Akt Protein Kinase Inhibits Non-apoptotic Programmed Cell Death Induced by Ceramide*

A growing body of evidence now suggests that programmed cell death (PCD) occurs via non-apoptotic mechanisms as well as by apoptosis. In contrast to apoptosis, however, the molecular mechanisms involved in the regulation of non-apoptotic PCD remain only poorly understood. Here we show that ceramide induces a non-apoptotic PCD with a necrotic-like morphology in human glioma cells. Characteristically, the cell death was not accompanied by loss of the mitochondrial transmembrane potential, cytosolic release of cytochrome c from mitochondria, or the activation of the caspase cascade. Consistent with these characteristics, this

suggested that apoptosis, cell death characterized by cell shrinkage as well as by early and prominent nuclear chromatin condensation, is one morphological manifestation of PCD. This idea is now firmly supported by a large body of evidence indicating that apoptosis is indeed a cell-autonomous, genetically regulated cell death, in the regulation of which a family of cysteine proteases with preference for an aspartate residue at the P1 position (caspases) plays the central role (5,6). Although it is thus established that caspase-regulated apoptosis is at least one entity of PCD, it remained unclear whether apoptosis is the only form of PCD or whether there are other forms, and if so, how they are regulated.
Recently, an increasing number of reports have suggested the existence of caspase-independent, non-apoptotic PCDs (7)(8)(9). For instance, the overexpression of c-Myc induces cell death with a necrotic-like morphology in the presence of pan-caspase inhibitors, which can be inhibited by treating cells with insulinlike growth factor I or by the overexpression of Bcl-2 (10). Similarly, pharmacological agents such as staurosporine and dexamethasone induced, in the presence of pan-caspase inhibitors, necrotic-like cell death inhibitable by Bcl-2 overexpression (11,12). Only recently, Apaf-1-null embryonic stem cells have been reported to undergo non-apoptotic cell death after treatment with various cytotoxic stimuli, a process that was independent of caspase activation but was yet inhibitable by Bcl-2 overexpression (13). These observations indicate that such non-apoptotic cell deaths can be controlled by genetic manipulation of the survival signaling pathway. In addition, recent studies indicated that the expression or activation of gene products such as PML (14), H-Ras (15), BNIP3 (16), and Bin1 (17) induces "atypical apoptosis" or non-apoptotic cell death that is neither accompanied by nor dependent on the activation of the caspase cascade. These lines of evidence together indicate that cells have genetic programs for cell death with necrotic-like appearance that can be regulated independently of caspases. However, in apparent contrast to apoptotic PCD where the central mechanism and molecules involved in the transduction of death signals have been largely delineated, information regarding the players participating in the regulation of caspase-independent, non-apoptotic PCDs is still limited.
Ceramide is one of the sphingosine-based second messenger molecules that are involved in the regulation of diverse cellular responses, including cell death, to exogenous stimuli (18). Ceramide is generated in response to a variety of cell death stimuli such as tumor necrosis factor (TNF)-␣, Fas, ionizing radiation, * This work was supported by a grant-in-aid for scientific research and by a grant-in-aid for scientific research on priority areas from the Ministry of Education, Sports, Science, and Culture of Japan. 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.
ʈ These authors contributed equally to this work as senior authors. ** To whom correspondence may be addressed. Dr. Kitanaka, Tel.: 81-3-3542-2511; Fax: 81-3-3546-1369; E-mail: ckitanak@ncc.go.jp; or Dr. Asai, Tel.: 81-3-5800-8853; Fax: 81-3-5800-8655; E-mail: asaisan-tky@umin.ac.jp. 1 The abbreviations used are: PCD, programmed cell death; TNF, tumor necrosis factor; TUNEL, terminal deoxynucleotidyl transferasemediated dUTP nick end labeling; myr-Akt, myristoylated Akt; ⌬⌿ m , mitochondrial transmembrane potential. ultraviolet C, heat shock, and oxidative stress, and the cell death-inducing effect of these stimuli is mimicked by cell-permeable analogs of ceramide (19). Interestingly, ceramide induces not only apoptosis but also caspase-independent and/or non-apoptotic cell death depending on the cell type. For instance, Mengubas et al. (20) showed that ceramide induces in resting T lymphocytes cell death with a non-apoptotic morphology that does not involve caspase activation. Belaud-Rotureau et al. (21) reported that ceramide-induced death of leukemic U937 cells was accompanied by minimal activation of the DEVD-cleaving caspases and was not inhibited by the pancaspase inhibitor zVAD-fmk, although it was not shown in this case whether the morphology was apoptotic or not. Importantly, ceramide-induced T lymphocyte and U937 cell deaths were inhibited by phytohemagglutinin-mediated mitogenic activation and by leupeptine, respectively, indicating that they are not a passive cell death (necrosis) but active, regulated cell death. However, at present, the genes or molecules involved in the regulation of the caspase-independent, non-apoptotic cell death triggered by ceramide remain unknown.
