Phosphatidylinositol 3′-Kinase, But Not S6-Kinase, Is Required for Insulin-Like Growth Factor-I and IL-4 To Maintain Expression of Bcl-2 and Promote Survival of Myeloid Progenitors¹

Phosphatidylinositol 3′-kinase (PI 3-kinase)⁴ is a membrane-associated lipid kinase that phosphorylates the 3′ carbon of phosphatidylinositol (1) and appears to be critically involved in the regulation of cellular growth (2) and differentiation (3). Although activation of PI 3-kinase is now known to also be associated with enhanced survival of both neurons (4, 5) and hemopoietic cells (6, 7, 8), the downstream effectors that are responsible for this PI 3-kinase-mediated inhibition of apoptosis are only beginning to be understood. One potential mediator is the serine/threonine kinase p70 S6-kinase (S6-kinase), which was originally discovered to phosphorylate the S6 component of the ribosomal complex and to promote RNA translation (9). In both murine T cells (10) and 32D cells transfected with a mutant insulin receptor that binds PI 3-kinase directly upon ligand stimulation (11), inhibition of PI 3-kinase blocks S6-kinase activity. In promyeloid cells transfected with the wild-type insulin receptor, the intracellular docking protein insulin receptor substrate-1 (IRS-1) is required to activate both PI 3-kinase and S6-kinase after insulin treatment, which leads to the proliferation of these cells (12). Similarly, activation of S6-kinase appears to be responsible for the proliferation of rat L6 fibroblasts in response to fibroblast growth factor (13) and the insulin-like growth factor-I (IGF-I)-promoted survival of rat pheochromocytoma cells differentiated into sympathetic neurons by nerve growth factor (5, 14). These same cells are dependent upon IGF-I activation of PI 3-kinase to prevent apoptosis (14). These data suggest that S6-kinase may play an important downstream role in regulating the biological activities induced by PI 3-kinase following activation of the IGF-I receptor.

A number of cytokines and growth factors can activate PI 3-kinase via diverse mechanisms (15). Intrinsic tyrosine kinase receptors, such as platelet derived growth factor (16), nonreceptor tyrosine protein kinases like JAK-1 (17, 18), the src family kinases involved in TCR signaling (19), and the high m.w. docking proteins IRS-1 and IRS-2 (7, 20), are all involved in the activation of PI 3-kinase. IL-4 activates PI 3-kinase in several types of hemopoietic cells, including T lymphocytes (18) and myeloid progenitors (21). This activity appears to be dependent upon JAK-1 tyrosine phosphorylation of IRS-1/IRS-2. Phosphorylated tyrosine residues in YXXM motifs of IRS-1 (22) and IRS-2 (23) are bound by the src homology domain 2 (SH2) domains of the p85 regulatory subunit of PI 3-kinase. Tyrosine phosphorylation of IRS-1 and the subsequent association with the p85 subunit of PI 3-kinase have now been shown to be critical for IL-4-mediated inhibition of apoptosis (7, 8). Similarly, activation of IGF-IR in human T cells leads to the phosphorylation of IRS-1 and the subsequent association with the p85 subunit of PI 3-kinase (24). Two recent studies have suggested that a member of the antiapoptotic Bcl-2 family, Bcl-x_L, is a downstream mediator of IGF-I-stimulated PI 3-kinase and the subsequent survival of neuronal cells (5, 25).

Activation of PI 3-kinase is not always required for cell survival. For example, we have shown that IL-3 increases the lipid kinase activity of PI 3-kinase, but this activity is not necessary for the survival of myeloid progenitor cells (6). This finding has been confirmed (7), clearly establishing that there are PI 3-kinase-dependent and -independent cell survival pathways. Here we show that IL-4 and IGF-I, but not IL-3, activate both S6-kinase and PI 3-kinase, which suggested that S6-kinase might play a role in cell survival via a pathway that is linked to PI 3-kinase. However, inhibition of S6-kinase activity did not affect the capability of either IGF-I or IL-4 to promote cell survival. Instead, IL-3, IGF-I, and IL-4 enhanced expression of Bcl-2. PI 3-kinase activity was required for IGF-I and IL-4, but not IL-3, to increase expression of Bcl-2 and to promote cell survival. Therefore, these data establish that, although Bcl-2 is a common target for cytokines that promote survival of promyeloid cells, only IL-4 and IGF-I increase expression of this protein via a pathway that is dependent upon PI 3-kinase.

