Role of the ASK1-SEK1-JNK1-HIPK1 signal in Daxx trafficking and ASK1 oligomerization.

Overexpression of JNK binding domain inhibited glucose deprivation-induced JNK1 activation, relocalization of Daxx from the nucleus to the cytoplasm, and apoptosis signal-regulating kinase 1 (ASK1) oligomerization in human prostate adenocarcinoma DU-145 cells. However, SB203580, a p38 inhibitor, did not prevent relocalization of Daxx and oligomerization of ASK1 during glucose deprivation. Studies from in vivo labeling and immune complex kinase assay demonstrated that phosphorylation of Daxx occurred during glucose deprivation, and its phosphorylation was mediated through the ASK1-SEK1-JNK1-HIPK1 signal transduction pathway. Data from immunofluorescence staining and protein interaction assay suggest that phosphorylated Daxx may be translocated to the cytoplasm, bind to ASK1, and subsequently lead to ASK1 oligomerization. Mutation of Daxx Ser667 to Ala results in suppression of Daxx relocalization during glucose deprivation, suggesting that Ser667 residue plays an important role in the relocalization of Daxx. Unlike wild-type Daxx, a Daxx deletion mutant (amino acids 501-625) mainly localized to the cytoplasm, where it associated with ASK1, activated JNK1, and induced ASK1 oligomerization without glucose deprivation. Taken together, these results show that glucose deprivation activates the ASK1-SEK1-JNK1-HIPK1 pathway, and the activated HIPK1 is probably involved in the relocalization of Daxx from the nucleus to the cytoplasm. The relocalized Daxx may play an important role in glucose deprivation-induced ASK1 oligomerization.

We previously observed that glucose derivation increases the intracellular levels of hydroperoxide and oxidized glutathione (1). Our recent studies have shown that increases in steadystate levels of hydrogen peroxide and glutathione disulfide are sensed through thioredoxin (TRX) 1 and glutaredoxin (GRX) and subsequently activate the ASK1-MEK-MAPK signal transduction pathway (2)(3)(4). TRX and GRX appear to act as physiological inhibitors of ASK1 by associating with the N-terminal and C-terminal portion of ASK1, respectively, and inhibiting ASK1 kinase activity (2,5). TRX and GRX contain two redoxactive half-cystine residues, -Cys-Gly-Pro-Cys-or -Cys-Pro-Tyr-Cys-, in an active catalytic center (5)(6)(7)(8). These sensor molecules may be converted to the intramolecular disulfide form of TRX-(S-S) and GRX-(S-S) during glucose deprivation. The oxidized form of TRX and GRX dissociates from ASK1 and consequently activates ASK1 (2,3,5). Recently, we observed that release of TRX and GRX from ASK1 occurs with different mechanisms: the glutathione-dependent GRX-ASK1 pathway and the glutathione-independent TRX-ASK1 pathway (4). Dissociation of either regulator from ASK1 is sufficient for ASK1 activation (4).
ASK1 is a member of the mitogen-activated protein kinase kinase kinase family that activates the JNK and p38 pathways by directly phosphorylating and thereby activating their respective mitogen-activated protein kinase kinases, MKK4 (SEK1)/MKK7 and MKK3/MKK6 (9). ASK1 is activated by oxidative stress (2,5), TNF-␣ (10,11), Fas ligand (12,13), and endoplasmic reticulum stress (14). Previous studies have shown that TNF-␣ activates ASK1 via TRAF2, a member of the TNF-receptor-associated factor (TRAF) family (10,11), whereas Fas ligand activates ASK1 via Daxx (12,13). Liu et al. (11) reported that TRAF2 activates ASK1 by enhancing and stabilizing the oligomerization of ASK1. Chang et al. (12) observed that Daxx activates ASK1 by displacing an inhibitory intramolecular interaction between the NH 2 and COOH termini of the kinase, thereby opening up the kinase into an active conformation. Previous studies have also shown that ASK1 is located in the cytoplasm and Daxx is mainly located in the nucleus (13,15). It is well known that Daxx relocalizes from the nucleus to the cytoplasm in response to stress (15). Thus, relocalization of Daxx is required prior to its interaction with ASK1. A fundamental question is what molecular change(s) regulate the relocalization of Daxx from the nucleus to the cytoplasm in response to stress? Here we provide a possibility that the initial activation of JNK1 during glucose deprivation induces Daxx relocalization through phosphorylation. The relocalized Daxx induces ASK1 oligomerization.

