Growth Factors Inactivate the Cell Death Promoter BAD by Phosphorylation of Its BH3 Domain on Ser155 *

The Bcl-2 family protein BAD promotes apoptosis by binding through its BH3 domain to Bcl-xL and related cell death suppressors. When BAD is phosphorylated on either Ser112 or Ser136, it forms a complex with 14-3-3 in the cytosol and no longer interacts with Bcl-xLat the mitochondria. Here we show that phosphorylation of a distinct site Ser155, which is at the center of the BAD BH3 domain, directly suppressed the pro-apoptotic function of BAD by eliminating its affinity for Bcl-xL. Protein kinase A functioned as a BAD Ser155 kinase both in vitro and in cells. BAD Ser155 was found to be a major site of phosphorylation induced following stimulation by growth factors and prevented by protein kinase A inhibitors but not by inhibitors of the phosphatidylinositol 3-kinase/Akt pathway. Growth factors inhibited BAD-induced apoptosis in both a Ser112/Ser136- and a Ser155-dependent fashion. Thus, growth factors engage an anti-apoptotic signaling pathway that inactivates BAD by direct modification of its BH3 cell death effector domain.

Bcl-2 family proteins are important regulators of apoptosis that function during development and other physiological processes but also contribute to pathological conditions associated with inappropriate cell survival such as cancer (1,2). This still expanding family contains both pro-and anti-apoptotic members characterized by the presence of one or more Bcl-2 homology (BH) 1 domains. Of these domains, BH3 has proven to be a key element in pro-apoptotic Bcl-2 homologs that mediate both protein binding and cell death functions (1,3). In Caenorhabditis elegans, the BH3-containing protein EGL-1 functions at the most proximal point in a pathway required for all programmed cell death, and in mammalian cells BH3 proteins transduce signals from cell surface receptors to a central cell death pathway regulated by Bcl-2 (4 -6). BAD is a "BH3-only" pro-apoptotic Bcl-2 family member whose function is regulated by phosphorylation in response to survival factors such as nerve growth factor, insulin-like growth factor-1, and interleukin-3 (7)(8)(9). In its unphosphorylated state, BAD forms heterodimers with anti-apoptotic Bcl-2 homologs and promotes cell death. These activities are inhibited by phosphorylation of BAD on either of two serine residues, Ser 112 or Ser 136 (8). The serine/threonine kinase Akt that is activated by growth factors through a PI 3-kinase-dependent mechanism phosphorylates BAD on Ser 136 (9,10). The ribosomal S6 kinases and a mitochondria-localized cAMP-dependent protein kinase (PKA) have been reported to phosphorylate BAD on Ser 112 (11,12) following stimulation by growth factors and interleukin-3, respectively. When BAD is phosphorylated on these sites, it is sequestered in the cytosol in a complex with 14-3-3 and fails to interact with Bcl-x L at mitochondrial membranes (8). The present study provides genetic, biochemical, and biological evidence that growth factor-induced phosphorylation on the novel site Ser 155 , which is within the functionally critical BH3 domain, directly blocks BAD binding to Bcl-x L and may represent another major regulatory mechanism of BAD inactivation.

EXPERIMENTAL PROCEDURES
Constructs-A cDNA for murine BAD was obtained by reverse transcriptase-PCR using mRNA isolated from FL5.12 cells and the following PCR primers: 5Ј-GCCTCCAGGATCCAAGATGGGAACC-3Ј and 5Ј-GG-AGCGGGTAGAATTCCGGGATG-3Ј. When this cDNA, which encodes the 204-amino acid BAD protein (7), was cloned into pcDNA3 (Invitrogen) and was expressed in cells, the protein product was significantly larger than the endogenous BAD protein (approximately 30 kDa versus 23 kDa, respectively) detected with an anti-BAD antibody (C-20, Santa Cruz Biotechnology, Inc.). A methionine residue at position 43 of the mouse sequence corresponds to the first methionine residue of the human BAD. We generated a cDNA for the shorter form of murine BAD by using an upstream PCR primer surrounding the second methionine residue (5Ј-TGGAGACCAGGATCCCAGAGTAGCT-3Ј) and the same downstream primer as above. The expressed construct co-migrated with the endogenous BAD of FL5.12 cells, suggesting that the shorter form of BAD (162 amino acids) is the major translation product of BAD in these cells.