RPMI 1640 (Media Tech, Herndon, VA) or DMEM (Life Technologies, Gaithersburg, MD) was prepared with 2 g/L of sodium bicarbonate, 100 U/ml penicillin, and 100 μg/ml streptomycin (Sigma; St. Louis, MO). Cultures were supplemented with 5% heat inactivated horse serum (Sigma; <25 pg endotoxin/ml, as assessed by Limulus amebocyte assay, Associates of Cape Cod, Woods Hole, MA) and 0.25 U/ml of recombinant murine IL-3 (Biosource, Camarillo, CA). FDCP-1 cells (FDCP; a gift from Dr. Lawrence Rohrschneider, Fred Hutchinson Cancer Center, Seattle, WA) were maintained at 37°C at 95% humidity and 7% CO₂. In experiments that measured apoptotic populations in FDCP cells treated with growth factors or cytokines, the cells were washed three times in RPMI (400 × g) and incubated in serum-free DMEM for 4 h with inhibitors of PI 3-kinase or S6-kinase, at the indicated concentrations, before the addition of growth factors. For the in vitro kinase assays, the cells were washed three times in RPMI and then treated for the indicated times with growth factors in serum-free medium. IL-3 and IL-4 were purchased from Biosource, IFN-γ was from Life Technologies, and IGF-I was obtained from Intergen (Purchase, NY). The ELISA kit used to measure IL-3 was from Biosource.

To determine the apoptotic population of FDCP cells, flow cytometric analysis was used (EPICS V; Coulter Instruments, Miami, FL). Cells (5 × 10⁵/ml) were incubated 24 h with IGF-I (100 ng/ml), IL-3 (25 U/ml), IL-4 (25 ng/ml), growth hormone (250 ng/ml), prolactin (250 ng/ml), or IFN-γ (250 U/ml) in the presence or absence of 1 nM rapamycin, 100 nM wortmannin, or 10 μM LY294002 in serum-free medium. Hoechst 33342 (7 ng/ml; Sigma) was added to the cells for 7 min at 37°C, and the cells were placed on ice. Immediately before analysis, 2 μg/ml of propidium iodide (PI; Sigma) was added to the cells. At least 10⁴ cells were analyzed for each sample by a double exclusion staining protocol using PI and Hoechst 33342, which excluded PI-positive cells and evaluated Hoechst 33342 staining vs forward angle light scatter (FALS) to determine the apoptotic populations, as previously described (6).

Specific activity of S6-kinase was determined by ³²P incorporation into S6 peptide (Upstate Biotechnology, Lake Placid, NY) as previously described (26). Cells (5 × 10⁶) were washed in RPMI 1640 and treated with indicated concentrations of IGF-I, IL-3, or IL-4 for 20 min in the presence or absence of 1 nM rapamycin, 100 nM wortmannin, or 10 μM LY294002 before lysis with 1 ml of lysis buffer (10 mM potassium phosphate, 1 mM EDTA, 5 mM EGTA, 10 mM MgCl₂, 50 mM β-glycerophosphate, 1 mM Na₃VO₄, 2 mM DTT, 40 μg/ml PMSF, and 0.1% Nonidet P-40, pH 7.4). Clarified lysates (14,000 × g) were added to a mixture containing a rabbit anti-p70 S6-kinase IgG specific for residues 511–525 of rat/human S6 kinase (2 μg/sample; Upstate Biotechnology) complexed to 50 μl of protein G Sepharose (Pharmacia, Piscataway, NJ) and incubated for 4 h at 4 C. The beads were washed once with lysis buffer and once with kinase buffer (20 mM Tris-HCl, pH 7.5, 10 mM MgCl₂, 0.4 mM DTT). Following the final wash, the immune complexes were suspended in 50 μl of kinase buffer containing 100 μM unlabeled ATP, 200 μCi/nmol [γ-³²P]ATP (Amersham, Arlington Heights, IL), and 125 μM S6 peptide. The reaction was allowed to proceed for 30 min at 30°C and was terminated by adding 20 μl of 250 mM EDTA and boiling for 5 min. Following a brief centrifugation, duplicate samples of the supernatant (25 μl) were applied to phosphocellulose paper (Upstate Biotechnology), and radioactivity was determined using a Beckman liquid scintillation counter, LS 6000IC (Irvine, CA). S6-kinase was not detected in immunocomplexes precipitated with a control rabbit Ig (data not shown).

