Systematic Evaluation of Anti-apoptotic Growth Factor Signaling in Vascular Smooth Muscle Cells

Peptide growth factors contribute to the pathogenesis of cardiovascular diseases by inducing a variety of cellular responses including anti-apoptotic effects. Several of the signaling molecules that are activated by growth factor receptors such as Src family kinases (Src), phosphatidylinositol 3′-kinase (PI3K), phospholipase Cγ (PLCγ), Ras, and SHP-2 were shown to mediate survival signals. We systematically investigated the relative contribution of each signaling molecule for growth factor-dependent cell survival in vascular smooth muscle cells (VSMC). Our approach was the use of mutated plateletderived growth factor (PDGF) β-receptors (βPDGFR) in which the tyrosine residues required for binding of each signaling molecule were individually mutated to phenylalanine. To bypass endogenous PDGFR in VSMC we used chimeric receptors (ChiRs), containing the extracellular domain of the macrophage colony-stimulating factor (M-CSF) receptor and the cytoplasmic domain of the wild type (WT) or mutated βPDGFR. Selective activation of the ChiR-WT with M-CSF significantly reduced apoptosis to the same extent as PDGF-BB in non-transfected cells. Deletion of the binding site for PI3K, but not for Src, RasGAP, SHP-2, or PLCγ, completely abolished the anti-apoptotic effect. Consistently, a ChiR mutant that only binds PI3K was fully able to mediate cell survival as efficiently as the ChiR-WT. Furthermore, the PDGF-dependent anti-apoptotic effect in non-transfected cells was completely abolished by the PI3K inhibitor wortmannin, whereas inhibitors of Src, PLCγ, ERK, or p38 MAP kinase had no effect. The exploration of downstream signaling events revealed that PDGF-BB activates the anti-apoptotic Akt signaling pathway in a PI3K-dependent manner. Moreover, Akt phosphorylates and thus inactivates the pro-apoptotic proteins BAD and Forkhead transcription factors (FKHR, FKHRL1). We conclude that growth factor-dependent cell survival in VSMC is mediated only by activation of the PI3K/Akt pathway, whereas all other receptor-associated signaling molecules do not play a significant role.

The pathogenesis of vascular diseases involves an abnormal accumulation of lipids, inflammatory cells, and vascular smooth muscle cells (VSMC) 1 within the intimal layer of the vessel wall. Recent studies have demonstrated that in addition to cell proliferation and migration, dysregulated apoptosis plays a prominent role in the pathogenesis and progression of atherosclerosis (1). Peptide growth factors such as plateletderived growth factor (PDGF) and insulin-like growth factor-I are potent inhibitors of cell death. These growth factors exert their biological responses via activation of transmembrane receptor tyrosine kinases. Upon ligand binding, receptor tyrosine kinases autophosphorylate on tyrosine residues and subsequently recruit and activate SH2 domain-containing cytoplasmic signaling molecules such as Src family kinases (Src), the GTPase activating protein of Ras (RasGAP), the phosphotyrosine phosphatase SHP-2, phosphatidylinositol 3Ј-kinase (PI3K), and phospholipase C␥1 (PLC␥) (2).
Recent studies indicated that all of these signaling molecules are implicated in survival signaling and therefore may contribute to growth factor-mediated cell survival. In PDGF-stimulated cells, Src family kinases are required for activation of STAT (signal transducer and activator of transcription) 3 (3,4), which protects fibroblasts from apoptosis and antagonizes the pro-apoptotic effects of activated STAT1 (5). In addition, Src was shown to mediate anti-apoptotic effects by activating PI3K (6). PI3K mediates anti-apoptotic signals of several growth factors via activation of its downstream target Akt (7). The serine/threonine kinase Akt (8,9) phosphorylates and thus inactivates several pro-apoptotic proteins such as BAD (10), Forkhead transcription factors or caspase-9 (11,12). Further downstream mediators of this pathway include NFB and glycogen synthase kinase-3␤ (GSK3␤) (9,13). Recently, SHP-2 was shown to be essential for insulin-like growth factor-I-or PDGF-mediated cell survival, as it suppresses caspase 3-dependent apoptosis in fibroblasts (14). Another well characterized survival pathway is the Ras/Raf/extracellular-regulated kinase (ERK) cascade, which leads to the phosphorylation of BAD (15,16). Finally, PLC␥ is required for insulin-like growth factor-I-mediated protection against apoptosis (17), as it medi-* This work was supported by Deutsche Forschungsgemeinschaft (DFG) (Ro 1306-2-1 (to S. R. and W. H. Z.)), the Center for Molecular Medicine of the University of Cologne (CMMC, TV84 (to S. R.)), and by the Marga and Walter Boll Stiftung, Germany (to M. V. and S. R.). 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 U.S.C. Section 1734 solely to indicate this fact.