Serine residues 112, 136, 134, and 155 were substituted with alanine, and serine 155 was also substituted with aspartic acid using PCR-mediated mutagenesis in the context of the shorter form of BAD. (Amino acid numbering was not changed to be consistent with previous conventions; serine positions are actually 70, 94, 92, and 113, respectively.) The nucleotide sequence of BAD and its mutants was confirmed by DNA sequencing. The cDNA of BAD and its mutants were cloned into a pcDNA3-HA vector, which introduces the HA epitope at the amino terminus of BAD coding sequences. The cDNA encoding the human protein kinase inhibitor for cAMP-dependent protein kinase (PKI) was isolated from total mRNA of HeLa cells by reverse transcriptase-PCR and was inserted in frame into pcDNA3-HA.
Cell Culture and Western Blot Analysis-Rat-1, COS-7, and HeLa cells were cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum. Transfections were carried out with the Superfect reagent from Qiagen or the GenePortor transfection reagent from Gene Therapy Systems. For generation of cell lysates, cells were washed with ice-cold phosphate-buffered saline and lysed in radioimmune precipitation buffer (50 mM HEPES, pH 7.5, 1% deoxycholate acid, 1% Nonidet P-40, 0.1% SDS, 150 mM NaCl, 1 mM Na 3 (VO) 4 , 1 mM NaO 3 P 7 , 1 mM phenylmethylsulfonyl fluoride, 10 g/ml leupeptin, and 20 g/ml aprotinin). Lysates were cleared by centrifugation at 16,000 ϫ g for 15 min at 4°C. For Western blot analysis, protein samples were electrophoresed on 16 or 18% SDS-polyacryl-amide gels and transferred to nitrocellulose membranes. Blots were blocked in Western wash buffer (40 mM Tris, pH 8.0, 150 mM NaCl, 0.2% Nonidet P-40) with 5% bovine serum albumin for 1 h at room temperature. Blots were incubated with primary antibody diluted in Western wash buffer with 3% bovine serum albumin at room temperature for 1-2 h and then were washed and incubated with secondary horseradish peroxidase-coupled antibody diluted in Western wash buffer with 1.5% bovine serum albumin and washed extensively. Proteins were detected by the ECL method according to the manufacturer's instructions (Amersham Pharmacia Biotech). Anti-HA antibody (3F10) was from Roche Molecular Biochemicals. Anti-Akt (sc1618) was from Santa Cruz Biotechnology, Inc., anti-p473-Akt (9271S) was from New England Bio-Labs, and anti-activated EGF receptor (E12120) was from Transduction Laboratories.
In Vitro Kinase Assay-DNA encoding BAD and various mutants was excised from pcDNA3 as a BamHI-EcoRI fragment and inserted into the BamHI-EcoRI cloning sites of pGEX-2T (Amersham Pharmacia Biotech). GST fusion proteins were expressed in Escherichia coli strain DH5␣. Cells (500 ml) were grown to an A 595 of 0.7-0.9 at 37°C and induced with 2 mM isopropyl ␤-D-thiogalactopyranoside at 30°C overnight. Cells were collected by centrifugation and suspended in 20 ml of HBS (10 mM HEPES, pH 7.5, 3.4 mM EDTA, 150 mM NaCl) plus 1% (v/v) Triton X-100 and 10 mM ␤-mercaptoethanol. Cells were lysed by two passages through a Microfluidizer M110S (Microfluidics), and cell debris was removed by centrifugation for 30 min at 20,000 ϫ g. A 20% ammonium sulfate precipitation was performed on the cell lysate to remove aggregated protein, and GST⅐BAD was purified from the supernatant by glutathione-agarose chromatography. The fractions containing GST⅐BAD were pooled and concentrated using a Centriprep 3 (Amicon), and the protein concentration was determined by the Bradford assay (Bio-Rad). In vitro kinase assays were carried out in 30-l volumes containing 10 mM HEPES, pH 7.5, 100 mM NaCl, 12 mM MgCl 2 , 1 mM dithiothreitol, 15 Ci of [␥-32 P]ATP (6000 Ci/mmol, NEN Life Science Products), 11 units of protein kinase A (catalytic subunit from bovine heart, Calbiochem), and 1 g of purified GST⅐BAD. Reactions were incubated for 30 min at 30°C and terminated by the addition of SDS-polyacrylamide gel electrophoresis sample buffer.