EXPERIMENTAL PROCEDURES
Cell Culture and Glucose Deprivation-Human prostate adenocarcinoma (DU-145) cells were cultured in Dulbecco's modified Eagle's medium with 10% fetal bovine serum (Invitrogen), and 26 mM sodium bicarbonate for monolayer cell culture. The cells were maintained in a humidified atmosphere containing 5% CO 2 and air at 37°C. For glucose deprivation, cells were rinsed three times with phosphate-buffered saline (PBS) and then exposed to glucose-free Dulbecco's modified Eagle's medium containing 10% dialyzed fetal bovine serum (Invitrogen).
Adenoviral Vector Construction-All recombinant adenoviruses were constructed by employing the Cre-lox recombination system (16). The selective cell line CRE8 has a ␤-actin-based expression cassette driving a Cre recombinase gene with an N-terminal nuclear localization signal stably integrated into 293 cells. Transfections were done by using LipofectAMINE Reagent (Invitrogen). 5 ϫ 10 5 cells were split into a 6-well plate 1 day before transfection. For the production of recombinant adenovirus, 2 g of SfiI/ApaI-digested Adlox/HA or Myc-ASK1 fragment or Adlox/FLAG-JBD or SfiI-digested Adlox/FLAG-Daxx including various Daxx deletion mutants or His-Daxx fragment or Adlox/ His-JNK1 and 2 g of 5 viral genomic DNA were co-transfected into CRE8 cells. The recombinant adenoviruses were generated by intermolecular homologous recombination between the shuttle vector and 5 viral DNA. A new virus has an intact packaging site and carries a recombinant gene. Plaques were harvested, analyzed, and purified. The insertion of HA-ASK1 or Myc-ASK1 or FLAG-JBD or various FLAG-Daxx or His-Daxx or His-JNK1 to adenovirus was confirmed by Western blot analysis, after infection of corresponding recombinant adenovirus into DU-145 cells.
In Vivo Binding of ASK1 with Daxx-To examine the interaction between ASK1 and Daxx, adenovirus of HA-tagged ASK1 (Ad.HA-ASK1) and FLAG-tagged Daxx (Ad.FLAG-Daxx) at an MOI of 10 were co-infected into DU-145 cells in 10-cm culture plates. For immunoprecipitation, after 48 h of infection, cells were lysed in buffer containing 150 mM NaCl, 20 mM Tris-HCl (pH 7.5), 10 mM EDTA, 1% Triton X-100, 1% deoxycholate, 1 mM phenylmethylsulfonyl fluoride (PMSF), 80 M aprotinin, 2 mM leupeptin, and the lysates were incubated with 3 g of anti-FLAG M2 mouse IgG1 (Sigma) or 0.5 g of rat anti-HA (clone 3F10; Roche Applied Science) for 2 h, respectively. After the addition of protein G-agarose (Santa Cruz Biotechnology, Inc., Santa Cruz, CA), the lysates were incubated for an additional 2 h. The beads were washed three times with the lysis buffer, separated by SDS-PAGE, and immunoblotted with rat anti-HA or mouse anti-HA (clone 12CA5; Roche Applied Science) antibodies or mouse anti-FLAG. Proteins in the membranes were then visualized using the ECL reagent as recommended by the manufacturer (Amersham Biosciences).