Each of the above studies investigated only one of the growth factor-activated signal relay enzymes, and most studies used either pharmacological inhibitors, which may lack specificity, or a dominant-negative approach, which may influence cell survival by overexpressed protein levels. Additionally, inhibitors as well as overexpressed dominant-negative proteins suppress the basal activity of the inhibited signaling molecule in unstimulated cells, limiting the significance about their role for ligand-induced cellular responses. Hence, the relative roles of Src, PLC␥, RasGAP, PI3K, and SHP-2 for growth factor-mediated cell survival are not clear from these studies. Therefore, we sought to systematically evaluate the relative contribution of each of the above signaling molecules for growth factormediated cell survival in one system. We chose the PDGF ␤-receptor (␤PDGFR), because this receptor tyrosine kinase, which is highly relevant for atherogenesis and tumorigenesis, binds each of the receptor-associated signal relay enzymes at one specific binding site. Our approach was to characterize ␤PDGFR mutants in which the binding sites for one or more of the receptor-associated signaling enzymes were individually deleted by selective tyrosine to phenylalanine substitutions. Importantly, we studied PDGFR signaling in VSMC, which are critically involved in atherogenesis, and in which PDGFR signaling is physiologically relevant, as these cells express endogenous PDGFR and thus respond to PDGF. However, to selectively stimulate PDGFR mutants in PDGFR-expressing cells, it is necessary to bypass endogenous ␤PDGFR. Therefore, we used chimeric M-CSF receptor (CSF1R)/␤PDGFR mutants (ChiR) containing the extracellular domain of the CSF1R, which is not expressed in naive VSMC, and the cytoplasmic domain of the ␤PDGFR. By these chimeras, ␤PDGFR signaling is selectively stimulated with M-CSF. The use of this highly efficient system allowed us to selectively activate individual signaling molecules that are recruited to the ligand-stimulated ␤PDGFR and to compare the ability of the various ChiR mutants to protect VSMC against apoptosis.
Generation and Expression of ChiR Mutants-The ChiR cDNA was constructed from DNA encoding amino acids 1-513 of the human CSF1R and amino acids 528 -1106 of the human ␤PDGFR joined at an EcoRI site (22). Mutations in the ␤PDGFR sequence are as described (23)(24)(25)(26) and were obtained by site-directed mutagenesis and subcloning. Constructs were assembled in the pLXSN retroviral expression vector (27), sequence-verified, and transfected into the 293T GPG packaging cell line as described (28). Forty-eight hours after transfection, retroviral supernatants were added to VSMC. Positive infectants were selected with 0.8 mg/ml G418.