In this paper, we report that ceramide also induces a caspase-independent cell death with necrotic-like morphology in human glioma cells. Significantly, this ceramide-induced non-apoptotic cell death was inhibited efficiently by the activation of the Akt/protein kinase B pathway, which has been shown to play a critical role in survival factor-mediated antiapoptotic signaling (22). The unexpected results not only reveal a novel player in the regulation of caspase-independent, nonapoptotic PCD, but also suggest that Akt kinase may be a key survival molecule implicated in both apoptotic and non-apoptotic PCD.
Cell Death Induction and Death Assays-Cells (2 ϫ 10 5 ) were seeded into 60-mm dishes, and 24 h later they were fed with medium containing 0.1% fetal calf serum and treated with 25 M C 2 -ceramide, 25 M C 2 -dihydroceramide, or 30 ng/ml TNF-␣ plus 20 g/ml cycloheximide. Cell viability was determined by the dye exclusion method using trypan blue or by the lactate dehydrogenase release method as described previously (23). Lactate dehydrogenase release (percent) was calculated as (lactate dehydrogenase activity released into the culture medium)/(lactate dehydrogenase activity retained within cells ϩ lactate dehydrogenase activity released into the culture medium) ϫ 100. For nuclear staining, cells were incubated with 25 M Hoechst 33342 and viewed with a fluorescence microscope (BX60, Olympus). For detection of apoptotic DNA fragmentation with 3Ј-OH ends, floating dead cells were collected, fixed with 4% formalin, and then subjected to the terminal deoxynucleotidyl transferase-mediated dUTP nick end labeling (TUNEL) assay using an apoptosis in situ detection kit (Wako) according to the manufacturer's instructions. In vitro measurement of caspase activity was done as described by Kagaya et al. (24).
Adenovirus Vectors and Infection-Myristoylated Akt (myr-Akt), which contains an src myristoylation signal sequence, was described previously (25,26). Akt(T308D/S473D) is a mutated Akt kinase in which Thr-308 and Ser-473 are replaced with aspartate residues. To obtain recombinant adenoviruses expressing these Akt mutants, the expression cosmid cassette pAdexCAwt was ligated with cDNAs encoding these mutants, and homologous recombination between the recom-binant cosmid cassette and its parental virus genome was carried out in 293 host cells, as described previously (25,27). Adenoviruses that underwent successful recombination were isolated, propagated in 293 cells, and subjected to infection of glioma cell lines after determination of the titer (plaque-forming unit/ml) of virus stocks. Adenovirus vectors expressing Bcl-xL (Adv-Bcl-xL) and ␤-galactosidase (Adv-LacZ) were kindly provided by Dr. Hirofumi Hamada (Department of Molecular Medicine, Sapporo Medical University). For infection, 0.8 ϫ 10 6 cells seeded into 100-mm dishes on the previous day were incubated with virus solutions, and 24 h later, 2 ϫ 10 5 of the infected cells were seeded into 60-mm dishes for further analysis, as described above under "Cell Death Induction and Death Assays." Immunoblot Analysis-For detection of Akt, FKHR, and FKHR phosphorylated at Ser-256, cells were lysed in a lysis buffer (20 mM Tris-HCl (pH 7.4), 150 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1% Triton X-100, 2.5 mM sodium pyrophosphate, 1 mM ␤-glycerophosphate, 1 mM Na 3 VO 4 , 1 g/ml leupeptin), and the cell lysates were separated on an SDSpolyacrylamide gel after equalization of protein contents. Proteins were transferred electrically to a nitrocellulose membrane, which was then immunoblotted with antibodies for Akt, FKHR, and FKHR phosphorylated at Ser-256, according to the manufacturer's instructions. The blots were visualized by the enhanced chemiluminescence method and analyzed by a Luminescent Image Analyzer (LAS-1000plus, FUJIFILM).
