Tumor Necrosis Factor Induces Phosphorylation and Translocation of BAD through a Phosphatidylinositide-3-OH Kinase-dependent Pathway*

Tumor necrosis factor (TNF) induced the phosphorylation of BAD at serine 136 in HeLa cells under conditions that are not cytotoxic. BAD phosphorylation by TNF was dependent on phosphatidylinositide-3-OH kinase (PI3K) and was accompanied by the translocation of BAD from the mitochondria to the cytosol. Blocking the phosphorylation of BAD and its translocation to the cytosol with the PI3K inhibitor wortmannin activated caspase-3 and markedly potentiated the cytotoxicity of TNF. Transient transfection with a PI3K dominant negative mutant or a dominant negative mutant of the serine-threonine kinase Akt, the downstream target of PI3K and the enzyme that phosphorylates BAD, similarly potentiated the cytotoxicity of TNF. By contrast, transfection with a constitutively active Akt mutant protected against the cytotoxicity of TNF in the presence of wortmannin. Phosphorylation of BAD prevents its interaction with the antiapoptotic protein Bcl-XL. Transfection with a Bcl-XL expression vector protected against the cytotoxicity of TNF in the presence of wortmannin. The mechanism by which the inhibition of the phosphorylation of BAD is likely linked to the induction of lethal mitochondrial damage in TNF-intoxicated cells is discussed.

its phosphorylation, growth factors lead to the dissociation of BAD from Bcl-X L and Bcl-2 and thereby promote cell survival by allowing the unhindered action of these proteins.
The PI3K inhibitor wortmannin enhanced the activation of caspase-3 that was induced by TNF or anti-Fas (8). Here we demonstrate that TNF acts like other growth factors to promote the phosphorylation of BAD at Ser-136. As a result, there is the translocation of BAD from the mitochondria to the cytosol. Moreover, the phosphorylation of BAD by TNF occurs by a PI3K-dependent pathway and is necessary to prevent the cytotoxicity of this cytokine. Inhibition of PI3K prevents both the phosphorylation of BAD and its translocation from mitochondria to the cytosol, effects that are accompanied by substantially enhanced cell killing by TNF.

EXPERIMENTAL PROCEDURES
Tissue Culture-HeLa cells (ATCC-CC-1, American Type Culture Collection) were maintained in 25 cm 2 polystyrene flasks (Corning Costar Corp., Oneonta, NY) with 5 ml of Dulbecco's modified Eagle's medium (DMEM) (high glucose, without pyruvate) (Life Technologies, Inc.) containing 100 units/ml penicillin, 0.1 mg/ml streptomycin, and 10% heat-inactivated fetal bovine serum, all incubated under an atmosphere of 95% air and 5% CO 2 . For transient transfections, HeLa cells were plated at one-third of confluency (3.0 ϫ 10 4 cells/cm 2 ) in 24-well plates. After overnight incubation, the cells were washed twice in phosphate-buffered saline (PBS). Transfections were performed using Lipofectamine-Plus (Life Technologies, Inc.) according to the manufacturer's instructions. Transfection efficiencies of 25-30% were routinely obtained, as assessed by ␤-galactosidase staining. For experiments, cells were transfected with 0.5 g of pCDNA-LacZ and 5 g of either pCDNA, pCDNA3-myc-⌬p85 (referred to as pCDNA-PI3K(Ϫ) (generously provided by Dr. Julian Downward, ICRF, London), pCDNA-Akt(ϩ) or pCDNA-Akt(Ϫ) (generously provided by Dr. Morris J. Birnbaum and Dr. Randall N. Pittman, Howard Hughes Medical Institute, University of Pennsylvania and Dr. Julian Downward, ICRF, London), or pCDNA-BclX L . After 4 h the cells were washed twice with PBS and placed in complete DMEM. After 48 h of further incubation, the cells were washed twice with PBS and placed in DMEM minus serum. After 30 min the cells were then treated with various reagents as described in the text.
