Inhibition of JNK by Cellular Stress- and Tumor Necrosis Factor α-induced AKT2 through Activation of the NFκB Pathway in Human Epithelial Cells*

Previous studies have demonstrated that AKT1 and AKT3 are activated by heat shock and oxidative stress via both phosphatidylinositol 3-kinase-dependent and -independent pathways. However, the activation and role of AKT2 in the stress response have not been fully elucidated. In this study, we show that AKT2 in epithelial cells is activated by UV-C irradiation, heat shock, and hyperosmolarity as well as by tumor necrosis factor α (TNFα) through a phosphatidylinositol 3-kinase-dependent pathway. The activation of AKT2 inhibits UV- and TNFα-induced c-Jun N-terminal kinase (JNK) and p38 activities that have been shown to be required for stress- and TNFα-induced programmed cell death. Moreover, AKT2 interacts with and phosphorylates IκB kinase α. The phosphorylation of IκB kinase α and activation of NFκB mediates AKT2 inhibition of JNK but not p38. Furthermore, phosphatidylinositol 3-kinase inhibitor or dominant negative AKT2 significantly enhances UV- and TNFα-induced apoptosis, whereas expression of constitutively active AKT2 inhibits programmed cell death in response to UV and TNFα stimulation with an accompanying decreased JNK and p38 activity. These results indicate that activated AKT2 protects epithelial cells from stress- and TNFα-induced apoptosis by inhibition of stress kinases and provide the first evidence that AKT inhibits stress kinase JNK through activation of the NFκB pathway.

Exposure of cells to environmental stress results in the activation of several signal transduction pathways including the MEKK4/MKK7/JNK, 1 MKK3/MKK6/p38, and IB kinase (IKK)/IB/NFB cascades. Stress-induced clustering and inter-nalization of cell surface receptors, such as those for plateletderived growth factor, tumor necrosis factor ␣ (TNF␣), epidermal growth factor, and insulin-like growth factor 1 (IGF1), mediate stress-kinase activation (1)(2)(3). Recent studies suggest that nearly all stress stimuli activate phosphatidylinositol 3-kinase (PI3K) (1), and of the downstream targets of PI3K, AKT is thought to play an essential role in the cellular response to stress.
AKT, also termed protein kinase B or RAC kinase, represents a family of PI3K-regulated serine/threonine kinases (4,5). Three different isoforms of AKT have been identified, AKT1/ protein kinase B␣ (〈⌲⌻1), AKT2/protein kinase B␤ (〈⌲⌻2), and AKT3/protein kinase B␥ (AKT3), all of which are activated by growth factors in a PI3K-dependent manner (4 -9). Full activation of the AKTs requires their phosphorylation at Thr 308 (AKT1), Thr 309 (AKT2), or Thr 305 (AKT3) in the activation loop and Ser 473 (AKT1), Ser 474 (AKT2), or Ser 472 (AKT3) in the C-terminal activation domain (9). AKT1, the most studied isoform, which was originally designated as AKT, suppresses apoptosis induced by a variety of stimuli, including growth factor withdrawal and loss of cell adhesion. Possible mechanisms by which AKT1 promotes cell survival include phosphorylation and inactivation of the proapoptotic proteins BAD and caspase-9 (10,11). AKT1 also phosphorylates and inactivates the Forkhead transcription factors, an event that results in the reduced expression of the cell cycle inhibitor, p27 Kip1 , and the Fas ligand (12)(13)(14). Via phosphorylation of IKK, AKT1 also activates NFB, a transcription factor that has been implicated in cell survival (15,16).
Two separate studies demonstrated that AKT1 is activated when NIH 3T3 fibroblasts are stressed in a variety of ways (17,18). Based on data showing that PI3K inhibitors do not prevent AKT1 activation by stress, these studies concluded that stressinduced AKT1 activation was PI3K-independent. Other studies, however, found that PI3K activity was required for AKT1 activation by heat shock or oxidative stress in Swiss 3T3 cells (19,20). It has been suggested that certain cellular stresses activate AKT1 and AKT3 but not AKT2 (19), a finding that is consistent with the different functions of the AKTs as revealed by studies of mice lacking AKT1 or AKT2 (21)(22)(23). Nevertheless, activation of AKT2 by stress and the role of AKT2 in the stress response have yet to be fully explored. The data presented here show that AKT2 is significantly activated by stress stimuli (e.g. UV irradiation, heat shock, and hyperosmolarity) and by TNF␣ in human epithelial cells but not in fibroblasts. Stress-induced AKT2 activation in epithelial cells is completely blocked by inhibitors of PI3K. When activated by stress, AKT2 inhibits JNK and p38 activities that are required for stressinduced apoptosis. In addition, AKT2 binds to and phosphoryl-ates IKK␣ and, consequently, activates NFB, resulting in inhibition of programmed cell death in response to stress stimuli. Moreover, AKT2-induced NFB activation is required for the inhibition of JNK, but not p38, activity.

