine, 100 M etoposide, or 100 ng/ml TNF-␣ plus 2.5 g/ml cycloheximide. At this time, untreated (control) plates of cells were counted using trypan blue exclusion to determine the number of cells at time of treatment. After various time intervals, treated plates were counted by trypan blue exclusion, and percent survival relative to untreated (control) was determined.
Microinjections-HeLa cells were seeded in 35-mm dishes (Becton-Dickinson Labware) and allowed to attach overnight. The next day, expression plasmids at 200 ng/l (pYFP C1) or 250 ng/l (Bcl-XL, p35, pcDNA3.1, p⌬caspase9) were injected into the cell nucleus using an Eppendorf microinjector, according to methods published previously (26). For each dish, pYFP C1 was either coinjected with the control plasmid (pcDNA3.1, control) or coinjected with cDNA encoding the protein of interest (Bcl-XL, p35, p⌬caspase9, test plasmid). Cells that had been microinjected successfully were identified by the fluorescence of the YFP protein using an Axiovert S100 fluorescent microscope and 450 -490/515-565 nm excitation/emission filters. To quantitate cell survival, cells were visualized and counted 4 -5 h after injection. The medium was then replaced with either control medium or medium containing drug, and incubation was continued for an additional 12 h for TNF/cycloheximide or 24 h for tachpyridine and etoposide. Percent survival of cells injected with the control plasmid or the test plasmid was determined by counting the number of surviving cells for each after treatment and dividing this number by the number of initial cells obtained from counting 4 -5 h after injection. In most cases, 50 -120 cells/dish/plasmid(s) were microinjected in each experiment.
Immunostaining of Caspase 3 and DAPI Staining-HeLa cells were seeded onto 35-mm dishes and allowed to attach overnight. The next day, control medium or medium containing drug was added, and cells were treated for the appropriate time. After treatment, nonadherent cells were collected by centrifugation at 200 ϫ g for 5 min at 4°C (untreated control cells remained adherent). The pellet was washed in PBS, resuspended in PBS, and plated onto 35-mm dishes coated with 1 mg/ml polylysine. For staining, both adherent (control) and nonadherent (drug-treated) cells were fixed in 10% formalin in PBS at room temperature for 10 min and then rinsed with PBS. 0.2% Triton X-100 in PBS was added at room temperature for 10 min, and plates were then rinsed again with PBS. Blocking was performed using 1% bovine serum albumin in PBS at room temperature for 10 min followed by incubation with an antibody that reacts selectively with activated caspase 3 (rabbit anti-human caspase 3 antibody, Cell Signaling Technology, Beverly, MA) (27) in 1% bovine serum albumin at room temperature for 30 min. Plates were washed with PBS and then incubated with rhodamineconjugated goat anti-rabbit antibody (Molecular Probes, Eugene, OR) at room temperature for 30 min. After washing in PBS, plates were fixed again in 10% formalin at room temperature for 10 min. DAPI in methanol was then added for 5 min, plates were washed with methanol and PBS and then visualized using fluorescence microscopy.

Tachpyridine-mediated Apoptosis Involves Activation of a
Capsase Cascade-Using a number of criteria, including membrane blebbing and blistering as visualized by time lapse video microscopy, terminal nucleotidyl transferase-mediated UTP nick end labeling (TUNEL) assays, chromatin condensation, and internucleosomal cleavage of DNA, we have shown previously that the iron chelator tachpyridine induces an apoptotic mode of cell death (20). To begin to explore molecular pathways utilized by tachpyridine to induce apoptosis, we first asked whether members of the cysteine protease family of caspases were involved. For these experiments we used HeLa cells, a cell line in which pathways of caspase activation in response to numerous agents, including chemotherapeutic drugs and death receptor ligands such as TNF, have been extensively characterized. Time lapse video microscopy indicated that these cells, like MCF7 and H1299 cells we have studied previously, undergo an apoptotic mode of cell death when treated with tachpyridine ( Fig. 1). Initially, we examined whether the caspase substrate PARP was cleaved in HeLa cells treated with 25 M tachpyridine (Fig. 2). As a positive control, HeLa cells were also treated with 100 M etoposide, a DNA-damaging agent known to induce apoptosis. As shown in Fig. 2, Western blotting revealed that 25 M tachpyridine induced discernible cleavage of PARP by 16 h, with more extensive cleavage of PARP at 18 h and beyond (Fig. 2).