Immune Complex Kinase Assay and in Vivo Labeling-The PCR product of human Daxx having restriction enzyme sites at the flanking sides (5Ј, NdeI; 3Ј, BamHI) was produced using the pFLAG/CMV2-Daxx as a template. Sense primer was 5Ј-GCTGCATATGGCCACCGCTAA-CAGCATCATC-3Ј, and antisense primer was 5Ј-CTGCGGATCCCTA-ATCAGAGTCTGAGAGCAC-3Ј. pET15b/Daxx was made by inserting NdeI/BamHI fragment from Daxx PCR product into NdeI/BamHI-cut pET15b (Novagen, Madison, WI). pET15b/Daxx was transformed into BL21(DE3)pLysS, and Daxx expression was confirmed by anti-Daxx (Sigma). His-Daxx was purified by using His-bind column (Novagen). For the immune complex kinase assay, DU-145 cells were infected with Ad.His-JNK1 at an MOI of 10. After 48 h of infection, cells were in glucose-free medium for 1 h and lysed in a buffer solution containing 20 mM Tris-HCl (pH 7.5), 150 mM NaCl, 5 mM EGTA, 10 mM NaF, 1% Triton X-100, 0.5% deoxycholate, 2 mM DTT, 1 mM sodium orthovanadate, 1 mM PMSF, and protein inhibitor mixture solution (Sigma). Cell extracts were clarified by centrifugation, and the supernatants were immunoprecipitated with mouse anti-His antibody (Qiagen, Valencia, CA) and protein G-agarose. The beads were washed twice with a solution containing 150 mM NaCl, 20 mM Tris-HCl (pH 7.5), 5 mM EGTA, 2 mM DTT, 1 mM sodium orthovanadate, 1 mM PMSF, and protein inhibitor mixture solution and washed once with the kinase buffer solution, and then they were subjected to kinase assays. To examine whether the Daxx was a direct substrate of JNK1, 1 g of His-Daxx or GST-c-Jun was incubated with immunoprecipitated His-JNK1 in kinase buffer containing 20 mM Tris-HCl (pH 7.5), 20 mM MgCl 2 , 1 mM sodium orthovanadate, 2 mM DTT, 20 M ATP, and 100 Ci/ml [␥-32 P]ATP at 30°C for 1 h. Finally, the reaction was stopped by adding 2ϫ Laemmli buffer. Phosphorylated proteins were resolved by SDS-PAGE and analyzed by autoradiography. For direct [ 32 P]orthophosphate labeling of His-Daxx, DU-145 cells were infected with Ad.His-Daxx at an MOI of 10. After 48 h of infection, cells were pre-equilibrated in phosphate-free medium for 3 h prior to the addition of 50 Ci/ml of [ 32 P]orthophosphate (ICN, Irvine, CA) in phosphate and glucose-free medium for 1 h and then lysed in a buffer solution containing 20 mM Tris-HCl (pH 7.5), 150 mM NaCl, 5 mM EGTA, 10 mM NaF, 1% Triton X-100, 0.5% deoxycholate, 0.1% SDS, 2 mM DTT, 1 mM sodium orthovanadate, 1 mM PMSF, and protein inhibitor mixture solution (Sigma). Cell extracts were clarified by centrifugation, and the supernatants were immunoprecipitated with mouse anti-His antibody and protein G-agarose. The beads were washed three times with a solution containing 150 mM NaCl, 20 mM Tris-HCl (pH 7.5), 5 mM EGTA, 2 mM DTT, 1 mM sodium orthovanadate, 1 mM PMSF, and protein inhibitor mixture solution. The immune complex was separated by SDS-PAGE, and dried gel was visualized by autoradiography.
