Phosphoinositide 3-Kinase-dependent Regulation of Interleukin-3-induced Proliferation

We have demonstrated previously that class IA phosphoinositide 3-kinases play a major role in regulation of interleukin-3 (IL)-3-dependent proliferation. Investigations into the downstream targets involved have identified the MAPK cascade as a target. Expression of Δp85 and incubation with LY294002 both inhibited IL-3-induced activation of Mek, Erk1, and Erk2. This was most pronounced during the initial phase of Erk activation. The Mek inhibitor, PD98059, blocked IL-3-driven proliferation, an effect enhanced by Δp85 expression, suggesting that inhibition of Mek and Erks by Δp85 contributes to the decrease in IL-3-induced proliferation in these cells but that additional pathways may also be involved. To investigate the mechanism leading to decreased activation of Erks, we investigated effects on SHP2 and Gab2, both implicated in IL-3 regulation of Erk activation. Expression of Δp85 led to a reduction in SHP2 tyrosine phosphorylation and its ability to interact with Grb2 and Gab2 but increased overall tyrosine phosphorylation of Gab2. LY294002 did not perturb SHP2 interactions, potentially related to differences in the effects of these inhibitors on levels of phosphoinositides. These results imply that the regulation of Erks by class IA phosphoinositide 3-kinase may contribute to IL-3-driven proliferation and that both SHP2 and Gab2 are possibly involved in this regulation.

We have demonstrated previously that class I A phosphoinositide 3-kinases play a major role in regulation of interleukin-3 (IL)-3-dependent proliferation. Investigations into the downstream targets involved have identified the MAPK cascade as a target. Expression of ⌬p85 and incubation with LY294002 both inhibited IL-3-induced activation of Mek, Erk1, and Erk2. This was most pronounced during the initial phase of Erk activation. The Mek inhibitor, PD98059, blocked IL-3-driven proliferation, an effect enhanced by ⌬p85 expression, suggesting that inhibition of Mek and Erks by ⌬p85 contributes to the decrease in IL-3-induced proliferation in these cells but that additional pathways may also be involved. To investigate the mechanism leading to decreased activation of Erks, we investigated effects on SHP2 and Gab2, both implicated in IL-3 regulation of Erk activation. Expression of ⌬p85 led to a reduction in SHP2 tyrosine phosphorylation and its ability to interact with Grb2 and Gab2 but increased overall tyrosine phosphorylation of Gab2. LY294002 did not perturb SHP2 interactions, potentially related to differences in the effects of these inhibitors on levels of phosphoinositides. These results imply that the regulation of Erks by class I A phosphoinositide 3-kinase may contribute to IL-3driven proliferation and that both SHP2 and Gab2 are possibly involved in this regulation.
The survival, proliferation, and differentiation of cells of the hemopoietic cell compartment is regulated by the actions of a diverse range of cytokines. We and others (1,2) have focused on the actions of interleukin-3 (IL-3), 1 which acts on cells of the myeloid lineage and is important for the survival and proliferation of mast cells and basophils. IL-3 induces the activation of a number of signaling cascades (reviewed in Ref. 3), including the Ras/Raf/MEK/MAPK module (4,5) and the class I A phosphoinositide 3-kinase family (6,7). Recent challenges have been to determine the functional requirement of these pathways in IL-3 action.
Phosphoinositide 3-kinases are a family of lipid kinases, whose products, phosphoinositide 3,4-bisphosphate (PI(3,4)P 2 ) and phosphoinositide 3,4,5-triphosphate (PI(3,4,5)P 3 ), are important intracellular second messengers (8). IL-3 activates members of the class I A family of PI3Ks, which consist of a regulatory subunit (p85) and a 110-kDa catalytic subunit (8). Three forms of p110 (␣, ␤, and ␦) have been identified, with the p110␦ isoform being largely restricted in its expression to cells of the immune system (7,9). PI3Ks have been implicated in regulating a broad range of physiological processes, including the control of proliferation, cell survival, vesicle trafficking, and glucose transport. We reported previously that expression of dominant negative p85 (⌬p85), which specifically targets class I A PI3Ks, results in a dramatic reduction in IL-3-induced proliferation (10), accompanied by reduced activation of protein kinase B and a concomitant decrease in the phosphorylation of the pro-apoptotic Bcl-2 family member, Bad. However, we observed only minimal effects on apoptosis, and we proposed that class I A PI3Ks play a major role regulating IL-3-dependent proliferation rather than cell survival (10). Regulated expression of active p110␣ (11) and microinjection studies (12)(13)(14) as well as studies using mice bearing knock outs of either the regulatory or catalytic subunits of this class of PI3Ks also support a role of these enzymes in regulating cell proliferation (15)(16)(17).