Antibody Development-The anti-phospho-Ser 155 -BAD antibody was produced at Bio-Synthesis Inc. (Lewisville, TX). A phosphopeptide of the sequence NH 2 -GCQRYGRELRRMpSDESVDSF-COOH where pS is phosphoserine was synthesized at Ͼ70% purity, conjugated to keyhole limpet hemocyanin, and injected into rabbits. Immune serum (10 weeks) reacted with Ser 155 -phosphorylated BAD in Western blot analysis (1:500 dilution). This antibody reacts exclusively with the Ser 155phosphorylated form of BAD. The antibody recognized the phosphorylated (upper) band of BAD S112A/S136A from transfected cell lysates, did not react with the lower unphosphorylated band (see Fig. 2C), and failed to recognize either Ser 112 -or Ser 136 -phosphorylated BAD (data not shown).
Apoptosis Assay-HeLa cells were plated in 12-well plates at 5 ϫ 10 4 cells/well 1 day prior to transfection. Cells were transfected with BAD (or BAD mutants) and ␤-galactosidase expression plasmids in a ratio of 4:1 (0.6 g:0.15 g) using the Superfect transfection reagent (Qiagen). Twenty-four hours after transfection, cells were cultured in serum-free medium or in serum-free medium plus EGF for an additional 12 h. ␤-Galactosidase activity was measured in extracts using a fluorogenic substrate (MUG, Bio-Rad FluorAce TM ␤-galactosidase reporter assay, 170 -3150). The loss of ␤-galactosidase activity in these assays reflects apoptosis and elimination of the transfected cells, and the ␤-galactosidase reductions were reversed by the addition of a broad spectrum caspase inhibitor, Z-VAD-fluoromethylketone (where Z is benzyloxycarbonyl, data not shown).
Competition Binding Assay-Immunlon 2 (Dynatech) microtiter plates were coated with 5 g/ml neutravidin (50 l/well, Pierce) in sodium bicarbonate buffer, pH 9.0, overnight at 4°C. All remaining steps were conducted at room temperature. Plates were washed two times with phosphate-buffered saline containing 0.1% Tween 20 (wash buffer) and blocked for 1 h with 0.2 ml/well 1% normal goat serum in phosphate-buffered saline. Following two additional washes, 1.25 g/ml Bak BH3 19-mer peptide (residues 71-89) that was biotinylated at the amino terminus was added to the wells in 50 l of 10 mM HEPES buffer, pH 7.2, containing 150 mM KCl, 5 mM MgCl 2 , 1 mM EGTA, and 0.2% Nonidet P-40 (Nonidet P-40 buffer). After 30 min, the wells were washed twice with wash buffer, and GST⅐Bcl-x L (0.25 M in 50 l of Nonidet P-40 buffer) was added in the absence or presence of BAD or Bak BH3 peptides. Following a 1-h incubation, the plates were washed twice with wash buffer, and the amount of bound GST⅐Bcl-x L was determined by enzyme-linked immunosorbent assay using an anti-GST primary antibody and a horseradish peroxidase-conjugated anti-mouse IgG secondary antibody (Jackson) with ABTS (Zymed Laboratories Inc.) as substrate. Five washes were conducted following each 1-h antibody incubation. GST⅐Bcl-x L fusion protein was produced in E. coli by a similar procedure that was described for the production of GST⅐BAD (see above).

Ser 155 Is a Third Phosphorylation Site of BAD-Immunoblot
analysis of a BAD mutant that was unable to be phosphorylated on Ser 112 and Ser 136 revealed that BAD was phosphorylated on a third site. BAD S112A/S136A expressed in COS-7 cells or in stably transfected MCF-7 cells (data not shown) migrated as a doublet on a large dimension 16% SDS-polyacrylamide gel (Fig. 1A). Phosphatase treatment of the transfected cell lysate completely eliminated the upper band of this doublet, confirming the existence of a phosphorylation site on BAD distinct from both Ser 112 and Ser 136 . The functional significance was examined by testing the impact of this phosphorylation on the ability of BAD to dimerize with Bcl-x L . GST⅐Bcl-x L bound only to the faster migrating form of BAD in lysate from BAD S112A/S136A-transfected cells (Fig. 1A), indicating that phosphorylation on the site distinct from Ser 112 and Ser 136 prevents interaction with Bcl-x L .