Measurement of Cell Viability-Experiments were performed 2 days after plating 1.0 ϫ 10 5 cells in 500 l of complete DMEM into 1.88 cm 2 wells of a 24-well plate. By the second day the cells were growing exponentially and had achieved a density of 2.5-3.0 ϫ 10 5 cells/well. Before treatment the cells were washed twice with Ca 2ϩ /Mg 2ϩ -free PBS, after which 500 l of DMEM without serum were added. Cell viability was determined by the ability of the cells to exclude trypan blue. 10 l of a 0.5% solution of trypan blue were added to the wells. Both viable and nonviable cells were counted for each data point in a total of eight microscopic fields. Cell viability for transiently transfected cells was assessed by counting the number of ␤-galactosidase-positive cells in wells transfected with pCDNA-LacZ and either pCDNA, pCDNA-PI3K(Ϫ), pCDNA-Akt(Ϫ), pCDNA-Akt(ϩ), or pCDNA-Bcl-X L as described above. The number of ␤-galactosidase-positive cells in untreated wells was then compared with the number in wells treated with either TNF alone, wortmannin alone, LY294002 alone, TNF plus wortmannin, or TNF plus LY294002. Cotransfection of 0.5 g of pCDNA-LacZ and 5 g of the above outlined constructs gave similar * This work was supported by National Institutes of Health Grant DK38305. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  transfection efficiencies (10 -15%) and number of ␤-galactosidase-positive cells/well (2500 -3000 cells/well) Detection of Caspase-3 Activity-The assay is based on the ability of the active enzyme to cleave the chromophore pNA from the enzyme substrate DEVD-p-nitroanilide. Cell extracts were diluted 1:1 with 2ϫ reaction buffer (10 mM Tris, pH 7.4, 1 mM dithiothreitol, 2 mM EDTA, 0.1% CHAPS, 1 mM phenylmethylsulfonyl fluoride, 10 g/ml pepstatin, 10 g/ml leupeptin). DEVD-p-nitroanilide was added to a final concentration of 50 M, and the reaction was incubated for 1 h at 37°C. The samples were then transferred to a 96-well plate, and absorbance measurements were made in a 96-well plate reader at 405 nm.
Isolation of Cytosol and Mitochondrial Fractions-Cells were plated in 25-cm 2 flasks at 5.0 ϫ 10 6 cells/flask. After treatment, the cells were harvested by trypsinization followed by centrifugation at 600 ϫ g for 10 min at 4°C. The cell pellets were washed once in PBS and then resuspended in 3 volumes of isolation buffer (20 mM Hepes, pH 7.4, 10 mM KCl, 1.5 mM MgCl 2 , 1 mM sodium EDTA, 1 mM dithiothreitol, and 10 mM phenylmethylsulfonyl fluoride, 10 M leupeptin, 10 M aprotinin in 250 mM sucrose). After chilling on ice for 3 min, the cells were disrupted by 40 strokes of a glass homogenizer. The homogenate was centrifuged twice at 2,500 ϫ g at 4°C to remove unbroken cells and nuclei. The mitochondria were then pelleted by centifugation at 12,000 ϫ g at 4°C for 30 min. The supernatant was removed and filtered through 0.2-m and then 0.1-m Ultrafree MC filters (Millipore) followed by centrifugation at 100,000 ϫ g at 4°C to give cytosolic protein. Mitochondrial and cytosolic fractions (25 g of protein) were separated on 12% SDS-polyacrylamide electrophoresis gels and electroblotted onto nitrocellulose membranes. Phospho-BAD-136, phospho-BAD-112, and BAD were detected by rabbit polyclonal antibody at a dilution of 1:500 (New England Biolabs). Secondary goat antirabbit horseradish peroxidase-labeled antibody (1:2000) was detected by enhanced chemiluminescence.