FDCP cells were washed three times in RPMI 1640 (400 × g) and were incubated for 4 h in serum-free medium with or without 1 nM rapamycin, 100 nM wortmannin, or 10 μM LY294002. Inducible phosphotyrosine-associated activity was determined by incubating 5 × 10⁷ cells with IGF-I (100 ng/ml), IL-4 (25 ng/ml), or medium alone for 30 s. Cells were then homogenized in 1 ml containing 1% Nonidet P-40, 50 mM Tris-HCl, 100 mM NaCl, 50 mM NaF, 10 mM tetrasodium pyrophosphate, 2 mM Na₃VO₄, 2.5 mM benzamidine, 1 mM PMSF, and 1 μM DTT, pH 7.4. Clarified lysates (20 min, 14,000 × g) were then added to 50 μl of protein G Sepharose (Pharmacia) conjugated with 2 μg of anti-phosphotyrosine Ab 4G10 (Upstate Biotechnology) that had been diluted with 100 μl of fast flow Sepharose CL4 beads (Sigma). The immune complexes were then washed twice with wash buffer A (1% Nonidet P-40, 1 mM DTT, and PBS, pH 7.4), buffer B (0.5 M LiCl, 1 mM DTT, and 100 mM Tris-HCl, pH 7.4), and buffer C (10 mM NaCl, 1 mM DTT, and 100 mM Tris-HCl, pH 7.4). PI 3-kinase activity was not detected in immunocomplexes precipitated with a control isotype-matched mouse IgG2bκ Ab (data not shown).

The lipid kinase assay was initiated by adding 60 μl of sonicated l-α-phosphatidylinositol (0.33 mg/ml) in kinase buffer containing 20 mM HEPES, 0.04 mM EGTA, 0.4 mM NaPO₄, 48 μM [γ-³²P]ATP (2.1 μCi/nmol; Amersham), and 10 mM MgCl₂ at pH 7.1. The reactions were performed at room temperature for 15 min and terminated by the addition of 15 μl of 4 N HCl. An equal volume mixture of chloroform/methanol (200 μl) was used to extract the lipids following vortexing and centrifugation (14,000 × g) for 10 min. The chloroform-containing lipid phase was reextracted with 150 μl of an equal volume mixture of 0.15 N HCl/methanol, followed by vortexing and centrifugation at 14,000 × g for 10 min. Ten microliters were then resolved by thin layer chromatography using a chloroform/methanol/ammonium hydroxide (75:58:17 (v/v)) running buffer, as previously described (6). Detection of phosphorylated phosphoinositol was performed by autoradiography on XAR film (Eastman Kodak, Rochester, NY), and measurement of band intensity was by Phosphorimager analysis (Molecular Dynamics, Sunnyvale, CA).

Cells were treated with medium alone, IGF-I, IL-3, or IL-4 in the presence or absence of 100 nM wortmannin or 10 μM LY294002 in serum-free medium for 24 h, and blots were performed with 50 μg of protein from lysates of 10⁶ cells in homogenization buffer (1% Nonidet P-40, 50 mM Tris-HCl, 100 mM NaCl, 1 mM PMSF). Protein determination was performed on the clarified lysates (20 min; 14, 000 × g) using the MicroBCA protein assay (Pierce, Rockford, IL). Protein from the lysates was then boiled in SDS PAGE loading buffer (10% SDS, 10 mM NaCl, 1 mM EDTA, 1% bromophenol blue, and 200 mM DTT) for 5 min and separated on 12% polyacrylamide gels. The proteins were then transferred to Trans-blot PVDF membrane (Bio-Rad, Hercules, CA), blocked for 1 h in 5% nonfat dry milk at room temperature, and incubated with hamster anti-murine Bcl-2 Ab (2.0 μg/ml) for 2 h at room temperature in PBS-0.1% Tween 20. The blot was then incubated with murine anti-hamster IgG (1 μg/ml) and subsequently with horseradish peroxidase-labeled sheep anti-mouse IgG (1:3000). The blots were then developed with enhanced chemiluminescence substrate (Amersham) and subsequently exposed to autoradiographic XAR film (Eastman Kodak).

In additional experiments, 50 μg of protein from cells treated with medium, IGF-I, IL-3, or IL-4 in the presence or absence of wortmannin (100 nM) or LY294002 (10 μM) was separated as above and transferred to PVDF for detection of Bax. The membranes were probed with 1:3000 dilution of rabbit anti-mouse Bax Ab (PharMingen) in PBS-0.1% Tween 20 for 2 h at room temperature. The membranes were subsequently incubated with horseradish peroxidase-labeled murine anti-rabbit IgG (Amersham) at room temperature for 4 h. Proteins were identified using enhanced chemiluminescence substrate and subsequent exposure to autoradiographic film (Eastman Kodak).