To localize the novel phosphorylation site in BAD, selected serine residues in addition to Ser 112 and Ser 136 were substi-FIG. 1. Ser 155 is a third phosphorylation site of BAD. A, lysates of COS-7 cells transiently transfected with HA-BAD S112A/S136A were electrophoresed on a large dimension 16% SDS-polyacrylamide gel and were followed by Western blot analysis with an anti-HA antibody. Samples of cell lysate were treated with phosphatase (ϩPP) or were left untreated (ϪPP). Untreated lysates were also incubated with either GST⅐Bcl-x L (X L ) or GST (at 10 g/ml) for 2 h at 4°C. GST fusion protein complexes were captured by incubation with glutathione beads and washed with radioimmune precipitation buffer (Pull-down). B, the BAD S112A/S136A double mutant and the HA-tagged triple serine substitution mutants of BAD, S112A/S134A/S136A and S112A/S136A/S155A, were transiently expressed in COS-7 cells. For the control, cells were transfected without plasmid DNA (mock). Cell lysates were prepared (treated either with phosphatase (PP) or left untreated) and HA⅐BAD proteins were detected by Western blot analysis using anti-HA antibody. C, HA-tagged wild-type BAD (WT) and BAD S155A were transiently expressed in HeLa cells. After serum starvation for 12 h, the cells were treated with EGF (50 ng/ml for 15 min) or fetal calf serum (FCS, 20%, 15 min). HA⅐BAD and its phosphorylated forms were detected in cell lysates by anti-HA Western blot analysis. tuted with alanine. Alanine substitution of Ser 155 but not Ser 134 eliminated the slower migrating band of the doublet when expressed in COS-7 cells (BAD S112A/S136A/S155A; Fig.  1B) suggesting that Ser 155 could be a third site of phosphorylation on BAD. A single Ser 155 to alanine mutation significantly reduced the overall extent of BAD phosphorylation in transfected HeLa cells and eliminated the slowest migrating phosphorylated form of BAD observed following stimulation with serum or EGF (Fig. 1C). Phosphorylation on Ser 155 was confirmed directly by demonstrating reactivity of BAD to a phospho-Ser 155 -specific antibody (see below). Interestingly, Ser 155 is located at the center of the BH3 domain of BAD, the structural element that mediates BAD/Bcl-x L dimerization (13)(14)(15).
Protein Kinase A Is a BAD Ser 155 Kinase in Vitro and in Vivo-The sequence surrounding Ser 155 , LRRMSD, matches the consensus phosphorylation site for mammalian PKA, XR-RXSX (16). To test whether PKA can phosphorylate BAD on Ser 155 in vitro, purified PKA was incubated with GST fusion proteins of BAD or different BAD mutants in the presence of [␥-32 P]ATP. PKA did phosphorylate wild-type GST⅐BAD ( Fig.  2A) although not in the presence of a PKI fragment that is a specific PKA inhibitor (17) (data not shown). Alanine substitution of Ser 112 and Ser 136 , either alone or in combination, did not substantially reduce the level of BAD phosphorylation by PKA in vitro (Fig. 2A). In contrast, mutation of Ser 155 dramatically reduced (S155A single mutant) or completely abolished (S112A/S136A/S155A triple mutant) phosphorylation ( Fig. 2A) indicating that Ser 155 is the major phosphorylation site of PKA in vitro.