EXPERIMENTAL PROCEDURES
Cell Lines, Transfection, and Stimulation-The human epithelial cancer cell lines, A2780, OVCAR3, and human embryonic kidney (HEK) 293 cells were cultured at 37°C and 5% CO 2 in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum. The cells were seeded in 60-mm Petri dishes at a density of 0.5 ϫ 10 6 cells/dish. After incubation overnight, the cells were transfected with 2 g of DNA/dish using LipofectAMINE Plus (Invitrogen). After 36 h of the transfection, the cells were serum-starved overnight and stimulated with UV-C irradiation, heat (45°C), 0.4 M NaCl, or 20 -50 ng/ml TNF␣.
Expression Constructs-The cytomegalovirus-based expression constructs encoding wild type HA-AKT2, constitutively active HA-Myr-AKT2, and dominant negative HA-E299K-AKT2 have been described (24). The HA-JNK1 construct was kindly provided by Michael Karin (School of Medicine, University of California at San Diego). GST-c-Jun-(1-79) and pCMV-FLAG-p38 were gifts from Roger J. Davis (School of Medicine, University of Massachusetts). The constructs used in the study of the NFB pathway were prepared as previously described (25).
Immunoprecipitation and Immunoblotting-Cells were lysed in buffer containing 20 mM Tris-HCl (pH 7.5), 137 mM NaCl, 15% (v/v) glycerol, 1% Nonidet P-40, 2 mM phenylmethylsulfonyl fluoride, 2 g/ml aprotinin and leupeptin, 2 mM benzamidine, 20 mM NaF, 10 mM NaPPi, 1 mM sodium vanadate, and 25 mM ␤-glycerol phosphate. Lysates were centrifuged at 12,000 ϫ g for 15 min at 4°C before immunoprecipitation or Western blotting. Aliquots of the cell lysates were analyzed for protein expression and enzyme activity. For immunoprecipitation, lysates were precleared with protein A-protein G (2:1)-agarose beads at 4°C for 20 min. After the removal of the beads by centrifugation, lysates were incubated with anti-HA monoclonal antibody 12CA5 (Roche Molecular Biochemicals), anti-FLAG antibody (Sigma), or anti-AKT2 antibody (Santa Cruz Biotechnology) in the presence of 30 l of protein A-protein G (2:1)-agarose beads for 2 h at 4°C. The beads were washed once with buffer containing 50 mM Tris-HCl (pH 7.5), 0.5 M LiCl, and 0.5% Triton X-10, twice with phosphate-buffered saline, and once with buffer containing 10 mM Tris-HCl (pH 7.5), 10 mM MgCl 2 , 10 FIG. 1. AKT2 is activated by cellular stress and TNF␣. A, in vitro kinase assay of AKT2 immunoprecipitates prepared from A2780 cells transiently transfected with HA-AKT2. Cells were exposed to 100 ng/ml IGF-1 (15 min), heat shock (45°C for 20 min), 0.4 M NaCl (15 min), 40 J/m 2 UV-C (254 nm), or TNF␣ 20 ng/ml (15 min), and AKT2 activity was determined by in vitro kinase assay using histone H2B as substrate. B, OVCAR3 cells were treated with the indicated stimuli and immunoprecipitated with anti-AKT2 antibody. The immunoprecipitates were subjected to in vitro kinase assay (upper) and Western blotting analyses with antiphospho-Ser473 AKT (middle), or anti-AKT2 (lower) antibody. The bottom panel shows relative AKT2 kinase activity quantified by phosphorimaging. Each experiment was repeated three times. mM MnCl 2 , and 1 mM dithiothreitol, all supplemented with 20 mM ␤-glycerol phosphate and 0.1 mM sodium vanadate. The immunoprecipitates were subjected to in vitro kinase assay or Western blotting analysis. Protein expression was determined by probing Western blots of immunoprecipitates or total cell lysates with the antibodies described above or with the appropriate antibodies as noted in figure legends. Detection of antigen-bound antibody was carried out with the ECL Western blotting Analysis System (Amersham Biosciences).