All experiments were repeated a minimum of three times. Results were analyzed using a general linear model with the Statistical Analysis System (SAS Institute, Cary, NC) (27), and differences between treatments were detected with Student’s t test.

Overexpression of IRS-1 in myeloid progenitors permits both IL-4 and insulin to activate S6-kinase and stimulate the proliferation of these cells (12), but it is unknown whether PI 3-kinase is associated with this activity. To determine the induction of S6-kinase by specific ligands in FDCP cells, optimal concentrations of IGF-I (6), IL-3 (6), and IL-4 (data not shown) that activate PI 3-kinase were added to 10⁶ cells in serum-free medium for 20 min, and cell lysates were immunoprecipitated with an Ab specific for S6-kinase. As shown in Fig. 1 A, both IGF-I (100 ng/ml) and IL-4 (25 ng/ml) increased the ability of these immunoprecipitates to phosphorylate a specific peptide substrate of S6-kinase by 10-fold ± 1 and 11-fold ± 1 (p < 0.01; n = 3). Interestingly, although we have previously established that IL-3 (25 U/ml) increases PI 3-kinase activity in FDCP cells (6), treatment with IL-3 did not significantly stimulate S6-kinase activity.

These experiments established that IGF-I and IL-4, but not IL-3, potently activate S6-kinase in FDCP cells, but they did not determine the potential association of PI 3-kinase in this process. To begin to address this possibility, we cultured FDCP cells with the immunosuppressive macrolide FKBP-12-rapamycin-associated protein (FRAP) kinase inhibitor rapamycin (28), the fungal metabolite wortmannin, which irreversibly binds the p110 catalytic subunit of PI 3-kinase (29), or LY294002, which is a synthetic competitive antagonist for the ATP-binding site of PI 3-kinase (30). When tested in vivo, we have established that the 50% inhibitory concentration (IC₅₀) of wortmannin in FDCP cells is approximately 10 nM (6). In the present experiments, a preliminary dose-response inhibition assay was performed to determine the optimal concentration of inhibitors that would result in substantial blockage (>85% instead of 50% inhibition) of the enzymatic activity (data not shown). This was done because we were interested in blocking most of the enzymatic activity so that we could evaluate its effect upon the in vivo readout of apoptotic cell death. Treatment of cells with rapamycin (1 nM; 27) for 4 h before stimulation with IGF-I or IL-4 inhibited S6-kinase activity by 89% ± 12 and 92% ± 9, respectively (Fig. 1,B; p < 0.01; n = 3). More significantly, wortmannin (100 nM; 28) also potently reduced both the IGF-I- and IL-4-mediated increase in S6-kinase activity by 82% ± 7 and 81% ± 8, respectively (Fig. 1,B; p < 0.01; n = 3). When the chemical inhibitor of PI 3-kinase activity was used, LY294002 (10 μM; 29), the inhibition of S6-kinase activity induced by IGF-I was 87% ± 9, whereas that for IL-4 was 84% ± 11 (Fig. 1 B; p < 0.01; n = 3). When cells were cultured in either medium alone or IL-3, none of the inhibitors affected S6-kinase activity (data not shown). Collectively, these data demonstrate that IGF-I and IL-4, but not IL-3, potently increase the activity of S6-kinase, an effect that may be dependent upon the activation of PI 3-kinase.

To determine the role of ligand-activated S6-kinase in inhibiting apoptosis of myeloid progenitor cells, FDCP cells (5 × 10⁵ cells/ml) were treated with IGF-I (100 ng/ml), IL-4 (25 ng/ml), or IL-3 (25 U/ml) for 24 h. The early apoptotic population was characterized as those cells negative for PI that expressed low forward angle light scatter and high Hoechst 33342 staining, as we previously described (6, 31). All three ligands significantly reduced the apoptotic population from 43% ± 6 in medium-treated cells to 17% ± 4, 6% ± 2, and 21% ± 5 in IGF-I-, IL-3-, and IL-4-treated cells, respectively (Fig. 2; p < 0.01; n = 3). Addition of a blocking dose of rapamycin (1 nM; Fig. 1) did not prevent either IGF-I or IL-4 from promoting the survival of these cells (Fig. 2). Similarly, rapamycin had no effect on the survival of cells treated with IL-3. These data demonstrate that, although IGF-I and IL-4 activate S6-kinase, this enzyme is not essential for IGF-I and IL-4 to enhance the survival of myeloid progenitor cells.

