A kinase-independent function of Ask1 in caspase-independent cell death.

Ask1 (apoptosis signal-regulating kinase 1) is activated as a consequence of cell exposure to a variety of stresses and can then initiate apoptosis. A known pathway of apoptosis downstream of Ask1 involves the activation of the stress-activated protein kinases, the release of cytochrome c from mitochondria, the activation of caspases, and the fragmentation of nuclei. Here, we characterized a novel mechanism of Ask1-mediated cell killing that is triggered by the interaction with Daxx. Co-transfection of Ask1 and Daxx induced a caspase-independent cell-death process characterized at the morphological level by distinctive crumpled nuclei easily distinguishable from the condensed and fragmented nuclei seen during classical caspase-dependent apoptosis. The kinase activity of Ask1 was not involved in this process, because mutants lacking kinase activity were as efficient as wild type Ask1 in mediating Daxx-induced cell death. Ask1N, a deletant that lacks the C-terminal half including the kinase domain of Ask1, was constitutively active in producing crumpled nuclei. In contrast, Ask1DeltaN, the reciprocal deletant that possesses constitutive kinase activity, produced fragmented nuclei typical of caspase-dependent death processes. We conclude that in addition to a caspase-dependent pro-apoptotic function that depends on its kinase activity, Ask1 possesses a caspase-independent killing function that is independent on its kinase activity and is activable by interaction with Daxx. In the physiological situation, such an activity is induced as a consequence of the translocation of Daxx from the nucleus to the cytoplasm, a condition that occurs following activation of the death receptor Fas.

Ask1 (apoptosis signal-regulating kinase 1) is a mitogenactivated protein kinase kinase kinase that activates the stress-activated protein kinase c-Jun N-terminal kinase (JNK) 1 and p38 pathways when stimulated by diverse stimuli such as tumor necrosis factor ␣, Fas, H 2 O 2 , serum withdrawal, and cisplatin (1)(2)(3)(4)(5). Several studies demonstrated that such activation can lead to apoptosis. Studies with Ask1-deficient cells revealed that apoptosis induced by tumor necrosis factor ␣ treatment and H 2 O 2 exposure is associated with an Ask1mediated sustained activation of the JNK and p38 pathways (6). It has also been shown that an interfering kinase mutant of Ask1 can in some cases block apoptosis (2,3,5), whereas a constitutively active kinase mutant of Ask1 can induce caspasedependent cell death (7). Possible mechanisms by which apoptosis is induced downstream of Ask1 involve the phosphorylation-mediated inactivation of Bcl-2 and the phosphorylation-mediated stabilization of c-Myc, both processes being catalyzed by activated JNK (8,9). c-Myc promotes the release of cytochrome c from mitochondria whereas Bcl-2 prevents this release (10 -12). Cytochrome c translocation leads to the activation of caspase activities and the cleavage of crucial targets such as the inhibitor of caspase-activated DNase, leading in turn to the typical manifestations of caspase-dependent apoptosis including internucleosomal DNA degradation and nuclear condensation and fragmentation (13).
One mechanism of activation of the pro-apoptotic function of Ask1 involves Daxx, a nuclear protein normally localized in the nuclear substructures called nuclear domain 10 or PML (promyelocytic leukemia) oncogenic domain (14 -17). Upon stimulation of the death receptor Fas, Daxx translocates from the nucleus to the cytoplasm and triggers caspase-independent cell death by association with Ask1 (3,17,18). In the present study we investigated the mechanism by which Ask1 mediates the toxic effect of cytoplasmic Daxx. The results revealed a novel function of Ask1 in regulated cell death.