Immunoprecipitation and Western Blot Analysis-Quiescent VSMC were left resting or stimulated with 50 ng/ml PDGF-BB or M-CSF for times indicated in the presence or absence of inhibitors as indicated. Cells were washed twice with HS (20 mM Hepes, pH 7.4, 150 mM NaCl) and then lysed in EB (10 mM Tris-HCl, pH 7.4, 5 mM EDTA, 50 mM NaCl, 50 mM NaF, 1% Triton X-100, 0.1% bovine serum albumin, 20 g/ml aprotinin, 2 mM Na 3 VO 4 , 1 mM phenylmethylsulfonyl fluoride). Lysates were centrifuged (20 min, 12,000 ϫ g), and the supernatants were subjected to immunoblot analysis or immunoprecipitation. The ␤PDGFR and the ChiRs were precipitated with a polyclonal ␤PDGFR antibody (97A) recognizing the kinase insert (amino acids 698 -797) of the human ␤PDGFR. ␤PDGFR immunoprecipitations representing ϳ3 ϫ 10 6 cells were resolved on a 7.5% SDS-polyacrylamide gel, and the proteins were transferred to Immobilon and subjected to Western blot analysis. To monitor the association of signaling molecules with the activated ␤PDGFR, receptor immunoprecipitations from resting or M-CSF-stimulated cells were subjected to Western blot analysis using antisera against the extracellular domain of the human CSF1R (anti-CSF1R), PLC␥, RasGAP, p85, and SHP-2. To monitor the association between Src and the ChiRs, Src was immunoprecipitated from resting or M-CSF-stimulated cells using the Src-2 antibody, and immunoprecipitates representing ϳ3 ϫ 10 6 cells were subjected to anti-CSF1R or anti-Src Western blot analysis. BAD was precipitated with a monoclonal BAD antibody. Immune complexes were bound to protein A-Sepharose, spun through an EB sucrose gradient, and then washed three times with EB. Bound proteins were resuspended in SDS sample buffer. BAD immunoprecipitations representing ϳ2 ϫ 10 7 cells were resolved on a 15% SDS-polyacrylamide gel. Proteins were subjected to Western blot analysis using a polyclonal BAD antibody or a phosphospecific (Ser-136) BAD antibody. To monitor the expression levels of ChiR mutants in VSMC and the phosphorylation state of downstream signaling molecules, whole cell lysates representing ϳ3 ϫ 10 6 cells were resolved on a 10% SDS-polyacrylamide gel and probed with anti-CSF1R, and antibodies were raised against RasGAP (lysate control), BAD and phosphorylated Akt, GSK3␤, BAD, FKHR, and FKHRL1. The results presented are representative of at least three independent experiments.
Induction and Detection of Apoptosis-Apoptosis was induced by either UV irradiation (254 nm, 70 mJ/cm 2 ) or H 2 O 2 (100 M). A cell death detection ELISA (Roche Diagnostics) was used to measure cell death by the detection of cytoplasmic histone-associated DNA fragments (mono-and oligonucleosomes). Cells (5,000 cells/well) were plated into 96-well plates in Dulbecco's modified Eagle's medium containing 10% fetal calf serum and starved for 24 h in Dulbecco's modified Eagle's medium. To assess the anti-apoptotic effect of PDGF-BB (50 ng/ml) or M-CSF (50 ng/ml), the cells were UV light-irradiated and subsequently cultured for 8 h in the absence or presence of ligand. Alternatively, the cells were incubated for 12 h in Dulbecco's modified Eagle's medium containing H 2 O 2 in the presence or absence of PDGF-BB or M-CSF. The cell lysates were collected, and apoptosis was measured using the cell death detection ELISA.
Materials-PDGF-BB and M-CSF were purchased from Promo Cell (Heidelberg, Germany). The polyclonal antibody raised against the extracellular domain of the human CSF1R was from Upstate Biotechnology. The anti-phosphotyrosine antibodies were purchased from Santa Cruz (PY20) and Upstate Biotechnology (4G10). The antibody against SHP-2 was obtained from Transduction Laboratories (dPTP1D), and the PLC␥ antibody was from Upstate Biotechnology. The antibodies against the ␤PDGFR (97A), RasGAP (69.3), and the p85 subunit of PI3K were kind gifts from Andrius Kazlauskas and Alex Toker (Harvard Medical School, Boston, MA). The Src-2 antibody used for immunoprecipitation of Src was purchased from Santa Cruz Biotechnology and recognizes the Src family members Src, Yes, and Fyn. For Western blot analysis of Src, a mouse monoclonal antibody, 327 (Oncogene Science Inc.), was used. The polyclonal phospho-Akt (Ser-473), phospho-FKHR (Thr-24)/FKHRL1 (Thr-32), phospho-Bad (Ser-136), Akt, and BAD antibodies were from Cell Signaling Technologies. The monoclonal BAD antibody was from BD Biosciences.