Nuclear Extract Preparation and Electrophoretic Mobility Shift Assay-At the end of the specified incubation time, cells (15-20 ϫ 10 6 / sample) were collected by centrifuging at 359 ϫ g for 5 min at 4°C. After washing once with PBS, the cell pellet was suspended in 0.5 ml of buffer A (10 mM Hepes-NaOH, pH 7.8, 15 mM KCl, 2 mM MgCl 2 , 0.1 mM EDTA, 1 mM dithiothreitol, and 1 mM phenylmethylsulfonyl fluoride). The suspension was transferred to a microcentrifuge tube and centrifuged at 750 ϫ g for 5 min at room temperature. The supernatant was removed by aspiration, and the pellet was resuspended in 0.2 ml of buffer A. After 10 min on ice, Nonidet P-40 was added to 0.5%, and the suspension was centrifuged at 1330 ϫ g for 15 min. The resultant nuclear pellet was suspended in 15 l of buffer B (20 mM Hepes-NaOH, pH 7.9, 1.5 mM MgCl 2 , 0.5 mM dithiothreitol, 0.42 M NaCl, 0.2 mM EDTA, 25% glycerol, and 0.5 mM phenylmethylsulfonyl fluoride). After 15 min on ice with vigorous stirring, the suspension was centrifuged at 16,300 ϫ g for 10 min. Fifteen l of the resultant supernatant was diluted with 75 l of buffer C (20 mM Hepes-NaOH, pH 7.9, 50 mM KCl, 0.2 mM EDTA, 0.5 mM dithiothreitol, and 0.5 mM phenylmethylsulfonyl fluoride) to obtain the final nuclear extract preparation. Electrophoretic mobility shift assays for NFB in nuclei was performed by using the gel shift assay system (Promega) according to the manufacturer's instructions. The reaction mixture contained 2 l of 5ϫ gel shift binding buffer, 1 l of 32 P-labeled NFB consensus oligonucleotide probe, 2 l of water, and 5 l of nuclear extract (1 g of protein). Relative band densities were obtained by densitometric analysis (Amersham Pharmacia Biotech).
Treatments-Human recombinant TNF (Sigma) was dissolved in PBS and added for the times indicated to a final concentration of 10 ng/ml. Wortmannin (Sigma) was dissolved in Me 2 SO and added at a 0.2% volume to a final concentration of 50 nM. LY294002 was dissolved in Me 2 SO and added at a 0.2% volume to a final concentration of 10 M. Rapamycin was dissolved in Me 2 SO and added at a 0.2% volume to a final concentration of 50 nM. In all experiments the cells were pretreated for 30 min with wortmannin, LY294002, or rapamycin before the addition of TNF. Control experiments demonstrated that Me 2 SO at the concentration used was without measurable effect on the parameters examined. Protein synthesis was measured as the incorporation of [ 3 H]leucine into an acid-insoluble precipitate as described previously (9).

RESULTS
Effect of TNF on BAD Phosphorylation-HeLa cells maintained a basal level of phosphorylation of BAD at Ser-136 (Fig.  1a, left panel). Within 30 min of treatment with TNF, however, there was a marked increase in BAD phosphorylation (phospho-BAD-136), an effect that increased further to reach a maximum within 2 h (Fig. 1a, left panel). The PI3K inhibitor wortmannin totally inhibited the induction by TNF of BAD phosphorylation at Ser-136 (Fig. 1a, right panel). In fact, there was a progressive decrease in the content of phospho-BAD-136 in the presence of both TNF and wortmannin over the 2 h time course of the experiment. Importantly, treatment of HeLa cells with either TNF alone or TNF together with wortmannin produced no change in the total content of BAD (Fig. 1b, left and  right panel, respectively). In addition, no staining was detected when an antibody was used against the Ser-112-phosphorylated form of BAD, either in control, TNF, or TNF and wortmannin-treated cells (data not shown).
Effect of TNF on the Localization of BAD-The HeLa cells were fractionated to assess simultaneously the content of BAD and phospho-BAD-136 in the cytosol and mitochondria. TNF induced a progressive increase in the level of phospho-BAD-136 in the cytosolic fraction over a 2-h time course (Fig. 2a, left  panel). As opposed to the situation in whole cell lysates (Fig.  1b), this change was mirrored in a concomitant increase in the total content of BAD in the cytosolic fraction (Fig. 2b, left  panel). Both of these effects were prevented by inhibition of PI3K (Fig. 2, a and b, right panel). With TNF alone, the mitochondria exhibited a marked decrease in the content of total BAD (Fig. 3, left panel) at the same time that the content of phospho-BAD-136 and total BAD increased in the cytosol. Again the decrease of mitochondrial-associated BAD induced by TNF was prevented by PI3K inhibition (Fig. 3, right panel). Neither the Ser-136 nor Ser-112 form of BAD was detected in the mitochondrial fraction of either TNF or TNF and wortmannin-treated cells. (data not shown).
