Osmotic Stress Activates the TAK1-JNK Pathway While Blocking TAK1-mediated NF-κB Activation

Osmotic stress activates MAPKs, including JNK and p38, which play important roles in cellular stress responses. Transforming growth factor-β-activated kinase 1 (TAK1) is a member of the MAPK kinase kinase (MAPKKK) family and can activate JNK and p38. TAK1 can also activate IκB kinase (IKK) that leads to degradation of IκB and subsequent NF-κB activation. We found that TAK1 is essential for osmotic stress-induced activation of JNK but is not an exclusive mediator of p38 activation. Furthermore, we found that although TAK1 was highly activated upon osmotic stress, it could not induce degradation of IκB or activation of NF-κB. These results suggest that TAK1 activity is somehow modulated to function specifically in osmotic stress signaling, leading to the activation of JNK but not of IKK. To elucidate the mechanism underlying this modulation, we screened for potential TAK1-binding proteins. We found that TAO2 (thousand-and-one amino acid kinase 2) associates with TAK1 and can inhibit TAK1-mediated activation of NF-κB but not of JNK. We observed that TAO2 can interfere with the interaction between TAK1 and IKK and thus may regulate TAK1 function. TAK1 is activated by many distinct stimuli, including cytokines and stresses, and regulation by TAO2 may be important to activate specific intracellular signaling pathways that are unique to osmotic stress.

MAPK 2 cascades play important roles in many extracellular stimuli-activated intracellular signaling pathways (1)(2)(3). These stimuli include not only growth factors and cytokines but also chemical and physical stresses such as ultraviolet light irradiation and osmotic stress. Among the three well characterized MAPK subfamilies, extracellular signal-regulated kinase (ERK), JNK, and p38 MAPK, JNK and p38 are the major regulators of stress-induced cellular responses. In mammalian cell cultures, high osmolarity generated by adding salt or sugar into the culture medium activates JNK and p38, which in turn up-regulates a number of genes involved in cellular stress responses, such as production of osmolytes in renal cells (4). In addition, the JNK and p38 pathways promote cell death, which is important for the elimination of dysfunctional cells (5)(6)(7). JNK and p38 are activated by the MAPK kinases (MAPKKs) MKK4/MKK7 and MKK3/MKK6, respectively. The MAPKKs are normally activated by members of the MAPKK kinase (MAPKKK) family that includes MEKKs, mixed lineage kinases (MLKs), apoptosis-stimulating kinase 1 (ASK1), and others in response to engagement of cell surface receptors or stresses. Although the MAPKKKs are believed to confer signal and stimuli specificity, the particular MAPKKKs that are essential to each chemical and physical stress have not yet been well defined.
Transforming growth factor-activated kinase 1 (TAK1) is a member of the MAPKKK family and plays an essential role in tumor necrosis factor (TNF), interleukin 1 (IL-1), and Toll-like receptor (TLR) signaling pathways (8 -13). TAK1 functions as a MAPKKK in these pathways and is indispensable for the activation of JNK and p38. Furthermore, TAK1 is also essential for TNF-, IL-1-, and TLR ligand-induced activation of IB kinase (IKK), which leads to the subsequent degradation of IB and activation of transcription factor NF-B. TAK1 Ϫ/Ϫ or TAK1⌬/⌬ (expressing a kinase-dead TAK1) ( Fig. 2A) cells do not activate JNK, p38, or NF-B upon treatment with TNF, IL-1, or TLR ligands (9,10). Several earlier studies have shown that TAK1 can be activated by chemical and physical stresses, including osmotic stress (14 -16). However, the role of TAK1 in stress signaling has not yet been defined.
In this study, we demonstrate the essential role of TAK1 in osmotic stress signaling using TAK1⌬/⌬ cells. We found that TAK1 is essential for osmotic stress-induced activation of JNK but not of p38. We also found that although TAK1 is highly activated upon osmotic stress treatment, this did not result in IB degradation or subsequent activation of NF-B. These results suggest that TAK1 is regulated to activate only the MAPK cascades and not the NF-B pathway in osmotic stress signaling. We attempted to elucidate the mechanism underlying this regulation of TAK1. We demonstrate here that a TAK1-binding kinase, TAO2 (thousand-and-one amino acid kinase 2) blocks the TAK1-mediated NF-B pathway by preventing the interaction of TAK1 with IKK, yet did not inhibit TAK1-induced activation of JNK. These results suggest that binding by TAO2 may modulate the specific cellular responses of TAK1.