We tested the possibility that IL-4 and IGF-I act indirectly to inhibit apoptosis by inducing the synthesis and release of IL-3. We measured the amount of IL-3 in supernatants from FDCP cells cultured with medium alone, IL-4 (25 ng/ml), or IGF-I (100 ng/ml) for 4, 12, or 24 h. In all cases, the level of IL-3 was below the sensitivity of the ELISA assay (<3 pg/ml). It is therefore unlikely that IL-4 and IGF-I inhibit apoptosis in FDCP cells by inducing the secretion of IL-3.

We (6) and others (7, 25) have established that there are at least two survival pathways in myeloid progenitor cells based upon their requirement for PI 3-kinase. To determine whether PI 3-kinase is required for the inhibition of apoptosis in IL-4-treated FDCP cells, we investigated whether IL-4 would induce PI 3-kinase and whether inhibition of this IL-4-induced enzymatic activity would prevent IL-4-mediated cell survival. We treated 5 × 10⁷ FDCP cells with IGF-I (100 ng/ml), a well-characterized inducer of PI 3-kinase activity in these cells (6), or an optimal concentration of IL-4 (25 ng/ml) that inhibits apoptosis (Fig. 2). Induction of PI 3-kinase activity was measured following immunoprecipitation of cell lysates with a phosphotyrosine Ab, followed by an in vitro lipid kinase assay that measures the phosphorylation of phosphatidylinositol. A representative example is shown in Fig. 3, A and B, and the summary of three independent experiments is given below. The positive control treatment, IGF-I, potently stimulated PI 3-kinase activity by 11-fold ± 2, and a similar 13-fold ± 2 increase in PI 3-kinase activity was observed following treatment with IL-4 (p < 0.01; n = 3). Pretreatment of FDCP cells with wortmannin (100 nM) for 4 h potently inhibited ligand-induced PI 3-kinase activity, reducing lipid phosphorylation by 90% ± 5 in the positive control treatment, IGF-I, and by 86% ± 8 in cells stimulated with IL-4 (p < 0.01; n = 3). A similar reduction in PI 3-kinase activity was obtained in IGF-I- and IL-4-treated cells (87% ± 6 and 84% ± 9, respectively) following a 4-h pretreatment with the chemical inhibitor of PI 3-kinase LY294002 (p < 0.01; n = 3). As expected, incubation of FDCP cells with rapamycin, at a concentration that inhibited S6-kinase activity (Fig. 1) but did not affect cell survival (Fig. 2), did not reduce the capability of either IGF-I (inhibition of 4% ± 1) or IL-4 (reduction of 6% ± 2) to stimulate the activity of PI 3-kinase. These data demonstrate that IL-4 stimulates anti-tyrosine-precipitable PI 3-kinase activity in FDCP cells and confirm that this activity is blocked by both wortmannin and LY294002, but not rapamycin.

PI 3-kinase is essential for IGF-I, but not IL-3, to protect FDCP cells from apoptosis (6). To determine whether IL-4-stimulated PI 3-kinase activity is critical for cell survival, we incubated FDCP cells with wortmannin (100 nM) or LY294002 (10 μM) at concentrations that block >85% of the PI 3-kinase activity (Fig. 3,A) and measured the apoptotic population. As expected, both IGF-I and IL-3 significantly reduced the apoptotic population from 43% ± 4 to 17% ± 2 and 6% ± 2, respectively (Fig. 4; p < 0.01; n = 3). The effect of IGF-I was blocked by both PI 3-kinase inhibitors, as demonstrated by the findings that wortmannin and LY294002 increased the apoptotic population in IGF-I-treated cells from 17% ± 2 to 39% ± 4 and 42% ± 4, respectively (p < 0.01; n = 3). Treatment of cells with IL-4 reduced the apoptotic population from 43% ± 4 to 23% ± 2 (Fig. 4; p < 0.01; n = 3), and this protection was totally blocked in the presence of either wortmannin (39% ± 4) or LY294002 (38% ± 4). Although we have previously demonstrated that IL-3 activates PI 3-kinase activity (6), neither wortmannin nor LY294002 increased the apoptotic population in IL-3-treated FDCP cells (5% ± 2 and 7% ± 2, respectively) compared with cells treated with IL-3 alone (6% ± 2; p > 0.10; n = 3). These data reinforce the original concept that there are both PI 3-kinase-dependent (IGF-I) and -independent (IL-3) pathways involved in inhibition of apoptosis and confirm the results of Zamorano et al. (7) by establishing that IL-4, like IGF-I, requires PI 3-kinase to enhance the survival of myeloid progenitor cells.