Inhibition of PI3-kinase Potentiates the Cytotoxicity of TNF-TNF alone killed few HeLa cells over a 6-h time course (Fig. 4a, closed squares). As measured by trypan blue uptake, less than 9% of the cells were dead after 4 h, and after 6 h, only 11% of the cells had died. By contrast, cells pretreated for 30 min with 50 nM wortmannin and then exposed to TNF exhibited substantial cell killing (Fig. 4a, closed circles). After 4 h, more than 40% of the cells were dead. Cell killing reached almost 70% after 6 h. Importantly, wortmannin alone (in the absence of TNF) had no effect on the viability of the HeLa cells after 6 h (Fig. 4a, closed  triangles).
The resistance of cells to the cytotoxicity of TNF has been attributed to the activation of the transcription factor NFB (10). The dependence of the killing produced by TNF on an inhibition of either transcription or translation is interpreted in turn as a consequence of the inhibition of the expression of factors induced by NFB that promote cell survival. Accordingly, it was possible that wortmannin potentiates the cytotoxicity of TNF by similarly inhibiting the activation of NFB. As shown in Fig. 5, within 30 min of exposure to TNF, activation of NFB was readily detectable. Wortmannin alone did not activate NFB, and a pretreatment with wortmannin followed by exposure to TNF did not inhibit the activation of NFB (Fig.  5). As noted above, the cytotoxicity of TNF is enhanced upon inhibition of protein synthesis by such agents as cycloheximide. Like wortmannin, LY294002 is an inhibitor of PI3K (11). Table II shows that 10 M of LY294002 produced only minimal toxicity in the absence of TNF. However, like wortmannin, in the presence of TNF and LY294002, there was substantial cell killing with more than 90% of cells dead after 6 h. In the same study, TNF alone killed only 15% of the cells (Table II). Rapamycin is an inhibitor of p70 S6 kinase, an enzyme that is a downstream target of Akt (12). Unlike wortmannin and LY294002, however, rapamycin did not potentiate the cytotoxicity of TNF (14% dead cells after 6 h of treatment with rapamycin and TNF). This result suggests that p70 S6 kinase is not responsible for exerting the antiapoptotic effect of Akt stimulation under the conditions of TNF treatment.
To further confirm that the potentiation seen with wortmannin is a consequence of the inhibition of PI3K, the effect was determined of transiently transfecting HeLa cells with a PI3K dominant negative mutant, as well as with mutants of Akt that are either constitutively active or kinase-defective (dominant negative). HeLa cells were cotransfected with pCDNA-LacZ and either pCDNA, pCDNA-PI3K(Ϫ), pCDNA-Akt(ϩ), or pCDNA-Akt(Ϫ). Cell death induced by TNF was measured by a reduction in the number of cells expressing ␤-galactosidase relative to that obtained in untreated transfected cultures. Cotransfection with pCDNA-LacZ and either the empty vector or one of the constructs gave similar transfection efficiencies and numbers of ␤-galactosidase-positive cells (data not shown), a result indicating that the constructs alone did not decrease cell viability. Transfection with a dominant negative inhibitor of PI3K resulted in only 37% of the number of ␤-galactosidasepositive cells after 6 h of exposure to TNF as compared with untreated control wells. By contrast, in cells cotransfected with the empty vector, treatment with TNF alone had little effect on the number of ␤-galactosidase-positive cells. Thus, transfection with a dominant negative mutant of PI3K potentiated like wortmannin the cytotoxicity of TNF.
Similarly, a dominant negative inhibitor of Akt, the downstream target of PI3K, also potentiated the cytotoxicity of TNF, with only 31% of the cells staining positive for ␤-galactosidase as compared with untreated controls (Table III). By contrast, transfection with a constitutively active Akt mutant protected against the cytotoxicity induced by TNF in the presence of wortmannin. Cotransfection with pCDNA-LacZ and pCDNA-Akt(ϩ) resulted in 86% of cells staining positive for ␤-galactosidase in the presence of TNF and wortmannin as compared with untreated wells. In cultures cotransfected with pCDNA-LacZ and pCDNA, only 27% of cells stained positive for ␤-galactosidase upon treatment with TNF and wortmannin as compared with the untreated wells. Thus expression of the constitutively active Akt is able to bypass the block at PI3K signaling brought about by wortmannin.