Statistical Analysis-All data are expressed as means Ϯ S.D. Statistical analysis was evaluated by non-parametric analysis. p Ͻ 0.05 was considered significant.

Characterization of Chimeric CSF1R/␤PDGFR Mutants-To
systematically evaluate the role of each of the ␤PDGFR signaling molecules (Src, PI3K, RasGAP, SHP-2, and PLC␥) for growth factor-mediated cell survival in a single cell type, we created a series of ChiR in which the tyrosine residues required for the association of each of the above signaling enzymes were individually mutated to phenylalanine (subtraction mutants) (Fig. 1A). A second set of ChiRs (add-back mutants) was based on the F5 receptor, in which the 5 tyrosine phosphorylation sites responsible for binding of PI3K, RasGAP, SHP-2, and PLC␥ were mutated. This panel of receptors included mutants that have the tyrosine residue(s) required for binding of one of the associated proteins restored (Fig. 1B). In this panel, the Src-binding sites were left intact, as their mutation in the juxtamembrane region was shown to attenuate the ligandinduced tyrosine phosphorylation of the full-length ␤PDGFR (25). All ChiR were stably expressed in VSMC. Additionally, an "F6" receptor was expressed, which carries an additional mu-tation at position 716 of F5. This tyrosine is reported to bind the Grb2 adaptor protein (29), which is involved in mediating Ras activation via binding of SOS. Finally, we generated a kinase-inactive receptor, referred to as R634, which carried a lysine to arginine mutation at position R634 within the kinase domain and served as a negative control. As demonstrated in Fig. 2, A and D, all ChiR were expressed at equal levels.
To investigate the ligand-induced autophosphorylation of the receptor and M-CSF-dependent binding of signaling molecules, ChiRs were immunoprecipitated from resting or M-CSF-stimulated cells, and the samples were subjected to Western blot analysis using antibodies against the extracellular domain of the CSF1R and against each of the signal relay enzymes. In addition, the receptor blot was stripped and reprobed with an anti-phosphotyrosine antibody. As shown in Fig. 2, B and E, M-CSF increased the phosphotyrosine content of all of the receptors, with the exception of the kinase-inactive mutant (R634). The level of receptor autophosphorylation was not sub-stantially decreased, even in mutants missing up to five or six phosphorylation sites. This may reflect the presence of additional phosphorylation sites, or the ability of the receptor to autophosphorylate at cryptic sites. A similar phenomenon is observed with mutants of the ␣PDGFR (28). Furthermore, ChiR-WT co-immunoprecipitated with Src, the p85 subunit of PI3K, RasGAP, SHP-2, and PLC␥ in a M-CSF-dependent manner. No ChiR expression or binding of receptor-associated proteins was detected in receptor immunoprecipitations from M-CSF-stimulated non-transfected VSMC. The mutation of tyrosine 1021 largely reduced the binding of PLC␥, the F40/51 mutant failed to associate with PI3K (p85), and the F771, F1009, and F79/81 mutants did not bind RasGAP, SHP-2, and Src upon M-CSF stimulation, respectively (Fig. 2, B and C). p85, respectively, whereas none of the other signal relay enzymes associated with these receptors. All add-back mutants associated with trace amounts of p85. Surprisingly, SHP-2 bound to the activated F5 receptor and to all add-back mutants, with the exception of the F6 and Y40/51 receptors. This suggests that SHP-2 is able to associate with at least one additional binding site besides tyrosine 1009 and, furthermore, that either binding or activation of PI3K prevents SHP-2 binding to this or these additional site(s). Rönnstrand et al. (30) have reported that SHP-2 is also capable to bind to phosphorylated tyrosine 763, which is relatively close to the PI3K-binding site. This suggests that PI3K binding may sterically hinder the association of SHP-2 with phosphorylated tyrosine 763. R634, which served as a negative control in the biological assays, was not autophosphorylated, and failed to recruit ␤PDGFR-associated signaling molecules upon M-CSF stimulation. These data demonstrate that the ChiR mutant system works sufficiently in VSMC and enabled us to reliably investigate anti-apoptotic ␤PDGFR signaling.