In Vitro Protein Kinase Assay-Protein kinase assays were performed as previously described (26,27). Briefly, reactions were carried out in the presence of 10 Ci of [␥-32 P] ATP (PerkinElmer Life Sciences) and 3 M cold ATP in 30 l of buffer containing 20 mM Hepes (pH 7.4), 10 mM MgCl 2 , 10 mM MnCl 2 , and 1 mM dithiothreitol. Histone H2B was used as exogenous substrate. After incubation at room temperature for 30 min, the reaction was stopped by adding protein loading buffer, and proteins were separated on SDS-PAGE gels. Each experiment was repeated three times, and the relative amounts of incorporated radioactivity were determined by autoradiography and quantitated with a PhosphorImager (Molecular Dynamics).
PI3K Assay-PI3K was immunoprecipitated from the cell lysates with pan-p85 antibody (Santa Cruz Biotechnology). The immunoprecipitates were washed once with cold phosphate-buffered saline, twice with 0.5 M LiCl, 0.1 M Tris (pH 7.4), and finally with 10 mM Tris/100 mM NaCl/1 mM EDTA. The presence of PI3K activity in the immunoprecipitates was determined by incubating the beads in reaction buffer (10 mM HEPES (pH 7.4), 10 mM MgCl 2 , 50 M ATP) containing 20 Ci [␥-32 P]ATP and 10 g L-␣-phosphatidylinositol 4,5-bisphosphate (Bi-omol) for 20 min at 25°C. The reactions were stopped by adding 100 l of 1 M HCl. Phospholipids were extracted with 200 l of CHCl 3 /MeOH, and phosphorylated products were separated by thin-layer chromatography as previously described (24). The conversion of phosphatidylinositol 4,5-bisphosphate to phosphatidylinositol 3,4,5-trisphosphate was detected by autoradiography and quantitated with a PhosphorImager.
NFB Transcriptional Activation Analysis-HEK293 cells were seeded in 60-mm dishes and transfected with 1.5 g of NFB reporter plasmid (pElam-luc), 0.8 g of pSV2-␤-gal, and different forms (wild type, constitutively active, or dominant-negative) of HA-AKT2 or vector alone. The total amount of DNA transfected was increased to 6 g with empty vector DNA. After serum starvation overnight, the cells were treated with UV (40 J/m 2 ) or TNF␣ (20 ng/ml) and lysed with 400 l/dish of reporter lysis buffer (Tropix). The cell lysates were cleared by centrifugation for 2 min at 4°C. Luciferase and ␤-galactosidase assays were performed according to the manufacturer's procedures (Promega and Tropix, respectively). Each experiment was repeated three times.
Terminal Deoxynucleotidyltransferase-mediated dUTP Nick End Labeling (TUNEL) Assay-AKT2 stably transfected A2780 cells were seeded into 60-mm dishes and grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum for 24 h and pretreated with or without LY294002 for 2 h before exposure to UV, heat shock, NaCl, or TNF␣. Apoptosis was determined by TUNEL using an in situ cell death detection kit (Roche Molecular Biochemicals). The cells were trypsinized, and cytospin preparations were obtained. Cells were fixed with freshly prepared paraformaldehyde (4% in phosphate-

FIG. 2. Activation of AKT2 by cellular stress and TNF␣ is PI3K-dependent.
