Regulation of the Pro-apoptotic Scaffolding Protein POSH by Akt*

POSH (Plenty of SH3 domains) binds to activated Rac and promotes apoptosis by acting as a scaffold to assemble a signal transduction pathway leading from Rac to JNK activation. Overexpression of POSH induces apoptosis in a variety of cell types, but apoptosis can be prevented by co-expressing the pro-survival protein kinase Akt. We report here that POSH is a direct substrate for phosphorylation by Akt in vivo and in vitro, and we identify a major site of Akt phosphorylation as serine 304 of POSH, which lies within the Rac-binding domain. We further show that phosphorylation of POSH results in a decreased ability to bind activated Rac, as does phosphomimetic S304D and S304E mutation of POSH. S304D mutant POSH also shows a strongly reduced ability to induce apoptosis. These findings identify a novel mechanism by which Akt promotes cell survival.

POSH 3 (Plenty of SH3 domains) is a recently discovered proapoptotic protein that appears to be widely expressed in multiple cell types, although at low levels. POSH was first identified as a binding partner of activated Rac and has been shown to act as a scaffolding protein in a kinase cascade signaling pathway that leads to apoptotic cell death (1,2). In this pathway, Rac activates one of the mixed lineage kinases (MLKs, a group of MAPKKKs), which in turn phosphorylate and activate MKK4 and/or MKK7 (which are MAPKKs) which then phosphorylate and activate c-Jun N-terminal kinases (JNKs, one group of MAPKs) (2). Activated JNKs induce release of cytochrome c from mitochondria and trigger subsequent apoptosis. POSH directly binds Rac, MLK, and another scaffold protein, JIP (JNK-interacting protein), which in turn binds MKK4/7 and JNK, to facilitate this pathway. This multiprotein signaling assembly has been termed PJAC, for POSH-JIP apoptotic complex (3). The role of POSH as a scaffold for this signaling complex appears to be critical; in an apoptotic model involving withdrawal of nerve growth factor from cultures of neuronally differentiated PC12 cells or rat primary sympathetic neurons, apoptosis was dramatically reduced by pretreatment with POSH short interfering RNA or antisense oligonucleotides (4).
The decision of a cell to undergo apoptosis is not undertaken lightly; apoptotic pathways are subject to regulation at many levels, and cells must integrate a variety of pro-apoptotic and anti-apoptotic signals. Regulation of the PJAC apoptotic pathway appears to follow this pattern, with multiple regulatory interactions. One form of regulation of PJAC may lie in the expression level of POSH protein within cells. POSH is maintained within healthy cells at very low levels, at least in part by POSH auto-ubiquitination and proteasomal degradation (5). Increasing the level of POSH protein by microinjection or ectopic expression induces apoptosis in a variety of cell types (1, 2, 4, 6 -9).
Another potential regulator of the POSH-JIP apoptotic complex appears to be the pro-survival kinase Akt, also known as protein kinase B. Three closely related Akt genes exist (AKT1-3) that have both significant functional overlap as well as specific unique functions (10,11). Akt activation can protect cells from a variety of pro-apoptotic signals, and Akt can phosphorylate an increasing number of intracellular targets (12)(13)(14)(15)(16)(17)(18)(19)(20)(21)(22). Recently a direct binding interaction between POSH and Akt2 was described, in which binding of Akt2 to POSH prevents MLK3 binding and thereby inhibits PJAC assembly and apoptotic signaling by JNK. This binding interaction was found to be specific to Akt2; Akt1 showed no such binding to POSH (23). However, despite this isoform-specific interaction of Akt2 and POSH, activated Akt1 is fully able to protect cells from apoptosis induced by overexpression of POSH (1,2). Thus, the precise physiologic importance of this isoform-specific interaction is unclear. It is also unclear whether Akt in general opposes POSH-induced apoptosis at the level of the PJAC complex or at more downstream sites. Furthermore, it has yet to be addressed whether Akt might regulate PJAC function by directly phosphorylating POSH and, if so, how this might affect POSH function. These questions are the subject of this study.

MATERIALS AND METHODS
Cell Culture-HEK293T cells were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, 100 units/ml penicillin, and 100 g/ml streptomycin. MDA-MB-231 cells were grown in the same media, additionally supplemented with 1 mM L-glutamine, minimum Eagle's medium nonessential amino acids, and 10 ng/ml insulin. Fetal bovine serum was obtained from HyClone (Logan, UT). All other media components were obtained from Invitrogen.