M-CSF Mimics PDGF-dependent Cell Survival in ChiR-WTexpressing VSMC-To confirm that the activated ChiR-WT mimics PDGF-BB-dependent survival responses we compared the apoptosis rate of PDGF-stimulated non-transfected VSMC with that of M-CSF-treated ChiR-WT-expressing VSMC upon UV irradiation or H 2 O 2 treatment. Irradiation and H 2 O 2 treat-ment led to a 2.13 Ϯ 0.54 and 5.31 Ϯ 1.12-fold increase in cell death, respectively (Fig. 3A). The supplementation of culture medium with 50 ng/ml PDGF-BB significantly decreased irradiation-and H 2 O 2 -induced cell death by 48 Ϯ 14 and 41 Ϯ 21%, respectively (both p Ͻ 0.05). Consistently, irradiation or treatment of ChiR-WT-expressing VSMC with H 2 O 2 caused a 1.59 Ϯ 0.17 and 2.51 Ϯ 0.27-fold increase in apoptotic cell death, respectively, compared with non-treated cells. Co-incubation with M-CSF (50 ng/ml) significantly reduced the rate of irradiation-or H 2 O 2 -induced apoptosis by 42 Ϯ 12 and 40 Ϯ 11%, respectively (both p Ͻ 0.05), whereas it had no effect on cell survival in non-transfected VSMC (not shown). Hence, stimulation of ␤PDGFR signaling by M-CSF in ChiR-WT-expressing VSMC was sufficient to mediate PDGF-induced responses to the same extent as PDGF-BB in non-transfected cells.
Systematic Evaluation of ␤PDGFR-mediated Anti-apoptotic Signaling in VSMC-To investigate the role of each of the signaling molecules for ␤PDGFR-induced protection from apoptosis, we compared this response in VSMC expressing either ChiR-WT or the various mutant receptors. We induced apoptosis by UV irradiation of quiescent cells. An analysis of the subtraction panel revealed that deletion of the binding site for PI3K (F40/51), but not for Src, RasGAP, SHP-2, or PLC␥, completely suppressed the anti-apoptotic effect (Fig. 3B). Consistently, binding and activation of only PI3K without activa- tion of RasGAP, SHP-2, and PLC␥ via the Y40/51 receptor was sufficient to mediate the complete M-CSF-induced anti-apoptotic effect. In contrast, all other add-back mutants were unable to mediate a significant anti-apoptotic effect (Fig. 3C).
To exclude the possibility that our findings were limited to one specific inducer of apoptosis, we repeated our experiments when apoptosis was induced by H 2 O 2 . Similar to the above studies, a lack of PI3K binding (F40/51) completely abolished the anti-apoptotic effect mediated by the ChiR-WT, whereas restoration of the PI3K-binding site to the F5 receptor (Y40/51) fully salvaged this cellular response (Fig. 3D).
Inhibition of Signaling Molecules by Pharmacological Inhibitors in Non-transfected Cells-A potential caveat of utilizing receptor mutants is the possibility that, in addition to the known binding partners of phosphotyrosines, other unknown signaling molecules may also interact with these sites. To confirm the results, which we obtained by the use of ChiR mutants, in a second approach, we repeated all measurements in PDGF-stimulated non-transfected VSMC in the presence of pharmacological inhibitors. Consistently, inhibition of PI3K by wortmannin completely suppressed the anti-apoptotic effect of PDGF-BB after UV irradiation, indicating that this signaling enzyme is involved in anti-apoptotic signaling, whereas inhibition of Src kinases (using SU6656), MEK1 (PD98059), and p38 MAPK (SB203580) had no effect on PDGF-dependent antiapoptotic signaling (Fig. 4). Interestingly, inhibition of PLC␥ by U-73122 also suppressed PDGF-mediated cell survival after UV irradiation but not when apoptosis was induced by H 2 O 2 . However, U-73122 appeared to affect cell survival by itself when used in higher concentrations (not shown). In summary, our data clearly demonstrate that anti-apoptotic signaling by the ␤PDGFR is mediated by only one of the receptor-associated signaling molecules, PI3K, whereas all other signal relay enzymes do not play a significant role.