A, in vitro PI3K assay. HA-AKT2transfected HEK293 cells were exposed to the indicated stimuli. Upper panel, PI3K immunoprecipitates were prepared with anti-pan-p85 antibody and assayed for PI3K activity. The middle panel shows the p85 protein level using anti-p85 antibody, and the bottom panel represents the relative PI3K activity quantified by phosphorimaging. B, HA-AKT2-transfected A2780 cells were treated with LY294002 for 30 min before exposure to indicated stimuli. HA-AKT2 immunoprecipitates were subjected to in vitro kinase assay. Results were confirmed by four independent experiments. PI-3,4,5-P 3 , phosphatidylinositol 3,4,5-trisphosphate; PI-4,5-P 3 , phosphatidylinositol 4,5-trisphosphate. buffered saline (pH 7.4)). Slides were rinsed with phosphate-buffered saline and incubated in permeabilization solution followed by TUNEL reaction mixture for 60 min at 37°C in a humidified chamber. After a rinse, the slides were incubated with converter-alkaline phosphatase solution for 30 min at 37°C and then detected with alkaline phosphatase substrate solution (Vector Laboratories, Burlingame, CA) for 10 min at 25°C. After an additional rinse, the slides were mounted and analyzed under a light microscope. These experiments were performed in triplicate.

RESULTS
AKT2 Is Activated by UV Irradiation, Heat Shock, Hyperosmolarity, and TNF␣-Previous studies showed that stress activates AKT1 and AKT3 but not AKT2 in fibroblasts (19). It has also been shown that TNF␣ receptor mediates UV-and heat shock-induced stress signaling (1)(2)(3). In agreement with these studies, we found that exposure of NIH 3T3 fibroblasts to UV-C, heat, or hyperosmotic conditions did not result in AKT2 activation (data not shown). It is possible, however, that stress might activate AKT2 in epithelial cells due to the fact of frequent alterations of AKT2, but not AKT1 and AKT3, in human epithelial tumors (7,24,27). For this reason we examined the effects of stress on AKT2 activation in two ovarian epithelial cancer cell lines, A2780 cells, which were transiently trans-fected with HA-AKT2, and OVCAR3 cells, which express high levels of endogenous AKT2 (7). The cells were exposed to UV-C, heat shock (45°C), 0.4 M NaCl, or 20 ng/ml TNF␣. IGF1stimulated cells were used as controls. As assessed by in vitro kinase and Western blot analyses of AKT2 immunoprecipitates, all the stimuli substantially increased AKT2 activity in both A2780 and OVCAR3 cells (Figs. 1, A and B). The levels of AKT2 activity induced by these agents, however, were variable. AKT2 activity induced by TNF␣ and UV was comparable with that stimulated by IGF-1, whereas the effect of heat shock and hyperosmolarity (NaCl) on AKT2 activity was relatively smaller (Fig. 1). Nevertheless, these findings suggest that stresses activate AKT2 in a cell type-specific manner.
Stress Simulates PI3K That Mediates AKT2 Activation-To show that stress does indeed activate PI3K in epithelial cells, A2780 or HEK293 cells were exposed to UV irradiation, heat shock, and 0.4 M NaCl or TNF␣, and cell lysates were immunoprecipitated with antibody to pan-p85, a regulatory subunit of PI3K. Assay of PI3K activity shows that these stress conditions as well as TNF␣ activated PI3K as efficiently as did IGF-1 ( Fig. 2A). As described above, stress has been shown to activate AKT1 by both PI3K-dependent and -independent pathways  (17,18). To assess the role of PI3K in the stress-induced activation of AKT2, A2780 cells transfected with HA-AKT2 were exposed to 25 M LY294002, a specific PI3K inhibitor, for 30 min before stress or TNF␣ treatments. LY294002 effectively inhibited stress-and TNF␣-induced AKT2 activation (Fig. 2B). These data provide direct evidences of stress-induced activation of AKT2 through a PI3K-dependent pathway in human epithelial cells.