DNA Constructs and GST-POSH Fusion Proteins-Human POSH cDNA was cloned by RT-PCR as follows. RNA was isolated from HEK293 cells using Trizol, as described by the manufacturer (Invitrogen), and used to prepare first strand cDNA using murine leukemia virus reverse transcriptase (PerkinElmer Life Sciences) with oligo(dT) as primer. This cDNA was then used as template in two different PCRs to separately amplify the 5Ј-and 3Ј-halves of POSH, which were subsequently reassembled using the internal BamHI site. The 5Ј-half of POSH was amplified by PCR using the primers POSH-5Ј-R1 (GGCG GAA TTC ACT AGT ACC ATG GAT GAA TCA GCC TTG) and POSH-(1303-1326) (CCG TAA ATG TGC TTA CTG GTC AGT). The product of this reaction contains engineered EcoRI and SpeI sites at the 5Ј-end and included the endogenous BamHI site. The 3Ј-half of POSH was amplified using POSH-5Ј-(1273-1296) (ATG GGA CCG AGG CCC ATG GCA) and POSH-myc-Not (GCG TGC GGC CGC TCA CAG GTC CTC CTC GCT GAT) to generate a product containing the endogenous BamHI site at its 5Ј-end and extending to include the remaining coding sequence, followed by an engineered Myc epitope tag, and a NotI site. The two POSH fragments were then subcloned and reassembled to generate a full-length Myc-tagged cDNA. The resulting cDNA was entirely sequenced and confirmed to be in agreement with GenBank TM sequences for human POSH.
To generate GST-POSH fusion proteins in mammalian cells, the Myc-tagged POSH cDNA described above was subcloned into the vector pEBG-SrfI (a generous gift from Yusen Liu (24)), using SpeI and NotI sites, to generate pEBG-POSH.
To construct GFP-POSH, this same EcoRI, NotI fragment was first cloned into pBS-KS(Ϫ) and then cut back out with EcoRI and SacII. This EcoRI, SacII fragment was then cloned into pEGFP-C1 (Clontech) to generate pGFP-POSH.
pcDNA3.1 myr-Akt1 was generously provided by Dr. Philip Tschilis. pcDNA3.1 myr-Akt2 was generated in our laboratory as follows. Human Akt2 cDNA was cloned by RT-PCR using cDNA prepared from HEK293 cells as described above and the primers myr-Akt2-HA F primer (ATC GAT AAG CTT ATG GGT TCT TCT AAA TCT AAA CCT AAA AAT GAG GTG TCT GTC ATC AAA GAA GGC) and myr-Akt2-HA-R (GAT ATC ACT AGT TCA GGC ATA GTC GGG CAC GTC ATA GGG ATA CTC GCG GAT GCT GGC CGA). The product of this reaction encodes a protein with a myristoylation sequence at the N terminus and a hemagglutinin epitope tag at the C terminus and was subcloned into the vector pCDNA3.1(ϩ), using engineered HindIII and EcoRV sites. The resulting plasmid was sequenced and confirmed to be in agreement with GenBank TM sequences for human Akt2.
Microinjections-MDA-MB-231 cells were counted and plated at a density of 1 ϫ 10 5 cells per 35-mm dish the day before micro-injections were to take place. The cells were injected with a control vector encoding GFP (pEGFP-C1, Clontech) or pGFP-POSH, plus or minus pcDNA3.1-myr-Akt1 or myr-Akt2. In some experiments separate vectors encoding GFP (pEGFP-C1) and POSH (pcDNA-POSH), rather than a GFP-POSH fusion protein, were employed and gave similar results. The concentration of each DNA was 0.25 g/l. Green fluorescent protein (GFP)-positive cells were counted 4 h post-injection. 18 h following injection GFP-positive cells were again examined, assessed for viability/apoptotic changes, and quantitated. Cells were visualized using an Axiovert 200 MOT microscope (Carl Zeiss, Inc., Thornwood, NY). Fluorescence images were captured using a Hamamatsu ORCA ER CCD camera run by Openlab (Improvision, Lexington, MA).
Transfections and Western Immunoblotting-Transfections were performed by the calcium phosphate method, as described by Xu and Greene (26), with HEK293T cells plated at 30,000/cm 2 in 10-cm dishes 20 -24 h prior to transfection, and fresh growth medium was added 3-4 h before transfection. Transfected cells were lysed 18 -24 h later in RIPA buffer (150 mM NaCl, 50 mM Tris, pH 7.4, 50 mM NaF, 2 mM EDTA, 1 mM Na 3 VO 4 , 1.0% Triton X-100, 0.5% deoxycholate, 0.1% SDS) supplemented with protease inhibitors (catalog number 1 697 498, Roche Applied Science). Lysates were clarified by centrifugation for 30 min at 13,000 ϫ g at 4°C, and protein concentrations were determined by the Bradford assay (Pierce). 1 mg of lysate protein was incubated with glutathione-Sepharose beads in binding buffer (50 mM NaCl, 50 mM NaF, 10 mM Tris, pH 7.4, 5 mM EDTA, 1 mM Na 3 VO 4 ) overnight at 4°C with rocking to recover GST-POSH. The beads were pelleted and washed three times with binding buffer. The proteins were then solubilized in SDS sample buffer and separated by SDS-PAGE, followed by transfer to Immobilon-P (Millipore, Bedford, MA) for immunoblotting. Antibodies used in immunoblots are as follows: anti-phospho-Akt substrate antibody (PAS) was obtained from Cell Signaling Technologies, anti-GST antibody was from Santa Cruz Biotechnology (Santa Cruz, CA), monoclonal anti-Rac antibody was from Sigma, and anti-POSH antibody was a generous gift of Dr. Yuval Reiss (Proteologics, Rehovot, Israel). Horseradish peroxidase-labeled secondary antibodies were obtained from Bio-Rad and visualized by chemiluminescence (Western Lightning, PerkinElmer Life Sciences).