PI3K/Akt Mediate Growth Factor-dependent Cell Survival via Phosphorylation of BAD and Forkhead Transcription Factors-Because activation of
Akt is thought to play a critical role in PI3K-dependent regulation of apoptosis, we investigated the role of the receptor-associated proteins for PDGF-induced Akt activation. The ligand-induced phosphorylation of Akt was monitored by Western blot analysis using a phospho-specific Akt antibody in M-CSF-stimulated ChiR-expressing VSMC as well as in PDGF-stimulated non-transfected VSMC in the presence of specific inhibitors. Akt was maximally phosphorylated 5 min after PDGF stimulation and remained activated for several hours. In addition to Akt phosphorylation, its activity was monitored by the ability to phosphorylate its substrate GSK3␤ at serine 9 (Fig. 5A).
We next explored the contribution of each signaling molecule to ␤PDGFR-mediated Akt activation. In ChiR-WT-expressing cells, Akt was phosphorylated upon M-CSF stimulation with a similar time course and to a comparable extent as in VSMC upon PDGF-BB-stimulation (Fig. 5B). The F40/51 receptor did not efficiently activate Akt, whereas the tyrosine to phenylalanine substitutions at all other sites had no effect on M-CSFdependent activation of Akt. In the ChiR-F6-expressing cells, M-CSF-induced Akt activation was completely abolished (Fig.  5C). Restoration of the PI3K-binding site rescued M-CSF-dependent activation of Akt, whereas the F5, Y771, Y1009, and Y1021 mutants induced Akt activation to a much lesser extent. The fact that these mutants are still able to weakly phosphorylate Akt may be explained by weak binding of PI3K to these mutants (see Fig. 2E). Nevertheless, the Akt-substrate GSK3␤ is exclusively phosphorylated when PI3K efficiently binds to the activated receptor, underlining that Akt is activated in response to PI3K binding to the ChiRs (Fig. 5, B and C). This result was confirmed with regard to the PDGF-stimulated endogenous PDGFR. Pharmacological inhibition of PI3K, but not of Src, PLC␥, MEK1 or p38, prevented PDGF-induced phosphorylation of Akt and GSK3␤ (Fig. 5D). These results indicate that only PI3K is able to efficiently activate the anti-apoptotic Akt pathway and suggest that the ␤PDGFR mediates its antiapoptotic effect exclusively via the PI3K/Akt survival pathway.
To further investigate the signaling mechanisms downstream of PI3K/Akt, we evaluated the Akt-dependent inactivation of a set of pro-apoptotic proteins. To this end, we immu-noprecipitated the pro-apoptotic protein BAD from resting or ligand-stimulated cells followed by the analysis of its phosphorylation status at serine residue 136 (Ser-136), which is the major Akt-dependent phosphorylation site, using a phosphospecific antibody. PDGF-BB caused a rapid and transient BAD phosphorylation at Ser-136 (Fig. 6A). This effect was dependent on PI3K-mediated signaling, because in ChiR-expressing VSMC the F40/51 receptor did not mediate M-CSF-dependent phosphorylation of BAD (Ser-136), whereas Y40/51-expressing cells did (Fig. 6B). In addition, inhibition of PI3K with wortmannin abrogated the PDGF-dependent BAD phosphorylation (Ser-136), whereas other inhibitors had no influence (Fig.  6C). Finally, the pro-apoptotic transcription factors of the Forkhead family (FKHR and FKHRL1), both substrates of Akt, showed the same phosphorylation pattern as BAD. Forkhead transcription factors were exclusively phosphorylated, and thus inactivated, by those ChiRs that were capable of binding PI3K in response to M-CSF (Fig. 6D) and upon PDGF-stimulation when PI3K was not inhibited (Fig. 6E).