Mitochondria Assays-To monitor the mitochondrial transmembrane potential (⌬⌿ m ), cells were incubated with 5 M rhodamine 123 for 15 min, and after being washed with phosphate-buffered saline, they were photographed under a laser scanning confocal imaging system (MicroRadiance, Bio-Rad). Subcellular fractionation for the detection of cytosolic and mitochondrial cytochrome c has been described by Asai et al. (23). In brief, cells were lysed in buffer A (20 mM Hepes-KOH (pH 7.5), 10 mM KCl, 1.5 mM MgCl 2 , 1 mM sodium EDTA, 1 mM sodium EGTA, 1 mM dithiothreitol, and 0.1 mM phenylmethylsulfonyl fluoride) containing 250 mM sucrose and homogenized with a Dounce homogenizer. The homogenates were centrifuged twice at 750 ϫ g for 10 min at 4°C. The supernatants were centrifuged at 10,000 ϫ g for 15 min at 4°C. The resulting mitochondrial pellets were retained for preparation of the mitochondrial fraction. The supernatants were centrifuged further at 100,000 ϫ g for 1 h at 4°C. The mitochondrial pellets were resuspended in buffer A containing 250 mM sucrose, sonicated, and centrifuged at 100,000 ϫ g for 1 h at 4°C. The final supernatants (cytosolic fraction and mitochondrial fraction) were separated on a 15% SDS-polyacrylamide gel and immunoblotted with an anti-cytochrome c mouse monoclonal antibody.
Electron Microscopy-Cells were fixed with 1% glutaraldehyde, postfixed with 1% osmium tetroxide, and embedded in epoxy resin. Ultrathin sections were cut, stained with uranyl acetate and lead citrate, and observed with a transmission electron microscope.

Ceramide Induces a Non-apoptotic Cell Death in Human
Glioma Cells-To examine the effect of ceramide on glioblastoma cells, the most malignant neoplasm of glial origin, we treated the U251 human glioblastoma cell line with a cellpermeable ceramide analog, C 2 -ceramide. Pilot experiments revealed that under normal culture conditions, C 2 -ceramide fails to induce the death of U251 cells at concentrations (ϳ50 M) commonly used to examine the biological effects of this ceramide analog. However, U251 cells became sensitive to as low as 25 M C 2 -ceramide when cultured under low serum conditions (serum concentration reduced from 10% to 0.1%), suggesting that serum factors may be responsible for ceramide resistance. As shown in Fig. 1A, when U251 cells were treated with 25 M C 2 -ceramide under the low serum conditions, they became rounded, showed retraction of cellular processes, and eventually detached from the dish. The morphological changes in U251 cells induced by ceramide treatment suggested that they were undergoing cell death, and this was confirmed by the dye exclusion method as well as by the lactate dehydrogenase release assay (Fig. 1, B and C). The induction of U251 cell death by C 2 -ceramide was apparently a specific event because C 2dihydroceramide, a biologically inactive structural analog of ceramide, failed to induce cell death (Fig. 1, A-C). Of note, cells dying from ceramide treatment were in general larger than those undergoing apoptotic death after TNF-␣ treatment (Fig.  1A). To examine whether ceramide-induced U251 cell death has features indicative of apoptosis, we explored evidence of nuclear condensation by staining the cells with a DNA dye Hoechst 33342. The results indicated that in contrast to the apoptotic nuclei of TNF-␣-treated cells showing prominent condensation and fragmentation, the nuclei of C 2 -ceramidetreated cells showed only modest changes compared with the control nuclei (Fig. 1D). Consistently, transmission electron microscopic analysis (Fig. 1E) revealed that whereas TNF-␣treated cells show cellular shrinkage and prominent chromatin condensation, the nuclei of C 2 -ceramide-treated cells remain almost intact. Nevertheless, cells dying as a result of C 2 -ceramide treatment appeared somewhat swollen compared with the control cells and apparently lost microvilli. TUNEL analysis indicated that apoptotic DNA fragments with 3Ј-OH ends are present in cells killed by TNF-␣ but not in cells killed by ceramide (Fig. 1F). These findings together demonstrate that ceramide induces a non-apoptotic cell death in U251 cells. We also confirmed that C 2 -ceramide similarly induces a non-apoptotic cell death in another human glioma cell line T98G (Fig.  1, B and C, and data not shown).