In Vitro Activation-In vitro activation was performed as described previously (24,25). Briefly, 293 cells were washed twice with 1ϫ phosphate-buffered saline and resuspended in hypotonic buffer (20 mM HEPES-KOH, 10 mM KCl, 1 mM MgCl 2 , 1 mM EDTA, 1 mM EGTA, 1 mM DTT (pH 7.5)) supplemented with protease inhibitors (10 mM NaF, 1 mM phenylmethylsulfonyl fluoride, and 20 M aprotinin). Resuspended cells were lysed by passing through a 22-gauge needle 10 times. Cellular debris was removed by centrifugation at 10,000 ϫ g for 20 min at 4°C followed by filtration at 0.45 m. Clear cell lysates were collected and incubated at 37°C for the indicated times or left untreated on ice. After incubation, an equivalent volume of hypotonic buffer containing 0.1% Nonidet P-40 and 300 mM NaCl was added.
NF-B Reporter Assay-293 cells (1.6 ϫ 10 5 cells/well) were seeded onto 6-well (35-mm) dishes 24 h before transfection. Cells in each well were transfected with 0.1 g of an Ig-B-luciferase reporter plasmid, 0.1 g of a control reporter plasmid pAct-␤-Gal (␤-galactosidase under the control of the ␤-actin promoter), and various expression constructs by the calcium phosphate precipitation method. The total amounts of DNA in each well were kept constant by supplementing with empty vector. After 36 h, cells were treated with or without TNF (20 ng/ml) for 12 h prior to harvesting using reporter lysis buffer (Promega), and luciferase activity was assessed via luminometry. Transfection efficiencies were normalized by ␤-galactosidase activities from pAct-␤-Gal. The NF-B luciferase activity was expressed as the mean of the normalized luciferase activity from duplicated experiments.
In Vitro Kinase Assays-TAK1 was immunoprecipitated, and the immunoprecipitates were incubated with 5 Ci of [␥-32 P]ATP (3,000 Ci/mmol) and 1 g of bacterially expressed His-MKK6 in 10 l of kinase buffer containing 10 mM HEPES (pH 7.4), 1 mM DTT, 5 mM MgCl 2 at 25°C for 2 min. IKK was immunoprecipitated, and the immunoprecipitates were incubated with 5 Ci of [␥-32 P]ATP (3,000 Ci/mmol) and 1 g of bacterially expressed GST-IB in 10 ml of kinase buffer containing 10 mM HEPES (pH 7.4), 1 mM DTT, 5 mM MgCl 2 , and 100 M ATP at 30°C for 30 min. Samples were then separated by 10% SDS-PAGE and visualized by autoradiography.
siRNAs-siRNAs were purchased from Ambion. siRNA used as control was Silencer negative control number 1 siRNA. Human TAO2 siRNAs used were pre-designed siRNAs 118287 (TAO2 siRNA 1) and 118285 (TAO2 siRNA 2) (Ambion). 293 cells were plated onto 6-well plates and transfected with siR-NAs (30 -50 nmol) using siPORT amine transfection agent (Ambion). Cells were incubated for 48 h post-transfection and were stimulated with 0.5 M NaCl. The reduction of TAO2 expression was monitored by real time PCR using human TAO2-specific primer (Qiagen).