To test whether PKA may function as a Ser 155 kinase of BAD in vivo, we examined whether known activators of PKA would stimulate BAD Ser 155 phosphorylation in cells. Forskolin, which stimulates PKA through the activation of adenylate cyclase, rapidly induced a sustained gel mobility shift of BAD S112A/S136A in transfected HeLa cells that was sensitive to phosphatase treatment (Fig. 2B). In contrast, forskolin had no effect on the BAD S112A/S136A/S155A mutant, indicating that Ser 155 is the likely site of phosphorylation by a cAMP-dependent kinase in vivo (Fig. 2B). An anti-phospho-Ser 155 -specific BAD antibody (see "Experimental Procedures") reacted with the phosphorylated bands of wild-type BAD and BAD S112A/ S136A induced by forskolin in transfected HeLa cells (Fig. 2C,  upper panels). This antibody is specific for phospho-Ser 155 -BAD; it failed to react with BAD S155A or the BAD triple mutant although the expression levels of BAD and the BAD mutants were at similar levels (Fig. 2C, lower panels). These results provide direct evidence for BAD phosphorylation on Ser 155 . Because adenylate cyclase is a target of activated G protein-coupled receptors (18), we also examined whether BAD Ser 155 phosphorylation would respond to ligands of G proteincoupled receptors. Of the ligands tested in HeLa cells, L-epinephrine rapidly induced a gel mobility shift of BAD S112A/ FIG. 3. Growth factors induce phosphorylation on BAD Ser 155 by a PKA-dependent mechanism, independent of PI 3-kinase/ Akt. A, HeLa cells were co-transfected with HA⅐BAD S112A/S136A and either pcDNA3 (ϩvector) or pcDNA3-HA-PKI (ϩPKI) at a ratio of 1:5. Cells were serum-starved for 12 h and were followed by treatment with EGF (50 ng/ml, 5 min). Cell lysates were prepared and analyzed with an anti-HA antibody. B, HeLa cells were transfected with HA⅐BAD S112A/ S136A and were serum-starved for 12 h. Cells were then pretreated with AG1478 (AG, 30 nM, 30 min), wortmannin (Wm, 100 nM, 15 min), or H89 (20 M, 60 min) followed by stimulation with EGF (50 ng/ml, 5 min). Western blot analysis of cell lysates was performed with antibodies against activated EGF receptor (EGFR), activated Akt (pS473 Akt followed by a reblot with anti-total Akt), and pS155BAD (followed by a reblot with anti-BAD). C, Rat-1 cells were serum-starved for 48 h and stimulated with platelet-derived growth factor (25 ng/ml, 15 min). Endogenous BAD was immunoprecipitated using an anti-BAD antibody (C-20) and subjected to anti-phospho-Ser 155 BAD (anti-pS155) Western blot analysis (left). As a control, the anti-BAD antibody was incubated with beads without cell lysates, and the eluate was subjected to the same anti-phospho-Ser 155 BAD analysis. The same blot was stripped and reprobed with an anti-BAD antibody (right). D, Rat-1 cells were serum-starved and stimulated with PDGF as in C with or without pretreatment with the PKA inhibitor H89 (20 M, 60 min) or the PI 3-kinase inhibitor wortmannin (100 nM, 15 min). Endogenous BAD was immunoprecipitated and blotted with the anti-phospho-Ser 155 BAD antibody.

FIG. 2. Protein kinase A is a BAD Ser 155 kinase in vitro and in vivo.
A, purified GST⅐BAD or the indicated GST⅐BAD mutant fusion proteins were incubated with [␥-32 P]ATP and PKA in vitro. Reaction samples were electrophoresed on a 4 -20% SDS-polyacrylamide gel and transferred to nitrocellulose, and 32 P-labeled GST⅐BAD proteins were detected by autoradiography (top). The nitrocellulose membrane was then subjected to Western blot analysis with anti-BAD antibody (bottom) to verify that equal amounts of GST⅐BAD substrates were present. B, HeLa cells transfected with HA⅐BAD S112A/S136A double mutant or S112A/S136A/S155A triple mutant were either untreated or stimulated with forskolin (ϩFK, 20 M, 1 min). Cell lysates were prepared and forskolin-treated BAD S112A/S136A-transfected cell lysates were further treated with phosphatase (ϩPP). HA⅐BAD was detected by Western blot analysis with an anti-HA antibody. C, HeLa cells were transiently transfected with wild-type HA⅐BAD (WT), HA⅐BAD S155A, HA⅐BAD S112A/S136A, or HA⅐BAD S112A/S136A/S155A and were treated with forskolin as described in B. Cell lysates were prepared and subjected to Western blot analysis with an antibody developed against phospho-Ser 155 of BAD (upper panels) followed by reprobing with an anti-BAD antibody (lower panels). D, HeLa cells transfected with HA⅐BAD S112A/S136A were treated at 24 h post-transfection with L-epinephrine (L-Epi., 20 M) for the times indicated. HA⅐BAD S112A/ S136A was detected in cell lysates by Western analysis with an anti-HA antibody. S136A (Fig. 2D) that was blocked by mutation of BAD Ser 155 to alanine or by the co-expression of PKI (17) (data not shown). Together, these results demonstrate that BAD Ser 155 can be phosphorylated in cells either by PKA directly or through a PKA-dependent pathway.