Immunoblot Analysis-Cell lysates were subjected to electrophoresis on 10% polyacrylamide gels containing SDS under reducing conditions, and the proteins in the gels were transferred onto a polyvinylidine difluoride membrane. The membranes were incubated with 7% (w/v) skim milk in PBST (PBS containing 0.1% (v/v) Tween 20) and then reacted with primary antibodies. Polyclonal rabbit anti-ACTIVE JNK was obtained from Promega (Madison, WI). Monoclonal mouse antiactin antibody was purchased from ICN. After washing three times with PBST, the membranes were incubated with horseradish peroxidase-conjugated anti-IgG. Then the proteins were detected with the ECL reagent.
Immunofluorescence-Cellular localization of FLAG-Daxx (or His-Daxx) was investigated using fluorescence microscopy. The cells were plated onto a Lab-Tek chamber slide (Nalge Nunc, Naperville, IL) and infected with Ad/FLAG-Daxx (or Ad/His-Daxx) at an MOI of 10. After 48 h of infection, cells were fixed in 100% cold methanol for 10 min at Ϫ20°C. After washing twice with cold PBS, the cells were blocked in 1% bovine serum albumin plus 10% rabbit or goat serum (depending on the source of second antibody) for 1 h at room temperature. They were then incubated with anti-FLAG (clone M2; mouse) or anti-His (penta-His; mouse) antibodies containing 1% bovine serum albumin plus 10% rabbit or goat serum for 1 h at room temperature, followed by three washes with cold PBS. Samples of FLAG-Daxx (or His-Daxx) were incubated for 1 h with fluorescein isothiocyanate-conjugated rabbit anti-mouse IgG. After washing three times with cold PBS, the slides were mounted in 90% glycerol.
Site-directed Mutagenesis-The QuikChange TM site-directed mutagenesis kit (Stratagene, La Jolla, CA) was used to make point mutations in Daxx protein. Two serine residues in Daxx (Ser 667 and Ser 670 ) were replaced with alanine. Sense primer oligonucleotide (5Ј-GTAC-CCTGCCCGCCCCACCTTCCC-3Ј) and antisense primer oligonucleotide (5Ј-GGGAAGGTGGGGCGGGCAGGGTAC-3Ј) were for S667A. Sense primer oligonucleotide (5Ј-GCCCAGCCCACCTGCCCCCTTA-GCTTC-3Ј) and antisense primer oligonucleotide (5Ј-GAAGCTAAGGG-GGCAGGTGGGCTGGGC-3Ј) were for S670A. PCR was prepared by adding 5 l of 10ϫ reaction buffer, 20 ng of double-stranded DNA template (pFLAGCMV2-Daxx), 125 ng of each sense primer, 125 ng of each antisense primer, 1 l of dNTP mix, double-distilled water to a final volume of 50 l, and 1 l of Pfu Turbo DNA polymerase (2.5 units/l). PCR was performed with 12 cycles (95°C for 30 s; 55°C for 1 min; 68°C for 14 min for S667A and 95°C for 30 s; 58°C for 1 min; 68°C for 14 min for S670A) with an initial incubation at 95°C for 30 s. Following temperature cycling, the reaction was placed on ice for 2 min to cool the reaction. After PCR, 1 l of DpnI restriction enzyme (10 units/l) was added directly to each amplification reaction and incubated at 37°C for 1 h to digest the parental supercoiled double-stranded DNA. The DpnI-treated double-stranded DNA was transformed into Epicurian Coli XL1-Blue supercompetent cells. Colonies were selected, and each plasmid (pFLAGCMV2-Daxx) was subcloned into pAdlox-Flag-Daxx, followed by digestion with HindIII/KpnI. Its fragment containing the mutation site was subcloned into pBluescript SK(Ϫ). Each pBluescript SK(Ϫ)-Daxx fragment was sequenced using T7 primer to confirm mutation.

