NF-κB Activation Mediates Doxorubicin-induced Cell Death in N-type Neuroblastoma Cells

Neuroblastoma is the most common extracranial solid tumor of childhood. N-type neuroblastoma cells (represented by SH-SY5Y and IMR32 cell lines) are characterized by a neuronal phenotype. N-type cell lines are generally N-myc amplified, express the anti-apoptotic protein Bcl-2, and do not express caspase-8. The present study was designed to determine the mechanism by which N-type cells die in response to specific cytotoxic agents (such as cisplatin and doxorubicin) commonly used to treat this disease. We found that N-type cells were equally sensitive to cisplatin and doxorubicin. Yet death induced by cisplatin was inhibited by the nonselective caspase inhibitor z-Val-Ala-Asp-fluoromethylketone or the specific caspase-9 inhibitor N-acetyl-Leu-Glu-His-Asp-aldehyde, whereas in contrast, caspase inhibition did not prevent doxorubicin-induced death. Neither the reactive oxygen species nor the mitochondrial permeability transition appears to play an important role in this process. Doxorubicin induced NF-κB transcriptional activation in association with I-κBα degradation prior to loss of cell viability. Surprisingly, the antioxidant and NF-κB inhibitor pyrrolidine dithiocarbamate blocked doxorubicin-induced NF-κB transcriptional activation and provided profound protection against doxorubicin killing. Moreover, SH-SY5Y cells expressing a super-repressor form of I-κB were completely resistant to doxorubicin killing. Together these findings show that NF-κB activation mediates doxorubicin-induced cell death without evidence of caspase function and suggest that cisplatin and doxorubicin engage different death pathways to kill neuroblastoma cells.

Neuroblastoma (NB), 1 the most common malignant sympa-thetic nervous system tumor of childhood, arises from the neural crest (1,2). Despite the array of chemotherapeutic agents presently available and current strategies employing intensive myeloablative chemotherapy with autologous bone marrow transplantation, most patients with high risk NB die of their disease (3,4). To improve this situation, the basis for treatment failures as well as the mechanisms that underlie successful responses to chemotherapy in NB cells must be understood. The molecular response of the tumor cells to cytotoxic agents has become the focus of these efforts because it is clear that these pathways can lead to tumor cell death, whereas their absence or failure leads to resistant disease (5).
Two major pathways control the death response to cytotoxic agents, both of which rely on the eventual activation of downstream effector caspases. One pathway that leads to caspase activation is initiated by engagement of cell surface death receptors such as Fas/CD95/Apo-1, TNFR1, and TRAIL receptors (6). Engagement of Fas/CD95/Apo-1 by FasL/CD95L or specific agonistic antibodies transduces a signal that leads to the recruitment of the adapter protein FADD/MORT-1 (7) and procaspase-8, resulting in formation of the death-inducing signaling complex (8,9), which leads to caspase-8 activation via autoproteolytic processing. A second major apoptosis-signaling pathway functions independent of surface receptor activation. In this pathway, diverse pro-apoptotic signals eventually provoke a change in mitochondrial function and the release of pro-apoptotic mediators including cytochrome c. Cytochrome c release initiates formation of apoptosome, a complex that includes procaspase-9, dATP, and Apaf-1. This allows proteolytic activation of caspase-9 as a result of induced proximity (10). In both the death receptor/caspase-8 and the mitochondria/casapase-9 pathways, the eventual activation of effector proteases caspases 3, 6, and 7 results in the dismantling of the cell as diverse cellular substrates are proteolyzed (11,12).
Understanding how these elements of the death machinery function in NB pathogenesis, particularly in the response of NB tumors to cytotoxic drugs, is complicated by the fact that the cells that comprise NB tumors are heterogeneous. Heterogeneity is observed in the histologic appearance of NB tumor specimens and is reflected in the phenotypes of cell lines isolated from NB tumor specimens. In vitro culture conditions allow two predominant phenotypes to emerge: N-and S-types, of which N-type cell lines are the most common. N-type cells grow as poorly attached aggregates of small round cells, exhibit neurite-like processes, and possess enzymatic activities associated with neurotransmitter synthesis (13). The N-myc oncogene is frequently amplified in N-type lines, and these cell lines are tumorigenic when xenografted into immunodeficient mice (14,15). In contrast, S-type cells grow flat, do not display neuritic processes, and are more adherent to substrate in culture. Stype cells have little or no ability to synthesize neurotransmitters, are not N-myc-amplified, and do not form tumors in mice (13)(14)(15).