Overexpression of Bcl-2 permits FDCP cells to survive in the absence of IL-3 (32, 33) and protects monocytic cells from apoptosis following treatment with inhibitors of PI 3-kinase (25). Similarly, recent evidence in neuroblastoma cells suggests that PI 3-kinase may be involved in the regulation of the Bcl-family members (34). To determine the role of PI 3-kinase in the regulation of the antiapoptotic protein Bcl-2 in hemopoietic cells, we compared the amount of Bcl-2 protein in IL-3-deprived FDCP cells that were subsequently incubated with IL-3 (25 U/ml), IL-4 (25 ng/ml), or IGF-I (100 ng/ml) in serum-free medium for 24 h. The amount of Bcl-2 in 50 μg of lysates was measured by Western analysis using a hamster anti-mouse Ab. A representative Western blot is shown in Fig. 5,A, and the densitometric results of three independent experiments are summarized in Fig. 5, B and C. Cells cultured in medium alone expressed little, but detectable, amounts of the 26-kDa Bcl-2 protein. IGF-I induced a 4.1-fold ± 0.5 increase (p < 0.01) in expression of Bcl-2, and similar results were observed with IL-4 (4.3-fold ± 0.3; p < 0.01) and IL-3 (5.1-fold ± 0.3; p < 0.01) (Fig. 5,B). All three cytokines also caused a similar increase in the expression of Bcl-2 in the experiments shown in Fig. 5 C.

IGF-I, IL-4, and IL-3 increase the activity of PI 3-kinase in FDCP cells (6). To determine whether this enzyme regulates expression of Bcl-2, both wortmannin (Fig. 5, A and B) and LY294002 (Fig. 5,C) were used to inhibit the activity of this enzyme. Wortmannin significantly reduced the induction of Bcl-2 in IGF-I-treated cells from 4.1-fold ± 0.5 to 0.9-fold ± 0.3 (Fig. 5,B), amounting to an inhibition of 90% ± 9 (p < 0.01; n = 3). Similarly, inhibition of PI 3-kinase activity by wortmannin reduced the capability of IL-4 to induce Bcl-2 expression from 4.3-fold ± 0.3 to 0.9-fold ± 0.3 (Fig. 5,B), or a 98% ± 12 decline (p < 0.01; n = 3). However, this inhibitor of PI 3-kinase activity did not affect the expression of Bcl-2 in cells treated with IL-3 (5.1-fold ± 0.3 vs 4.7-fold ± 0.3; n = 3). Similarly, wortmannin did not affect expression of Bcl-2 in cells incubated in medium alone (1-fold ± 0.4 vs 0.8-fold ± 0.3). When the LY294002 inhibitor of PI 3-kinase activity was used, nearly identical results were obtained for IGF-I (92% ± 11 inhibition; p < 0.01; n = 3) and IL-4 (94% ± 11 inhibition; p < 0.01; n = 3), with little affect on Bcl-2 expression in FDCP cells cultured in either medium or IL-3 (Fig. 5 C). Collectively, these data establish that, even though all three proteins increase expression of Bcl-2, only IGF-I and IL-4 do so by a pathway that requires PI 3-kinase. Since neither inhibitor affected the ability of IL-3 to induce Bcl-2, these data further suggest that there exists an alternative pathway for cytokine-mediated maintenance of this antiapoptotic protein.

The 21-kDa protein Bax has recently been shown to act as a tumor suppressor by stimulating apoptosis in vivo (35). To determine whether PI 3-kinase also might regulate expression of this apoptosis-inducing protein, FDCP cells (10⁶) were cultured in medium, IGF-I, IL-3, or IL-4 for 24 h, and Bax was measured in 50 μg of cell lysates using a rabbit anti-mouse Bax Ab (Fig. 6,A). When averaged over three independent experiments, treatment with IGF-I, IL-3, or IL-4 did not significantly alter Bax protein expression (1.1-fold ± 0.5, 1.0-fold ± 0.4, or 1.1-fold ± 0.3 increase, respectively, above medium-treated cells; Fig. 6,B), suggesting that this apoptotic-inducing protein is independent of growth factor or cytokine stimulation. To determine whether PI 3-kinase might regulate Bax expression, we pretreated cells with either wortmannin or LY294002. At concentrations that effectively inhibit PI 3-kinase activity, neither wortmannin (100 nM; Fig. 6,B) nor LY294002 (10 μM; Fig. 6 C) significantly affected Bax levels in cells cultured in medium or with IGF-I, IL-4, or IL-3. These data establish that Bax is not subject to the same regulatory mechanisms as those for Bcl-2, supporting the idea that IL-3, IL-4, and IGF-I increase expression of antiapoptotic proteins rather than reducing the amount of proapoptotic proteins.

