Activation of apoptosis signal-regulating kinase 1 by reactive oxygen species through dephosphorylation at serine 967 and 14-3-3 dissociation.

Oxidative stress has been indicated in a variety of pathological processes such as atherosclerosis, diabetes, and neurodegenerative diseases. Understanding how intracellular signaling pathways respond to oxidative insults such as hydrogen peroxide (H(2)O(2)) would have significant therapeutic implications. Recent genetic studies have placed apoptosis signal-regulating kinase 1 (ASK1) in a pivotal position in transmitting H(2)O(2)-initiated signals. How ASK1 is activated by H(2)O(2), though, remains a subject of intense investigation. Here we report a mechanism by which H(2)O(2) induces ASK1 activation through dynamic control of its phosphorylation at serine 967. We found that treatment of COS7 cells with H(2)O(2) triggers dephosphorylation of Ser-967 through an okadaic acid-sensitive phosphatase, resulting in dissociation of the ASK1.14-3-3 complex with concomitant increase of ASK1 catalytic activity and ASK1-mediated activation of JNK and p38 pathways.

Oxidative stress has been indicated in a variety of pathological processes such as atherosclerosis, diabetes, and neurodegenerative diseases. Understanding how intracellular signaling pathways respond to oxidative insults such as hydrogen peroxide (H 2 O 2 ) would have significant therapeutic implications. Recent genetic studies have placed apoptosis signal-regulating kinase 1 (ASK1) in a pivotal position in transmitting H 2 O 2 -initiated signals. How ASK1 is activated by H 2 O 2 , though, remains a subject of intense investigation. Here we report a mechanism by which H 2 O 2 induces ASK1 activation through dynamic control of its phosphorylation at serine 967. We found that treatment of COS7 cells with H 2 O 2 triggers dephosphorylation of Ser-967 through an okadaic acid-sensitive phosphatase, resulting in dissociation of the ASK1⅐14-3-3 complex with concomitant increase of ASK1 catalytic activity and ASK1-mediated activation of JNK and p38 pathways.
Reactive oxygen species (ROS) 1 produced through a variety of cellular processes or derived from exogenous sources play important roles in the regulation of normal physiology, including cell proliferation, survival, senescence, and apoptotic cell death (1). However, unchecked and excessive production of ROS may result in severe damage to cellular components including DNA, proteins, and lipids, causing oxidative stress or injury. Oxidative stress has been implicated in a variety of human diseases including atherosclerosis, diabetes, arthritis, cancer, and neurodegenerative disorders (1). Thus, understanding how cellular signaling pathways respond to oxidative insults such as hydrogen peroxide (H 2 O 2 ) may lead to novel strategies for therapeutic interventions. Accumulating evidence suggests that apoptosis signal-regulating kinase 1 (ASK1) plays a pivotal role in mediating the H 2 O 2 -induced stress response.
ASK1 is a multifunctional serine/threonine protein kinase involved in the regulation of diverse physiological processes, including cell differentiation and apoptosis (2). It was originally discovered as a mitogen-activated protein kinase kinase kinase with proapoptotic activity (3)(4)(5). The kinase activity of ASK1 is activated by many stress signals and proinflammatory cytokines, including H 2 O 2 , tumor necrosis factor-␣, endoplasmic reticulum stress, serum withdrawal, and chemotherapeutic agents (5)(6)(7)(8)(9)(10). ASK1 stimulation in turn activates both the MKK4/MKK7-JNK pathway and the MKK3/MKK6-p38 kinase pathway, leading to stress responses or apoptosis (3,5). In support of its role in apoptotic signaling, overexpression of ASK1 or a constitutively active fragment, ASK1-KC, can trigger apoptotic cell death in several cell types via a mitochondriadependent caspase 3 pathway (5,11). Furthermore, the expression of a catalytically inactive mutant of ASK1 exhibits a dominant negative effect, inhibiting apoptosis induced by stress signals such as tumor necrosis factor-␣ and H 2 O 2 (5-7). Significantly, it has been established that ASK1 plays a critical role in mediating oxidative stress signaling (6,7,12). Mouse embryonic fibroblast cells from ASK1 Ϫ/Ϫ mice are resistant to oxidant-and tumor necrosis factor-␣-induced apoptosis and fail to maintain sustained levels of JNK and p38 kinase activity upon treatment with H 2 O 2 or tumor necrosis factor-␣ (12). These lines of evidence suggest that ASK1 is a key player in apoptotic signaling, and in particular, ASK1 may function as a pivotal mediator of H 2 O 2 -induced stress response. However, the molecular mechanism by which ASK1 transmits H 2 O 2 signals remains a subject of intense investigation.