NB cell apoptosis has been most thoroughly characterized in S-type cell lines. The S-type cell line SH-EP1 undergoes apoptosis in response to many anticancer agents, and these effects are blocked by Bcl-2 (16). SH-EP1 cells express Fas/CD95, and treatment with CDDP, Dox, or etoposide (VP-16) induces apoptosis that is associated with up-regulation of Fas/CD95 and activation of caspase-8 (17). Whether this mechanism appropriately characterizes the arguably more malignant N-phenotype and, more generally, NB tumor behavior is uncertain. Recent studies have identified potentially important variations in the expression and function of elements of the apoptosis signaling pathways between S-and N-type NB cell lines. For instance, N-type cells express higher levels of Bcl-2, and increased expression is associated with a drug-resistant phenotype (18,19). N-type cells fail to express caspase-8 as a result of silencing through DNA methylation, which is implicated in the relative resistance of some lines to Dox (20). However, the Fas/CD95 pathway appears to be less important in these cells. Some N-type cells are resistant to Fas-mediated apoptosis (21), potentially as a result of expression of Bcl-2 and the caspase-8 inhibitory protein cellular FLICE-inhibitory protein or downregulation of caspase-8 (22). The N-type SH-SY5Y cell line treated with staurosporine undergoes apoptosis associated with activation of caspase-3 (23) that does not require the Fas/CD95-L-receptor pathway (24). Despite all of these findings many N-type cell lines are sensitive in vitro to cytotoxic agents (19,25). The present study was designed to determine the mechanism by which specific cytotoxic agents kill N-type NB cell lines in the hopes of defining elements of the death machinery that are functional and can therefore be targeted to treat high risk malignant NB tumors.
Cell Culture and Transfection-Human NB cell lines SH-EP1, SH-SY5Y, and IMR32 were cultured in MEM supplemented with 10% FBS, 100 units/ml penicillin, and 100 g/ml streptomycin. The cells were maintained at 37°C in a humidified 5% CO 2 incubator. SH-SY5Y cell lines stably expressing dominant negative I-B␣ (I-B␣M) as well as vector-transfected cells were generated as described previously (26) and maintained in routine MEM supplemented with 500 g/ml G418.
Transfections were carried out using LipofectAMINE PLUS (Invitrogen) according to the manufacturer's instructions. Briefly, the cells were seeded in tissue culture plates to achieve 50% confluence. Twentyfour hours later, the cells were transfected using a mixture of DNA and LipofectAMINE PLUS in OPTI-MEM (Invitrogen). Eight hours following transfection, the cells were supplemented with serum at a final concentration of 10%. The cells were treated and analyzed for luciferase activity as described below.
Cell Viability Assays-NB cells were plated in triplicate at 1.2 ϫ 10 4 cells/well in 96-well culture plates in MEM. On the following day, the cells were treated with Dox (0.5 g/ml) or CDDP (3 mol/liter). On consecutive days, the cells were incubated with MTT dye (1 mg/ml) at 37°C for 4 h and lysed in a buffer containing 20% (w/v) SDS, 50% (v/v) N,N-dimethylformamide (pH 4.5). Absorbance at 600 nM (OD 600 ) was determined for each well using an ELX 808 automated microplate reader (Bioteck Instruments, Winnoski, VT). In assays of cell death inhibition, the cells were preincubated with one of the following caspase inhibitors: 100 M z-VAD-fmk, 200 M Ac-DEVD-CHO, or 50 M Ac-LEHD-CHO in MEM supplemented with 5% FBS prior to the addition of Dox or CDDP. After subtraction of background absorbance, the OD 600 of the cells treated with Dox was divided by that of the untreated cells to obtain the percentage of viable cells. In some experiments, cell viability was determined by trypan blue exclusion. For these assays, cells (2 ϫ 10 5 cells/well) were plated on 24-well plates and treated with Dox (0.5 g/ml). After variable times, the cells were harvested by trypsinization, washed with phosphate-buffered saline, resuspended in MEM containing 0.2% trypan blue (Invitrogen), and enumerated (living and dead) by light microscopy. The Student's paired t test was used to test the significance of the difference in cell viability for these assays.