Overexpression of Bcl-X L Protects against TNF-induced Cytotoxicity in the Presence of Wortmannin-It is hypothesized that the increased cytotoxicity of TNF upon inhibition of PI3K relates to the inhibition of BAD phosphorylation by Akt and its subsequent dissociation from antiapoptotic proteins such as Bcl-X L . It was of interest, therefore, to explore the effect of the overexpression of Bcl-X L in HeLa cells on the cytotoxicity of TNF in the presence of the PI3K inhibitors used above. Table  IV shows the results of the transient transfection of HeLa cells with a Bcl-X L expression vector (pCDNA-BclX L ). Treatment of cells cotransfected with pCDNA-LacZ and pCDNA with TNF in the presence of either wortmannin or LY294002 resulted in only 33 and 29%, respectively, of the number of ␤-galactosidase-positive cells as compared with untreated wells. By contrast, overexpression of Bcl-X L afforded significant protection against the cytotoxicity induced by TNF in the presence of either wortmannin or LY294002. In cultures cotransfected with pCDNA-LacZ and pCDNA-BclX L and treated with TNF and either wortmannin or LY294002, 85 and 81%, respectively, of the cells stained positive for ␤-galactosidase as compared with untreated wells.
Inhibition of PI3-kinase Potentiates the Activation of Caspase-3-The cytotoxicity data were mirrored by the activity of the apoptotic protease caspase-3. Treatment of HeLa cells with TNF alone resulted in a 1.5-fold increase over the control value in the activity of caspase-3 (Fig. 4b, closed squares). By contrast, the combination of TNF and wortmannin produced a 12-fold increase in the activity of caspase-3 after 6 h (Fig. 4b,  closed circles). Again, wortmannin alone (in the absence of TNF) had no effect on caspase-3 activity (Fig. 4b, closed  triangles). DISCUSSION We have shown that BAD is phosphorylated on Ser-136 in response to the treatment of HeLa cells with TNF. This phosphorylation is accompanied by the translocation of BAD from the mitochondria to the cytosol. Importantly, the phosphorylation and translocation of BAD occurs in the absence of any cell killing by TNF. In turn, inhibition of PI3K by wortmannin prevented both the phosphorylation and translocation of BAD, a result that was now reflected in substantial cell killing in response to TNF. Transient transfection with a PI3K dominant negative mutant or a dominant negative mutant of the serinethreonine kinase Akt, the downstream target of PI3K and the FIG. 2. The translocation of phosphorylated BAD from the mitochondria to the cytosol is inhibited by wortmannin. HeLa cells (5.0 ϫ 10 6 ) were either treated with TNF (10 ng/ml) alone or pretreated for 30 min with 50 nM wortmannin followed by the addition of TNF. At the times indicated, cytosolic fractions were prepared, and the levels of BAD-P136 and BAD were determined by Western blotting.

FIG. 3. The depletion of BAD associated with the mitochondria induced by TNF is prevented by wortmannin.
HeLa cells (5.0 ϫ 10 6 ) were either treated with TNF (10 ng/ml) alone or pretreated for 30 min with 50 nM wortmannin followed by the addition of TNF. At the times indicated, mitochondria were prepared, and the levels of BAD-P136 and BAD were determined by Western blotting. enzyme that phosphorylates BAD, similarly potentiated the cytotoxicity of TNF. By contrast, transfection with a constitutively active Akt mutant protected against the cytotoxicity of TNF in the presence of wortmannin. Phosphorylation of BAD prevents its interaction with the antiapoptotic protein Bcl-X L . Transfection with a Bcl-X L expression vector protected against the cytotoxicity of TNF in the presence of wortmannin.
Two concerns, at least, are raised by these results. First, the mechanism of PI3K activation by TNF needs to be considered. Second, we need to address the mechanism by which BAD phosphorylation promotes cell survival or, alternatively, the mechanism by which inhibition of BAD phosphorylation promotes cell killing.