In Vitro Kinase Reactions-1 g of purified GST or GST-POSH protein was incubated with 20 ng of active Akt1 (Upstate catalog number 14-276, Millipore) or Akt2 (Upstate catalog number 14-447) in 40 l of a kinase buffer consisting of 50 mM NaCl, 50 mM Tris, pH 7.4, 10 mM MgCl 2 , 0.1 mM EGTA, 0.1% ␤-mercaptoethanol, 0.01% Brij-35, and 0.25 mM ATP. For experiments involving 32 P, ATP was reduced to 0.1 mM, and 1 Ci of [␥-32 P]ATP was added. The reactions were incubated at 30°C for 1 h and terminated by the addition of SDS sample buffer, followed by boiling.
Phosphopeptide Analysis by Mass Spectrometry-5 g of GST-POSH-(53-888) was phosphorylated in vitro with either Akt1 or Akt2 as described above. After the reaction GST-POSH was recovered on glutathione-Sepharose beads and washed twice in 100 mM NaCl, 20 mM Tris, pH 8.0, 1 mM CaCl 2 , 0.1% Thesit, followed by one wash in 100 mM NaCl, 20 mM Tris, pH 8.0, 1 mM CaCl 2 , 5 mM DTT. Bound proteins were then denatured and reduced by boiling for 5 min in the above buffer containing 1% SDS. Samples were cooled and alkylated by the addition of iodoacetamide to a final concentration of 10 mM for 30 min in the dark at room temperature. Samples were then diluted 5-fold with 100 mM NaCl, 20 mM Tris, pH 8.0, 1 mM CaCl 2 , 2% Thesit and digested with 0.5 g of trypsin (sequencing grade, Promega, Madison, WI) overnight at 37°C. Acetic acid was then added to 0.25 M followed by acetonitrile to 30%, and the tryptic digests were incubated with 10 l of immobilized Fe 3ϩ resin (PHOS-Select, Sigma) for 1 h at room temperature with rotation. Fe 3ϩ beads were recovered by centrifugation, washed three times with 250 mM acetic acid, 30% acetonitrile, once with H 2 O, and then eluted with 0.4 M NH 4 OH. Eluted peptides were analyzed by high resolution MALDI-TOF mass spectrometry performed on a Bruker Ultraflex MALDI-TOF/TOF instrument.
For identification of minor phosphorylation sites, 10 g of S304A-GST-POSH-(1-888) was phosphorylated in vitro with either Akt1, Akt2, or no added kinase at 30°C for 2 h, and the reaction products were subjected to SDS-PAGE and Coomassie staining. The bands containing GST-POSH were excised and digested in-gel with trypsin, and phosphopeptides were recovered by binding to a titanium dioxide matrix, as directed by the manufacturer (Phos-trap, PerkinElmer Life Sciences). Eluted peptides were analyzed by MALDI-TOF mass spectrometry, as above.
Rac Binding-35 S-Labeled L61Rac was prepared by in vitro transcription and translation using [ 35 S]methionine (PerkinElmer Life Sciences) and the TNT T7 Quick Coupled Transcription/Translation System as described by the manufacturer (Promega), and used without further purification.
For the phosphorylation time course experiment, kinase reactions were performed in duplicate containing 2 g of purified GST-POSH-(53-888) protein and 40 ng of active Akt in 20 l of 50 mM KCl, 10 mM HEPES, pH 7.4, 5 mM MgCl 2 , 1 mM DTT, and 1 mM ATP. The reactions were set up on ice and transferred to 30°C for varying times (0 -60 min), following which the reactions were returned to ice, and 1 unit of apyrase was added to remove residual ATP. Glutathione-Sepharose (30 l of 50% slurry in phosphate-buffered saline, 1% BSA, 0.1% Thesit) and 35 S-labeled L61 Rac (1 l, diluted 1:10 in phosphate-buffered saline, 0.1% Thesit) were then added, and the reactions were rotated at 4°C for 2 h to allow protein binding. The samples were then centrifuged, and the Sepharose pellets were washed four times with ice-cold phosphate-buffered saline, 0.1% Thesit, to remove unbound protein. Washed samples were then resuspended in SDS sample buffer, followed by boiling. Aliquots of each sample were analyzed by scintillation counting, SDS-PAGE followed by Coomassie staining and autoradiography, and SDS-PAGE followed by immunoblotting with phospho-Akt substrate antibody (Cell Signaling Technology, Danvers, MA).