SDS-PAGE and Immunoblot Analysis-The cells were washed twice with cold phosphate-buffered saline and lysed in SDS buffer containing 50 mM Tris-HCl, 100 mM dithiothreitol, 2% SDS, 0.1% bromphenol blue, and 10% glycerol. Aliquots of cell lysates containing 30 g of total protein were resolved on 12% SDS-PAGE and transferred to Hybond-P membranes (Amersham Biosciences, Inc.). The membranes were incubated with antibodies in 5% milk in phosphate-buffered saline with 0.1% Tween 20. Immunoreactivity was detected using the ECL detection system (Amersham Biosciences, Inc.). The films were exposed at multiple time points to ensure that the images were not saturated.
5-Aza-2Ј-deoxycytidine Treatment-The NB cell lines SH-SY5Y and IMR32 that show no expression of caspase 8 were subjected to demethylating agent 5-aza-2Ј-deoxycytidine treatment. The cells were plated into 60-mm tissue culture dishes and incubated with 0.5, 1.5, 5, and 10 M 5-aza-2Ј-deoxycytidine for a week. At days 4 and 7, cell lysates were collected in SDS sample buffer and subjected to immunoblot analysis.
Flow Cytometric Analysis for Cell Membrane Permeability and Apoptosis-The cells were plated at a density of 2 ϫ 10 5 cells/well in 24-well plates and cultured overnight. The cells were then treated with Dox for designated time points and harvested by trypsinization. Cell membrane permeability was assessed by flow cytometry. Duplicate samples were mixed with hypotonic lysis buffer (0.1% sodium citrate, 0.01% Triton X-100, and 0.1 mg/ml propidium iodide) and incubated at 4°C for 4 h. Subsequently, DNA content was determined by flow cytometry (Becton Dickinson, Mountain View, CA), and the data were analyzed using CellQuest software (Becton Dickinson). For the cell protection experiments, vitamin C (100 M), vitamin E (1 mM), cyclosporin A (1 M), or PDTC (100 M) was freshly prepared and added 2 h before Dox treatment.
Measurement of the Mitochondrial Permeability Transition and Reactive Oxygen Species-MPT and ROS generation were measured by flow cytometry. NB cells were plated in 24-well plates at 2 ϫ 10 5 cells/well and incubated in MEM overnight. The cells were treated with Dox (0.5 g/ml) for indicated times. During the last 30 min of treatment, 100 nM 3,3Ј-dihexyloxacarbocyanine iodide was added to the medium. At the end of the experiment, the medium was removed, and the cells were washed with phosphate-buffered saline. The cells were trypsinized and resuspended in 500 l of 2% FBS-supplemented MEM and analyzed by flow cytometry. Intracellular ROS was measured using two different cell-permeable fluorescent dyes: dihydroethidium (10 M) and dichlorodihydrofluorescein diacetate (100 M). Stock solutions of dihydroethidium (10 mM) and dichlorodihydrofluorescein diacetate (100 mM) were freshly made in Me 2 SO prior to each use. The data were analyzed using CellQuest software (Becton Dickinson).
Determination of NF-B-dependent Reporter Gene Activation-SH-SY5Y and IMR32 cells were transfected with 1 g of the reporter plasmid pBVIx-Luc as described above. To assure identical transfection efficiency in control and treated cells, the cells were replated 12 h after transfection into 12-well plates and after attachment were treated with or without Dox (0.5 g/ml) and CDDP (3 mol/liter) for the designated times. The cells were harvested by trypsinization, and luciferase activity was determined according to the manufacturer's instructions using the dual luciferase reagent kit (Promega, Madison, WI) and a Monolight 2010 luminometer (Analytical Luminescence Laboratory, Ann Arbor, MI). An aliquot of the same samples was subjected to protein concentration determination (Bio-Rad). The NF-B-dependent luciferase activity was then normalized by protein concentration.