The TNF superfamily of receptors includes, in addition to the two TNF receptors (TNF-R1 and TNF-R2), the Fas receptor (CD95), CD40, the lymphotoxin ␤-receptor, and nerve growth factor (13). NGF and anti-CD40 have also been shown to activate PI3K (14 -16). The exposure of PC-12 cells to NGF causes their differentiation (14). Inhibition of PI3K with wortmannin or with a dominant negative mutant of PI3K inhibited the neurite outgrowth in PC-12 cells that was induced by NGF, an effect that was accompanied by the death by apoptosis of the cells. In this regard, it is noteworthy that interleukin-2 and -3 and insulin-like growth factor-I promoted cell survival by a mechanism that depended on activation of PI3K (1,6,17).
Expression of a dominant negative inhibitor of PI3K was able to potentiate TNF-induced cytotoxicity as did wortmannin or LY294002. In addition, the ability of a dominant negative inhibitor of Akt and constitutively active Akt to potentiate and inhibit TNF-induced cell killing, respectively, most likely means that it is Akt activation by PI3-kinase that is critical for

FIG. 5. Wortmannin does not inhibit NFB activation by TNF.
HeLa cells were either left untreated, treated with 50 nM wortmannin or TNF (10 ng/ml) or pretreated with wortmannin for 30 min followed by treatment with TNF. After 15 min of incubation, cells were harvested, and nuclear extracts were prepared. NFB activity was assessed by electrophoretic mobility shift assay as described under "Experimental Procedures." Relative density of the major band: Control, 1.75; TNF, 10; wortmannin, 1.87; TNF ϩ wortmannin, 10.8. oligo, oligonucleotide. Rapamycin (50 nM) 9 5 Ϯ 7 9 1 Ϯ 9 LY294002 (10M) 9 3 Ϯ 4 3 0 Ϯ 8 inhibition of TNF-induced cytotoxicity and not PI3-kinase activity per se. The precise mechanism by which the cell surface receptors for any of these agents (TNF, NGF, interleukin-2) trigger the activation of PI3K is not fully understood. The increase in PI3K upon the activation of the CD40 receptor was independent of the activation of NF-B and, thus, would not seem to depend on the action of Traf2 (15). Recent work suggests that PI3K may be activated by Ras (18 -20). However, there is at present little evidence of a connection between TNF and Ras. Thus, the mechanism of PI3K activation by TNF remains enigmatic and a focus of our current efforts.
By contrast, a mechanism can be readily proposed by which the phosphorylation of BAD promotes the survival of cells exposed to TNF. We have shown previously that the cytotoxicity of TNF depends on induction of the mitochondrial permeability transition (MPT) (21). Thus, we can rephrase our concern as the mechanism by which BAD phosphorylation prevents induction of the MPT in cells exposed to TNF. The overexpression of the proapoptotic protein Bax killed Jurkat cells as a consequence of induction of the MPT (22), and purified Bax induced the MPT in isolated mitochondria in vitro. 2 Bax, Bcl-X L , and Bcl-2 are present constitutively in HeLa cells. 3 Bax is located in the cytosol and upon the initiation of the apoptotic process, has been reported to translocate to the mitochondria (23,24). We hypothesize that upon treatment with TNF the phosphorylation of BAD prevents its interaction with Bcl-X L or Bcl-2, thereby allowing these proteins to complex with Bax (Fig. 6). When Bax is bound to either Bcl-X L or Bcl-2, it would be incapable of inducing the MPT and, thus, of killing the cells. Conversely, when the phosphorylation of BAD is inhibited, the continued interaction of BAD with either Bcl-X L or Bcl-2 prevents the interaction of these proteins with Bax. Bax then moves to the mitochondria and induces the MPT, a result that causes the death of the cells. In this way, there is a kind of apoptotic thermostat that determines the fate of the cell. Cell survival or cell death is determined by the relative level of BAD phosphorylation and, hence, by the amount of Bcl-X L or Bcl-2 that is available at the mitochondria to bind and neutralize the proapoptotic protein Bax, thereby inhibiting mitochondrial dysfunction and the initiation of an apoptotic cascade. constitutively active Akt inhibits TNF-induced cytotoxicity HeLa cells were cotransfected with 0.5 g of pCDNA-LacZ and 5 g of one of the indicated constructs. 48 h post-transfection, cells were either left untreated or treated with TNF alone or TNF and wortmannin for 6 h. Cells were then stained for ␤-galactosidase, and the number of blue cells was counted. Results are the mean ϮS.D. from two experiments.