For the experiments with Ser-304 mutants, binding reactions containing 5 g of GST or full-length GST-POSH and 10 g of BSA in 55 l of 150 mM NaCl, 20 mM Tris, pH 7.4, 1 mM DTT, and 0.1% Thesit were assembled on ice, and then 35 S-L61-Rac and glutathione-agarose were added as above. Binding, washing, and further analysis were as above.
Caspase Assays-24 h prior to transfection, MDA-MB-231 cells were plated at 20,000 cells/well in 96-well plates. The next day cells were transfected with pEBG or pEBG-POSH-S304 mutants using Superfect (Qiagen, Valencia, CA) according to manufacturer's protocol. 24 h post-transfection, cells were lysed and assayed for caspase activity by the addition of 100 l of Caspase-Glo 3/7 reagent to each well according to the manufacturer's protocol (Promega, Madison, WI). Reactions were done in triplicate, including no treatment, transfection vehicle alone, and vector alone controls. The contents of the wells were mixed using a plate shaker for 30 s and incubated at room temperature for 1 h. Luminescence was measured using a luminometer.

RESULTS
Akt Prevents POSH-induced Apoptosis-Overexpression of POSH induces apoptosis in a variety of different cell types, including Swiss 3T3 and NIH 3T3 fibroblasts, neuronally differentiated PC12 cells, primary rat sympathetic neurons, and HEK293 cells (1, 2, 4, 6 -9). Here we examined the effects of Phosphorylation of POSH by Akt JULY 27, 2007 • VOLUME 282 • NUMBER 30 POSH overexpression in an additional cell type, MDA-MB-231 human breast cancer cells. These cells were chosen because they have low levels of endogenous activated Akt, 4 and we wished to examine POSH-induced death in the absence of activated Akt. Preliminary experiments indicated that transfection of POSH or a GFP-POSH fusion construct induced rapid cell death in MDA-MB-231 cells (not shown). In Fig. 1B we quantitate this effect by micro-injecting cells with an expression vector encoding either GFP or GFP-POSH and monitoring subsequent cell survival at 4 and 18 h after injection. Greater than 90% of cells injected with GFP-POSH undergo cell death by 18 h. As seen in other cell types, POSH-induced cell death in MDA-MB-231 cells appears to occur by apoptosis, as evidenced by cell rounding, increased refractivity, nuclear condensation, surface blebbing, and detachment from the plate (see Fig. 1A), whereas control cells injected with GFP alone do not show such morphological changes.
However, when these cells are injected with both GFP-POSH and activated forms of either Akt-1 or Akt-2, apoptosis is dramatically reduced. These constitutively active Akt forms contain a myristoylation sequence (abbreviated as myr-Akt), which directs them to the plasma membrane and results in their activation. Similar results demonstrating rescue of POSH overexpressing neuronal PC12 cells by co-expression of myr-Akt-1 have also been reported (2).
Akt Directly Phosphorylates POSH-Akt interacts with several members of the PJAC complex to exert a variety of antiapoptotic effects. We wanted to ask whether Akt might directly phosphorylate POSH. We examine this question in Fig. 2. In Fig. 2A HEK293T cells were transfected with pEBG-POSH, encoding a GST-POSH fusion protein, plus and minus an expression vector for myr-Akt1. Cells were treated with the caspase inhibitor benzyloxycarbonyl-VAD to prevent apoptosis. After 24 h the cells were lysed, and GST-POSH was recovered on glutathione-agarose. Analysis of recovered protein by Western blot using antibodies to GST shows similar levels of GST-POSH protein expression with and without Akt ( Fig. 2A,  lower panel, compare lanes 2 and 3). To examine possible POSH phosphorylation, we made use of a commercially available phospho-specific antibody that recognizes a variety of Akt substrates in their phosphorylated forms (PAS; Cell Signaling Technology). This PAS antibody shows strong reactivity with GST-POSH precipitated from cells co-expressing myr-Akt1, as compared with cells expressing GST-POSH alone ( Fig. 2A,  upper panel, compare lanes 2 and 3), indicating that expression of active Akt1 results in POSH phosphorylation in vivo.
To confirm that POSH is directly phosphorylated by Akt1 (versus other kinases activated by Akt1), we attempted to reconstitute Akt phosphorylation of POSH in vitro, using purified proteins. A full-length GST-POSH fusion protein was produced in E. coli, purified, and incubated with purified preparations of active Akt1 or Akt2 in the presence of [␥-32 P]ATP. Reaction products were then analyzed by SDS-PAGE followed by autoradiography. Fig. 2B shows the results of this experiment. It can be seen that full-length GST-POSH is indeed directly phosphorylated by both Akt1 and Akt2 in vitro, whereas GST alone is unmodified (Fig. 2B, compare lanes 5 and  6 with 2 and 3).