Oligomerization of ASK1 and Interaction between Daxx and ASK1 during Glucose Deprivation-Previous studies have
shown that oligomerization of ASK1 occurs during treatment with TNF in human embryonic kidney 293 cells (11). Activation of ASK1 requires reactive oxygen species-mediated dissociation of TRX from ASK1 followed by the binding of TRAF2 and consequent ASK1 homo-oligomerization (11). Thus, we examined whether Daxx also plays a role in the oligomerization of ASK1 during glucose deprivation. Fig. 1A shows that homooligomerization of ASK1 occurred during glucose deprivation or H 2 O 2 treatment. Unlike ASK1, Daxx existed as an oligomer form irrespective of oxidative stress (Fig. 1B). Several researchers have shown that Daxx, a Fas-binding protein, binds to ASK1, thereby activating the ASK1 kinase (12,17). We investigated whether glucose deprivation induces interaction between Daxx and ASK1. Fig. 2 shows that Daxx associated with ASK1 during oxidative stress (glucose deprivation or H 2 O 2 treatment).
Daxx Binding Site to ASK1 and Localization of Various Deletion Mutant Types of Daxx-We further examined which domain of Daxx is responsible for interacting with ASK1. Cells were co-infected with Ad.HA-ASK1 and adenoviral vectors containing wild-type Daxx or its various Daxx deletion mutants (amino acids 1-625, 1-500, 501-625, and 626 -739). Fig. 3A shows that wild-type Daxx and Daxx deletion mutants (aa 1-625 and 501-625) interacted with ASK1 during glucose deprivation. Fig. 3B shows that wild-type Daxx, which contains two nuclear localization signals (18), is mainly located in the nucleus. Wild-type Daxx relocalized to the cytoplasm during glucose deprivation. In contrast, the Daxx deletion mutant (aa 501-625) is mainly located in the cytoplasm irrespective of the extracellular glucose concentration. Other deletion mutants (aa 1-625, 1-500, and 626 -739), which contain one nuclear localization signal, localize to both the cytoplasm and the nucleus (Fig. 3B). Glucose deprivation did not significantly alter the intracellular distribution of any of the deletion mutants (Fig. 3B). These results suggest that intracellular location of Daxx as well as amino acid residues of the Daxx domain play an important role in the interaction between Daxx and ASK1. This possibility was further examined. As shown previously in Fig. 2, wild-type Daxx associated with ASK1 during glucose deprivation (Fig. 4A, lane 3). However, unlike wild-type Daxx, Daxx deletion mutant (aa 501-625) interacted with ASK1 regardless of whether glucose was present or absent from the medium (Fig. 4A, lanes 4 and 5). We also investigated whether Daxx is involved in the homo-oligomerization of ASK1.  shows that oligomerization of ASK1 induced by glucose deprivation was unaffected by overexpressed wild-type Daxx (Fig.  4B, lanes 3 and 5). Unlike wild type Daxx, expression of Daxx deletion mutant (aa 501-625) caused oligomerization of ASK1 even in the presence of glucose (Fig. 4B, lane 6). The differential roles of wild-type Daxx and deletion mutant-type Daxx (aa 501-625) can be explained by differential localization of these proteins (Fig. 3B). Previous studies have shown that Daxx deletion mutant (aa 501-625) is largely responsible for JNK activation (17). Fig. 4C shows that overexpression of Daxx deletion mutant (aa 501-625) indeed activates JNK1 even in the presence of glucose. These results suggest that physical interaction between Daxx and ASK1 is sufficient to cause ASK1 oligomerization and subsequent JNK1 activation.