One of the major pathways that has been widely implicated in the regulation of proliferation is the Ras/Raf/MEK/MAPK module (18). This protein kinase cascade is activated by a wide variety of agents, including IL-3 (4,5). Activation of this pathway in fibroblasts culminates in translocation of activated Erk1 and -2 to the nucleus, where they phosphorylate members of the Ets family of transcription factors, including Elk-1 and components of the AP1 complex (19). This leads to transcriptional activation of genes important for cell cycle progression, DNA synthesis, and cell division. Recent evidence has shown that G 1 cyclins, including cyclin D1, are transcriptionally regulated by MAPK-dependent pathways (20,21), providing a direct link between Erks and cell cycle regulation.
PI3Ks have been implicated in the regulation of the MAPK cascade. Inhibitors of PI3K (the fungal metabolite wortmannin or the LY294002 compound) lead to inhibition of agonist-stimulated Erk1 and -2 activation (22)(23)(24)(25), and different forms of dominant negative p85 mutants also inhibit agonist-stimulated Erk activation (25)(26)(27). However, it has also been reported that PI3K inhibition does not influence Erk activity (28,29). This issue has been more closely addressed, and it appears that there may be cell-and receptor-specific requirements for the effects observed. For example, Erk activation by EGF in COS7 cells can be inhibited by PI3K inhibitors at low doses of * This work was supported by project grants from The Medical Research Council and The Wellcome Trust (to M. J. W.). 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.
Our demonstration that expression of dominant negative p85 inhibits IL-3-driven proliferation of BaF/3 cells, without significantly affecting apoptosis, led us to investigate in greater detail the downstream targets of PI3K that may contribute to this decrease in proliferation. Here, we demonstrate that expression of ⌬p85, or treatment with the PI3K inhibitor, LY294002, led to consistent decreases in the ability of IL-3 to induce Mek, Erk1, and Erk2 activation. The Mek activation inhibitor, PD98059, blocked IL-3-induced proliferation, and this effect was enhanced in combination with ⌬p85 expression. The alterations in Mek and Erk activation in ⌬p85-expressing cells were coupled with alterations in tyrosine phosphorylation of SHP2 and its association with both Grb2 and the scaffolding protein Gab2. However, LY294002 did not lead to altered SHP2 interactions, which could be related to differences in the effect of ⌬p85 expression versus LY294002 on in vivo levels of phosphoinositides. We propose that PI3K activity is involved in the initial phase of activation of Erks by IL-3 and that the decrease in Erk activation observed upon ⌬p85 expression may contribute to the decrease in IL-3-induced proliferation we observe.

EXPERIMENTAL PROCEDURES
Cell Culture-Murine IL-3-dependent BaF/3 cells and derivatives expressing the tetracycline transactivator from the plasmid pUHD15-1, containing a puromycin-selectable marker, were the kind gifts of DNAX, Palo Alto, CA (32). The generation, characterization, and growth of clones inducibly expressing Myc epitope-tagged ⌬p85 and control cells containing empty vector have been described previously (10). All cells were maintained at 37°C, 5% (v/v) CO 2 in a humidified incubator in RPMI 1640 medium supplemented with 10% (v/v) fetal bovine serum (Life Technologies, Inc.), 20 M 2-mercaptoethanol, 100 units of penicillin/streptomycin, and 2 mM glutamine (RPMI media), with the addition of 5% (v/v) conditioned media from WEHI3B cells as a source of murine IL-3. ⌬p85-expressing clones and empty vector controls were cultured in the presence of 2 g/ml tetracycline.