Cell Culture and Transfection-293 cells were maintained at 37°C in a humidified 5%-CO 2 atmosphere in Dulbecco's modified Eagle's medium containing 2.2 g/liter NaHCO 3 , 4 mM L-glutamine, and 4.5 g/liter glucose (Life Technologies, Inc., catalog number 12800-017) and supplemented with 10% fetal calf serum. In some experiments as indicated, the cells were incubated in L-glutamine-free medium (Life Technologies, Inc., catalog number 11960-044). Transfection was done as described before by calcium-phosphate precipitation in the presence of 50 M chloroquine (17).
Immunofluorescence Microscopy-293 cells that were plated on gelatin-coated 6-well dishes were fixed for 20 min in 3.7% formaldehyde made in saline (137 mM NaCl, 5 mM KCl, 10 mM Na 2 HPO 4 , and 11 mM glucose, pH 7.2) and permeabilized for 15 min in 0.1% saponin. EL-68 antibody diluted 1/100 and HA.11 antibody diluted 1/300 was used to detect Daxx and HA-tagged proteins, respectively. Protein-antigen-antibody complex was revealed with AlexaFluor 488-labeled anti-rabbit IgG and AlexaFluor 594-labeled anti-mouse antibodies (Molecular Probes) diluted 1/2000. For fluorescence microscopy of green fluorescent protein-expressing cells, the cells were fixed in 3.7% formaldehyde as described above and then postfixed in 70% ethanol overnight. In all cases, the cell nuclei were stained with DAPI (2.5 g/ml). Epi-fluorescence microscopy was done on a Nikon Eclipse E600 microscope. Photographs were obtained with a CoolSnap digital camera system (RS Photometrics).
Internucleosomal DNA Fragmentation-Exponentially growing 293 cells were plated onto 6-well dishes at 1.5 ϫ 10 5 cells per well and transfected or treated 24 h later as described in the legend for Fig. 1. 24 h after transfection or treatment, floating and adherent cells were pooled, washed in saline, and lysed at 50°C for 48 h in buffer containing 10 mM TRIS, pH 7.4, 1 mM EDTA, 150 mM NaCl, 1% SDS, and 0.2 mg/ml proteinase K. The lysate was subjected to two phenol-chloroform extractions. DNA was precipitated with isopropanol and resuspended in 10 mM TRIS, pH 8.0, 1 mM EDTA, and 0.5 mg/ml RNase. The DNA was separated into a 1.0% agarose gel and revealed by ethidium-bromide staining.

RESULTS
Transfection of plasmid pcDNA3-HA-Ask1 yielded the expression of HA-Ask1 that co-localized in the cytoplasm with endogenous Ask1. Transfection of plasmid pCINHuDaxx resulted in an enhanced expression of Daxx in the nucleus, which like endogenous Daxx was concentrated in nuclear bodies with the PML protein (see Ref. 14, and see Fig. 1A). However, co-transfection of the Daxx and Ask1 plasmids resulted in more than 50% of the cells in the localization of Daxx with Ask1 in the cytoplasm and in more than 95% of these cells, in severe morphological alterations of the nuclei (Fig. 1, A and B). The nuclei looked as if they had been crushed out of shape, showing frequent invaginations and irregularly condensed chromatin regions forming intranuclear wrinkles. These crumpled nuclei were easily distinguishable from the condensed and fragmented nuclei seen in classical caspase-dependent apoptosis as, for example, induced by Fas in the same cells (see later in Fig. 3). In contrast to the known mechanisms of Ask1-mediated apoptosis, this Daxx-Ask1-mediated cell-death process was insensitive to the caspase inhibitor zVAD-fmk (17) and was not accompanied by internucleosomal DNA degradation as seen otherwise after treatment with puromycin or activation of Fas in the same cells (Fig. 1C).