Effect of JBD Overexpression on Relocalization of Daxx and Interaction between ASK1 and Daxx during Glucose Deprivation-A fundamental question that remains unanswered in
this study is how Daxx translocates to the cytoplasm during glucose deprivation. One possibility is that oxidative stressinduced ASK1-SEK1-JNK1 signal transduction is involved in the relocalization of Daxx. We hypothesized that the ASK1-SEK1-JNK1 pathway must already be activated to get increased Daxx binding to ASK1, and then increased Daxx binding would maintain the activated ASK1-SEK1-JNK1. To test this possibility, we overexpressed the JNK binding domain (JBD), a negative regulator of JNK. As shown in Fig. 5, overexpression of JBD inhibited JNK1 activation during glucose deprivation. In contrast, JBD overexpression did not affect the activation of p38 during glucose deprivation (data not shown). Overexpression of JBD prevented relocalization of Daxx to the cytoplasm and the binding of Daxx to ASK1 during glucose deprivation or H 2 O 2 treatment (Fig. 6, A, B, and D). However, Daxx relocalization and interaction between Daxx and ASK1 were not prevented by treatment with SB203580, a specific p38 inhibitor (Fig. 6, C and D). We confirmed the inhibitory effect of SB203580 on p38 by using a p38 MAP kinase assay kit (Cell Signaling Technology, Inc., Beverly, MA) (data not shown). Our results suggest that the oxidative stress-activated ASK1-SEK1-JNK1 signal transduction pathway, but not the ASK1-MKK3/MKK6-p38 pathway, plays an important role in the relocalization of Daxx to the cytoplasm and its subsequent interaction with ASK1.
Glucose Deprivation-induced Daxx Phosphorylation and Effect of JBD Overexpression on Daxx Phosphorylation-To investigate whether Daxx is phosphorylated during glucose deprivation, DU-145 cells were infected with Ad.His-Daxx. Studies from in vivo labeling with [ 32 P]orthophosphate show that Daxx was phosphorylated during glucose deprivation (Fig. 7, lane 3). The phosphorylation of Daxx was suppressed by overexpression of JBD, a negative regulator of JNK (Fig. 7, lane 4). These results suggest that JNK1 is involved in Daxx phosphorylation.
Phosphorylation of Daxx in the ASK1-SEK1-JNK1-HIPK1 Signal Transduction Pathway Is Mediated during Glucose Deprivation-It is well known that activated JNK can phosphoryl-  (23). Thus, we examined whether JNK1 directly phosphorylates Daxx. Fig. 8A shows that activated JNK1 directly phosphorylated c-Jun but not Daxx. These results suggest that activated JNK1 is indirectly involved in Daxx phosphorylation. Recent studies show that homeodomain-interacting protein kinase (HIPK1) physically interacts and directly phosphorylates Daxx (24). We hypothesized that JNK1 activates HIPK1, which then consequently phosphorylates Daxx. An immune complex kinase assay indeed demonstrated that activated JNK1 directly phosphorylated HIPK1 (Fig. 8B) and consequently phosphorylated Daxx (Fig. 8C). These results suggest that glucose deprivationinduced Daxx phosphorylation is mediated through the JNK1-HIPK1 signal transduction pathway.
Role of Serine Residue in the Relocalization of Daxx during Glucose Deprivation-Ecsedy et al. (24) reported that HIPK1 phosphorylates murine Daxx on Ser 669 , and this amino acid residue plays an important role in the relocalization of Daxx. Unlike murine Daxx, human Daxx contains two serine residues in positions 667 and 670. Based on previous results, we postulated that either serine residue of Daxx plays a role in its relocalization during glucose deprivation. To identify the serine residue that plays an important role in the relocalization of Daxx, we employed site-directed mutagenesis techniques to create one point mutant at two serine residues (Ser 3 Ala) and evaluated its role in the relocalization of Daxx during glucose deprivation. Fig. 9 shows that S667A mutant type Daxx, but not S670A mutant type Daxx, did not relocalize to the cytoplasm during glucose deprivation.
Effect of JBD Overexpression on ASK1 Oligomerization during Glucose Deprivation-We further investigated the effect of JBD overexpression on ASK1 oligomerization. Fig. 10 shows that glucose deprivation-induced ASK1 oligomerization was inhibited by JBD overexpression (Fig. 10, lane 3

versus lane 5).
In contrast, Daxx deletion mutant (aa 501-625)-induced ASK1 oligomerization was not affected by JBD overexpression, regardless of whether glucose was present or absent from the medium. These results suggest that Daxx relocalization is essential for the ASK1 oligomerization during glucose deprivation.