Because of its important role in cell death signaling, the activity of ASK1 is tightly regulated by multiple mechanisms, including phosphorylation, oligomerization, and protein-protein interactions. Phosphorylation of ASK1 at Ser-845 appears to be required for activity because dephosphorylation at this site by PP5 leads to ASK1 inactivation (13). Phosphorylation at Ser-83 by Akt/protein kinase B, on the other hand, attenuates ASK1 activity (14). It has also been demonstrated that intramolecular interaction, probably between the NH 2 -terminal and COOH-terminal domains of ASK1, may be required to maintain ASK1 in its inactive state (15), whereas oligomerization of its COOH-terminal domains is correlated with ASK1 activation (6,13). The most commonly observed means of ASK1 regulation, however, is through protein-protein interactions. Numerous proteins have been shown to bind ASK1 to exert their regulatory function. For example, binding of TRAF2 (16) or Daxx (15) promotes ASK1 function, whereas the kinase and proapoptotic activities of ASK1 are inhibited by many other associated proteins, including reduced thioredoxin (7), glutaredoxin (17), Cdc25A (18), Hsp 72 (19), ASK1-interacting protein 1 (20), and 14-3-3 proteins (21). It seems probable that ASK1 is suppressed from promoting apoptosis under normal survival or proliferative conditions by specific phosphorylation and regu-lated protein-protein interactions but becomes proapoptotic when suppression is relieved. One mechanism that mediates the H 2 O 2 effect appears to involve redox-regulated dissociation of ASK1 from bound thioredoxin, leading to ASK1 oligomerization and activation (6,7,22). The inhibitory effect of bound thioredoxin on ASK1, however, may be redox-independent, as shown in an endothelial cell system (23).
Among ASK1 regulatory events, the interaction of ASK1 with 14-3-3 proteins is particularly intriguing because this complex formation is controlled by phosphorylation at Ser-967 (21). 14-3-3 belongs to a family of phosphoserine/phosphothreonine-binding molecules that recognize the motif Arg-Ser-X-pSer/pThr-X-Pro or its derivatives (pSer/pThr represents phosphorylated Ser or Thr, and X denotes any amino acid) (24 -27). They can bind to many signaling molecules through specific pSer/pThr motifs, including the proapoptotic protein Bad (28) and transcription factor FKLR1 (29). In fact, the control of 14-3-3/Bad binding by Akt/protein kinase B represents the first molecular link by which a major survival signaling pathway is coupled to the death machinery (30). Functionally, the 14-3-3⅐Bad complex formation results in inhibition of Bad-mediated apoptosis (28,31,32). Through such phosphorylation-dependent protein-protein interactions, 14-3-3 plays an important role in the regulation of many cellular processes including cell proliferation and survival signaling. We have demonstrated previously that 14-3-3 binding to ASK1 through phosphorylated Ser-967 effectively suppresses ASK1-induced apoptosis (21). However, how upstream signals control Ser-967 phosphorylation and ASK1⅐14-3-3 association remains elusive. Stress signals may utilize the sensitive kinase and phosphatase signaling network to control ASK1 function, achieving a balanced biological outcome such as cell suicide or survival.
In this study, we have examined the dynamic regulation of ASK1 by H 2 O 2 and uncovered a critical link between H 2 O 2 signaling and the phosphorylation status of Ser-967. Our findings define a novel mechanism whereby H 2 O 2 triggers the activity of an okadaic acid-sensitive phosphatase upstream of ASK1, resulting in dephosphorylation of Ser-967 and subsequent activation of ASK1-mediated stress pathways.