Akt Phosphorylates POSH within the Rac-binding Domain-To investigate how phosphorylation might affect POSH function, we sought to identify the precise site(s) of POSH phosphorylation by Akt. Toward this end, we made use of a slightly truncated form of POSH lacking the RING domain (residues 1-52), because previous experiments had indicated that this form was more efficiently phosphorylated in vitro (not shown). GST-POSH-(53-888) was phosphorylated by either Akt1 or Akt2 and digested with trypsin, and phosphopeptides were recovered by immobilized Fe 3ϩ affinity chromatography. Eluted peptides were then analyzed by high resolution MALDI-TOF mass spectrometry, with the results presented in Fig. 3A. In both the Akt1-and the Akt2-phosphorylated sam-  ples, a total of four peaks were observed with m/z ratios that closely matched the values predicted for phosphopeptides derived from POSH. All correspond to peptides containing a single phosphate group derived from same region of POSH, residues 301-313 with the sequence KRHSFTSLTMANK. The four peaks observed correspond to phosphopeptides either containing or lacking the first lysine residue (Lys-301) and with the methionine residue (Met-310) either oxidized or unmodified. No other phosphopeptides were detected in this experiment, although additional studies have revealed several minor phosphorylation sites (described in more detail below). The large, unoxidized form of this peptide (m/z ratio of 1600.8) from the Akt2 sample was subjected to further analysis by fragmentation and tandem MS-MS, which is shown in Fig. 3B. This analysis yielded the predominant peak expected for loss of phosphate (observed m/z of 1503.1, predicted 1502.8), confirming the phosphorylation status of the peptide. This peptide contains four potential phosphorylation sites (KRHSFTSLT-MANK), and the theoretical ion series expected for phosphorylation at each site were compared with the observed MS-MS spectrum. This analysis strongly supports serine 304 as the site of phosphate addition. Akt Phosphorylates POSH at Multiple Sites-To determine whether Akt may phosphorylate POSH at other sites in addition to Ser-304, we constructed mutant forms of GST-POSH in which Ser-304 was mutated to either alanine, S304A, or aspartate, S304D. These mutant GST-POSH proteins were then tested as Akt substrates in vitro. It can be seen in Fig. 4A that phosphorylation of the Ser-304 mutants is substantially reduced compared with wild type, confirming the mass spectrometry identification of this residue as the major phosphorylation site. S304A and S304D mutants of GST-POSH were still able to incorporate reduced levels of 32 P, however, indicating that although Ser-304 constitutes the major phosphorylation site for Akt, additional minor phosphorylation sites are also present.
To begin to map these additional sites, we constructed a series of smaller GST-POSH proteins containing either N-terminal or C-terminal halves of the protein, or isolated POSH domains, and we tested these as Akt substrates. Fig. 4C shows the results of this analysis. GST proteins containing the N-terminal half of POSH, as well as the isolated Rac-binding domain are efficiently phosphorylated, as is consistent with the mass spectroscopy studies described above (Fig. 4C, lanes 3 and 9). None of the four isolated SH3 domains are found to be phosphorylated under these conditions; however, an additional significantly phosphorylated site (or sites) appears to be present within the C-terminal half of POSH (Fig. 4C, lane 4). Note that in the experiment shown, GST-N-POSH was inadvertently underloaded compared with the other proteins, and thus the phosphorylation of N-POSH appears (misleadingly) low (Fig. 4, C  and D, lane 3). In Fig. 4E, phosphorylation of N-POSH and C-POSH is directly compared with more equal protein loadings, and it can be seen that N-POSH is actually phosphorylated to a much greater extent than C-POSH (Fig. 4E, compare lanes  1 and 2). Quantitation of these data indicate that N-POSH incorporated 4.17 times as much 32 P as C-POSH under these conditions, and this may in fact be an underestimate, because Coomassie staining reveals a slightly higher level of C-POSH protein. In addition to the C-terminal half of POSH, a linker region between the RING and the first SH3 domain, residues 53-134, also incorporates [ 32 P]phosphate, although somewhat weakly (Fig. 4C, lane 10).