DISCUSSION
In the present study we have systematically investigated the relative importance of the signaling enzymes that are recruited to the activated ␤PDGFR for protection against apoptosis in VSMC. Using receptor mutants and pharmacological inhibitors, we demonstrate that PDGF-dependent cell survival is mediated by only one of the ␤PDGFR-associated signaling molecules, PI3K. Although the other receptor-associated signaling molecules such as Src, PLC␥, RasGAP, and SHP-2 were previously shown to mediate survival signals, their contribution to growth factor-induced anti-apoptotic effects is not significant when compared with PI3K/Akt signaling. In addition, we characterized the signaling mechanisms downstream from PI3K/ Akt, which includes phosphorylation/inactivation of pro-apoptotic BAD and Forkhead transcription factors (FKHRL1 and FKHR).
The complex methodology of ChiR mutants was used to reliably investigate the signaling pathways that contribute to cell survival in a relevant cell line that expresses endogenous PDGFR, namely VSMC. To circumvent endogenous PDGFR, the cytoplasmic domains of mutated ␤PDGFR were selectively stimulated via binding of M-CSF to the extracellular CSF1R domain. Using this highly efficient system we were able to selectively activate ␤PDGFR mutants in PDGFR-expressing cells and thereby prevent the ligand-induced activation of multiple signaling pathways without affecting their basal activity.
The systematic characterization of the "add-back" and "subtraction" panels of chimeric ␤PDGFR mutants as well as the use of pharmacological inhibitors consistently revealed that although the ␤PDGFR recruits multiple signaling enzymes, only PI3K is responsible for protection against apoptosis. Deletion of the PI3K-binding sites in the F40/51 receptor completely abolished the anti-apoptotic effect, whereas restoration of the PI3K-binding sites to the F5 receptor was fully sufficient to rescue ␤PDGFR-mediated cell survival. These data clearly demonstrate that binding and activation of PI3K to the ␤PDGFR is necessary and sufficient for ␤PDGFR-mediated suppression of apoptosis. Consistently, inhibition of PI3K with wortmannin led to a total loss of PDGF-BB-induced cell survival in non-transfected cells. In contrast, mutation of all other binding sites in the ChiR mutants as well as pharmacological inhibition of Src, PLC␥, or MAP kinases (ERK, p38) did not influence ␤PDGFR-mediated protection against apoptosis. Activation of the PI3K/Akt pathway has been shown to inhibit apoptosis in a variety of cell types (31)(32)(33)(34). The evaluation of downstream signaling events revealed that Akt and the Akt effectors BAD, Forkhead transcription factors, and GSK3␤ were only phosphorylated when PI3K was active. Upon Aktdependent phosphorylation at specific sites, pro-apoptotic proteins such as BAD, Forkhead transcription factors, and GSK3␤ are inactivated and sequestered in the cytosol bound to 14-3-3 (11,35). Taken together, these findings indicate that growth factor-induced protection from apoptosis is exclusively mediated via PI3K, and the PI3K-dependent Akt signaling pathway leading to inactivation of pro-apoptotic proteins is crucial for cellular survival.
Based on previous work demonstrating that Src, RasGAP, SHP-2, and PLC␥ all mediate survival signals (3-5, 14 -20), it appears quite surprising that neither of them significantly contributed to growth factor-mediated cell survival in our system. The discrepancy between these studies and our data may be explained by the fact that these signaling enzymes under certain conditions exert anti-apoptotic effects indirectly by contributing to PI3K/Akt activation rather than by inducing independent survival signals, as indicated by some studies (14). When their relative contribution to growth factor-mediated survival signaling is compared with PI3K in one system, our data clearly show that of the ␤PDGFR-associated signaling molecules, only PI3K is sufficient to mediate an anti-apoptotic effect.