Growth Factors Induce Phosphorylation of BAD Ser 155 by a PKA-dependent Mechanism That Is Independent of PI 3-Kinase/Akt-Growth factors suppress the pro-apoptotic function of BAD in part through the activation of Akt, which functions as a BAD Ser 136 kinase (9,10). We examined whether growth factor stimulation would also induce Ser 155 phosphorylation of BAD. EGF stimulated the phosphorylation of BAD on Ser 155 in transfected HeLa cells, which was blocked by co-expression of PKI (Fig. 3A). To test whether phosphorylation of Ser 155 is likewise dependent on the PI 3-kinase/Akt pathway, we treated transfected HeLa cells with wortmannin, an inhibitor of PI 3-kinase, and analyzed BAD Ser 155 phosphorylation following EGF stimulation. Wortmannin treatment prevented the activation of endogenous Akt by PI 3-kinase following EGF stimulation but did not impair EGF-induced phosphorylation of BAD on Ser 155 (Fig. 3B). In parallel assays, BAD Ser 155 phosphorylation was prevented by the addition of the PKA inhibitor H89 without affecting the phosphorylation of Akt. The EGF receptor kinase inhibitor AG1478 blocked the phosphorylation of both Akt and BAD Ser 155 . These results demonstrate that EGF stimulates the phosphorylation of BAD Ser 155 through a PKA-dependent mechanism that is distinct from the PI 3-ki-nase/Akt pathway.
To examine whether endogenous BAD is phosphorylated on Ser 155 in response to growth factors, serum-starved Rat-1 fibroblasts were stimulated with PDGF, endogenous BAD was immunoprecipitated, and Ser 155 phosphorylation was analyzed by Western blot with the anti-phospho-Ser 155 antibody. PDGF induced the phosphorylation of endogenous BAD on Ser 155 , which was significantly reduced by pretreatment of cells with the PKA inhibitor H89 but not the PI 3-kinase inhibitor wortmannin (Fig. 3, C and D).
Phosphorylation of BAD on Ser 155 Correlates with Cell Survival-The pro-apoptotic function of BAD requires its ability to heterodimerize with Bcl-x L and related proteins (13)(14)(15). Thus, the phosphorylation of Ser 155 in response to growth/survival factors (Fig. 3) and its apparent inhibitory effect on binding to Bcl-x L (Fig. 1A) suggest that it may suppress the pro-apoptotic activity of BAD. In keeping with this possibility, mutation of BAD Ser 155 to a residue that cannot be phosphorylated enhanced the proapoptotic function of BAD in transient transfection assays. BAD S155A showed a modest but highly reproducible enhancement of toxicity compared with wild-type BAD, and the S112A/S136A/ S155A triple mutant consistently exhibited greater toxicity than the S112A/S136A double mutant (Fig. 4A, open bars). Western blot analysis demonstrated that the enhanced toxicity of these mutants could not be attributed to elevated levels of protein expression in these assays (Fig. 4A, inset).
EGF stimulation, which induces BAD Ser 155 phosphoryla- FIG. 4. Phosphorylation of BAD on Ser 155 correlates with cell survival. A, HeLa cells were co-transfected with ␤-galactosidase and either wild-type (WT) BAD or indicated BAD mutants (BAD S112A/S136A/S155A, AAA), incubated overnight, and then cultured in either serum-free medium (SFM) or serum-free medium and EGF (SFM ϩ EGF, 50 ng/ml) for 12 h. Cell lysates were prepared, and BAD-induced cell death was detected by the loss of ␤-galactosidase activity in a fluorescence-based assay. Results shown are the average and standard deviation of triplicate transfections. (This experiment was repeated two more times with similar results.) Expression levels of HA-tagged BAD and BAD mutants in transfected cell lysates were detected by anti-HA Western blot analysis. B, HeLa cells were co-transfected with ␤-galactosidase and either the control vector pcDNA3-HA (CTL) or wildtype BAD or indicated BAD mutants, and BAD-induced cell death was detected as described in A.