All members of the hemopoietic receptor superfamily activate the nonreceptor JAK protein tyrosine kinases (36), so we tested the ability of other members of this family to promote the survival of FDCP cells. IFN-γ, growth hormone, and prolactin are members of this receptor superfamily, and both IFN-γ and growth hormone have also been reported to tyrosine phosphorylate IRS proteins and recruit PI 3-kinase (37). Using flow cytometry, we measured the low forward angle light scatter, high Hoechst 33342 early apoptotic population of growth factor-deprived FDCP cells (5 × 10⁵ cell/ml) treated with IFN-γ (250 U/ml), IL-3 (25 U/ml), growth hormone (500 ng/ml), or prolactin (500 ng/ml) for 24 h. IFN-γ was nearly as effective as IL-3 in enhancing the survival of FDCP cells, reducing the apoptotic population from 43% ± 4 to 9% ± 2 (p < 0.01; n = 3). Neither growth hormone nor prolactin inhibited apoptosis in these cells (data not shown). These data establish that not all members of the hemopoietic receptor superfamily are capable of protecting promyeloid cells.

Inhibition of cell death in both the neuronal and hemopoietic systems by members of the Bcl-2 family is now clearly established, probably by inhibiting the efflux of cytochrome c from the mitochondria into the cytosol (38) by linking ced-4 to ced-3 and therefore inhibiting the activity of the IL-1β-converting enzyme (ICE) family of proapoptotic cysteine proteases (39). While these new data point to potential effector mechanisms that permit Bcl-2 to inhibit apoptosis, the intracellular signals that regulate expression of Bcl-2 and that permit Bcl-2 to integrate competing survival and death signals continue to remain unclear (40). Here we present data that establish that both IL-4 and IGF-I regulate expression of Bcl-2 and the survival of FDCP promyeloid cells via a PI 3-kinase-dependent, S6-kinase-independent pathway. We demonstrate that both IL-4 and IGF-I, but not IL-3, stimulate S6-kinase activity and that this enzymatic activity is directly inhibited by rapamycin and indirectly blocked by two different PI 3-kinase inhibitors, wortmannin or LY294002 (Fig. 1). Although these data suggest that S6-kinase is downstream of IL-4- and IGF-I-stimulated PI 3-kinase, the activation of S6-kinase is not critical for biological responses of FDCP cells to IL-4 or IGF-I, as assessed by an increase in cell survival (Fig. 2). Both IL-4 and IGF-I stimulate PI 3-kinase activity (Fig. 3), and addition of wortmannin or LY294002 abrogates IL-4- and IGF-I-mediated inhibition of apoptosis (Fig. 4) and potently reduces IL-4- and IGF-I-stimulated expression of Bcl-2 in these cells (Fig. 5). Addition of either inhibitor does not prevent IL-3 from promoting the survival of FDCP cells or the expression of Bcl-2 protein, suggesting that IL-3 increases the expression of Bcl-2 and the subsequent survival of myeloid progenitor cells via a PI 3-kinase-independent pathway. Not all ligands that bind to members of the hemopoietic receptor superfamily are able to protect myeloid progenitors from cell death (Fig. 7). Finally, none of the cytokines or inhibitors that we tested reduced expression of the apoptotic inducer Bax (Fig. 6), suggesting that Bax is not regulated via a PI 3-kinase-dependent pathway. Collectively, these data indicate that increased expression of Bcl-2 is critical for IL-4, IGF-I, and IL-3 to inhibit apoptosis in FDCP cells. However, only IGF-I and IL-4 require PI 3-kinase to enhance expression of Bcl-2 and promote the survival of myeloid progenitors.

Although PI 3-kinase has recently been shown to promote survival of both neurons (5) and hemopoietic cells (6, 25), the downstream mediators of PI 3-kinase are only beginning to be elucidated. We suspected Bcl-2 might be a target for IGF-I because overexpression of this protein permits FDCP cells to survive in the absence of IL-3 (32). Data presented here are consistent with this idea in promyeloid cells. Recent studies have demonstrated that a role for activated PI 3-kinase in neurons may be to regulate the expression of Bcl family members, including Bcl-2 and Bcl-X_L, and to subsequently inhibit the activation of ced-3/ICE-like proteases (34, 39). Our data are also in accord with those of Erhardt and Cooper (25), who demonstrated that inhibition of PI 3-kinase by wortmannin or LY294002 results in the apoptotic demise of U937 cells.