RESULTS
We have demonstrated previously that TAK1 activates JNK, p38, and NF-B in the TNF and IL-1 signaling pathways (8,9,26). Ablation of TAK1 markedly impairs TNF-and IL-1-induced activation of JNK, p38, and NF-B. TAK1 is activated not only by proinflammatory cytokines but also in response to physical and chemical stresses, including osmotic stress. We found that among several physical and chemical stresses, including oxidative stress, arsenic, and ultraviolet light irradiation, osmotic stress is the most potent activator of TAK1 in human embryonic kidney 293 and mouse embryonic fibroblasts (MEFs) (data not shown). To begin to define the role of TAK1 in osmotic stress signaling, we examined activation of TAK1, JNK, p38, and NF-B following treatment by IL-1 or osmotic stress. 293 IL-1RI cells that stably express the IL-1 receptor were treated with IL-1 or high osmolarity. The activation of TAK1 was monitored by in vitro kinase assay using MKK6 as an exogenous substrate (Fig. 1A). TAK1 was activated by osmotic stress at higher levels relative to activation by IL-1. Activation of TAK1 can be also assessed by the appearance of phosphorylated forms of TAK1 and its binding partner TAK1-binding protein 1 (TAB1), resulting in slower migration on SDS-PAGE (27) (Fig. 1A, 2nd panel). Phosphorylation of TAK1 and TAB1 was also observed upon osmotic stress. JNK and p38 were activated following TAK1 activation at higher levels in cells subjected to osmotic stress compared with IL-1-treated cells (Fig. 1A, 3rd and 4th panels). Activa-tionofNF-Bwasmonitoredbyphosphorylation and degradation of IB␣ (Fig. 1A, 5th and bottom panels). In contrast to JNK and p38, only minor phosphorylation and degradation of IB␣ were observed in osmotic stress-treated cells. To further confirm this lack of NF-B activation by osmotic stress, we examined DNA binding of NF-B by electrophoretic mobility shift assay (EMSA) in 293 IL-1RI and parental 293 cells (Fig. 1B). We could not detect any activation of NF-B in cells subjected to osmotic stress (Fig. 1B, top panel ). Under these conditions, IB␣ phosphorylation was minor, and no IB␣ degradation was observed (Fig. 1B, middle  and bottom panels). These results demonstrate that osmotic stress activates JNK and p38 but not NF-B, even though TAK1 is activated.
To define the essential role of TAK1 in osmotic stress signaling, we utilized TAK1⌬/⌬ MEFs that express a truncated form of TAK1 lacking kinase activity ( Fig. 2A). This cell line was established by infecting a Cre-expressing retroviral vector into MEFs in which both TAK1 alleles are flanked by loxP (9). We examined activation of JNK and p38 following osmotic stress (Fig. 2B). We found that activation of JNK was greatly impaired in these TAK1-deleted cells, whereas p38 was activated at similar or slightly lower levels. These results suggest that TAK1 plays a major role in the activation of JNK in osmotic stress signaling. To determine whether this role of TAK1 is specific to fibroblasts or if TAK1 functions generally as a mediator of JNK activation in osmotic stress, we used TAK1⌬/⌬ keratinocytes, which were isolated from mice having a skin epidermis-specific deletion of TAK1 (13). Similar to our results with MEFs, we found that osmotic stress could not activate JNK in TAK1-deficient keratinocytes, whereas p38 was activated at slightly lower levels compared with wild type keratinocytes (Fig. 2C). These results demonstrate that TAK1 is an indispensable intermediate in osmotic stress-induced JNK activation. TAK1 and other signaling were lysed and incubated for the indicated times at 37°C to allow spontaneous formation of the TAK1-TAO2 complex. Anti-HA was used for a control precipitation. Anti-TAK1 and anti-HA immunoprecipitates were analyzed for the presence of TAO2 by immunoblotting.