tion, suppressed the pro-apoptotic function of BAD in these assays. Mutation of Ser 155 to alanine alone did not eliminate the protective effect of EGF consistent with the ability of EGF to inactivate BAD through the phosphorylation of Ser 112 and/or Ser 136 . However, the pro-apoptotic activity of a BAD S112A/ S136A double mutant was also suppressed by EGF stimulation revealing the ability of EGF to inactivate BAD through a distinct mechanism. Our data indicate that this mechanism involves Ser 155 because mutation of Ser 155 to alanine in the context of S112A/S136A completely inhibited the anti-apoptotic effect of EGF stimulation (S112A/S136A/S155A triple mutant, Fig. 4A, AAA, filled bars). The ␤-galactosidase reporter assay was valid as a measurement of apoptosis because BAD-induced reduction of ␤-galactosidase activity was reversed by the treatment of cells with Z-VAD, a broad spectrum caspase inhibitor (data not shown). These results indicate that the anti-apoptotic effects of EGF can be mediated through Ser 155 phosphorylation independently of Ser 112 /Ser 136 phosphorylation and are consistent with the distinct EGF-activated signaling pathways that lead to the phosphorylation of BAD on Ser 136 and on Ser 155 (Fig. 3, B and D).
To further evaluate the impact of Ser 155 phosphorylation on the pro-apoptotic activity of BAD, we substituted Ser 155 with aspartic acid (S155D) to mimic the negatively charged phospho-Ser 155 residue. Upon expression in cells, BAD S155D showed no pro-apoptotic activity compared with wild-type BAD (Fig. 4B). Consistent with this result, the BAD S155D mutant did not associate with co-expressed Bcl-x L in a co-immunoprecipitation experiment although wild-type BAD and BAD S155A did (data not shown). These results further support the hypoth-esis that phosphorylation of Ser 155 leads to the functional inactivation of BAD in cells.
Phosphorylation of BAD BH3 Domain on Ser 155 Directly Blocks BAD/Bcl-x L Heterodimerization-Ser 155 within a PKA consensus site is evolutionarily conserved at the center of the BH3 domain of BAD whereas BH3 domains of other known pro-apoptotic Bcl-2 homologs contain glycine at this position (Fig. 5A). NMR studies have revealed that BH3 forms an ␣-helical structure, which binds to a hydrophobic cleft on the surface of Bcl-x L (19). We tested whether the addition of a phosphate group on BAD Ser 155 interferes with this interaction by measuring the affinity of synthetic BAD BH3 peptides for Bcl-x L in an in vitro competition binding assay. A BH3 peptide encompassing BAD residues 143-168 bound to Bcl-x L with an affinity similar to a Bak BH3 control peptide. However, the affinity of a matched peptide that was phosphorylated on Ser 155 was reduced by greater than 100-fold (Fig. 5B). Moreover, in the context of full-length BAD produced by in vitro translation phosphorylation by PKA induced a gel mobility shift of wildtype (but not S155A mutant) BAD and blocked the ability of BAD to bind to Bcl-x L (Fig. 5C). These results demonstrate that phosphorylation on Ser 155 is sufficient to directly inactivate the heterodimerization function of BH3 and provide a biochemical mechanism for how the pro-apoptotic function of BAD is suppressed by Ser 155 phosphorylation. DISCUSSION Mutational studies have provided strong evidence that BAD promotes apoptosis by dimerizing with Bcl-x L and related proteins (13), emphasizing the importance of understanding how FIG. 5. Phosphorylation of the BAD BH3 domain on Ser 155 directly blocks BAD/Bcl-x L heterodimerization. A, the structure of murine BAD is shown schematically with BH3, and the sites of serine phosphorylation are indicated (top). The BH3 domain sequences from BAD orthologs are shown in alignment with BH3 domains from other known pro-apoptotic Bcl-2 family proteins. The sequence of zebra fish BAD was deduced from an expressed sequence tag clone (GenBank™ accession no. AI332008). B, synthetic BAD BH3 peptides were incubated at the indicated concentrations with GST⅐Bcl-x L . The binding of BH3 peptides to Bcl-x L was measured by the ability of the peptides to block the subsequent binding of GST⅐Bcl-x L to a Bak BH3 peptide immobilized on a microtiter plate as detected by enzyme-linked immunosorbent assay (triplicate