Activation of the Caspase Cascade Is Not Involved in Ceramide-induced Glioma Cell Death-We next asked whether caspases, the critical regulator of apoptotic cell death as well as of a certain type of necrosis (28), are involved in the regulation of the non-apoptotic glioma cell death induced by ceramide. When the activity of effector caspases recognizing and cleaving after the DEVD sequence was monitored in vitro using a synthetic fluorogenic substrate, no significant increase of DEVDcleaving activity was observed in cells dying as a result of C 2 -ceramide treatment in contrast to the untreated and the C 2 -dihydroceramide-treated controls ( Fig. 2A). That TNF-␣ treatment of the same cells induced a substantial increase of DEVD-cleaving activity indicated that glioma cells do have functional caspases (Fig. 2A). To exclude the possibility that caspases are only transiently activated in the course of ceramide-induced cell death, we examined the history of caspase activation by immunoblot analysis using an antibody against poly(ADP-ribose) polymerase, a representative in vivo substrate for effector caspases (5). Again, we found no significant caspase-specific processing of poly(ADP-ribose) polymerase (Fig. 2B), strongly suggesting that the caspase cascade is not activated during ceramide-induced cell death. To determine whether ceramide-induced glioma cell death requires caspase activity, the effect of a pan-caspase inhibitor was examined against ceramide-induced glioma cell death. As shown in Fig.  2C, a pan-caspase inhibitor zVAD-fmk failed to inhibit ceramide-induced glioma cell death at the same concentration that efficiently inhibited TNF-␣-induced apoptosis. Thus, the re- sults indicate that the activation of caspase essential for apoptosis is not involved in ceramide-induced glioma cell death.
Lack of Mitochondrial Alterations in Ceramide-induced Nonapoptotic Cell Death-Evidence is accumulating that mitochondria play a key role in the regulation of cell death, be it either apoptotic or non-apoptotic (29 -31). Among the sequence of events taking place in mitochondria during the course of cell death, loss of the ⌬⌿ m and/or the release of soluble proteins in the intermembrane space such as cytochrome c into the cytosol as a result of outer membrane permeabilization appear to be the major events closely associated with cell death (29 -31). We therefore examined the involvement of mitochondria in ceramide-induced non-apoptotic glioma cell death by monitoring ⌬⌿ m and the cytosolic release of cytochrome c. During TNF-␣-induced apoptosis, we could detect loss of ⌬⌿ m (Fig. 3A) as well as cytochrome c release into the cytosol (Fig. 3B). However, we failed to obtain evidence of ⌬⌿ m loss (Fig. 3A) and cytochrome c release (Fig. 3B) during ceramide-induced glioma cell death, which suggests that mitochondria may not be involved in the regulation of ceramide-induced non-apoptotic glioma cell death.
The Akt Kinase Inhibits Ceramide-induced Non-apoptotic Cell Death That Cannot Be Blocked by Bcl-xL-The experimental findings thus far presented indicated that ceramide treatment induces cell death in human glioma cells which is completely distinct from apoptosis in terms of both morphology and mechanism of regulation. Here, a critical question remains to be answered: is ceramide-induced non-apoptotic cell death a passive process caused by forced destruction of vital cellular machinery (so-called "necrosis"), or is it an active, programmed death that is dependent on the functions of gene products and hence can be controlled by genetic manipulation? To discriminate between these two possibilities, we examined the effect of the expression of genes that have been shown to modulate PCDs. Because anti-apoptotic Bcl-2 family members such as Bcl-2 and Bcl-xL inhibit not only apoptosis but some types of non-apoptotic cell death (32), we examined whether overexpression of Bcl-xL rescues U251 cells from ceramide-induced cell death. Adenovirus-mediated transfer of Bcl-xL achieved a high level of Bcl-xL expression in U251 cells (Fig. 4A), and the U251 cells overexpressing Bcl-xL were apparently resistant to TNF-␣-induced apoptosis compared with the control U251 cells infected with the LacZ-expressing adenovirus (Fig. 4, B and C). However, although there was no significant difference between the expression levels of Bcl-xL under C 2 -ceramide treatment and TNF-␣ treatment (Fig. 4D), Bcl-xL failed to confer significant protection against ceramide-induced cell death (Fig. 4, B and C). Given that Bcl-xL is in general more potent than Bcl-2 as an anti-cell death molecule (33,34), these results suggest that ceramide-induced glioma cell death may not be sensitive to cell death inhibition by anti-apoptotic Bcl-2 family members.