intermediates likely function redundantly to activate p38 in osmotic stress signaling. TAK1 can activate JNK, p38 and NF-B when activated by IL-1 and TNF treatment or when overexpressed. However, our results show that activation of TAK1 by osmotic stress does not lead to the activation of NF-B. This prompted us to speculate that under some conditions TAK1 activity may be modulated so that it does not activate NF-B. One possibility is that TAK1 forms a complex with a molecule that interferes with the interaction between TAK1 and the IKK-NF-B pathway. To identify such a putative molecule, we performed a yeast two-hybrid screen with TAK1 as bait as described previously (11,28). Along with TAB1 and TAB2, which are well characterized TAK1-binding partners, we isolated TAO1 as a binding partner of TAK1. TAO1 shares substantial homology with a closely related kinase TAO2 (also called prostate-derived STE20-like kinase, PSK1) (18 -20). TAO1 and TAO2 are of particular interest because they have been reported to be involved in osmotic stress signaling via activation of the MAPK pathway (20). We conducted a co-immunoprecipitation assay to assess the interaction of TAK1 with TAO1 and TAO2 (Fig. 3, A and  B). Both TAO1 and TAO2 could associate with TAK1. Interaction of TAO2 with TAK1 was independent of its catalytic activity, because a kinase-dead version of TAO2 (TAO2D169A) could also bind to TAK1. Because TAO2 showed a relatively higher affinity to TAK1 compared with TAO1 in several independent experiments, we conducted subsequent experiments using TAO2. We next investigated whether TAO2 and TAK1 could form a complex under more physiological conditions. We raised antibody against TAO2 and attempted to examine the interaction of endogenous TAK1 with TAO2. Although our TAO2 antibody could weakly detect endogenous TAO2 (see Fig. 6A), the efficiency was not sufficient to examine the coprecipitation assay. We therefore ectopically expressed TAO2 and examined its interaction with endogenous TAK1. We could not detect any co-precipitation of TAO2 with TAK1 in the presence or absence of osmotic stress when we used a lysis buffer containing 0.5% Triton X, which we usually use for immunoprecipitation and immunoblotting. We anticipated that the TAK1-TAO2 complex is unstable under this lysis condition. It has been reported that some endogenous protein complexes can assemble spontaneously in cell lysates after incubation at 37°C, such as in the case of the apoptosome and caspase 2-containing complex (24,25,29). We performed coprecipitation assay with endogenous TAK1 following a similar incubation step at 37°C. TAO2 was found in the TAK1 complex after incubation for 3 h (Fig. 3C). The interaction of TAO2 with TAK1 was specific, as no TAO2 was found in the control precipitates. These results suggest that TAO2 interacts with TAK1 in vivo.
It has been demonstrated that TAO2 can activate JNK and p38 (19,20,30). However, the role of TAO2 in the NF-B pathway has not been examined. We next tested whether TAO2 influences TAK1-mediated NF-B activation. TAK1 is highly active when ectopically expressed in culture cells together with its activator subunit TAB1 and can activate JNK, p38, and NF-B (8,27,28,31). We therefore examined NF-B-dependent reporter activity in 293 cells co-expressing TAO2 with TAK1 and TAB1. TAO2 alone was unable to activate NF-B. Furthermore, activation of NF-B by ectopic expression of activated TAK1 was markedly reduced by co-expression of TAO2 (Fig. 4A). To assess whether TAO2 kinase activity is required for its effect on TAK1, we used a kinase-dead mutant of TAO2. We found that TAO2 could inhibit TAK1-mediated activation of NF-B independent of its kinase activity. We next tested whether TAO2 could inhibit the TNF-induced NF-B pathway by examining TNF activation of IKK and NF-B-dependent transcription (Fig. 4, B and C). We found that ectopic expression of TAO2 inhibited activation of IKK and subsequent transcription. These results suggest that TAO2 associates with TAK1 and prevents activation of the TAK1-mediated NF-B pathway. To determine whether the action of TAO2 on TAK1 is specific to the TAK1-mediated NF-B pathway, we next examined JNK activation. We co-expressed TAO2 with an activated form of TAK1 (TAK1 ϩ TAB1) and monitored activation of JNK (Fig. 4D). Consistent with the earlier study, we found that TAO2 alone could activate JNK (30). Furthermore, TAO2 enhanced TAK1-mediated activation of JNK. These results suggest that the TAK1-TAO2 complex can activate JNK but that binding by TAO2 blocks the TAK1-mediated NF-B pathway.
To determine the mechanism by which TAO2 interferes with the TAK1-NF-B pathway, we asked whether TAK1 activity is altered when TAO2 is co-expressed (Fig. 5A). We found that TAK1 activity was not changed by co-expression of TAO2, suggesting that association of TAO2 does not directly alter the activity of TAK1. This is consistent with results showing that TAO2 did not inhibit the TAK1-mediated JNK activation described above (Fig. 4D). We have demonstrated previously that TAK1 forms a complex with the IKKs (22) and that this is important for TAK1mediated activation of the IKK-NF-B pathway. We therefore speculated that TAO2 might affect the interaction of TAK1 with IKK. We transfected expression vectors for TAK1, IKK, and increasing amounts of TAO2 and conducted co-precipitation assays in 293 cells to examine the relative amounts of TAK1 present in complexes with IKK (Fig. 5B). We found that increasing TAO2 blocked the formation of the TAK1-IKK complex. To further confirm the effect of TAO2 under more physiological conditions, we examined endogenous interaction of TAK1 with IKK in the presence and absence of ectopic expression of TAO2 (Fig. 5C). Because the interaction of endogenous TAK1 with IKK is unstable, we conducted co-precipitation assays using the 37°C incubation  -TAO2 and T7-TAK1ϩ T7-TAB1 (active TAK1). Activated JNK was detected with antiphospho-specific JNK (P-JNK) (top panel). The amounts of JNK, T7-TAK1 ϩ T7-TAB1, and HA-TAO2 in the whole cell extracts are shown in the 2nd, 3rd, and 4th panels, respectively. Signal intensity of phospho-JNK was quantitated and normalized to the levels of total JNK amounts. Relative activation levels of JNK are shown (fold). step as described above. We found that TAO2 blocked the interaction of IKK with TAK1. These results collectively suggest that TAK1, when associated with TAO2, is sequestered away from the IKK complex, thereby activating the JNK but not the NF-B pathway.