samples). Bars indicate mean Ϯ S.D. BAD peptides corresponded to residues 143-168 that were either unphosphorylated (BADBH3) or phosphorylated on Ser 155 (BADBH3-P). A Bak BH3 peptide (residues 71-89) was used as a positive control for competition binding. C, [ 35 S]methionine-labeled wild-type BAD (WT) or BAD S155A was produced by in vitro translation (IVT), and incubated with PKA as described in the legend to Fig. 2 except that unlabeled ATP (200 M) was used. Following the kinase reaction, aliquots were incubated for 1 h with either GST or GST⅐Bcl-x L (1 M), and protein complexes were captured on glutathione-agarose beads. Proteins bound to beads (Bound) and samples of reactions prior to incubation with beads (Total) were analyzed by SDS-polyacrylamide gel electrophoresis followed by autoradiography. phosphorylation regulates this interaction. Our findings suggest that phosphorylation on different sites within BAD have mechanistically distinct consequences; Ser 155 phosphorylation directly prevents heterodimerization by abolishing the affinity of BAD BH3 for Bcl-x L whereas phosphorylation on Ser 112 and Ser 136 that are located outside the BH3 domain may inhibit dimerization with Bcl-x L indirectly. In particular, phosphorylation on Ser 112 or Ser 136 , but not Ser 155 , generates a consensus binding site for the cytosolic protein 14-3-3 that may bind and alter the subcellular distribution of BAD to prevent interaction with Bcl-x L at mitochondrial membranes (8).
Phosphorylation of BAD provides an important link between extracellular survival factors and the intrinsic cell death pathway regulated by Bcl-2. The prevailing model for anti-apoptotic signaling by growth/survival factors emphasizes a pathway involving the PI 3-kinase-dependent activation of Akt, phosphorylation of BAD on Ser 136 , and subsequent dissociation from Bcl-x L . Our findings identify a distinct anti-apoptotic pathway that leads to the biochemical inactivation of the BAD BH3 domain through a PKA-dependent but PI 3-kinase/Aktindependent mechanism. PKA-mediated phosphorylation of BAD Ser 155 may explain, at least in part, the anti-apoptotic effects of PKA activation that have been observed in some settings (20 -22) and PI 3-kinase-independent survival signals triggered by growth factors (23,24). Precisely how growth factors activate PKA in cells is not clear although there have been reports showing that PDGF activates PKA by stimulating its release from cell membrane (25) and that EGF activates PKA via Grb-2-mediated recruitment of PKA to the EGF receptor (26). Mitochondrial membrane-localized PKA was identified as an interleukin-3-induced kinase of BAD Ser 112 (12). This is a paradox because our results indicate that Ser 112 is a poor substrate for PKA relative to Ser 155 both in vitro and in cells. It is possible that Ser 155 may prove to be an important target for the mitochondria-localized PKA because the BAD substrate used in that study was a mutant lacking Ser 155 (12).
The convergence of multiple growth factor-stimulated pathways on BAD underscores the important role of this BH3 protein in anti-apoptotic signaling by cell surface receptors. It has been generally proposed that BAD kinases modify the unphosphorylated form of BAD bound to Bcl-x L at mitochondrial membranes, thereby releasing free active Bcl-x L . Both PKA and Akt, however, are unable to phosphorylate BAD in vitro on residues Ser 155 and Ser 136 , respectively, when BAD is in a pre-existing complex with Bcl-x L . 2 Although substrate accessibility may be different in cells, this finding suggests that PKA and Akt may inactivate BAD not by dissociating existing BAD⅐Bcl-x L complexes but rather by preventing its accumulation in a dephosphorylated "active" state. The relative significance of PKA-and Akt-mediated pathways to the inactivation of BAD in different settings remains to be determined. In tumor cell lines, our preliminary results indicate that basal BAD Ser 155 phosphorylation is higher than in non-transformed cells and is more refractory to dephosphorylation upon growth factor deprivation. 3 This raises the possibility that the inactivation of BAD BH3 through Ser 155 phosphorylation makes an important contribution to the suppression of apoptosis in cancer cells.