NB Cells Are Killed in the Presence or Absence of Caspase-8 -One
S-type (SH-EP1) and two N-type (SH-SY5Y and IMR32) NB cell lines with differences according to type in morphology, expression of neurochemical markers, and N-myc amplification were selected to compare their sensitivities to killing by CDDP and Dox (13)(14)(15). Viability after exposure to these agents was measured by the MTT assay. Time course (Fig. 1, A and B) and dose-response analysis (data not shown) revealed no meaningful differences between these cell lines with respect to either CDDP or Dox. The lack of a difference with respect to Dox was surprising. In part, a difference was expected because Dox killing of NB cells has previously been shown to operate through Fas/CD95-induced caspase-8 activation (17), and many N-type cells do not express caspase-8 (19). Moreover, the higher level of Bcl-2 found in the N-type lines used compared with the S-type SH-EP1 cells would predict greater resistance to both of these agents.
To definitively exclude the involvement of caspase-8 in the death of these particular N-type cells, we examined whether caspase-8 is expressed or whether the gene is functionally silenced. Caspase-8 protein was present in the SH-EP1 cells but was not detected by immunoblotting in either the SH-SY5Y or IMR32 cells, despite their sensitivity to Dox (Fig. 1C). By way of comparison, all three cell lines express caspase-9 and caspase-3, although the levels of caspase-9 are relatively low in SH-SY5Y cells (Fig. 1C). We further confirmed that the lack of caspase-8 expression results from DNA methylation by treating the cells with 5-aza-2Ј-deoxycytidine. Within 4 days of this treatment, caspase-8 protein became detectable in the N-type cells ( Fig. 1D and data not shown). In aggregate, these results argue that despite a lack of caspase-8 expression, N-type cells are as sensitive to CDDP and Dox as caspase-8 expressing S-type cells. Therefore, a caspase-8-independent death mechanism must exist in N-type cells.
CDDP and Dox Killing Are Differentially Sensitive to Caspase Inhibition-Having found that Dox and CDDP each kill N-type cells in the absence of caspase-8, we sought to determine whether there is any evidence of caspase involvement in this process. We first examined treated cells to determine whether pro-caspase-3 becomes processed. Caspase-3 activation is an important indicator of apoptosis because of its apical position in the execution arm of caspase function and its regulation by initiator caspases 8 and 9. A substantial portion of pro-caspase-3 is processed to its active forms  in N-type cells treated with CDDP ( Fig. 2A). In contrast, treatment with Dox results in minimal processing ( Fig. 2A) even after 72 h of treatment (data not shown). When caspase activity was inhibited using z-VAD-fmk, CDDP killing was almost entirely blocked (Fig. 2B). Identical results were obtained when a caspase-3-selective inhibitor (Ac-DEVD-CHO) was used (data not shown). Similar experiments in which caspase-9 was selectively inhibited using Ac-LEHD-CHO demonstrated almost complete protection against CDDP (Fig. 2C). When these same inhibitors were used to probe the caspase dependence of Dox-induced death, the findings were quite different. z-VAD-fmk (or caspase-3 inhibition; data not shown) minimally increased SH-SY5Y and IMR32 cell survival (ϳ10 and 20%, respectively) (Fig. 2B). Inhibition of caspase-9 had no statistically distinguishable effect on Dox killing (Fig. 2C).
These results indicate that caspase-9 and caspase-3 are functional in N-type cells and that their activity is necessary for CDDP-induced death. However, cell death induced by Dox appears largely independent of caspase activity and results in only minimal caspase-3 processing.
The findings presented above used MTT conversion to reflect cell viability. Because this method assesses viability indirectly (dependent on mitochondrial function and chemical reducing activity), we verified the conclusion that caspases are of little importance in this death process by assessing cell morphology. As shown in Fig. 3B, SH-SY5Y cells treated with Dox for 24 h show morphologic evidence of cellular destruction, which includes membrane blebbing, vacuolization of the cytoplasm, and nuclear condensation. Even without affecting morphological evidence of cell death, z-VAD-fmk prevented the hypodiploid DNA changes associated with apoptosis (Fig. 3C). Dox treatment resulted in 21% of cells having hypodiploid DNA content (Fig. 3B) compared with only 2% in the presence of z-VAD-fmk (Fig. 3C). Thus whereas some apparently caspase-dependent apoptotic DNA damage is detected following Dox treatment, morphologic evidence (Fig. 3) and quantitative evidence (Fig. 2) of cell death confirms that Dox killing is substantially inde-pendent of caspase activity.