MATERIALS AND METHODS
Plasmids, Cell Culture, and DNA Transfection-Expression vectors for ASK1 and its mutants have been described (5,21). COS7 cells were cultured in Dulbecco's modified Eagle's medium with fetal bovine serum (10%; Mediatech, Washington, D. C.). Transfection was performed with FuGENE 6 (Roche Applied Science) according to the manufacturer's instructions.
Reagents-Hydrogen peroxide, N-acetyl-L-cysteine, catalase, okadaic acid, calyculin A, and cyclosporin A were obtained from Sigma. Calyculin and cyclosporin A were diluted in ethanol, and okadaic acid was prepared in methanol. All other reagents were diluted in water. Calf intestinal phosphatase was purchased from New England Biolabs.
Generation of Phospho-specific pSer-967 Antibodies-Polyclonal antibodies were produced by immunizing two rabbits with a synthetic phospho-peptide (keyhole limpet hemocyanin-coupled) corresponding to residues around Ser-967 of human ASK1. From each rabbit, antibodies were purified by protein A and a nonphospho-peptide affinity chromatography (antibodies 2 and 4). The flow-through of non-phosphopeptide chromatography, in which the non-phospho activity was depleted, was collected and applied to a phosphopeptide chromatography for final purification (antibodies 1 and 3).

RESULTS
Phosphorylation of ASK1 at Ser-967 in Vivo-We have shown previously that Ser-967 of ASK1 is harbored in a critical recognition motif for the pSer/pThr-binding protein 14-3-3 (21). Phosphorylation of this residue enables ASK1 to bind 14-3-3, which leads to suppression of ASK1-induced apoptosis. Thus, Ser-967 of ASK1 may serve as an integration point that couples survival signaling pathways to this death-promoting kinase. To understand how Ser-967 of ASK1 is regulated, we developed antibodies that allow us to monitor the phosphorylation status of this critical site in vivo. Four batches of rabbit antisera were raised against keyhole limpet hemocyanin-conjugated Ser-967containing phosphopeptides from ASK1. Purified antibodies from these sera showed strong cross-reactivity with ASK1/WT expressed in COS7 cells (Fig. 1A). Two appeared to be specific to the phosphorylated Ser-967 (pSer-967) epitope in ASK1 because they were unable to recognize ASK1/S967A, a mutant of ASK1 which is unphosphorylatable at Ser-967. However, it is possible that the inability of the pSer-967 antibodies to recognize ASK1/S967A might be the result of mutation-induced conformational or structural changes of the epitope. To address this concern, we employed an alternative approach to dephosphorylate Ser-967 by treating ASK1/WT with a broad spectrum protein phosphatase, calf intestinal phosphatase. As shown in Fig. 1B, pSer-967 antibodies failed to recognize dephosphorylated ASK1/WT, confirming the specificity of the antibody for phosphorylated Ser-967.
To define the utility of the pSer-967 antibody, we tested its sensitivity. A testament to its usefulness for the monitoring of Ser-967 phosphorylation in vivo, the anti-pSer-967 antibody could recognize pSer-967 when lysate with 0.25 g of total protein was immunoblotted (Fig. 1C). This level of sensitivity is comparable with that of the pan-ASK1 antibody. Furthermore, antibody titration showed that the phospho-antibody can recognize phosphorylated Ser-967 with a 1:20,000 dilution (Fig.  1D). These experiments demonstrate the sensitivity and specificity of the antibody for pSer-967 of ASK1, providing us with a powerful tool for monitoring ASK1 phosphorylation under various physiological conditions.
ROS Induce Dephosphorylation at Ser-967-The above tests also demonstrate that Ser-967 is phosphorylated in cells under growth conditions in the presence of serum (Fig. 1), which is consistent with its role in cell survival signaling. It is possible that the death promoting function of ASK1 is suppressed under growth conditions but becomes proapoptotic when suppression is relieved. To test this hypothesis, we sought to determine whether the phosphorylation status of Ser-967 is responsive to stress signals. It has been well established that ASK1 plays an important role in the apoptotic response induced by ROS, in particular H 2 O 2 (6,7,12). It has been shown that cells derived from ASK1 knock-out mice are resistant to H 2 O 2 -induced apoptosis (12). However, the mechanism by which ASK1 mediates the ROS effect remains unclear. It is possible that ROS induces ASK1 activation and cell death through dephosphorylation of Ser-967. To test this model, we monitored the in vivo phosphorylation status of Ser-967 in COS7 cells. Indeed, treatment of cells with a high (5 mM), intermediate (1 mM), or low (0.5 mM) dose of H 2 O 2 resulted in marked dephosphorylation of Ser-967 (Fig. 2, A and C). In support of a role of ROS in inducing Ser-967 dephosphorylation, pretreatment of cells with N-acetyl-L-cysteine, a ROS scavenger, effectively blocked the H 2 O 2 effect (Fig. 2B).