Our earlier observations indicated that U251 cells were resistant to higher concentrations of C 2 -ceramide in the presence of 10% serum. Although we cannot entirely exclude the possibility that serum factors may simply interfere with the action of C 2 -ceramide extracellularly, for instance, by preventing cel-

FIG. 2. Ceramide-induced glioma cell death is caspase-independent.
Human glioma cells were treated as described in Fig. 1. The cells were harvested 24 h after treatment, and the cell lysates were subjected to in vitro caspase assay using Ac-DEVD-MCA as a substrate (A) or to immunoblot analysis using anti-poly(ADP-ribose) polymerase antibody (B). In C, cells deprived of serum were treated with C 2ceramide or vehicle alone (control), or with TNF-␣ plus cycloheximide, in the presence (ϩ) or absence (Ϫ) of a pan-caspase inhibitor zVAD-fmk (50 M). Cell viability was determined by the dye exclusion method 24 h after treatment and normalized to the vehicle control. The data in A and C represent the means Ϯ S.D. from three separate experiments. Data for U251 glioma cells are presented unless otherwise indicated.

FIG. 3. Lack of mitochondrial alterations in ceramide-induced non-apoptotic cell death.
A, U251 cells were deprived of serum and treated with C 2 -ceramide, C 2 -dihydroceramide, or vehicle control for 18 h, or with TNF-␣ plus cycloheximide for 8 h. The cells were then subjected to assessment of the mitochondrial transmembrane potential using a potentiometric dye rhodamine 123. Identical fields are presented as a fluorescence image to visualize rhodamine 123 fluorescence (top panels) and as a bright field image (bottom panels). B, U251 cells treated as in Fig. 1 were harvested and subjected to immunoblot detection of cytochrome c redistribution after subcellular fractionation. Cytochrome c blots of the mitochondrial and cytosolic fractions are presented in the top and bottom panels, respectively lular interaction through absorption, the observations suggested that serum survival factors may confer such resistance through the activation of an intracellular survival signaling. Because the Akt kinase is a key anti-cell death molecule that integrates intracellular signals from various survival factors (22), we asked whether the activation of the Akt signaling pathway might inhibit cell death induced by ceramide. To activate the Akt kinase pathway constitutively, we used an Akt kinase tagged at the N terminus with a myristoylation signal motif (myr-Akt). Myristoylated Akt is relocalized to the plasma membrane and is thereby activated constitutively through phosphorylation by Akt-activating kinases (26). As shown in Fig. 5A, adenovirus-mediated expression at varying levels revealed that myr-Akt, but not LacZ control, protects U251 cells from ceramide-induced death in an expression level-dependent manner. To establish that the protection conferred by myr-Akt is not a nonspecific consequence of protein overexpression or adenovirus infection but is actually dependent on its ability to activate the downstream pathways, we used another Akt mutant in which Thr-308 and Ser-473 are replaced by aspartate residues. This Akt mutant was originally designed to function as a constitutively active kinase (aspartates were expected to mimic the activating phosphorylation at Thr-308 and Ser-473), but turned out to be incapable of phosphorylating the known substrate of Akt at least in our experimental system. In Fig.  5B, U251 cells were infected with adenovirus vectors expressing myr-Akt, Akt(T308D/S473D), and LacZ, and the activity of the Akt signaling pathway was assessed by monitoring the phosphorylation status of the FKHR transcription factor, a representative in vivo substrate of Akt (35). The results indicated that myr-Akt, but not Akt(T308D/S473D) and LacZ, caused significant phosphorylation of FKHR as demonstrated by phospho-FKHR-specific immunoblotting, although myr-Akt and Akt(T308D/S473D) were expressed at comparable levels. We then compared the effect of these Akt kinases on ceramideinduced U251 cell death. The results clearly indicated that myr-Akt, but not Akt(T308D/S473D) and LacZ, rescued U251 cells from ceramide-induced death (Fig. 5, C-E). Essentially similar