Cytokine-dependent Proliferation Assays-For sodium 3Ј-(1-[(phenylamino)-carbonyl]-3,4-tetrazolium]-bis(4-methoxy-6-nitro)benzenesulfonic acid hydrate (XTT) dye reduction assays, recombinant murine IL-3 (rmIL-3, R & D Systems) was set up in triplicate at a range of doses (0.5 pg/ml to 2 ng/ml), in serum-free AIM-V media (Life Technologies, Inc.) in the presence or absence of 2 g/ml tetracycline and in flat-bottomed 96-well trays (Nunc). Transfectants were washed three times in HBSS, resuspended at 2 ϫ 10 4 cells in AIM-V media, and plated at 1000 cells/well in 100 l total volume. Cells were incubated for 72 h at 37°C. 25 l of a solution containing 1 mg/ml XTT and 25 M phenazine methosulfate was added/well for the final 4 h of incubation. The soluble formazan product was measured at 450 nm on a Dynatech MR5000 plate reader.
Induction of ⌬p85 Expression-Transfectants were washed three times with Hanks' buffered saline solution, pH 7.5, and resuspended in IL-3-containing RPMI at 1 ϫ 10 5 cells/ml in the absence (to induce expression) or the presence of 2 g/ml tetracycline (Tet, to suppress expression) and were incubated at 37°C for 16 -20 h. In some cases cells were incubated for a further 14 h in media containing 0.5%(v/v) WEHI3B conditioned media in the presence or absence of Tet. This IL-3 deprivation was used to increase IL-3 receptor cell surface expression.
Cell Stimulations and Immunoprecipitations-Treatment of all cells with IL-3 was carried out as described previously (33). Unless otherwise stated, recombinant murine IL-3 was used at a concentration of 20 ng/ml, which we had determined previously to induce maximal levels of tyrosine phosphorylation of cellular substrates. Cell pellets were lysed in solubilization buffer (50 mM Tris-HCl, pH 7.5, 10% (v/v) glycerol, 1% (v/v) Nonidet P-40, 150 mM NaCl, 5 mM EDTA, 1 mM sodium orthovanadate, 1 mM sodium molybdate, 10 mM sodium fluoride, 10 g/ml aprotinin, 10 g/ml soybean trypsin inhibitor, 10 g/ml leupeptin, 0.7 g/ml pepstatin, 40 g/ml phenylmethylsulfonyl fluoride) at concentrations between 1 and 2 ϫ 10 7 cells/ml. Nuclei and insoluble material were removed by centrifugation for 3 min at full speed in a microcentrifuge at 4°C. Clarified supernatants were used for immunoprecipitations, which were performed as described previously (33,34). The following antibodies were used: 5 g of 9E10 (anti-Myc) monoclonal antibody; 2 g of rabbit polyclonal antibody against SHP-2 (sc-280, Santa Cruz Biotechnology); 5 g of rabbit polyclonal antibody against Gab2 (06-967, Upstate Biotechnology, Inc), 2 g of rabbit polyclonal antibody against Grb2 (sc-255, Santa Cruz Biotechnology). The glutathione S-transferase fusion protein containing the SH2 domain of Grb2 has been described previously (34). Precipitations were carried out using 10 g of fusion protein per sample.
Preparation of Nuclear Extracts-Following stimulation, cell pellets were resuspended in 400 l of ice-cold nuclear extract buffer 1 (10 mM HEPES, pH 7.9, 10 mM KCl, 0.1 mM EDTA, 0.1 mM EGTA, 1 mM sodium orthovanadate, 1 mM sodium molybdate, 10 mM sodium fluoride, 40 g/ml phenylmethylsulfonyl fluoride, 10 g/ml aprotinin, 10 g/ml soybean trypsin inhibitor, 10 g/ml leupeptin, 0.7 g/ml pepstatin) and left on ice for 15 min. 15 l of 10% Nonidet P-40 was added to each sample. The samples were vortexed briefly and then the nuclei were pelleted by centrifugation for 5 min at full speed in a Hereaus Biofuge at 4°C. The supernatant was retained as the cytosolic extract. Nuclear pellets were rinsed briefly in nuclear extract buffer 1 and then resuspended in 30 -50 l of nuclear extract buffer 2 (20 mM HEPES, pH 7.9, 0.4 M NaCl, 1 mM EDTA, 1 mM EGTA, 1 mM sodium orthovanadate, 1 mM sodium molybdate, 10 mM sodium fluoride, 40 g/ml phenylmethylsulfonyl fluoride, 10 g/ml aprotinin, 10 g/ml soybean trypsin inhibitor, 10 g/ml leupeptin, 0.7 g/ml pepstatin) for examination of nuclear proteins alone, or 400 l when relative distribution experiments were performed, and rotated at 4°C for 30 min. Debris was removed by centrifugation for 5 min as described previously. Clarified supernatants were retained and used for immunoblotting.