To explore whether phosphorylation of Ser-967 is a regulated process responsive to varying conditions in the cell, we tested the effects of H 2 O 2 using a prolonged time course that allowed for the metabolic clearance of ROS. COS7 cells expressing ASK1 were treated with a low (0.5 mM) or intermediate (1 mM) dose of H 2 O 2 for various times. We show that the dephosphorylation induced by ROS is reversible, with marked Ser-967 dephosphorylation after 30 min followed by increased phosphorylation as early as 60 min post-treatment (Fig. 2C). This reversibility suggests that phosphorylation at Ser-967 is a dynamic process, presumably regulated by a kinase(s) and phosphatase(s), and could be a major mode of ASK1 regulation.
Ser-967 Dephosphorylation Is Associated with Increased ASK1 Catalytic Activity-We demonstrated previously that disruption of the ASK1/14-3-3 interaction dramatically accelerated ASK1-induced cell death (21). The mechanism by which this increased cell death occurred, however, remains unknown.
Because the catalytic activity of ASK1 is required for its death promoting function, it is possible that removing 14-3-3 from ASK1 may relieve an inhibitory mechanism, leading to ASK1 activation and the subsequent activation of the JNK and p38 kinases. Thus, having established that Ser-967 dephosphorylation is correlated with dissociation of 14-3-3, we examined whether this H 2 O 2 -induced dephosphorylation affects the kinase activity of ASK1. COS7 cells were transfected with HA-ASK1/WT or HA-ASK1/K709R, a kinase-inactive mutant, and treated with H 2 O 2 for various times. Lysates from treated cells were probed for the phosphorylation status of ASK1 at Ser-967 and examined for the kinase activity of ASK1 as well as its downstream kinases, JNK and p38. Again, treatment of COS7 cells with H 2 O 2 leads to decreased phosphorylation levels at Ser-967 of ASK1 (Fig. 4, A and B). Significantly, the dephosphorylation of Ser-967 was accompanied by increased autokinase activity of ASK1. This is further accompanied by the increased activity of the JNK and p38 kinases, downstream effectors of ASK1. As a control, ASK1/K709R showed only basal level of activity. These results clearly suggest an inverse relationship between the phosphorylation status of ASK1 at Ser-967 and the kinase activity of ASK1 (Fig. 4B). Dephosphorylation of Ser-967 and 14-3-3 dissociation induced by H 2 O 2 may directly impact the catalytic activity of ASK1. In support of this notion, S967A, the unphosphorylatable mutant of ASK1, which cannot bind 14-3-3, exhibited both enhanced autokinase activity and increased transkinase activity (Fig. 4, C and D). Therefore, H 2 O 2 may induce activation of ASK1 and its downstream effectors at least in part through dephosphorylation of Ser-967 and dissociation of 14-3-3 proteins.

Ser-967 Is a Major Site for Mediating H 2 O 2 -induced ASK1
Activation-Two major redox-sensitive proteins in the cell, thioredoxin and glutaredoxin, have been shown to bind directly to ASK1 and inhibit its kinase activity (7,17). To determine the contribution of Ser-967 to H 2 O 2 -induced ASK1 activation, we compared the responses of ASK1/WT with ASK1/S967A, a phosphorylation-and 14-3-3 binding-defective mutant, to H 2 O 2 treatment. Consistent with previous reports, treatment of COS7 cells with H 2 O 2 induced drastic increases in ASK1 kinase activity (Fig. 4, E and F). Interestingly, ASK1/S967A showed only slight increases in kinase activity with increasing time of H 2 O 2 treatment, a sharp contrast to the increasing activity observed with ASK1/WT (Fig. 4, E and F). These data suggest that phosphorylation of Ser-967 and possibly 14-3-3 binding play a critical role in H 2 O 2 -induced activation of the ASK1 pathway.