To identify more precisely the identity of these minor phosphorylation sites, we again turned to mass spectrometry. Fulllength GST-POSH S304A protein was phosphorylated in vitro with either Akt1 or Akt2 or no added kinase. Following the reaction, GST-POSH was re-isolated by SDS-PAGE, digested with trypsin, and phosphopeptides purified by binding to a tita-  JULY 27, 2007 • VOLUME 282 • NUMBER 30 JOURNAL OF BIOLOGICAL CHEMISTRY 21991 nium dioxide matrix. Peptides eluted from this matrix were then analyzed by MALDI-TOF spectrometry, with the results shown in Fig. 5. Several prominent peaks are evident in the sample incubated with Akt2 that are absent in the control (Fig.  5, peaks labeled A-E) and are thus candidates for phosphopeptides. All five of these peaks are present in both Akt1 and Akt2 samples, although at significantly higher signal intensity in the Akt2 sample. Peak A corresponds to the peptide VQSWSPPVR, residues 123-131 of POSH, containing a single phosphate group (m/z predicted 1135.53, observed 1135.52). This peak was subjected to tandem MS-MS analysis, which confirmed the presence of phosphate, and supported Ser-125 as the site of phosphorylation (not shown). Serine 125 is also predicted to be a possible Akt phosphorylation site by the program Scansite. Peak C corresponds to the peptide KASSLDSAVPIAPPPR, residues 797-812 of POSH, containing a single phosphate group (m/z predicted 1685.86, observed 1685.86). Tandem MS was also performed on this peptide, which confirmed the presence of phosphate, and supported Ser-799 as the site of phosphorylation (not shown). Peak E appears to correspond to the peptide VSPPASPTLEVELGSAELPLQGAVGPELPPGGGHGR, resi-dues 733-768 of POSH, containing a single phosphate group (m/z predicted 3552.77, observed 3552.77). MS-MS analysis of this peak was unsuccessful; however, Scansite predicts Ser-734 as a possible Akt phosphorylation site. Peaks B and D do not appear to be derived from POSH and were not analyzed further.

Phosphorylation of POSH by Akt
Residue Ser-125 lies in the linker domain between the RING domain and the first SH3 domain of POSH, and likely explains the weak phosphorylation of this domain seen in Fig. 4C (lane 10). The other two phosphopeptides containing Ser-734 and Ser-799, are located in the C-terminal half of POSH, between the third and fourth SH3 domains. Although no function has yet been linked to this region of POSH, this result can explain the phosphorylation of GST-C-POSH seen in Fig. 4, C and E.
Akt Phosphorylation of POSH Reduces Rac Binding-We have identified serine 304 in POSH as a major phosphorylation site for both Akt1 and Akt2. Because this residue lies within the Rac-binding domain of POSH, we wished to determine how phosphorylation might affect Rac binding. This question is addressed in Fig. 6. In this experiment, GST-POSH-(53-888) was used as a substrate for in vitro phosphorylation by Akt2. Phosphorylation reactions were incubated at 30°C for different lengths of time (5-60 min, as indicated) or remained on ice (indicated as time 0 in Fig. 6). In addition, control reactions lacking either ATP or Akt2 were employed. Following the incubation period, apyrase was added to remove residual ATP, followed by addition of 35 S-labeled L61-Rac. Glutathione-agarose was then added to recover GST-POSH and any 35 S-Rac bound to it. The samples were washed, and bound proteins were analyzed by SDS-PAGE followed by Coomassie staining and autoradiography. In addition, aliquots of each sample were analyzed by scintillation counting to quantitate 35 S-Rac binding and by immunoblotting with PAS antibody to reveal POSH phosphorylation. It can be seen that as phosphorylation of POSH increases, the ability to bind 35 S-Rac decreases (Fig. 6, lanes 5-12). We note also that Rac binding is reduced even in the samples incubated with Akt at 0°C (Fig. 6,  time 0, lanes 5 and 6). We have observed that some phosphorylation of POSH by Akt occurs at 0°C under these conditions (data not shown). Furthermore, it appears that the control sam-  ples lacking Akt may show slightly higher Rac binding than the controls containing Akt but lacking added ATP (Fig. 6, top and  middle panels, compare lanes 1 and 2 with 3 and 4). A possible explanation for this discrepancy may be that the "No ATP" reactions in fact contain low levels of ATP introduced with the 35 S-Rac, because the 35 S-Rac added was produced by in vitro translation (in a reaction mixture containing ATP) and used without further purification. Our findings here indicate that phosphorylation of POSH by Akt strongly reduces the ability of POSH to bind Rac.