Stress-induced AKT2 Activation Inhibits UV-and TNF␣induced JNK and p38 Activities-Previous studies demonstrated that two groups of mitogen-activated protein kinases, the JNK and p38, are activated by environmental stress and TNF␣ (28). Therefore, we examined the effects of stress-induced AKT2 activation on the JNK and p38 to determine whether stressed-induced AKT2 activation could target these two stress kinases. A2780 cells were transfected with constitutively active AKT2 or pcDNA3 vector alone. Thirty-six hours after transfection, cells were treated with TNF␣ or UV and analyzed by Western blot for JNK and p38 activation using anti-phospho-JNK and anti-phospho-p38 antibodies. Both JNK and p38 were activated by TNF␣ and UV irradiation. The maximal activation was observed at 10 min of stimulation. Expression of constitutively active AKT2, however, exhibited inhibitory effects on the activation of JNK and p38 that was induced by TNF␣ and UV irradiation. Notably, the activation of JNK and p38 in constitutively active AKT2-transfected cells does not significantly differ from that of the cells transfected with pcDNA3 vector at 10 min of TNF␣ treatment. However, the phosphorylation levels of JNK and p38 in the cells expressing constitutively active AKT2 declined much more than that of Immunoprecipitates were prepared with anti-IKK␣ antibody or IgG and immunoblotted with antibody to AKT2 (top) or IKK␣ (bottom). B, in vitro kinase assay analyses of immunoprecipitates prepared from A2780 cells transfected with indicated plasmids using immunopurified FLAG-IKK␣ as substrates (upper). Expression of FLAG-IKK␣ was confirmed by immunoblotting analysis with anti-FLAG antibody (middle). The bottom panel shows the relative phosphorylation levels of IKK␣ by AKT2. C, in vivo labeling of IKK␣ from COS7 cells transfected with indicated DNA constructs treated with or without TNF␣ and incubated with [␥-32 P]orthophosphate for 4 h. IKK␣ immunoprecipitates were separated by SDS-PAGE, transferred to nitrocellulose, exposed to film (top), and then detected with anti-IKK␣ antibody (bottom). D, AKT2 induces IB␣ degradation. HEK293 cells were transfected with indicated plasmids and treated with cycloheximide (50 g/ml) for 1 h before treatment with 50 ng/ml TNF␣ for up to 60 min. Cell lysates were immunoblotted (IB) with antibody to IB␣ (left panels) or ␤-actin (right panels). Degradation of IB␣ was quantified with a densitometer. The bottom panel shows the degradation rate of IB␣ by normalizing density of IB␣ bands at 0 time point as 100%. E, reporter assays. HEK293 cells were transfected with 2ϫNFB-Luc, ␤-galactosidase and WT-AKT2, Myr-AKT2, or DN-AKT2 pretreated with or without LY294002 and subsequently exposed to 40 J/m 2 UV-C or 20 ng/ml TNF␣. Cell lysates were assayed for luciferase activity and normalized by ␤-galactosidase activity. Error bars represent S.D. Data were obtained from triplicate experiments. pcDNA3-transfected cells after 30 min of stimulation ( Fig. 3 and data not shown). We therefore conclude that the activation of AKT2 does not activate but rather inhibits TNF␣and UVinduced JNK and p38 activities.
AKT2 Interacts With and Phosphorylates IKK␣, but Not NIK, Leading to IB␣ Degradation and NFB Activation-The capacity of both cellular stress and TNF␣ to activate the NFB pathway is well documented (29). Previous studies also show that AKT1 induces activation of the NFB by interaction with IKK␣ (13,14). However, to date there are no reports addressing the potential role of AKT2 in the activation of the NFB pathway. To determine whether AKT2 associates with IKK␣, HEK293 cells were treated with or without TNF␣, immunoprecipitated with anti-AKT2, and immunoblotted with anti-IKK␣ antibody or vice versa. In both instances, the association of AKT2 with IKK␣ was observed (Fig. 4A). Additional studies showed that AKT2-IKK␣ interaction was unaffected by treatment of cells with PI3K inhibitor, wortmannin, or LY294002 (Fig. 4A). These findings indicate that AKT2 constitutively associates with IKK␣. In addition, we have identified putative AKT2 phosphorylation sites in the IKK␣ ( 18 RERLGT 23 ) and in NFB-inducing kinase (NIK, 366 RSREPS 371 ) (bold residue letters represent Akt consensus sequence). To determine whether IKK␣ and/or NIK are phosphorylated by AKT2, A2780 cells were transfected with different forms of AKT2 and treated with LY294002 and TNF␣. In vitro AKT2 kinase assays were performed using FLAG-IKK␣ or HA-NIK, purified from the trans-fected COS7 cells, as substrate. Repeated experiments show that TNF␣-induced AKT2 and constitutively active AKT2 phosphorylated IKK␣ (Fig. 4B) but not NIK (data not shown). Phosphorylation of IKK␣ induced by TNF␣ was largely attenuated by PI3K inhibitor LY294002. Quantification analyses revealed that approximate 70% of TNF␣-induced IKK␣ phosphorylation was inhibited by pretreatment with LY294002 (Fig. 4B). Furthermore, we assessed AKT2 to determine if it phosphorylates IKK␣ in vivo. COS7 cells were transfected with FLAG-IKK␣ together with either constitutively active or dominant-negative AKT2 or vector alone and labeled with [␥-32 P]orthophosphate. IKK␣ immunoprecipitates prepared using anti-FLAG antibody were separated by SDS-PAGE and transferred to nitrocellulose. The phospho-IKK␣ was detected by autoradiography. As shown in Fig. 4C, IKK␣ was highly phosphorylated in cells expressing constitutively active AKT2 but not in the cells transfected with pcDNA3 and dominant-negative AKT2. Collectively, these data indicate that IKK␣ is an AKT2 physiological substrate.