results were obtained with T98G cells (Fig. 5E). These results collectively demonstrate that the constitutive activation of the Akt signaling pathway blocks transduction of the caspase-independent, non-apoptotic cell death signal elicited by ceramide in human glioma cells. DISCUSSION We have shown in this study that ceramide induces cell death with a non-apoptotic morphology in human glioma cells. This ceramide-induced cell death occurred without the apparent involvement of mitochondria and caspases and was inhibited by a constitutively active Akt kinase but not by the overexpression of Bcl-xL. These observations, together with previous reports (20,21), provide compelling evidence that ceramide is capable of activating a genetically regulatable cell death program that is totally distinct from apoptosis. On the other hand, it has been well established that ceramide can induce typical caspase-dependent apoptotic cell death in vari- FIG. 4. Ceramide-induced non-apoptotic cell death is insensitive to Bcl-xL, which efficiently inhibits TNF-induced apoptosis. U251 cells were infected with adenovirus vectors expressing Bcl-xL (Adv-Bcl-xL) or ␤-galactosidase (Adv-LacZ). The cells were lysed 48 h after infection and subjected to immunoblot analysis using an anti-Bcl-xL antibody (A). In B and C, U251 cells were infected in triplicate with either Adv-Bcl-xL or Adv-LacZ. The infected cells were deprived of serum and then treated with C 2 -ceramide or the vehicle alone (control) or with TNF-␣ plus cycloheximide 48 h after infection. 24 h after C 2ceramide or TNF-␣ treatment, the cells were photographed under a phase contrast microscope (B), or cell viability was determined by the dye exclusion method (expressed as relative to control, the means Ϯ S.D. from three separate experiments; C). In D, U251 cells were infected with Adv-Bcl-xL; 48 h after infection, they were deprived of serum and treated as indicated. After 24 h, the cells were harvested for immunoblot analysis to determine the expression level of Bcl-xL. ous cell types including a certain glioma cell line (19,36). Thus, apparently, whether ceramide induces apoptotic or non-apoptotic cell death (or both) is determined by complex factors including the cell type and the environment surrounding the cells.
We found in this study that ceramide-induced non-apoptotic cell death is accompanied neither by loss of the ⌬⌿ m nor by a redistribution of cytochrome c. Because these mitochondrial alterations are associated with both apoptotic and non-apoptotic cell death whereas caspase activation is unique to apoptotic cell death, it has been suggested that alteration of the mitochondrial physiology is the key step that determines whether a cell lives or dies (29 -31). This idea is consistent with the observation that anti-apoptotic Bcl-2 family proteins, which are presumed to function mainly through maintenance of the mitochondrial physiology, are capable of inhibiting not only apoptosis but also some types of non-apoptotic (necrotic-looking) cell death (10 -12, 32). In this sense, that Bcl-xL failed to inhibit ceramide-induced non-apoptotic cell death in this study may be in line with the lack of mitochondrial alterations during ceramide-induced cell death. Similarly, a lack of mitochondrial alterations and the inability of Bcl-2 to inhibit cell death have been recently reported in Bin1-induced caspase-independent, non-apoptotic cell death (17). These findings may suggest that , or with TNF-␣ plus cycloheximide. After 24 h, cell viability was determined by the dye exclusion method and expressed as relative to that of control (D). Representative phase contrast photomicrographs of cells treated with C 2 -ceramide are shown in C. In E, U251 and T98G cells were infected with the indicated adenovirus vectors (multiplicity of infection ϭ 10). 48 h after infection, the cells were deprived of serum and then treated either with C 2 -ceramide or with TNF-␣ plus cycloheximide. Cell death was quantified 24 h later by the lactate dehydrogenase (LDH) release assay. In D and E, the data represent the means Ϯ S.D. from three and two separate experiments, respectively. non-apoptotic PCDs are heterogeneous, some involving mitochondria and others not.