We finally assessed whether TAO2 plays an essential role in osmotic stress signaling by using TAO2 siRNAs. Two siRNAs showed ability to reduce expression of TAO2 by about 60% (Fig. 6A). We examined the JNK and NF-B activation upon osmotic stress in the TAO2 knockdown cells using these siRNAs (Fig. 6, B and C). Two independent TAO2 siRNAs reduced osmotic stress-induced JNK activation, which suggest that TAO2 plays at least partially an essential role in the osmotic stress-JNK pathway. Activation of TAK1 was determined by detecting the essential phosphorylation of its activity (16). We found that osmotic stress-induced activation of TAK1 was not altered by TAO2 knockdown (Fig. 6B, left  side, bottom panel). These results suggest that TAO2 is not an upstream kinase of TAK1 in the osmotic stress signaling pathway but rather functions in parallel with TAK1 to lead to JNK activation. NF-B was slightly activated upon osmotic stress in TAO2 knockdown cells (Fig. 6C). These results further support our conclusion that TAO2 facilitates the TAK1-JNK pathway but blocks the TAK1-NF-B pathway. However, the induction of NF-B was marginal in TAO2 knockdown cells. This suggests that NF-B activation might be blocked by not only TAO2 but also through other mechanisms.

DISCUSSION
TAK1 can be activated by diverse stimuli, including the proinflammatory cytokines TNF, IL-1, Toll-like receptor ligands, and physical and chemical stresses. The role of TAK1 in each of these signaling pathways has been investigated using cells from genetically engineered TAK1deficient mice, as well as with siRNA gene knockdown methods. TAK1 has been shown to be an indispensable intermediate of TNF, IL-1, and Toll-like receptor signaling in mammalian cells (9,10,13,26). In Drosophila, TAK1 is essential for activation of MAPK in response to innate immune stimuli and plays a redundant role together with MEKK and MLK in stress signaling pathways (32). We have demonstrated here that TAK1 is an essential intermediate in osmotic stress induction of JNK in mammalian cells. We also showed that TAK1 is not an exclusive mediator of osmotic stress-induced p38 activation. TAK1 and other MAP- A, 293 cells were transiently transfected with fixed amounts of expression vectors for T7-TAK1 (1 g) and T7-TAB1 (1 g) and increasing amounts of HA-TAO2 as indicated. TAK1 was immunoprecipitated from the lysates with anti-TAK1, and the immunoprecipitates were subjected to in vitro kinase assay using His-MKK6 as substrate (top panel). The amount of T7-TAK1 in the immunoprecipitates is shown in the middle panel. The amount of HA-TAO2 in the whole cell extracts (WCE) is shown in the bottom panel. TAK1 activity was quantitated and normalized to the levels of TAK1 amounts. Relative activation levels of TAK1 are shown (fold). IP, immunoprecipitation; IB, immunoblotting. B, 293 cells were transiently transfected with fixed amounts of expression vectors encoding T7-TAK1 (0.5 g) and HA-IKK␣ (0.1 g) and increasing amounts of FLAG-TAO2. HA-IKK␣ was immunoprecipitated from the lysates with anti-HA (2nd panel), and co-precipitated T7-TAK1 was detected with anti-T7 (top panel). The amounts of T7-TAK1 and FLAG-TAO2 in the whole cell extracts (WCE) are shown in the 3rd and 4th panels, respectively. C, 293 cells were transfected with an empty vector or an expression vector for HA-TAO2. Cells were lysed and incubated for the indicated times at 37°C to allow spontaneous formation of the TAK1-IKK or TAK1-HA-TAO2 complexes. Endogenous TAK1 was immunoprecipitated, and the precipitates were analyzed for the presence of endogenous IKK␣ or HA-TAO2 by immunoblotting. FIGURE 6. Effect of TAO2 siRNA. A, 293 cells were transfected with control siRNA and TAO2 siRNAs (TAO2 siRNA #1 and #2) using increasing amounts of transfection reagent. TAO2 expression was determined by