Dox-induced Death Is Blocked by PDTC and Is Independent of the MPT or ROS-Whereas prior work with the S-type SH-EP1 cell line demonstrated that the mechanism of Doxinduced cell death depends on signaling from Fas/CD95 leading to caspase-8 activation, disruption of the ⌬ M , cytochrome c release, and caspase-3 activation (17,27), the results presented above caused us to suspect that an alternative pathway(s) might function in the N-type cells. In other well characterized models, Dox-induced death requires the generation of ROS (28) or collapse of the MPT (29). To determine the potential involvement of these signals in N-type cells, we looked for evidence of an increase in ROS and collapse of the MPT over the first 8 h after exposure of SH-SY5Y and IMR32 cells to Dox. Using flow cytometry to detect signals from the fluorescent indicators 3,3Јdihexyloxacarbocyanine iodide to measure ⌬ M , dihydroethidium to detect superoxide, and dichlorodihydrofluorescein diacetate to detect other ROS, we found no evidence of changes in these cells within this time period (data not shown).
NF-B activation has also been implicated in Dox signaling (30), although in many cases NF-B activation protects cells against death (31). To evaluate the potential involvement of NF-B in Dox-induced N-type cell death, we blocked NF-B activation by pretreating cells with the NF-B inhibitor and antioxidant PDTC. PDTC has at least two chemical properties: a chelating activity for heavy metals and an antioxidant activity of its dithiocarboxy group. It appears that PDTC functions to suppress a reaction required for the release of I-B from NF-B, which may involve a ROS. PDTC has no effect on the DNA binding properties of NF-B, the nuclear uptake of NF-B, the amount of NF-B-I-B complex, or the release of I-B (32). After 48 h of treatment with Dox in the presence or absence of PDTC, the cells were evaluated by flow cytometry to determine viability on the basis of plasma membrane permeability to PI. PI permeability provided a more direct measure of cell death or survival than MTT conversion, and preliminary experiments demonstrated that it correlated well with morphologic evidence of death. As shown in Fig. 4, PDTC completely blocked cell death induced by treatment with Dox for 24 h in both SH-SY5Y and IMR32 cells. In contrast, PDTC had no effect on CDDP-induced death (data not shown). To further define whether the mechanism of Dox resistance by PDTC involves general antioxidant properties or a specific inhibitory action against NF-B activation, other antioxidants were studied as potential inhibitors. SH-SY5Y cells were pretreated for 30 min with either one of the antioxidants (vitamin C, vitamin E, and Trolox) or 1 M cyclosporin A to inhibit the MPT, prior to Dox exposure. Viability was determined on the basis of PI exclusion. Table I summarizes the results of these studies. 51% of SH-SY5Y cells treated with Dox alone demonstrate loss of membrane integrity by 24 h. Pretreatment with cyclosporin A to inhibit the MPT has no effect on viability. Similarly, none of the other antioxidants (vitamin C, vitamin E, and Trolox) prevented cell death like PDTC. Thus only PDTC was able to block this process. Similar results were obtained with IMR32 cells (data not shown). These results suggest the possibility that NF-B activation is involved in Dox-induced cell death, whereas we found no convincing evidence to implicate either the collapse of ⌬ M or an increase in ROS.
Dox Induces I-B␣ Degradation and NF-B Activation Prior to the Onset of Cell Death-Although PDTC is recognized as an inhibitor of NF-B, its mechanism of action is not well defined, and its antioxidant properties make it unlikely to be entirely specific for NF-B. As such, further experiments were necessary to definitely implicate NF-B in this process. First, considering that I-B␣ levels are of paramount importance in the regulation of NF-B activity under most conditions (33), we determined the level of I-B␣ after treatment with Dox. In SH-SY5Y cells treated with Dox, I-B␣ levels decreased significantly between 4 and 8 h after treatment (Fig. 5A), which is before significant loss of cell viability (Fig. 1A). Similar results were obtained with IMR32 cells, although the degradation of I-B␣ occurred earlier, within 4 h of exposure (data not shown). These same membranes were also immunoblotted to detect I-B␤, but no appreciable change in I-B␤ levels was detected (Fig. 5A). Treatment of these lines with CDDP over an identical time course demonstrated no change in the levels of either I-B␣ or I-B␤ (data not shown).