Given that tachpyridine induced PARP cleavage, we next assessed activation of the "effector" caspase, caspase 3. Activation of caspase 3 was measured by its conversion from a 32-kDa zymogen to an active enzyme of 17 and 20 kDa using Western blotting. As a positive control, HeLa cells were treated with 100 M etoposide for 48 h. As seen in Fig. 3, treatment of HeLa cells with 25 M tachpyridine caused minor cleavage of caspase 3 by 14 h (Fig. 3). Pronounced cleavage of caspase 3 was seen after 16 h, and active caspase 3 continued to increase throughout the 30-h time course (Fig. 3). (Decreased sensitivity of the Western blot assay relative to the single cell morphological assay shown in Fig. 1 likely accounts for the slightly different kinetics in Figs. 1 and 3.) As expected, the appearance of the cleavage product was accompanied by a gradual loss of the zymogen precursor.
To date, two major "initiator" caspases have been described which result in the activation of effector caspases such as caspase 3: caspase 8, triggered by binding of ligands to death receptors, and caspase 9, activated by a mitochondrial pathway (21). To test whether either of these pathways was activated in chelator-treated cells, we used Western blotting to assess activation of caspases 9 and 8 in HeLa cells treated with 25 M tachpyridine (Fig. 3). As a positive control for caspase 8 activation, HeLa cells were treated with 50 ng/ml TNF-␣ in the presence of 1 g/ml cycloheximide for 4 h; treatment with 100 M etoposide for 48 h was used as a positive control for caspase 9 activation (Fig. 3). As seen in Fig. 3, caspase 9 was activated as early as 10 h, consistent with the early recruitment of a mitochondrial pathway in tachpyridine-mediated apoptosis. Caspase 8 was also activated in tachpyridine-treated cells; however, activation of caspase 8 was delayed relative to the activation of both caspase 9 and caspase 3, and substantial activation of caspase 8 was not observed until 30 h (Fig. 3).
To confirm activation of caspase 9, 3, and 8 in chelatortreated cells, we used enzymatic assays to measure cleavage of caspase substrates by extracts prepared from control and treated cells. Cells were incubated with either 100 M etoposide as a positive control for caspase 3-like and caspase 9 activity or with 50 ng/ml TNF-␣ in the presence of 1 g/ml cycloheximide as a positive control for caspase 8 activity. As shown in Fig. 4, enzymatic activity of all three caspases rose after treatment with tachpyridine; the increase in activity was comparable with that induced by treatment with etoposide or TNF. These results demonstrate that caspase zymogens are both cleaved and enzymatically activated in response to tachpyridine.
Activation of a Caspase Cascade by Tachpyridine Is Dependent on Its Metal Binding Activity-To verify that induction of caspases by tachpyridine was related to its metal binding properties rather than some other feature of the molecule, we treated HeLa cells with an N-alkylated derivative of tachpyridine which lacks an ability to react with iron and is no longer toxic (10). Activation of all caspases (3, 9, and 8) was completely absent in cells treated with this N-alkylated derivative (Fig. 5).
These results indicate that activation of caspases by tachpyridine is dependent on its metal binding ability.
Caspase Activation Is Also Observed in Cells Treated with Dipyridyl and DFO-To determine whether caspase activation is a general feature of iron chelator-mediated apoptosis, we treated cells with dipyridyl and DFO, two structurally unrelated iron chelators that are chemically distinct from tachpyridine. As shown in Fig. 6 caspase 3 activity, both classical characteristics of apoptosis (Fig. 6). As expected, treatment of HeLa cells with tachpyridine or TNF-␣ also resulted in nuclear fragmentation and positive caspase 3 staining (Fig. 6). Because this assay demonstrated induction of the executioner caspase 3 in cells treated with dipyridyl and DFO, we used Western blotting to examine further the ability of these chelators to induce the initiator caspases 9 and 8. As shown in Fig. 7, caspase 9, caspase 3, and caspase 8 were activated in response to both chelators, although with delayed kinetics relative to tachpyridine (Fig. 7).