real time PCR using human TAO2 primers. The amount of TAO2 mRNA relative to that from control siRNA transfected cells is shown. The TAO2 protein was detected with anti-TAO2. ␤-Catenin was used as a loading control. IB, immunoblotting. B, 293 cells were transfected with control siRNA and TAO2 siRNAs (TAO2 siRNA #1, left panels; TAO2 siRNA #2, right panels) using 10 l of transfection reagent. Cells were subjected to osmotic stress (0.5 M NaCl), and cell extracts were immunoblotted with anti-phospho-JNK (top panel) and anti-phospho-TAK1 (3rd panel). The amounts of endogenous JNK and TAK1 are shown in the 2nd and bottom panels, respectively. Signal intensity of phospho-JNK was quantitated and normalized to the levels of total JNK amounts. Relative activation levels of JNK are shown (fold). Representative experiments of four are shown. C, 293 cells were transfected with control siRNA and TAO2 siRNAs. Cells were subjected to osmotic stress (0.5 M NaCl), and cell extracts were subjected to monitor NF-B activation by EMSA. The amount of p65 NF-B subunit is shown as a loading control. A representative result of three is shown.
KKK family members such as MEKK1 and MLK are likely to function redundantly in the p38 pathway in response to osmotic stress in mammalian cells.
TAK1 is activated in response to many extracellular stimuli, yet TAK1 can initiate cellular responses that are unique to each stimulus. This suggests that TAK1 may be modified by some mechanism to selectively activate different downstream pathways. We have demonstrated here that TAO2 can interact with TAK1 and is responsible for the selective activation of JNK versus the NF-B pathway. TAO2 is activated by osmotic stresses (19,20); therefore, the TAK1-TAO2 complex may function to mediate osmotic stress signaling. Previously, Mochida et al. (33) reported that the MAP-KKK ASK1 (apoptosis-stimulating kinase 1) interacts with and inhibits TAK1 by blocking the interaction of TAK1 with tumor necrosis factor receptor-associated factor 6 in the IL-1 signaling pathway. ASK1 is activated by oxidative stress and participates in stress-induced apoptosis (34,35). Therefore, it may be possible that the TAK1-ASK1 complex mediates oxidative stress signaling that leads to apoptosis by activating JNK while inhibiting NF-B.
In this study, we show that TAO2 knockdown only marginally induces NF-B activation upon osmotic stress. This suggests that TAO2 is not solely responsible for blocking the NF-B pathway under osmotic stress conditions. The TAO2related kinase TAO1 may play redundantly to sequester TAK1 from the IKK complex. In addition, to securely block NF-B activation upon osmotic stress, it is likely that the NF-B pathway is negatively regulated through several other mechanisms.
NF-B is a major activator of cell survival signaling (36,37). In contrast, activation of JNK and p38 has been generally correlated with apoptotic as well as necrotic cell death (5-7, 38, 39). Proinflammatory cytokines such as TNF activate both NF-B and JNK pathways through TAK1. Activation of both pathways is important for gene expression involving inflammatory responses. In inflammatory signaling, cell death is not a desired outcome. Therefore, it is believed that the proinflammatory cytokine-induced NF-B pathway functions not only to activate inflammatory genes but also aid to cell survival by masking the pro-apoptotic actions of JNK. In contrast, cell death may be a desired outcome in response to excessive physical and chemical stress, and in these cases NF-B is not activated and JNK is able to induce cell death. Such cell death is important to eliminate damaged and dysfunctioning cells from the body. Thus, TAK1 regulation in osmotic stress signaling, which activates only JNK while blocking the NF-B pathway, is important for osmotic stress-induced cell death.