Measurement of D3 Phosphoinositides-In vivo lipid labeling and measurement of levels of deacylated lipids were performed essentially as described (35). Cells were washed and preincubated for 30 min in phosphate-free Dulbecco's modified Eagle's medium containing 0.5% (v/v) dialyzed fetal bovine serum and then incubated for 3 h at 2 ϫ 10 7 /ml in phosphate-free Dulbecco's modified Eagle's medium containing 0.5 mCi/ml 32 P i . After extensive washing 10 7 cells were either left untreated as controls or stimulated for 1 min with 20 ng/ml rmIL-3. Reactions were terminated by addition of 500 l (chloroform/methanol/ water; 32.6:65.3:2.1 v/v) and vortexing. Phases were separated first by addition of 200 l of chloroform containing 10 g/ml Folsch lipids and 200 l of 2.4 M HCl, 5 mM tetrabutylammonium sulfate, and centrifugation. A second separation was carried out with 400 l of 1 M HCl, 5 mM EDTA. Dried lipids were deacylated at 53°C for 40 min in 25% (w/v) methylamine/methanol/butanol-1 (4:4:1) and re-dried. Water-soluble lipids were washed twice with butanol-1/petroleum ether (b.p. 40 -60°C)/ethyl formate (20:4:1 v/v), re-dried, and deacylated 32 P-labeled lipids analyzed by HPLC using a 12.5-cm Whatman Partisphere SAX column. Samples were eluted using a gradient of buffers A (water) and B (1.25 M (NH 4 ) 2 HPO 4 adjusted to pH 3.8 with phosphoric acid at a flow rate of 1 ml/min: 0 min, 0% B; 5 min, 0% B; 45 min, 12% B; 60 min, 30% B; 61 min, 100% B; 66 min, 0% B; 90 min, 0% A and B. Counts were analyzed using an on-line radiodetector. Elution of lipids was calibrated using CD28-activated Jurkat cells (35,36). When effects of ⌬p85 expression on D3 phosphoinositide levels were examined, cells were incubated for 18 h ϮTet prior to labeling. When the effects of LY294002 were examined, cells were labeled, separated, and half-incubated for 30 min with 30 M LY294002, or Me 2 SO alone, prior to IL-3 stimulations.

RESULTS
⌬p85 Expression Reduces Activation of Erk1 and Erk2 in Response to IL-3-Our previous demonstration (10) that class I A PI3Ks play a major role in controlling IL-3-driven proliferation prompted us to investigate the downstream targets of PI3Ks involved in this response. The extracellular regulated/ mitogen-activated protein kinases (Erk1 and -2) have been linked to regulation of cellular proliferation in many systems (18), and a number of reports support a role for PI3Ks in regulating the activation of Erks (22)(23)(24)(25)(26)(27). Therefore, we investigated whether expression of ⌬p85 influenced activation of the MAPKs, Erk1 and Erk2, by IL-3.
To address the involvement of PI3Ks in regulation of Erk by IL-3 we used IL-3-dependent BaF/3 transfectants that express dominant negative p85 (⌬p85) in a tetracycline-regulated manner (10). Transfectants were induced to express ⌬p85 by removal of tetracycline or left uninduced in the presence of Tet. Cells were subsequently stimulated for different times with maximal doses of IL-3 (previously determined as 20 ng/ml), and the activity of Erk1 and -2 was assessed by immunoblotting extracts with antibodies that detect the dual phosphorylated (activated) forms of both Erk1 and Erk2.