Activation of Caspase 8 by Tachpyridine Is a Late Event and Is Independent of FADD-Classically, caspase 8 has been viewed as an initiator caspase involved in death receptor sig-naling (21). However, our experiments using Western blots of cells treated with tachpyridine demonstrated that caspase 8 occurred after activation of caspase 9 and caspase 3 (Fig. 3, A  and B). This time course suggested that caspase 8 was not the primary initiator caspase involved in tachpyridine-mediated apoptosis. To test this, we used HeLa cells that had been stably transfected with cDNA encoding a FADD fusion protein in which the N-terminal death effector domain of FADD was replaced with GFP (28). This fusion protein acts as a dominantnegative by blocking activation of caspase 8 at the death receptor, rendering cells resistant to apoptosis after exposure to TNF-␣ plus cycloheximide (28). HeLa cells expressing GFP alone or the GFP-FADD dominant-negative fusion protein were treated with either tachpyridine, TNF, or the DNA-damaging drug etoposide, and effects on caspase 8, PARP cleavage, and survival were measured. As expected, activation of caspase 8 after 100 ng/ml TNF-␣ plus 2.5 g/ml cycloheximide treatment was blocked completely in cells expressing FADD-DN (Fig. 8A). In contrast, 25 M tachpyridine caused activation of caspase 8 in both HeLa cells transfected with the control vector and in HeLa cells transfected with the GFP-FADD-DN fusion protein (Fig. 8A). Similarly, the GFP-FADD-DN fusion protein did not block activation of caspase 8 by 100 M etoposide (Fig. 8A).
We also examined PARP cleavage as a measure of caspase activity. As shown in Fig. 8B, the FADD-DN did not block cleavage of PARP after treatment with 25 M tachpyridine or 100 M etoposide (Fig. 8B). In contrast, PARP cleavage was blocked completely in FADD-DN transfectants after treatment with 100 ng/ml TNF-␣ plus 2.5 g/ml cycloheximide (Fig. 8B). Consistent with these results, although the FADD dominantnegative blocked cell death after TNF-␣ plus cycloheximide treatment, cell death was still observed after treatment with either tachpyridine or etoposide (Fig. 8C). These results confirm that the activation of caspase 8 after tachpyridine treatment is independent of FADD and death receptor signaling.
Cell Death in Response to Tachpyridine Is Blocked by the Caspase Inhibitor p35-To determine whether caspase activation was required for tachpyridine-mediated apoptosis, we used the baculovirus p35 protein, a broad caspase inhibitor (29). Cells were coinjected with YFP and control plasmid or coinjected with YFP and cDNA encoding the baculovirus p35 protein. YFP served as a marker of expression after injection. As shown in Fig. 9, p35 increased survival almost 3-fold (17.4 Ϯ 1.2% for control vector versus 45.1 Ϯ 7.5% for p35) after tachpyridine treatment (Fig. 9). The cytotoxicities of TNF-␣ Protein extracts were prepared and subjected to SDS-PAGE. Western blotting with an anti-caspase 9, anti-caspase 3, or anti-caspase 8 antibody was performed as described under "Materials and Methods." As positive controls, cells were treated with 100 M etoposide (VP) for caspase 9 and caspase 3 activity or 50 ng/ml TNF-␣ plus 1 g/ml cycloheximide (TNF) for caspase 8 activity. For caspase 8, the arrowhead points to the procaspase zymogen (p55/50) and arrows to the cleaved intermediates p40/36. For caspase 3, the procaspase p32 (arrowhead) and p20/p17 cleaved forms (arrows) are shown. For caspase 9, the procaspase p46 (arrowhead) and cleaved intermediate p35 (arrow) are shown.  7. DFO and dipyridyl activate caspase 3, 8, and 9. HeLa cells were treated with 250 M DFO or 250 M dipyridyl (DP) for the indicated times. Protein extracts were prepared and subjected to SDS-PAGE. Western blotting with an anti-caspase 9, anti-caspase 3, or anti-caspase 8 antibody was performed as described under "Materials and Methods." As positive controls, cells were treated with 100 M etoposide (VP) for caspase 9 and caspase 3 activity or 50 ng/ml TNF-␣ plus 1 g/ml cycloheximide (TNF) for caspase 8 activity. For caspase 8, the arrowhead points to the procaspase zymogen (p55/50) and arrows to the cleaved intermediates p40/36. For caspase 3, the procaspase p32 (arrowhead) and p20/p17 cleaved forms (arrows) are shown. For caspase 9, the procaspase p46 (arrowhead) and cleaved intermediate p35 (arrow) are shown. plus cycloheximide and etoposide were also inhibited by p35 (Fig. 9).