Our results on the role of PLC␥ are inconsistent with a recent study reporting that PLC␥ is required for the prevention of cell death in response to growth factors such as insulin-like growth factor-I (17). These authors investigated a special kind of cell death, anoikis, and it is likely that PLC␥-mediated signaling plays a limited role for cell survival under these special conditions. Our data obtained by the ChiR mutant system clearly indicate that PLC␥ is not required for efficient growth factor-mediated cell survival (F1021 and Y1021 mutants). However, our findings obtained with the PLC␥ inhibitor U-73122 are somewhat inconclusive, as pretreatment of nontransfected VSMC with U-73122 abolished the anti-apoptotic effect of PDGF in the case of UV irradiation-induced apoptosis, whereas it did not under the condition of oxidative stress. This discrepancy may be explained by differences in the experimental conditions. VSMC were incubated with U-73122 before and after UV irradiation because of one additional medium change after irradiation. In contrast, no medium change was necessary during H 2 O 2 -treatment. Thus, U-73122 may have accumulated within the cells after irradiation and in turn induced apoptosis by itself. Consistent with this idea, we observed that high doses of U-73122 induced cell death even in the absence of any other apoptosis inducer (not shown). When viewed together, our data strongly indicate that PLC␥ has no significant function in anti-apoptotic growth factor signaling.
The Ras/Raf/MAPK pathway was shown to mediate cell survival during Drosophila development (36) and in the neuron system (15). Because neither deletion of the Grb2-binding site in the F6 mutant nor mutation of the RasGAP-binding site (F771) had any influence on cell survival, it is unlikely that Ras activation is involved in anti-apoptotic signaling in VSMC. Jung et al. (37) reported recently that ERK activation suppresses apoptosis in VSMC. Because the ␤PDGFR activates ERK via activation of PI3K, it is possible that PI3K-mediated ERK activation is required for cell survival. Our data demonstrated that co-incubation with the specific MEK-inhibitor PD98059 did not reverse PDGF-BB-mediated prevention of apoptosis. This suggests that protection from apoptosis by growth factors such as PDGF is at least in VSMC not ERK-dependent or demands a much higher degree of ERK activation. Jung et al. (37) incubated VSMC with fetal calf serum, which activates many types of receptors and thus provides excessive cross-talk. Thus, it seems that ERK activation is not sufficient to rescue cells from apoptosis and that co-activators derived from other pathways are necessary for full induction of ERKdependent survival effects.
Another effector of the ␤PDGFR, STAT3, was discussed to protect from UV irradiation-induced apoptosis (5). In our model we cannot completely rule out anti-apoptotic effects of activated STAT3, because every ChiR mutant was able to activate STAT3 (not shown). However, a ChiR mutant, which mediated a stronger induction of STAT3 activity than the ChiR-WT (Y771, in which the RasGAP-binding site is restored in the F5 receptor), did not prevent UV irradiation-or H 2 O 2 -induced apoptosis indicating that STAT3 is at least not able to mediate anti-apoptotic effects in the absence of other signaling pathways. It is nevertheless possible that constitutively active or overexpressed STAT3 is capable of releasing survival effects.
Because physiological apoptosis inducers like tumor necrosis factor ␣ have only little effect on VSMC apoptosis (38) we used two alternative approaches to induce apoptosis. UV irradiation was used as a general inducer of apoptosis, which acts by DNA damage and activation of death receptors (39) and is commonly used to induce apoptosis in various cell types. To rule out that our observations were limited to one specific apoptosis inducer, we additionally used a second unrelated approach, H 2 O 2 , which activates the mitochondrial death pathway to induce apoptosis in VSMC. This may even represent a more relevant approach, because large amounts of H 2 O 2 are released from circulating macrophages during vascular conditions such as atherosclero-sis, which influence the progression of these pathologies (40,41). However, a potential disadvantage of H 2 O 2 is its capability to activate many signaling pathways like PLC␥ (42). Because we obtained identical results with both apoptosis inducers, our findings indicate that growth factor-dependent activation of PI3K/Akt inhibits multiple death pathways.
In summary, the direct comparison of the various signaling pathways that are turned on in a PDGF-stimulated cell demonstrates that only one signaling enzyme recruited to the ␤PDGFR, PI3K, accounts for ␤PDGFR-mediated cell survival, whereas all other receptor-associated signaling molecules do not play a significant role. Only PI3K is able to efficiently activate the anti-apoptotic Akt signaling pathway and mediates its survival effects via inactivation of pro-apoptotic proteins including BAD and Forkhead transcription factors.