Activation of NFB requires its dissociation from its cytosolic inhibitor, IB, a process dependent on the phosphorylation and consequent degradation of IB by IKK. Thus, we next examined AKT2 to determine if it induces IB degradation. Immunoblotting analyses revealed that constitutively active AKT2 significantly promoted IB␣ degradation (Fig. 4D). To assess the involvement of AKT2 in NFB activation, HEK293 cells were co-transfected with a NFB-luciferase reporter and either vector alone, wild type, or constitutively active or dominant negative AKT2 treated with or without LY294002 before UV or TNF␣ stimulation. As shown in Fig. 4E, ectopic expression of wild-type AKT2 significantly enhanced UV-and TNF␣-induced NFB activity, which was abolished by treatment of cells with LY294002 or dominant negative AKT2. Constitutively active AKT2 alone was able to induce NFB activity to a level comparable with UV-or TNF␣-treated cells transfected with wildtype AKT2. These data show that PI3K/AKT2 mediates both stress-and TNF␣-activated NFB pathway.
To determine AKT2 phosphorylation site of IKK␣, GST fusion proteins containing either wild type IKK␣ ( 18 RERLGT 23 ; termed GST-WT-IKK␣) or mutant IKK␣ ( 18 RERLGA 23 ; termed GST-IKK␣T23A) were prepared and used as substrates in in vitro AKT2 kinase assays. As seen in Fig. 5A, UV-and TNF␣-activated AKT2 as well as constitutively active AKT2 phosphorylated GST-WT-IKK␣ but not GST-IKK␣T23A. We next assessed the capacity of AKT2-induced IKK␣ to phosphorylate IB␣. Constitutively active AKT2 was expressed in HEK293 cells, and cell lysates were immunoblotted with an antibody that specifically recognizes phosphorylated IB␣ at Ser 32 . The results of these experiments show that constitutively active AKT2 increased IB␣ phosphorylation ϳ2-fold and that this increase was abolished by cotransfection of pcDNA3-IKK␣T23A. Expression of IKK␣T23A also blocked IB␣ phosphorylation induced by TNF␣ or UV (Fig. 5B). Additional luciferase reporter experiments demonstrated that expression of IKK␣T23A inhibited the TNF␣-or constitutively active AKT2-induced NFB activation (Fig. 5C). These data indicate that phosphorylation of IKK␣ at Thr 23 is required for AKT2-mediated NFB activation.
IKK␣ Phosphorylation by AKT2 Is Required for Inhibition of JNK but Not p38 Activation-Recent studies showed that NFB exerts its cell survival function by inhibition of JNK activation in response to extracellular stress (30,31). However, it is currently unknown whether AKT-induced NFB activation results in inhibition of JNK. Therefore, we next attempted to determine if AKT2-activated IKK␣ is required for AKT2 inhibition of JNK and p38 activities induced by stress and TNF␣. The activation of JNK and p38 was examined in HEK293 cells transfected with IKK␣ or IKK␣T23A together with or without constitutively active AKT2. Western blotting analyses with phospho-JNK and -p38 antibodies revealed that wild type IKK␣ did not significantly enhance AKT2 inhibition of JNK (Fig. 6). However, expression of IKK␣T23A abrogated the effects of constitutively active AKT2 on inhibition of JNK (Fig. 6). Similar to the results shown in Fig. 3, TNF␣-induced JNK activation reached the maximal level at 10 min of stimulation, which was neither significantly inhibited by constitutively active AKT2 nor affected by expression of IKK␣T23A (Fig. 6). Therefore, these data indicate that inhibition of JNK activation by AKT2/NFB could be via a mechanism of induction of dephosphorylation of JNK by the AKT2/IKK␣/NFB cascade.