The Antiapoptotic Protein Bcl-XL Blocks Tachpyridine-mediated Cell Death-Anti-apoptotic Bcl-2 family members such as Bcl-XL can prevent apoptosis by blocking cytochrome c release (22). Because cytochrome c release is involved in the activation of caspase 9, we next wanted to determine whether Bcl-XL could protect cells from chelator-mediated cell death. Microinjection of Bcl-XL into HeLa cells blocked killing by both tachpyridine and etoposide (Fig. 10). Bcl-XL also blocked killing by TNF-␣ plus cycloheximide treatment (Fig. 10).
Tachpyridine-mediated Cell Death Requires Caspase 9 Activation-If caspase 9 plays an important role in chelator-mediated apoptosis, it should be possible to protect cells from cell death by inhibiting caspase 9. To test this prediction, we microinjected cells with a catalytic mutant of caspase 9 that acts as a dominant-negative form of procaspase 9 (23). Cells were then treated with etoposide, TNF-␣ plus cycloheximide, or tachpyridine. As seen in Fig. 11, cells were substantially protected from killing by all three treatments and fully protected from tachpyridine cytotoxicity (Fig. 11).

DISCUSSION
To elucidate mechanisms of iron chelator-mediated cell death, we explored the role of caspases, cysteine proteases that mediate apoptosis, in HeLa cells treated with iron chelators. We found that tachpyridine, an apoptotic iron chelator, activates multiple caspases, including those involved in the mitochondrial pathway (caspase 9) and the death receptor pathway (caspase 8) as well as the executioner caspase 3. A time course analysis revealed a sequential induction of caspase 9, 3, and 8 (Fig. 3, A and B). Robust cleavage of the caspase 3 substrate PARP commenced shortly after the activation of caspase 3 (Fig.  2). These results suggest that the most proximate event in tachpyridine-mediated apoptosis is activation of caspase 9, followed by caspase 3, and ultimately by activation of caspase 8.
To confirm that the induction of caspases by tachpyridine was related to its metal binding properties, we tested the ability of N-alkylated tachpyridine to activate caspases (Fig. 5). Alkylation of nitrogen donor ligands in tachpyridine blocks its ability to react with iron through steric hindrance and inhibition of oxidative dehydrogenation (18). N-Alkylated tachpyridine was unable to activate caspases (Fig. 5) or induce cell death (10). Thus, induction of caspases and apoptosis are Tachpyridine is a hexadentate chelator that binds iron via a process of oxidative dehydrogenation (18). Tachpyridine functions as an iron chelator in cells, as demonstrated by its ability to repress synthesis of ferritin, a protein whose translation is dependent on intracellular pools of chelatable iron (10). However, tachpyridine also binds zinc and copper (10), and a role for chelation of these metals in tachpyridine-mediated apoptosis and caspase activation cannot be excluded. We therefore also tested the ability of the chemically unrelated chelators, dipyridyl and DFO, to activate caspases. Dipyridyl exhibits a 10,000fold preference for iron over zinc (log cumulative stability constant ϭ 17.2 for Fe(II); 13.2 for Zn(II)) (30). DFO is even more highly selective for iron, demonstrating a 10 20 -fold preference for iron over zinc (log stability constant ϭ 30.6 for Fe(III) versus 11.1 for Zn(II)) (30). As seen in Figs. 6 and 7, all three chelators activated caspases 3, 8, and 9, suggesting that iron chelation can indeed activate caspases. These results are consistent with those obtained in other laboratories: using a pan-caspase assay that does not discriminate among individual caspases, DFO was shown to induce caspase activation in neuroblastoma cells (16). The chelators O-Trensox and hinokitol have also been shown to induce the executioner caspase 3 in F9 and HepG2 cells (9,13). Thus, although one cannot completely exclude a role for zinc chelation in the effects of these three chemically unrelated iron chelators, taken together these results suggest that depletion of iron is at least an important component of the mechanism whereby these diverse iron chelators induce caspases and apoptosis.
DFO is a hydrophilic, hexadentate chelator that preferentially binds Fe(III) while inducing both growth arrest (8,31) and apoptosis (11-16; Fig. 6). In contrast, dipyridyl, which also induces apoptosis (Fig. 6), is a hydrophobic Fe(II) chelator that partitions into cell membranes and binds iron as it passes through this lipid environment (32). Tachpyridine has a lipophilicity intermediate between these two chelators (log p ϭ 0.4) and binds both Fe(II) and Fe(III). However, despite these differences in lipophilicity and metal binding preferences, all three chelators activated caspase 3, 9, and 8 (Figs. 3, 10, and 11), induced PARP cleavage (not shown for DFO, dipyridyl), and triggered apoptosis. Thus, fundamental pathways elicited by these different chelators overlap, arguing that their ability to activate caspases and induce apoptosis derives from their metal binding properties rather than other ancillary chemical features. However, the rate of caspase activation and dose of drug required to elicit this response varied substantially among the three chelators (Figs. 3 and 7 and data not shown), suggesting that it may be possible to tailor chelators either to augment or diminish their cytotoxic properties.