Phosphomimetic Mutation of POSH Ser-304 Reduces Rac Binding and Apoptosis-To further confirm that serine 304 of POSH is the major phosphorylation site used by Akt, and to investigate the functional consequences of phosphorylation by Akt at this site, we further studied the behavior of POSH containing Ser-304 mutations. In Fig. 7A, we compare the ability to bind 35 S-L61-Rac of wild type POSH with mutant versions containing either alanine (S304A), aspartate (S304D), or glutamate (S304E). We find that mutation of Ser-304 to alanine does not appear to disrupt Rac binding (Fig. 7A, compare lanes 5 and 6  with lanes 7 and 8). In contrast, mutation to either aspartate or glutamate, which contain fixed negative charge and may there-fore mimic phosphorylation at this site, significantly reduces Rac binding to POSH (compare lanes 9 -12 with 5-8). Similar results are observed in vivo with endogenous Rac. In Fig. 7B, expression vectors for GST-POSH containing either the S304A or S304D mutations were transfected into HEK 293T cells, and 24 h later GST-POSH and any proteins bound to it were recovered on glutathione-agarose. Immunoblotting of bound proteins shows that the POSH S304A mutant retains the ability to bind Rac, whereas Rac binding to the S304D mutant, expressed at similar levels, is drastically reduced (Fig. 7B, compare lanes 2 and 3). The functional consequences of reduced Rac binding are evident in Fig. 7C. Here MDA-MB-231 human breast cancer cells were transfected with expression vectors for Ser-304 POSH mutants or empty vector, and caspase activity was measured 24 h later as an indication of apoptosis. It can be seen that the POSH S304A mutant retains the ability to induce apoptosis, as measured by caspase activation, whereas the POSH S304D mutant showed only background levels of caspase activity for this assay (Fig. 7C, compare  lanes 1-3). Taken together, our findings indicate that phosphorylation of POSH at serine 304 by Akt strongly reduces the ability of POSH to bind Rac and thereby reduces the ability of POSH to promote apoptosis.

DISCUSSION
POSH participates in apoptotic signaling by binding activated Rac and linking it to a kinase cascade pathway that results in JNK activation. POSH acts as a scaffold protein in a multiprotein signaling complex that has recently been termed "PJAC" (for POSH-JIP apoptotic complex). In this signal transduction pathway POSH directly binds both active Rac and one of the mixed lineage kinases (MLK1, -2, -3, or DLK), and the subsequent interaction of active Rac with MLK induces autophosphorylation and activation of MLK. POSH also directly binds a second scaffold protein JIP (JNK-interacting protein 1 or 2), and JIP completes the complex by binding MKK (4 or 7) and JNK (1, 2, or 3). Activated MLK can then phosphorylate and activate MKK4/7, and active MKK4/7 in turn phosphorylates and activates JNK. JNKs activated through this pathway go on to trigger apoptosis through release of cytochrome c from mitochondria and the subsequent activation of caspases.
Apoptosis is highly regulated through multiple mechanisms and at multiple sites. The pro-survival kinase Akt has emerged FIGURE 5. Identification of minor phosphorylation sites in POSH. Full-length GST-POSH-(1-888) containing an S304A point mutation was phosphorylated in vitro by either Akt1, Akt2, or no added kinase, and the reaction products were separated by SDS-PAGE. GST-POSH protein bands were excised and digested with trypsin, and phosphopeptides were purified by affinity chromatography with a titanium dioxide matrix. Eluted peptides were analyzed by MALDI-TOF mass spectrometry. Prominent peaks observed in the sample treated with Akt2 that are absent in the control are labeled as peaks A-E. These peaks are also present in the Akt1 sample, although at lower intensity, and are candidates for phosphopeptides. Tandem MS-MS analysis was successfully performed for peaks A and C and confirmed their identities as phosphopeptides (not shown). Peak A corresponds to the peptide VQSWSPPVR, residues 123-131 of POSH, containing a single phosphate group (m/z predicted 1135.53, observed 1135.52). Peak C corresponds to the peptide KASSLDSAVPIAPPPR, residues 797-812 of POSH, containing a single phosphate group (m/z predicted 1685.86, observed 1685.86). Peak E appears to correspond to the peptide VSPPASPTLEVELGSAELPLQGAVGPELPPGGGHGR, residues 733-768 of POSH, containing a single phosphate group (m/z predicted 3552.77, observed 3552.77), however MS-MS was unsuccessful for this peak. Peaks B and D do not appear to have originated from POSH, and they were not analyzed further.
as a key regulator of cell survival and can prevent apoptosis induced by a variety of death signals (27,28). Cellular targets of Akt are numerous and include directly proapoptotic proteins such as BAD and caspase-9 as well as transcription factors such as members of the Forkhead family/FOXO that regulate expression of death proteins such as FAS ligand (12)(13)(14)(15)(16).
In this study, we have shown that Akt1 and Akt2 can suppress the pro-apoptotic actions of POSH in MDA-MB-231 human breast cancer cells, confirming earlier findings in PC12 cells and extending them to an epithelial cell type. We have identified POSH as a novel substrate for phosphorylation by both Akt1 and Akt2, and we mapped a major phosphorylation site to serine 304, which lies within the Rac-binding domain. Serine 304 of POSH occurs within a context (KKNTKKRHS) that is not an exact match for the classic "consensus" Akt phosphorylation sequence (R/K)X(R/K)XX(S/T) (29). Nonetheless, serine 304 is identified as a potential Akt phosphorylation site by the program Scansite. We demonstrate that Ser-304 mutation of POSH strongly reduces POSH phosphorylation by Akt but does not completely eliminate it. Although serine 304 is a major site of phosphorylation by Akt, other minor sites are also present. We identify Ser-125, Ser-799, and possibly Ser-734 as minor phosphorylation sites. We further show that phosphorylation of POSH by Akt leads to loss of its ability to bind active Rac, and would thereby be predicted to block transmission of the apoptotic signal from Rac to the kinase cascade that results in activation of JNK. Support for this prediction is obtained by observing the behavior of POSH Ser-304 mutants. POSH S304A retains the ability to bind Rac, whereas S304D and S304E mutants show reduced Rac binding, indicating these mutants may mimic the effects of phosphorylation at that site. Although S304A POSH is still able to induce apoptosis as measured by caspase activation, the POSH S304D mutant did not show any increase in apoptosis above background levels in the assay.