NF-B transcriptional activity was then directly measured in cells after treatment with Dox. This was accomplished by transiently transfecting SH-SY5Y and IMR32 cells with a luciferase reporter plasmid that contained within its promoter region six tandemly placed NF-B consensus-binding sites. NF-B-dependent transcriptional activation was detected within 8 h of exposure to Dox, but when cells were pretreated with PDTC, there was no luciferase response to Dox (Fig. 5B). In accord with the findings above, CDDP treatment did not increase NF-B transcriptional activity (Fig. 5B). These results suggest Dox induces NF-B activation in a time-dependent manner and suggest that this activation is associated with degradation of I-B␣ before the onset of cell death.

NF-B Activation Is Essential for Dox-induced NB Cell
Death-To confirm that NF-B activation is required for Doxinduced NB cell death, we utilized SH-SY5Y cells stably transfected to express a dominant negative mutant of I-B␣, which functions as a super-repressor of NF-B activation (26). These cells (designated IϪB␣M/SH-SY5Y) and corresponding vector transfected controls were treated with Dox and then studied to determine I-B␣ levels as well as viability. Vector control cells responded to Dox with evidence of I-B␣ degradation (Fig. 6A). As anticipated, in the I-B␣M/SH-SY5Y cells there was no change in the level of I-B␣ following Dox treatment (Fig. 6A). The NF-B-sensitive reporter plasmid was transfected into each of these cell lines, and the luciferase response following Dox treatment was measured. The vector control line responded similarly to the parental cell type with a Ͼ10-fold induction of luciferase activity at 8 h, whereas the I-B␣M/SH-SY5Y cells showed no induction (Fig. 6B). As well, suppression of NF-B caused the cells to resist Dox killing. IϪB␣M/SH-SY5Y cell viability measured by trypan blue exclusion in the absence of Dox was 85 Ϯ 9%, whereas control cell viability was 99 Ϯ 2%. 24 h after Dox treatment IϪB␣M/SH-SY5Y viability remained at 87 Ϯ 2%, whereas vector control cell viability was 54 Ϯ 4%. Morphologic changes were obvious in control cells within 24 h of Dox treatment as the cells became rounded and detached from the tissue culture plate (Fig. 6C, left panels). The I-B␣MSH-SY5Y cells displayed morphology after treatment indistinguishable from untreated cells (Fig. 6C, right panels). These results confirm that I-B␣ degradation and subsequent NF-B activation is required for Dox-induced cell death in these N-type NB cell lines. DISCUSSION To define the mechanisms contributing to the development of NB drug resistance and treatment failure, it is necessary to first understand the pathways that mediate the death response to CDDP and Dox used in the treatment of this disease. For our studies, we focused on the response of cell lines derived from primary tumor specimens that display the N phenotype. The importance of studying these responses in N-type cells stems not only from the fact that the N phenotype is the most common to emerge from high risk tumor explants but also because N-type cells display morphologic and biologic characteristics of aggressive tumor cell behavior. In the present study, we found that N-type cells are as sensitive to treatment with Dox and CDDP as an S-type cell line. This finding was somewhat unexpected because S-and N-type cells differ in the expression of several elements of the cell death machinery. In particular N-type cells express higher levels of Bcl-2 (18) and fail to express caspase-8 secondary to DNA methylation (20). The N-type lines under study here fulfill both of these characteristics. These results mean that caspase-8 is not required for killing by either agent and that even the relatively high level of Bcl-2 is insufficient to block death induced by CDDP or Dox.