An Okadaic Acid-sensitive Phosphatase Is Implicated in Dephosphorylation at Ser-967 of ASK1-It appears that phosphorylation of Ser-967 is responsive to ROS signals. However, the signaling pathways that control ASK1 function through Ser-967 remain unresolved. It is conceivable that H 2 O 2 may activate a protein Ser/Thr phosphatase that specifically dephosphorylates Ser-967, leading to ASK1 activation. To test this possibility, we took a pharmacological approach by using phosphatase inhibitors to inhibit major classes of serine-threonine phosphatases in eukaryotic cells, protein phosphatase 1 (PP1), protein phosphatase 2A (PP2A), and protein phosphatase 2B (PP2B)/calcineurin (34). To determine the phosphatases involved, we used okadaic acid and calyculin A for inhibition of PP1 and PP2A and cyclosporin A for PP2B and examined the effect of these inhibitors on H 2 O 2 -induced dephosphorylation of ASK1. Treatment of COS7 cells with okadaic acid or calyculin caused hyperphosphorylation of ASK1 at Ser-967 in the absence of H 2 O 2 . Pretreatment of cells with these inhibitors prevented Ser-967 dephosphorylation by H 2 O 2 (Fig. 5, A and B). The overall levels of ASK1 were not changed upon these treat-

FIG. 4. Ser-967 dephosphorylation is correlated with increased ASK1 catalytic activity.
A, H 2 O 2 -induced Ser-967 dephosphorylation is associated with increased ASK1 activity. COS7 cells (4 ϫ 10 5 ) transfected with ASK1/WT or ASK1/K709R were treated with 1 mM H 2 O 2 and 25 mM aminotriazole for various times. The degree of Ser-967 phosphorylation in the lysates was determined by immunoblotting (WB) with an anti-pSer-967 (pS967) antibody. At each time point, HA-ASK1 was immunoprecipitated (IP) with an anti-ASK1 monoclonal antibody (Santa Cruz Biotechnology). The kinase activity of ASK1 (IVK) was determined by its autophosphorylation, by its ability to phosphorylate p38 in a coupled MKK6-p38 assay, and by its activation of JNK as measured by the immunoprecipitated JNK to phosphorylate c-Jun. The reactions were stopped by adding SDS sample buffer and boiling for 5 min. After resolution on SDS-PAGE (12.5%), phosphorylation of ASK1, c-Jun, or p38 was visualized by a Typhoon PhosphorImager. Immunoprecipitated ASK1 and JNK levels were determined by immunoblotting with anti-ASK1 or anti-JNK antibodies, respectively. B, quantification of data presented in A by ImageQuant. The quantification of phosphorylated Ser-967 is normalized to the 60-min treatment sample (solid line). The kinase activity for each sample is normalized to the 0-min sample for each respective assay. C, ASK1/S967A exhibits higher kinase activity than ASK1/WT. Autokinase activity of ASK1 and its activation of the JNK pathway were measured as described in A. D, quantification of results presented in C, normalized to the WT activity. E, ASK1/S967A shows diminished response to H 2 O 2 treatment. Autokinase activity of ASK1/WT, ASK1/S967A, and ASK1/K709R was measured upon treatment of COS7 cells with H 2 O 2 as in A. The amount of ASK1 proteins used was determined by immunoblotting. F, quantification of results presented in E. The activity is normalized to the 0-min treatment of the K709R sample. All results presented are representative of at least three independent experiments. ments. On the other hand, cyclosporin A pretreatment had no effect on H 2 O 2 -induced Ser-967 dephosphorylation and did not cause Ser-967 hyperphosphorylation in the absence of H 2 O 2 (Fig. 5A), suggesting that PP2B is unlikely involved in the regulation of Ser-967 dephosphorylation. These results indicate that phosphorylation at Ser-967 of ASK1 is a dynamic process regulated in part by a PP1/PP2A-like phosphatase.