Model for the Role of the ASK1-MAPK-MEK Signal Transduction in Daxx Trafficking during Glucose Deprivation- Fig.  11 shows a schematic diagram of a theoretical model based on the literature and data presented here. According to the model, glucose deprivation elevates the intracellular level of reactive oxygen species, in particular H 2 O 2 . Reactive oxygen species activate the ASK1-SEK1-JNK1-HIPK1 signaling pathway, which subsequently signals the relocalization of Daxx from the nucleus to the cytoplasm. The relocalization of Daxx may require its phosphorylation on Ser 667 through activated HIPK1. The cytoplasmic Daxx then binds to ASK1 and leads to ASK1 oligomerization. After 48 h of infection, cells were exposed to glucose-free medium for various times (10 -120 min). Cell lysates containing equal amounts of protein (20 g) were separated by SDS-PAGE and immunoblotted with anti-active JNK antibody, anti-FLAG antibody, or antiactin antibody. Actin was used to confirm the amount of protein loaded in each lane. WB, Western blot.

DISCUSSION
The goal of our studies was to examine whether ASK-SEK1-JNK1 signaling is responsible for Daxx trafficking. Previous studies have shown that Daxx, which contains two nuclear localization signals, is mainly located in the nucleus of an unstressed cell (13). It interacts with nuclear proteins such as centrometric protein, Pax, and promyelocytic leukemia protein (25). Our data show that Daxx relocalizes from the nucleus to the cytoplasm during glucose deprivation and H 2 O 2 treatment (Figs. 3B and 6D). Overexpression of JBD, which inhibits glucose deprivation-induced JNK1 activation, also suppresses the relocalization of Daxx to the cytoplasm (Figs. 6D). Moreover, overexpression of JBD prevents the glucose deprivation-induced association of Daxx with ASK1 (Fig. 6A). These results suggest that glucose deprivation-activated ASK1-SEK1-JNK1 play an important role in the relocalization of Daxx to the were infected with Ad.HA-ASK1 at an MOI of 10, Ad.FLAG-Daxx at an MOI of 10, and Ad.FLAG-JBD/Ad.EGFP at an MOI of 100. After 48 h of incubation, cells were exposed to glucose-free medium for 1 h (A) or cytoplasm and its subsequent interaction with ASK1. Recent studies also demonstrate that Daxx requires ASK1 for its cytoplasmic localization (13). Overexpression of ASK1 stimulates the redistribution of Daxx to the cytoplasm (13). The relocal-ization of Daxx is probably due to an elevated level of JNK1 activity (26). Our data show that Daxx was phosphorylated during glucose deprivation, and its phosphorylation was mediated through HIPK1 activation (Fig. 8). These results are consistent with recent studies that show that HPIK modulates Daxx relocalization and phosophorylation (24). Our data clearly demonstrate that JNK1 activates HIPK1, which then consequently phosphorylates Daxx during glucose deprivation. Our studies also show that Daxx relocalization was inhibited by inhibiting JNK1 activation by JBD overexpression (Figs. 5 and 6D). It is possible that Daxx phosphorylation is associated with Daxx export from the nucleus. Recently, several researchers reported that yeast transcription factor, Yap1, a subfamily of AP-1, which is a sensor of the redox state of the cell, is activated by oxidative stress such as H 2 O 2 (27,28). Activated Yap1 (oxidized form) leads to disulfide bond formation in the C-terminal cysteine-rich region, which contains 3 conserved cysteines and the nuclear export signal. Formation of an intramolecular disulfide linkage leads to a conformational change of Yap1 and consequently conceals the nuclear export signal from the export receptor Crm1p/Xpo1p, resulting in the localization of Yap1p to the nucleus. In contrast to Yap1p in yeast, Daxx in mammalian cells could be