Consistent with the results reported in other tumor models, we found that caspases are activated by CDDP in N-type NB cells, and we showed they are required for CDDP-induced cell death. CDDP has been shown previously to kill cells by forming DNA adducts causing G 2 cell cycle arrest (34). In HeLa cells and ovarian and squamous carcinoma cell lines the apoptotic pathway mediating CDDP killing requires caspase-9 (35)(36)(37), and in accord with these studies we found that a selective caspase-9 inhibitor strongly protected NB cells from CDDP killing. In part, the mechanism by which caspase-9 initiates this process is through inactivation of the nuclear transport system, which increases the diffusion limit of the nuclear pores, allowing caspase-3 and other molecules to enter the nucleus (38).
In contrast, our results with Dox fail to demonstrate an important role for caspases in the death response to this agent. Dox killed these cells even though (a) they do not express caspase-8, (b) there was only minimal pro-caspase-3 process-  ing, and (c) specific inhibitors of caspase function were present in high concentrations. Notwithstanding, Dox induced DNA fragmentation characteristic of apoptosis; and this change (unlike plasma membrane permeability and morphologic destruction) was inhibited by z-VAD-fmk. These characteristics are consistent with the criteria suggested by Blagosklonny to define an alternative form of cell death termed "slow death" (39). In slow death, the cells that normally die with evidence of caspase activation still die even when caspase inhibition interferes with the biochemical and morphological changes associated with apoptosis. In many cases, this process may have been misinterpreted as drug resistance in that evidence of apoptosis has been used (instead of loss of cell viability) as a measure of the effectiveness of an agent. How does Dox act to kill cells? Despite extensive clinical experience with this agent against a wide range of malignant disorders, some of the most fundamental aspects of its mechanism of action are not defined. For example, the direct molecular target, which it binds to in cells, is ambiguous. Although anthracyclines are known to intercalate with DNA, the tumoricidal actions of Dox may not even require that the agent enter into cells. For example, covalent linkage of Dox to agarose beads has revealed that even in this non-cell-permeable form, it effectively kills human leukemia cells in culture (40). Above all, however, Dox is believed to induce DNA damage. This action itself has been attributed to at least two mechanisms. In the first, it has been shown that Dox can accept an electron to become a highly reactive semiquinone free radical that reacts with molecular oxygen to form superoxide. The ROS generated subsequently damages DNA (28). In our experiments, we found no evidence that superoxide or other ROS are generated in the N-type cells following Dox treatment. A second proposed mechanism involves the direct action of Dox on the nuclear enzyme DNA topoisomerase II (41). The drug is found in covalent complexes with this enzyme and DNA that are associated with double strand breaks.
Our experiments were primarily directed at uncovering Doxinduced signals in N-type cells, which then engage the death response. We found that neither of the most prominent initiator caspases (caspase-8 and -9) was required for Dox killing, which immediately distinguished the death mechanism in Ntype cells from that reported for the S-type SH-EP1 NB cell line as well as for leukemic cells. In these other examples, Fas/ CD95-dependent caspase-8 activation and further mitochondrial perturbation are of paramount importance (17,27).
Mechanistically, our data point to the fact that NF-B plays a key role in mediating Dox-induced death of N-type NB cell lines. NF-B is activated in cells after exposure to cytotoxic agents including ligands of the cell surface death receptors 24 h later, growth medium was replaced with medium containing 0.5 g/ml Dox, and the cells were cultured for a further 24 h. The cells were photographed (400ϫ) using interference contrast microscopy. Cell viability was measured by trypan blue exclusion and is shown. such as tumor necrosis factor and Fas, as well as by genotoxic agents and chemotherapeutic drugs. In many instances, NF-B activation mediates resistance against cell death or protection from apoptosis (42)(43)(44)(45). Indeed, it is well established that Dox and other anthracyclines induce NF-B activation. As examples, inhibition of NF-␤ enhances apoptosis induced by Dox in T leukemia cell lines (46) and in pancreatic carcinoma cells (47). NF-B suppresses tumor necrosis factor-␣-induced apoptosis through the transcription of gene products that function to block apoptosis, such as the cellular inhibitors of apoptosis (cIAP1, cIAP2, and XIAP), the tumor necrosis factor receptorassociated factors TRAF1 and TRAF2, the Bcl-2 homologue A1/Bfl-1, and the A20 zinc finger protein (43, 48 -52).