Having shown that dephosphorylation at Ser-967 is correlated with increased ASK1 kinase activity, we reasoned that pretreatment with okadaic acid would suppress H 2 O 2 -induced activation of ASK1 if an okadaic acid-sensitive phosphatase were involved in this process. To test this premise, COS7 cells expressing ASK1 were pretreated with okadaic acid or the vehicle control, MeOH, for 2 h before adding H 2 O 2 for 60 min. Cell lysates were prepared for determination of Ser-967 phosphorylation status, ASK1 kinase activity, and the activity of the ASK1 effectors JNK and p38. In support of our hypothesis, pretreatment of cells with okadaic acid drastically blocked H 2 O 2 -induced ASK1 activation as well as H 2 O 2 -stimulated JNK and p38 activity (Fig. 5, B and C). Together, these data suggest that an okadaic acid-sensitive phosphatase(s) may be responsible for H 2 O 2 -induced dephosphorylation of Ser-967 and the subsequent activation of ASK1-mediated pathways. DISCUSSION ASK1 plays an essential role in mediating H 2 O 2 -induced activation of the JNK/p38 pathways and subsequent apoptosis (2,5,6). Here we reveal a novel mechanism by which H 2 O 2 activates the ASK1-JNK/p38 pathways through controlling the phosphorylation status of ASK1 at Ser-967. Specifically, H 2 O 2 action appears to require the stimulation of an okadaic acidsensitive phosphatase(s) that catalyzes the dephosphorylation of ASK1 at Ser-967, which leads to dissociation of an ASK1 inhibitor, 14-3-3, and activation of ASK1 and its downstream effector kinases. We found that (i) treatment of cells with H 2 O 2 triggers dephosphorylation of ASK1 at Ser-967 ( Figs. 1 and 2); (ii) dephosphorylation of Ser-967 is correlated with 14-3-3 dissociation from ASK1 (Fig. 3); (iii) in support of a redox-dependent effect, the addition of an ROS scavenger, N-acetyl-L-cysteine, retains Ser-967 phosphorylation in the presence of H 2 O 2 (Fig. 2); (iv) a single mutation in ASK1 which abolishes the Ser-967 phosphorylation site, S967A, renders ASK1 largely resistant to H 2 O 2 treatment (Fig. 4), supporting a critical role for Ser-967 phosphorylation/14-3-3 binding in mediating the H 2 O 2 effect; (v) functionally, dephosphorylation of Ser-967 is correlated with the increased catalytic activity of ASK1 (Fig. 4); (vi) significantly, inhibitors of protein phosphatase PP1/PP2A families block H 2 O 2 -induced dephosphorylation of Ser-967 and activation of ASK1 and the JNK/p38 pathways (Fig. 5). These findings establish a mechanistic link between the ROS signaling and a kinase/phosphatase network that converges on Ser-967 of ASK1.
Elevated ROS in response to diverse stress signals may serve as a second messenger to control a broad range of physiological and pathological processes, including cell proliferation, inflammation, and apoptosis. ASK1 has emerged as a major player in mediating ROS signaling. In support of this notion, H 2 O 2induced sustained activation of JNK and p38 and consequent apoptotic cell death is drastically impaired in ASK1 Ϫ/Ϫ mouse embryonic fibroblasts (12). However, how ROS signal through ASK1 remains to be elucidated. It is well established that ASK1 forms a complex with reduced thioredoxin, which may maintain an inactive or quiescent state of ASK1 (7). This complex is disrupted upon ROS treatment, resulting in ASK1 activation. It has recently been shown that the COOH-terminal sequence of ASK1 binds glutaredoxin in a redox-dependent manner and that this binding inhibits kinase activity and ASK1-mediated apoptosis (17). These results suggest an important role for redox-sensitive proteins in controlling ROS-induced activation of ASK1. The mechanisms by which dissociation of these proteins induces apoptosis, however, remain to be determined. Our work may provide such a mechanism. We propose that ROS-induced dissociation of thioredoxin may ini- tiate a conformational change in ASK1, exposing phosphorylated Ser-967 to a bound Ser/Thr phosphatase. This Ser-967 phosphatase dephosphorylates the site, preventing reassociation of 14-3-3. It is conceivable that the induced conformational change of ASK1 may also decrease its affinity to 14-3-3, thus facilitating dephosphorylation of Ser-967. 