The most novel and important finding of this study is that Akt inhibited caspase-independent, non-apoptotic cell death induced by ceramide. Although the Akt kinase has been established as a key molecule that mediates anti-"apoptotic" survival signals from various growth factors and cytokines (22), this is to our knowledge the first study to demonstrate the ability of the Akt kinase to inhibit "non-apoptotic" cell death. Although it remains to be seen whether Akt inhibits non-apoptotic PCDs triggered by death stimuli other than ceramide, our results suggest that the Akt kinase, like Bcl-2 and Bcl-xL, may be a key molecule for survival implicated in both apoptotic and non-apoptotic PCD. The molecular mechanism by which Akt inhibits ceramide-induced non-apoptotic PCD remains unclear, but our data support the notion that Akt inhibits non-apoptotic PCD through phosphorylation of its substrate(s) as it inhibits apoptosis. Substrates of the Akt kinase that mediate its antiapoptotic effect identified to date include Bad, caspase-9, IKK, Forkhead transcription factors, GSK-3, and ASK1 (22,37). Some of these Akt targets may mediate the anti-non-apoptotic as well as anti-apoptotic signal from Akt, although Bad and caspase-9 are unlikely to be candidates given the caspaseindependence and lack of Bcl-xL-mediated inhibition of ceramide-induced cell death. Alternatively, there may be yet unidentified Akt substrate(s) specifically involved in the regulation of non-apoptotic PCD. Identification of the substrate(s) will also contribute to an understanding of the signaling pathway connecting ceramide and non-apoptotic cell death.
Our observation that serum factors conferred resistance against ceramide-induced non-apoptotic PCD is of interest in light of the fact that the Akt kinase is known to be activated by various factors presumed to be present in serum (22). This suggests that Akt may mediate the inhibitory effect of serum factors on ceramide-induced non-apoptotic PCD. However, the level of activity, as monitored from the phosphorylation level of the FKHR transcription factor, achieved by the constitutively active Akt (myr-Akt) was much higher than that maintained in the presence of serum. Akt therefore may not be solely responsible for the serum effect, and there may be other molecules mediating survival signaling from serum factors.
The fact that the Akt kinase inhibits both apoptotic and non-apoptotic PCD may have relevance in carcinogenesis. It is now widely accepted that the inhibition of programmed death of potentially oncogenic cells exhibiting oncogene activation and/or tumor suppressor gene inactivation is a critical step toward carcinogenesis. Although "inhibition of apoptosis" has been used in place of "inhibition of PCD," "apoptosis" and "PCD" should be distinguished in this context because it is now apparent that representative oncogenes such as myc and ras may trigger not only apoptotic but also non-apoptotic PCDs (10,15,38,39). If inhibition of apoptosis is sufficient for prevention of the premature death of potentially neoplastic cells, it is expected that the core machinery of apoptosis regulation such as Apaf-1 and caspases should be a favored target of mutation or deletion in human cancer. However, this is unlikely to be the case because inactivation of caspases or Apaf-1 has been observed rather exceptionally in human malignancies (40 -42). In this sense, it is of note that Akt and Bcl-2 having the ability to inhibit both apoptotic and non-apoptotic PCD were originally identified as oncogenes. The Akt genes (Akt1 and Akt2) were originally identified as cellular homologs of v-Akt (43), the oncogene of the AKT-8 murine leukemia virus (44). Subsequent studies revealed that cellular Akt genes are amplified and/or overexpressed in human cancers including gastric, ovarian, breast, and pancreatic carcinomas (43,(45)(46)(47) and that the Akt kinase activity is elevated in glioblastoma cells (48). In addition, cellular transformation has been successfully induced by the activation of the Akt kinase pathway (49,50). Similarly, Bcl-2 was originally isolated from the chromosomal breakpoint of follicular lymphoma, and its oncogenic potential has been well documented in bcl-2 transgenic mice (51). Although Akt and Bcl-2 may contribute to oncogenesis through other mechanisms in addition to cell death inhibition (e.g. cell cycle regulation), these findings suggest that inhibition of not only apoptosis but also non-apoptotic PCD may be essential during carcinogenesis. Conversely, they provide indirect evidence that cells may activate non-apoptotic as well as apoptotic cell suicide programs in response to genetic alterations that are potentially neoplastic.