AKT2 Activation Inhibits Stress-induced Apoptosis-It is documented that various stresses and TNF␣ are capable of inducing apoptosis in different cell types through activation of JNK and p38 pathways (29). Because PI3K/AKT is essential for cell survival and activated AKT2 inhibits JNK/p38 and induces NFB pathway, we investigated the role of PI3K/AKT2 in stress-and TNF␣-induced programmed cell death. AKT2 stably transfected A2780 cells were pretreated with or without LY294002 for 2 h before exposure to UV, heat shock, NaCl, or TNF␣. As determined by the TUNEL assay, inhibition of PI3K activity dramatically increased the percentage of cells undergoing apoptosis in response to UV or TNF␣ (Fig. 7). Moreover, inhibition of AKT2 activity by expression of dominant-negative AKT2 increased the percentage of apoptotic cells in the UVand TNF␣-treated populations by ϳ2-fold. On the other hand, cells expressing constitutively active AKT2 were resistant to UV-and TNF␣-induced apoptosis. These data show that the PI3K/AKT2 pathway plays a key role in protecting cells from apoptosis induced by extracellular stress or TNF␣.

DISCUSSION
In this report, we have provided evidence that AKT2 is activated by extracellular stress and TNF␣ through a PI3K-dependent pathway in human epithelial cells. Most importantly, the activation of AKT2 inhibits stress-and TNF␣-induced JNK and p38 activities and activates the NFB cascade, leading to protection of cells from stress-and TNF␣-induced apoptosis.
Previous studies show that stress activates cell membrane receptors, including those for epidermal growth factor, plateletderived growth factor, and IGF. As a result, receptors associate with numerous proteins that activate downstream signaling molecules (1)(2)(3). One such protein is PI3K, which has been implicated in the regulation of nearly all stress signaling pathways (1). Because the AKTs are major downstream targets of PI3K, their role in the stress response has been recently investigated. In Swiss 3T3 cells, both oxidative stress and heat shock were shown to induce a marked activation of AKT1 and AKT3 but not AKT2 (19). AKT1 activation by hyperosmotic stress in COS7 and NIH 3T3 cells has also been demonstrated (17). In this study, we show that AKT2 is activated by different stress conditions including UV irradiation, hyperosmolarity, and heat shock as well as by TNF␣ in several human epithelial cell lines.
Three isoforms of AKT display high sequence homology and share similar upstream regulators and downstream targets as identified so far. However, there are clear differences between them in terms of biological and physiological function. In addition to the more prominent role of AKT2 in human malignancy and transformation (7,32), the expression patterns of AKT1, AKT2, and AKT3 in normal adult tissues as well as during development are quite different (4,8,33). Recent studies suggest that AKT1, AKT2, and AKT3 may interact with different proteins and, thus, may play different roles in signal transduction. For instance, the Tcl1 oncoprotein preferentially binds to and activates AKT1 but not AKT2 (34). Gene knockout studies revealed that AKT1-deficient mice display defects in both fetal and postnatal growth but, unlike AKT2 Ϫ/Ϫ mice, do not exhibit a type II diabetic phenotype; these differences suggest that the functions of AKT1 and AKT2 are non-redundant with respect to organismic growth and insulin-regulated glucose metabolism (21-23). It has been also shown that AKT2 but not AKT1 plays a specific role in muscle differentiation (35). 2 In this study, we demonstrated that AKT2 is activated by a variety of stress conditions in human epithelial cells but not in fibroblasts, suggesting that activation of different isoforms of AKT is cell type-specific in response to extracellular stress.
It is controversial whether stress-induced AKT1 activation is mediated by the PI3K pathway (17)(18)(19). Two previous reports showed that PI3K inhibitors did not block heat shock-or H 2 O 2induced activation of AKT1 and, thus, suggested that stress (unlike growth factors) activates AKT1 in a PI3K-indpendent manner (17,18). However, the opposite results were observed by other groups (19,20). Konishi et al. also provide evidence of AKT1 activation by H 2 O 2 and heat shock through both PI3Kdependent and -independent pathways (18). We previously demonstrated that activation of AKT2 by growth factors required PI3K activity, whereas both PI3K-dependent and -independent pathways contributed to AKT2 activation by Ras (26). In this report, we show that PI3K inhibitors completely block AKT2 activation induced by UV-C, heat shock, and hyperosmolarity, indicating that stress activates AKT2 via the PI3K pathway.