14-3-3 seems to protect pSer-967 from dephosphorylation because overexpression of 14-3-3 delays the H 2 O 2 -induced dephosphorylation of Ser-967 on ASK1 (Fig. 3). In support of our model, a conformational change of ASK1 has been documented in cells treated with H 2 O 2 (22). This model places the disruption of the ASK1⅐14-3-3 complex downstream of thioredoxin dissociation in the signaling events leading to ASK1 activation. Indeed, the unphosphorylatable mutant of ASK1, S967A, in interesting contrast to WT, exhibits only modest increases in kinase activity upon exposure to H 2 O 2 at a concentration that has been shown to induce dissociation of thioredoxin (Fig. 4). These data suggest that dephosphorylation of ASK1 at Ser-967 and release of 14-3-3 from ASK1 may be a critical downstream regulatory mechanism for thioredoxin to mediate the H 2 O 2 effect. Alternatively, phosphorylation of Ser-967 and the 14-3-3 binding may be mechanistically linked to the binding of ASK1 to redox-sensitive proteins, thioredoxin and glutaredoxin. Elimination of the phosphorylation state of ASK1 at Ser-967 may diminish the binding of thioredoxin or glutaredoxin to ASK1, thus decreasing the responsiveness to ROS insult. We also have not ruled out a Ser-967-independent mechanism for the H 2 O 2 -induced ASK1 activation. The modest increase in the kinase activity of ASK1/S967A in response to H 2 O 2 may suggest the involvement of other redox-sensitive events independent of the Ser-967, which could be mediated by thioredoxin or glutaredoxin (Fig.  4E). How 14-3-3, thioredoxin, and glutaredoxin coordinate the control of ASK1 in response to H 2 O 2 requires further investigation. Here we have established that controlling the phosphorylation status of ASK1 at Ser-967, and thus 14-3-3 association, is a major mechanism for H 2 O 2 to activate ASK1.
The above model envisions the removal of 14-3-3 as a pivotal mechanism for the activation of ASK1 upon exposure of cells to ROS. Therefore, it is important to note that dephosphorylation of Ser-967 on ASK1 and 14-3-3 dissociation is functionally correlated with increased kinase activity of ASK1 and its downstream effectors, JNK and p38 (Fig. 4). Consistent with these results, the S967A mutant of ASK1, which is deficient in 14-3-3 binding, exhibits enhanced catalytic activity and enhanced activation of JNK and p38 kinases (Fig. 4). These data support an inhibitory role of 14-3-3 in ASK1 activation. Functionally, 14-3-3 binding has been shown to suppress ASK1-induced apoptosis (21). Expression of a dominant negative mutant of 14-3-3, DN-14-3-3 (R56A/R60A), induces an increase in ASK1 activity in cells as well as increased apoptotic response in a transgenic animal model (35). It is possible that through binding to ASK1, 14-3-3 effectively suppresses ASK1 function, which could be accomplished by several nonmutually exclusive mechanisms (26): (i) 14-3-3 may act as an ASK1 inhibitory cofactor, directly suppressing the catalytic function of ASK1; (ii) 14-3-3 binding may maintain an inactive conformation of ASK1 under nonstressed conditions, as with Raf-1 (36); (iii) 14-3-3 may determine the subcellular localization of ASK1, leading to its sequestration, a well documented property of 14-3-3 (37); (iv) 14-3-3 may interfere with the activation process of ASK1, such as its ability to homo-oligomerize, an effect seen with other ASK1binding proteins, such as Hsp72 and glutathione S-transferase Mu (19) (38); (v) 14-3-3 association may cause steric hindrance effect, blocking the interaction of ASK1 with its activators or with its substrates. Thus, disruption of the ASK1/14-3-3 interaction may release a negative regulatory mechanism, leading to ASK1 activation. Our work demonstrates that the ROS signaling pathway targets the ASK1/14-3-3 interaction to activate ASK1. The ASK1/14-3-3 axis may function as a molecular controller that funnels ROS signals to the JNK and p38 pathways to induce subsequent stress responses.