Once activated Ask1 leads to the phosphorylation and activation of the stress-activated protein kinases JNK and p38, two kinases whose activation has been often associated with apoptosis (see Refs. 2 and 21, and see Ref. 22 and references therein). In initial experiments, however, we observed that transfection of Daxx, alone or together with Ask1, had no effect on the activity of JNK1 or p38 (data not shown) suggesting that Daxx did not lead to the activation of the kinase activity of Ask1 and thus that cell death was not induced as a consequence of such an activation. This was confirmed by looking at the potential of a kinase-inactive mutant of Ask1 (Ask1KM) to induce nuclear crumpling when co-transfected with Daxx. We found that co-transfection of Ask1KM with Daxx was as efficient as wild type Ask1 to cause the re-localization of Daxx to the cytoplasm and to induce the typical nuclear alterations (Fig. 1B).
The kinase domain of Ask1 is centrally located in the protein and is regulated both by the N terminus, which contains the sites of interaction of numerous Ask1 regulators (3,(23)(24)(25)(26), and by intramolecular interactions between the N and C termini. Such interactions keep the kinase in a repressed state, and deletion of the N-terminal domain of Ask1 leads to a constitutively activated kinase (2, 3). We investigated which domain of Ask1 was responsible for the caspase-independent cell-death activity by generating a number of deletants of the N terminus (Ask1⌬N, lacking residues 5-649), the kinase domain (Ask1⌬K, lacking residues 649 -940), or the kinase domain and the C terminus (Ask1N, lacking residues 649 -1375) ( Fig. 2A). All the mutant proteins with an intact N-terminal domain were able to re-localize Daxx to the cytoplasm, and conversely, the mutant lacking this domain (Ask1⌬N) was unable to retain Daxx in the cytoplasm (Fig. 2B). This result is consistent with a previous demonstration that Daxx associates with the Nterminal domain of Ask1 (3). The potential of the deletants to induce cell death either when expressed alone or with Daxx was tested also (Fig. 2C). All deletants that could maintain Daxx in the cytoplasm also produced crumpled nuclei in the is not associated with internucleosomal DNA degradation. One negative and two positive controls were included. NT, untransfected cells; NTϩpuro, untransfected cells treated with puromycin (5 g/ml for 24 h); Fasϩ␣Fas, Fas-transfected cells exposed to the activating Fas antibody (100 ng/ml) for 18 h. kb, kilobase pair.
A Kinase-independent Function of Ask1 36072 presence of Daxx. Interestingly, in contrast to transfection of wild type Ask1, transfection of Ask1N alone already resulted in a significant constitutive cell killing activity, i.e. co-transfection with Daxx was not required although it enhanced the toxicity. The results clearly demonstrated that the kinase activity was not required for Ask1 to induce nuclear crumpling. Furthermore, the results suggested that the N-terminal domain of Ask1 possesses a cell-death activity that is normally repressed in the wild type protein and de-repressed by Daxx.
Ask1⌬N, a constitutively kinase-active form of Ask1, is also known to induce cell death. When transfected, Ask1⌬N induced a slow caspase-dependent cell-death process that requires some 2-3 days and a sustained activation of JNK (6, 7). We confirmed these results by showing that transfection of Ask1⌬N induced by 3 days a condensed/fragmented nucleus pattern typical of caspase-dependent apoptosis (Fig. 2D). No fragmentation (nor crumpling) was seen at 2 days. Furthermore, we found that Ask1⌬N-induced apoptosis was dependent on the absence of glutamine in the medium (data not shown), consistent with the results that t-glutamyl RNA synthetase is a re-pressor of this activity (27). Cell death induced by transfection of Ask1⌬N contrasted with death induced by transfection of Ask1N. The Ask1N-induced process occurred by 24 h, was observed both in the presence and absence of glutamine (data not shown), and produced crumpled rather than fragmented nuclei (Fig. 2E). The results thus implied that Ask1 possesses two domains active in inducing cell death and thereby modulates two distinct mechanisms of cell death.