Akt appears to regulate the PJAC apoptotic pathway at multiple sites and by multiple mechanisms, as summarized diagrammatically in Fig. 8. We report here that phosphorylation of POSH by Akt at serine 304 reduces the ability of POSH to bind to Rac. Previous studies by other authors have reported Akt phosphorylation of other PJAC components. Kwon et al. (30) have described Akt phosphorylation of Rac1 at serine 71, which prevents Rac from being activated by blocking its ability to bind GTP. Akt has also been reported to phosphorylate MLK3 on serine 674 and MKK4 on serine 78, both of which result in decreased kinase activity and subsequent reduced JNK activation (31,32). In addition to modulation of PJAC components by phosphorylation, it appears that Akt may also regulate PJAC by nonenzymatic mechanisms. As discussed earlier, Akt2 (but not Akt1) can directly bind to POSH, and Akt2 binding appears to block binding of MLK3, inhibiting the apoptotic pathway by preventing PJAC assembly (23). Finally, in another isoformspecific nonenzymatic mechanism, Akt1 (but not Akt2) can bind directly to JIP1, an interaction that has been reported to reduce JNK activation and apoptotic signaling (33).
In our experiments examining the ability of Akt-phosphorylated POSH to bind Rac, it is possible that some phosphorylation of Rac might also have occurred. We believe this is unlikely because apyrase was added to deplete ATP prior to the addition of 35 S-Rac. Furthermore, Akt-mediated phosphorylation of Rac is unlikely to contribute to the results we have obtained here because we have employed a constitutively active form of Rac that already contains bound GTP and lacks the ability to hydrolyze it (Rac Q61L).
The multiple regulatory actions of Akt upon PJAC suggest that in order for JNK to become activated through the PJAC pathway and trigger apoptosis, Akt activity must be low, or these multiple levels of regulation by Akt must be overcome. A degree of cross-talk between the JNK and Akt pathways appears to exist. There is evidence that JNK can directly oppose actions of Akt, and furthermore that under some conditions JNK activation may lead to a reduction in the levels of activated Akt. For example, Akt phosphorylation of Bad and FOXO3a provides a binding site for 14-3-3 proteins and results in sequestration of these pro-apoptotic proteins in the cytoplasm, where they are unable to act. JNK can phosphorylate 14-3-3 and cause release of bound Bad and FOXO3a, antagonizing the effect of Akt (34). In another apparent example of cross-talk, Sunters et al. (35) have reported that JNK activation following treatment with paclitaxel leads to reduced levels of active Akt. Other examples of cross-talk between these pathways likely await discovery.
Our findings here identify POSH phosphorylation as a novel  lanes 3 and 4), or remained on ice (0, lanes 5 and 6). Following the 30°C incubation, apyrase was added to remove residual ATP, followed by 35 S-L61-Rac and glutathione-agarose. After 2 h at 4°C, agarose beads were washed to remove unbound proteins, and bound proteins were analyzed by SDS-PAGE followed by Coomassie staining and autoradiography. In lanes 13 and 14, GST was used instead of GST-POSH as a control for Rac binding. The top panel shows quantitation of 35 JULY 27, 2007 • VOLUME 282 • NUMBER 30  A, Akt regulates PJAC through phosphorylation at multiple sites. We report here that Akt phosphorylates POSH on Ser-304 (shown by large arrow) and other minor sites (shown by smaller arrows) to result in a loss of its ability to bind Rac. Others have reported that Akt also phosphorylates Rac, MLK3, and MKK4 (additional small arrows), in each case resulting in decreased JNK activation and reduced apoptosis. B, Akt apparently also regulates PJAC through nonenzymatic binding interactions. Akt2 specifically binds POSH and prevents MLK3 binding, and Akt1 specifically binds JIP1. Both interactions lead to reduced JNK activation. See "Discussion" for more details.

Phosphorylation of POSH by Akt
mechanism for regulation of apoptosis by the pro-survival kinase Akt. The extensive regulation of the PJAC apoptotic pathway by Akt at multiple sites and through multiple mechanisms suggests that the interplay between Akt and the PJAC pathway is of significant physiological importance.