The phosphorylation-dependent nature of the ASK1/14-3-3 interaction subjects this complex formation to dynamic control by an upstream kinase(s) and phosphatase(s). Oxidative stress signals such as H 2 O 2 may utilize a Ser-967-specific phosphatase or kinase to control ASK1 function to induce a cellular response. It is likely that H 2 O 2 acts through a phosphatase that binds to ASK1. Three protein phosphatases have been found to associate with ASK1, Cdc25A (18), stress-activated protein kinase pathway-regulating phosphatase 1 (SKRP1) (39), and PP5 (13). However, none of these is likely to be the Ser-967 phosphatase. Cdc25A binding interferes with ASK1 oligomerization, and PP5 dephosphorylates phosphorylated Thr-845 in the activation loop of ASK1 and inhibits H 2 O 2 -induced ASK1 activation. SKRP1 appears to act as a scaffold protein to promote the association of ASK1 with its substrate MKK7, but suppresses JNK signaling (39). None of these phosphatases causes ASK1 activation in response to H 2 O 2 , an expected effect for a specific Ser-967 phosphatase. On the other hand, our results are consistent with the model that H 2 O 2 may target an okadaic acid/calyculin A-sensitive phosphatase. Pretreatment of cells with inhibitors of the PP1 and PP2A family phosphatases, okadaic acid or calyculin A, abolishes the H 2 O 2 -induced dephosphorylation of Ser-967, whereas a PP2B inhibitor shows no effect. Importantly, okadaic acid also blocks H 2 O 2 -induced FIG. 6. A working model for the H 2 O 2 -mediated activation of ASK1. ASK1 is dynamically regulated by reversible phosphorylation at Ser-967. In the presence of survival factors, a kinase is active which phosphorylates ASK1 at Ser-967, enabling formation of the ASK1⅐14-3-3 complex, thus inactivating ASK1. The inactive state of ASK1 is also maintained by its binding to other inhibitors, including reduced thioredoxin. Cellular stresses such as H 2 O 2 , however, induce dissociation of thioredoxin and activate a phosphatase and/or suppress a Ser-967 kinase. This results in dephosphorylation of ASK1 at Ser-967, dissociation of the ASK1⅐14-3-3 complex, and activation of ASK1 and its downstream signal transduction pathways, leading to stress responses and apoptosis.
activation of ASK1 and its downstream JNK and p38 kinase pathways. Because of the difficult nature of distinguishing the specific contributions of PP1 and PP2A using these inhibitors in cells, the exact phosphatase involved requires further investigation. Taken together, these results indicate that H 2 O 2 induces the activation of the ASK1-mediated stress response in part by targeting the pSer-967-dependent ASK1/ 14-3-3 interaction through activation of a PP1-or PP2A-like phosphatase(s).
Our results suggest a model whereby the catalytic activity of ASK1 is controlled by reversible phosphorylation at Ser-967 (Fig. 6). Integrated actions of a kinase and phosphatase are likely to regulate Ser-967 phosphorylation and 14-3-3 association. In nonstressed cells, ASK1 is highly phosphorylated by a Ser-967-specific kinase and bound to 14-3-3. Upon stimulation by oxidative stresses, a Ser-967 phosphatase is activated either in a thioredoxin-dependent or independent manner, resulting in dephosphorylation of ASK1, dissociation of 14-3-3, and stimulation of ASK1 catalytic activity. It is also possible that H 2 O 2 signaling plays a role in inhibiting the Ser-967 kinase. The Ser-967 phosphatase is sensitive to okadaic acid and calyculin A, implicating the involvement of a PP1/PP2A-like phosphatase. Identification of the Ser-967 phosphatase and kinase will lead to identification of the signal transduction pathways that control ASK1 activity in response to survival and death signals.
ROS have been implicated in more than 50 pathogenic conditions and participate as fundamental components of tissue injury in many human diseases (1 and references therein). The downstream target of H 2 O 2 , ASK1, has also been associated with a number of disease states, including human immunodeficiency virus infection (40), cardiac hypertrophy (41), and poly(Q) expansion-related neurodegenerative disorders such as Huntington's disease (8). Thus, understanding how ASK1 is regulated by ROS may have significant implications for the development of novel therapeutics, and the ASK1/14-3-3 interaction may provide a molecular target for pharmacological interventions.