JNK and p38 are stress mitogen-activated protein kinases that are activated by cytokines and a variety of cellular stresses (28). Like the classical mitogen-activated protein kinase kinase (MEK), direct activators for JNK and p38 have been identified. JNK is activated by phosphorylation of tyrosine and threonine by the dual specificity kinases, MKK4/SEK1 and MKK7. Similarly, p38 is activated by MKK3 and MKK6. However, biochemical studies have documented the existence of other JNK 2 S. Kaneko, S. V. Nicosia, Z. Wu, T. Nobori, and J. Q. Cheng, submitted for publication. and p38 activators or inhibitors in cells stimulated by a variety of cellular stresses (28). Although previous reports showed that AKT, JNK, and p38 are downstream targets of PI3K and represent parallel pathways in response to stress (17-20, 37, 38), the data presented in this study indicate that stress-and TNF␣-induced activation of AKT2 inhibits the JNK and p38 activities, suggesting that AKT2 cross-talks with JNK and p38 stress pathways.
NFB is another critical stress response pathway (29). Activation of NFB is achieved through the signal-induced proteolytic degradation of IB, which is associated with and inhibits the activity of NFB in the cytoplasm. The critical event that initiates IB degradation is the stimulus-dependent activation of the IB kinases IKK␣ and IKK␤, which phosphorylate IB at specific N-terminal serine residues (Ser 32 and Ser 36 for IB␣; Ser 19 and Ser 23 for IB␤). Phosphorylated IB is then selectively ubiquitinated by an E3 ubiquitin ligase and degraded by the 26 S proteasome, thereby releasing NFB for translocation to the nucleus where it initiates the transcription of target genes (29). Moreover, two mitogen-activated protein kinase kinase kinase (MAPKKK) members, NIK and MEKK1, have been reported to enhance the activity of the IKKs and consequently trigger the phosphorylation and destruction of the IBs and induce the activation of the NFB pathway (29). Recent studies also showed that AKT1 induces the NFB cascade through activation of IKK and degradation of IB (13,14). In this report, we show that AKT2 physically binds to and phosphorylates IKK␣ but not NIK even though NIK contains an AKT2 phosphorylation consensus sequence. When activated by stress or TNF␣, AKT2 degrades IB and activates NFB-mediated transcription, indicating that stress-activated AKT2 targets the NFB pathway.
Importantly, we have provided evidence that activation of AKT2 induced by stress and TNF␣ inhibits JNK activity through activation of the NFB pathway to protect cells from apoptosis in response to these stimuli. Previous studies showed that the AKT2 pathway is important for cell survival and malignant transformation (7,24,32). The data presented here show that cells expressing constitutively active AKT2 are resistant to stress-and TNF␣-induced apoptosis and that dominant-negative AKT2 and LY294002 sensitize cells to stressand TNF␣-induced programmed cell death. These findings indicate that stress-induced AKT2 activation promotes cell survival. Among the stress-activated kinases are JNK; recent studies demonstrated that activation of JNK and p38 plays an important role in triggering apoptosis in response to extracellular stress and TNF␣ (36, 39 -41), whereas activation of NFB protects cells from programmed cell death (29). Although a number of downstream targets of AKT2 have been identified, our data indicate that AKT2-inhibited JNK and p38 activities and AKT2-induced NFB activation could play, at least in part, an important role in the AKT2 pathway that protects cells from stress-and TNF␣-induced apoptosis. Recent reports demonstrate that NFB-up-regulated Gadd45␤ and Xiap inhibited JNK activation and abrogated TNF␣-induced programmed cell death (30,31). Our cDNA microarray experiments showed that constitutively active AKT2 induces Xiap. 3 Thus, AKT2 inhibition of JNK activity could be due to up-regulation of Xiap by NFB pathway (Fig. 8). Further studies are required to characterize the mechanism of inhibition of p38 stress pathway by AKT2 and involvement of Xiap in AKT2/NFB inhibition of the JNK activation.