The stimulation of the death receptor Fas represents one physiological situation in which this novel function of Ask1 may participate in cell-death induction. Indeed, Fas induces the translocation of Daxx from the nucleus to the cytoplasm and a caspase-independent cell-death process, which becomes important when the Fas-induced caspase-dependent pathway is inhibited either with zVAD-fmk treatment or expression of a dominant negative form of FADD (17). Such a situation may occur in normal physiology, for example, during viral infection where inhibitory proteins like v-FLIP or CrmA are produced (28). We looked at whether the specific nuclear alterations obtained upon transfection of Ask1N or co-transfection of Daxx and Ask1 were also induced during Fas stimulation. Transfection with Fas followed by exposure to the Fas-activating anti- A Kinase-independent Function of Ask1 36073 body yielded within 24 h characteristic features of caspase-dependent apoptosis such as internucleosomal DNA degradation (Fig. 1C) and with times up to 72 h an increasing number of cells with a condensed and fragmented nucleus (Fig. 3, A and C). Treatment with zVAD-fmk totally inhibited at all times the Fas-induced nuclear condensation/fragmentation phenotype. However, in the presence of zVAD-fmk, an increasing number of cells showed with times up to 72 h the same crumpled nuclear phenotype as seen after co-transfection of Daxx and Ask1 or Ask1N alone (see Fig. 3B, and compare Fig. 3D to Fig.  3E and Fig. 2E). Overexpression of the kinase-dead mutant of Ask1 (Ask1KM) enhanced by 2-to 3-fold Fas-induced caspaseindependent nuclear crumpling (Fig. 3B) presents further evidence that the two phenomena are similar. Nuclear crumpling was produced much faster after transfection of Ask1N than after activation of Fas, suggesting that Ask1N acted as a constitutive mutant short-circuiting early signaling events (e.g. Daxx re-localization) required in the case of Fas.

DISCUSSION
Previous reports demonstrated that Ask1 can induce through its kinase activity a caspase-dependent apoptosis pathway. This was well illustrated in the case of caspase-dependent apoptosis induced by genotoxic stress and tumor necrosis factor ␣ where overexpression of a kinase-dead mutant of Ask1 was shown to inhibit cell death (2,5). We reported here the existence of a novel mechanism of cell death signaling by Ask1. The process does not involve the kinase activity of Ask1 and activates a caspase-independent pathway leading to a nuclear crumpling morphology distinct from the caspase-dependent nuclear condensation/fragmentation process induced downstream of kinase active Ask1 and other classical pathways of apoptosis. The mechanism involves the N-terminal half of the protein for which no catalytic activity is known. The existence of two independent functions for Ask1 is not so surprising considering the possibility for many distinct functional domains in a protein of the size of Ask1 (ϳ155 kDa) and the already known large number of different targets to which Ask1 binds (3,(23)(24)(25)(26). There are in fact several protein kinases, including Ask1 itself, for which kinase-independent activities have been characterized (29 -35).
Our finding that Ask1N possesses a constitutive cell-death activity suggests that this activity is repressed in the wild type protein and that Daxx de-represses this activity when it interacts with the N terminus. This resembles the way the kinase activity of Ask1 is regulated, being repressed by intermolecular interaction between the C and N termini and activated by de-repressors such as TRAF2 (3,26). In this respect it is interesting that Daxx has also been reported to activate the kinase activity of Ask1 (3). The similarity of the mechanisms involved in the activation of the two functions of Ask1 makes it plausible that in some cellular context the interaction of Daxx with Ask1 results in the activation of the kinase activity rather than in the cell-death activity linked to the N terminus.
The results presented here, together with previous findings, strongly suggest that the caspase-independent cell-death pathway mediated by the N terminus of Ask1 is activated and can play a role upon stimulation of the Fas receptor (17). It is tempting to suggest that de-repression of this Ask1-induced cell-death pathway by Daxx may also be implicated importantly in the response to several stressful stimuli inasmuch as HSP27, a protein that protects cells against a large number of stress such as heat shock and toxic chemicals and has been demonstrated to bind Daxx and to prevent its translocation into the cytoplasm and its interaction with Ask1 (17,36).