Given that caspase 8 was originally identified as an initiator caspase triggered by death receptors, its involvement as a late event in chelator-induced apoptosis was unanticipated. The sequence of caspase activation we observed suggests an activation of caspase 8 via caspase 3, which in turn may be triggered by the mitochondrial pathway via caspase 9 (Fig. 3). This order is consistent with results obtained in cell-free extracts, which indicate that caspase 8 can be activated downstream of caspase 3 (38), although additional caspases may also contribute to the downstream activation of caspase 3 by chelators. Once activated, caspase 8 may participate in an amplification mechanism, further activating effector caspases to accelerate chelator-mediated cell death. Indeed, recent studies have shown the involvement of the CD95 death receptor pathway in apoptosis mediated by cytotoxic drugs (33,34) and demonstrated the ability of chemotherapeutic drugs to elicit processing of caspase 8 (35)(36)(37). Furthermore, Wieder et al. (36) and Wesselborg et al. (35) showed that a FADD dominant-negative did not block caspase 8 activation or apoptosis induction by chemotherapeutics (35,36) and that caspase 8 activation can occur downstream of caspase 9 and 3 in response to anti-cancer drugs (35,36). In our experiments, tachpyridine similarly activated caspase 8 independent of FADD signaling because a FADD dominant-negative did not block caspase 8 processing or inhibit tachpyridine-mediated apoptosis (Fig. 8). In this regard, pathways of apoptosis induced by tachpyridine overlap those induced by DNA-damaging chemotherapeutic drugs.
The important role of caspases in chelator-mediated apoptosis is supported by the ability of the baculovirus caspase inhibitor p35 to inhibit cell death induced by tachpyridine (Fig. 6). Because p35 acts as an irreversible stoichiometric caspase inhibitor, the inability of p35 to provide complete protection against tachpyridine in this experiment likely reflects the requirement for high levels of expression of p35 for effective caspase inhibition (39). These high levels may not have been attained in all microinjected cells. Protection against TNF/ cycloheximide and etoposide was also partial.
Tachpyridine-mediated cytotoxicity was blocked by Bcl-XL and prevented completely by a caspase 9 dominant-negative mutant (Figs. 10 and 11), reinforcing the critical role played by the mitochondrial pathway in chelator-mediated apoptosis. Cell death after TNF-␣ in the presence of cycloheximide was also blocked by both Bcl-XL and a caspase 9 dominant-negative mutant. These results are consistent with previous reports that Bcl-2 and Bcl-XL can inhibit TNF-and Fas-induced apoptosis in selected cell types, including HeLa cells (40 -42). Although the mechanism of this effect is incompletely understood, based on results obtained in hematopoetic cells it has been suggested that two CD95 (APO-1/FAS) signaling pathways exist (43). In type I cells, strong caspase 8 activation occurs at the deathinducing signaling complex, leading to direct caspase 3 activation independent of mitochondrial activity (43). In type II cells, only a little death-inducing signaling complex is formed, and a mitochondrial amplification loop initiated by caspase 8dependent cleavage of Bid is required for effective activation of caspase 3 and apoptosis (43,44). Bcl-2 and Bcl-XL inhibit apoptosis mediated by CD95 (APO-1/FAS) and the death receptor pathway in type II cells. Thus, HeLa cells represent epithelial type II cells in which TNF-␣ signaling and commitment to apoptosis are dependent on mitochondrial alterations (44). In our experiments, the ability of Bcl-XL and a dominant-negative caspase 9 mutant to inhibit tachpyridine-mediated cell death coupled with the negligible contribution of FADD signaling to chelator-mediated apoptosis (Fig. 8) points to an important contribution of mitochondrial pathways in chelator-mediated apoptosis.
Studies of iron chelators have largely focused on their ability to mobilize iron from iron-overloaded patients and tissues. More recently, additional applications of iron chelators have been considered, including protection from oxidative stress (3) and treatment of neoplastic disease (8 -10;45). A systematic study of factors that augment or diminish chelator-induced caspase activation may be useful in the optimization of chelators as either cytotoxic or cytoprotective agents.