The serine/threonine S6-kinase may be a critical substrate of ligand-activated PI 3-kinase (41). However, inhibition of S6-kinase activity with rapamycin does not affect survival of FDCP cells cultured with either IGF-I or IL-4. Similarly, survival of neurons (4), as well as skeletal muscle cell differentiation (42), requires the activity of PI 3-kinase but not S6-kinase. It therefore appears that S6-kinase is not required for IL-3, IGF-I, or IL-4 to protect cells from apoptosis. The recent identification of novel high-molecular mass serine/threonine kinases has revealed a new family of proteins that contain a carboxyl-terminal catalytic domain that is closely related to PI 3-kinase (43). One of these PI 3-kinase homology of mammalian origin is mTOR (mammalian targets of rapamycin), and wortmannin and LY294002 have been shown to directly inhibit the activation of mTOR (44). Our results in vivo are in accord with this observation (Fig. 1). More importantly, however, the mTOR PI 3-kinase family member does not appear to be involved in the survival-promoting activity of IGF-I or IL-4 because rapamycin, which also totally blocked the activation of S6-kinase but not that of PI 3-kinase, had no effect upon the ability of IGF-I or IL-4 to promote cell survival. These data are consistent with the idea that closely related PI 3-kinase family members can have separate and distinct biological functions, at least in myeloid progenitor cells. We have previously established that IGF-I can increase cell survival in the presence of actinomycin D (6), so it is possible that IGF-I and IL-4, acting via PI 3-kinase, maintain expression of Bcl-2 by reducing the degradation of Bcl-2 mRNA or protein. We are now exploring the idea that PI 3-kinase maintains Bcl-2 and cell survival via an alternative mechanism that does not utilize S6-kinase, such as activation of Akt/protein kinase B that has been reported for the IGF-I-promoted survival of fibroblasts (45). Indeed, preliminary results are consistent with the possibility because IGF-I increases the activity of Akt-1 more effectively than IL-3 and because this IGF-I-induced Akt-1 serine kinase activity is inhibited by both wortmannin and LY294002, but not rapamycin.

PI 3-kinase may potentially down-regulate the expression of proapoptotic members of the Bcl-family and subsequently promote cell survival (40, 46). Bax has recently been shown to act as a tumor suppressor, inhibiting tumor growth (35). Here we measured the expression of Bax in FDCP cells and showed that addition of IL-3, IL-4, or IGF-I does not affect the expression of Bax protein. A similar finding has been reported by Akbar et al. (47), who demonstrated that Bax expression remains unchanged in IL-2-deprived T lymphocytes, again suggesting that Bax is not under cytokine control. Since here we show that survival factors act to increase expression of Bcl-2, our findings are not inconsistent with earlier experiments that established that induction of apoptosis following cytokine deprivation is related to a reduction in the amount of Bcl-2 relative to Bax (48).

In this report we significantly extend our previous findings that there are two mechanisms that mediate the survival of myeloid progenitor cells (6). Enhanced expression of Bcl-2 in IGF-I- and IL-4-treated cells requires PI 3-kinase activity, while IL-3 maintains Bcl-2 expression in FDCP cells via a PI 3-kinase-independent pathway. IGF-I stimulates the intrinsic tyrosine kinase activity of its receptor and induces the direct phosphorylation of IRS-2 in myeloid cells (49) whereas IL-4 is dependent upon JAK-1 to mediate tyrosine phosphorylation of IRS-2 (20, 50, 51). These data suggest that other ligands that stimulate JAK activity, such as those in the hemopoietic receptor superfamily that have been shown to affect a number of immune events (52, 53), may inhibit apoptosis in FDCP cells. Here, we demonstrate that IFN-γ, but not growth hormone or prolactin, enhances the survival of myeloid progenitor cells. IFN-γ may act through either the PI 3-kinase-dependent or -independent pathway.

In summary, we have established that IL-4 shares with IGF-I the requirement for PI 3-kinase to promote the survival of myeloid progenitor cells and that this process is independent of S6-kinase. Expression of Bcl-2 is maintained by IL-3 as well as IGF-I and IL-4, but neither cell survival nor expression of Bcl-2 is regulated by PI 3-kinase in IL-3-treated cells. In contrast, both IGF-I and IL-4 depend upon PI 3-kinase activity to promote cell survival and to increase expression of Bcl-2. This PI 3-kinase regulation is specific for Bcl-2 because the antiapoptotic protein Bax is unaffected by either of the cytokines or their downstream inhibitors. Collectively, these data demonstrate that there are at least two pathways that regulate Bcl-2 protein expression and the subsequent inhibition of apoptosis in myeloid progenitor cells and that neither pathway requires the activation of S6-kinase.