Our data provide compelling evidence for an opposing function of NF-B in N-type NB cells. In our model, drug-induced NF-B activation actually mediates killing by Dox. This conclusion was first established by our results showing that Dox induced I-B␣ degradation and NF-B-dependent transcriptional activation prior to cell death. Second, the small molecule antioxidant PDTC suppresses Dox-induced I-B␣ degradation, NF-B activation, and cell death. Furthermore, a highly specific NF-B inhibitor (I-B␣M) prevented I-B␣ degradation and cell death. Other recent studies have found evidence that NF-B is a mediator of cell death in PC12 cells exposed to dopamine (53), in neurons undergoing focal ischemia (54), in Jurkat T cells exposed to either VP-16 or UV irradiation, and in hepatocellular carcinoma cell lines treated with Dox (55). Of particular note, Ryan et al. (56) have shown that p53-induced apoptosis requires NF-B activation. Further, they demonstrated that mouse embryo fibroblasts derived from p65-deficient mice that are sensitized to undergo p53-dependent apoptosis by E1A transformation completely resist Dox-induced death, whereas E1A-transformed wild-type p65 controls were killed. In this context it is worthwhile noting that most NB tumors have wild-type p53, that the particular N-type cells used in our experiments have wild-type p53, and that Dox has been shown to increase p53 levels in SH-SY5Y cells (57). Thus our results with N-type cells appear to offer strong support for the hypothesis advanced by Ryan et al. (56) that inhibition of NF-B in tumor cells that retain wild-type p53 diminishes rather than augments a therapeutic response. Finding that NF-B mediates Dox-induced death of N-type cells opens a new opportunity for experimental strategies to treat this disease. For example mitozantrone has been shown to induce NF-B in HL-60 cells by targeting topoisomerase II (58). This and other agents with the capacity to activate NF-B should be carefully considered as candidates for further study in NB. Conversely, resistance to Dox (and potentially other drugs) may result from disturbances of this signaling pathway.
NF-B activation has been observed in NB cells as part of the response to agents that cause cellular differentiation. SH-SY5Y NB cells when induced to differentiate with retinoic acid or 12-O-tetradecanoylphorbol 13-acetate transiently activate NF-B before the morphological signs of differentiation (26). In this case NF-B activation was associated with a significant reduction in the amount of I-B␤, and cells transformed with I-B␣M failed to differentiate into neuronal cell types when treated with retinoic acid or 12-O-tetradecanoylphorbol 13acetate. In our experiments we found a reduction in I-B␣, with no detectable change in I-B␤, suggesting that the signaling in SH-SY5Y cells that leads to NF-B activation may vary depending on the stimulus applied. These differences may provide some specificity to the signal response along with the myriad other signals that are concurrently generated by these agents.
We have yet to define the exact signals that lie upstream and downstream of NF-B activation in response to Dox in these cells. Given the findings of Ryan et al. (56), it will be reasonable to determine whether NF-B is activated by a p53-dependent mechanism, which also involves signals through the MAPK cascade. This is presently being explored using a series of p53 mutants and specific inhibitors of MEK activity. It is tempting to speculate on the mechanism by which NF-B mediates the death response to Dox. One possibility is NF-B activates Fasmediated apoptosis. Support for this notion is provided by studies of T cells and T cell lines indicating that DNA-damaging agents activate the stress-activated protein kinase/c-Jun N-terminal kinase pathway to activate AP-1 and NF-B resulting in expression of Fas/CD95-L-and Fas-mediated apoptosis (59). This possibility seems unlikely in our system because N-type cells do not express caspase-8, and caspase-mediated death was not a primary mechanism of Dox-induced death. Another possibility is suggested by studies on endothelial cell death induced by hypoxia (60) and on hydrogen peroxide-induced apoptosis of HeLa cells (61). In both systems, NF-B is induced and increases p53 expression. Hypoxia-induced cell death is additionally associated with a decrease in Bcl-2 expression (60). In any case, efforts are underway in our laboratory to identify the genes induced by NF-B specifically following Dox treatment to define this portion of the signaling cascade. It will be important to determine whether the prosurvival genes associated with NF-B expression in other cell types are absent from this expression profile.