exported to the cytoplasm during oxidative stress through an export receptor, FIG. 8. Phosphorylation of Daxx is mediated by JNK1-HIPK1 during glucose deprivation. DU-145 cells were infected with Ad.His-JNK1 at an MOI of 10. After 48 h of infection, cells were exposed to glucose-free medium for 1 h and lysed. Cell lysates were divided into two portions. One portion was immunoprecipitated with mouse anti-His antibody. A, to examine whether the Daxx is a direct substrate of JNK1, 1 g of His-Daxx or GST-c-Jun was incubated with immunoprecipitated His-JNK1 in kinase buffer containing 100 Ci/ml [␥-32 P]ATP at 30°C for 1 h. B, to examine whether the HIPK1 is a direct substrate of JNK1, 0.5 g of GST-HIPK1 or 1 g of GST-c-Jun was incubated with immunoprecipitated His-JNK1 in kinase buffer containing 100 Ci/ml [␥-32 P]ATP at 30°C for 1 h. C, to examine whether the Daxx is a sequential substrate of HIPK1, 0.1 g of GST-HIPK1 was incubated with immunoprecipitated His-JNK1 in kinase buffer containing 100 M ATP at 30°C for 30 min and subsequently with 1 g of His-Daxx with 100 Ci/ml [␥-32 P]ATP at 30°C for an additional 30 min. Phosphorylated proteins were resolved by SDS-PAGE and analyzed by autoradiography (upper panel). The other portion was immunoblotted with antiactive JNK or anti-His antibody (lower panels). WB, Western blot; IP, immunoprecipitation. which is similar to Crm1p/Xpo1p in yeast. We postulate that phosphorylation of Daxx results in conformational changes, exposing the nuclear export signal, which would be recognized by an export receptor. Interaction between the nuclear export signal of Daxx and the export receptor may thus be the critical step in redirecting nuclear Daxx to the cytoplasm. Obviously, this model requires substantiation.
It is well known that the C-terminal 112 amino acids of Daxx (aa 626 -739) are necessary for Fas binding. However, prior to our study, it was not clear which portion of Daxx is responsible for ASK1 binding. Our data demonstrate that the ASK1 binding site of Daxx resides in the region of amino acids 501-625 (Fig. 3A). Interestingly, we observed that this deletion mutant is mainly localized to the cytoplasm (Fig. 3B). Moreover, overexpression of Daxx 501-625 promotes ASK1 oligomerization as well as JNK1 activation, even in the presence of glucose (Figs. 4C and 10). These results are consistent with previous observations that Daxx 501-625 induces JNK activation as well as apoptotic death (12). These results also suggest that the ASK1 binding site of Daxx plays a role in ASK1 oligomerization and that oligomerization of ASK1 is sufficient for activation of the ASK1-SEK1-JNK1 signal transduction pathway. This observation is supported by data from previous studies, which have shown that TRAF2 enhances ASK1 homo-oligomerization and consequently promotes ASK1 activation (11). Although we are far from understanding how Daxx regulates ASK1 oligomerization, we present the possible role of Daxx in the ASK1-SEK1-JNK1-HIPK1 signal transduction pathway. We hypothesize that activation of the ASK1-SEK1-JNK1-HIPK1 signal promotes relocalization of Daxx, which stabilizes ASK1 oliogomerization and maintains the activation of ASK-SEK1-JNK1 signal. If Daxx does not bind to ASK1, activated ASK1 may be quickly inactivated through degradation. We believe that this model will provide a framework for future studies. FIG. 11. A theoretical model for the mechanism by which Daxx is involved in glucose deprivation-induced ASK1-SEK1-JNK1-HIPK1 signal transduction. This model illustrates that Daxx is phosphorylated and relocalized from the nucleus to the cytosol through the ASK1-SEK1-JNK1-HIPK1 signal transduction pathway during glucose deprivation. Daxx binds to ASK1 and may stabilize the form of ASK1 oligomer.