The results in Fig. 1A clearly demonstrate that expression of ⌬p85 reduces the activation of Erk1 and Erk2 following IL-3 simulation. Interestingly, the effect is most pronounced at early time points (Fig. 1A, ␣-pErk, compare lanes 3 and 9, left panel) with little to no effect at later time points (Fig. 1A, compare lanes 5 and 11, left panel). These results suggest either that (i) PI3K is required during the early phase of Erk activation by IL-3 or (ii) at early time points maximal occupancy of IL-3 receptors has not been achieved, and the reduced activation of Erks is due to the ability of ⌬p85 expression to affect IL-3 signaling at suboptimal levels of receptor engagement. No effect of removing tetracycline on IL-3-induced Erk1 and -2 activation was observed in cells containing empty vector (see Fig. 1A, right panel, CONTROL). These membranes were stripped and reprobed with anti-Erk (␣-Erk) antibodies to assess loading of the gels (Fig. 1B).
⌬p85 Expression Reduces Activation of Mek in Response to IL-3-The immediate upstream activators of Erks are the MAPK kinases, Mek1 and Mek2. To investigate whether inhibition of class I A PI3K by expression of ⌬p85 affects IL-3induced activation of Mek, we utilized an antibody that specifically detects the phosphorylated and hence active form of Mek1 and -2. The same immunoblots as in Fig. 1, A and B, were stripped and reprobed with the anti-phospho-Mek antibody (␣-pMek), with the results shown in Fig. 1C. Expression of ⌬p85 clearly reduces the activation of Mek in response to IL-3. As with the effect on Erk1 and -2 activation, Mek phosphorylation is significantly reduced at early time points following IL-3 treatment. No effect on Mek activation was observed in control cells incubated in the presence or absence of Tet (Fig. 1C, right panel, CONTROL). Fig. 1D shows expression of ⌬p85 in this experiment.
Effect of ⌬p85 Expression on Activation of Mek, Erk1, and Erk2 at Different Doses of IL-3-It has been demonstrated in some systems that the requirement for PI3K activity in activation of Erks is dependent on the dose of stimulating factor used and level of receptor expression (30,31). Given that we observed a decrease in Erk activation upon ⌬p85 expression at early time points following IL-3 treatment, we wanted to examine if this was related to the dose of cytokine used. IL-3 dose-response analyses were performed using a stimulation time of 5 min, and the effects of ⌬p85 expression on IL-3induced Erk1, Erk2, and Mek activation were examined. Maximal activation of all three kinases was consistently observed at concentrations of IL-3 of 5 ng/ml and above (see Fig. 2). Expression of ⌬p85 resulted in significant inhibition of Erk1, Erk2, and Mek activation at all concentrations of IL-3 tested (see Fig. 2, A and C). When time course analyses using 5 ng/ml IL-3 were performed (Fig. 3), maximal activation in the presence of Tet was observed after 10 min of IL-3 treatment, with significant inhibition of Mek, Erk1, and -2 activation observed when ⌬p85 was expressed at both early and later time points (2-, 10-, and 20-min treatment times; Fig. 3, A and C). At times of 30 min, little difference was observed in activation of Erk1 and Erk2, but Mek was still clearly inhibited, suggesting it is very sensitive to expression of ⌬p85. The same membranes were stripped and reprobed with anti-Erk (Figs. 2B and 3B) and/or anti-Mek antibodies (Fig. 3D) for loading controls and p85 antibodies to monitor induction of p85 expression (Figs. 2D and 3E). In Fig. 2B mobility shifts are apparent for both Erk1 and Erk2 in the ϩTet samples treated with IL-3, due to their phosphorylation (see Fig. 2A). This complicates the assessment of loading, although in the untreated cells, the levels appear similar, as do the levels of endogenous p85 observed (see Fig.  2D). Also, we consistently noted that the anti-Mek antibody appeared to recognize the phosphorylated forms of Mek better than the unphosphorylated forms. These results suggest that expression of ⌬p85 can inhibit IL-3-induced activation of Mek and Erks across a range of doses of IL-3, but at sub-optimal doses, the inhibition appears to be more sustained. We have attempted to investigate if expression of a constitutively active form of p110␣ (p110*, see Ref. 37) potentiates IL-3-induced proliferation and Erk activation, but after repeated rounds of transfection using the same Tet-off regulated expression system, we were unable to isolate clones reproducibly expressing full-length p110*. In addition, we have also been unable to express kinase-dead versions of p110␣, suggesting the cells require extremely tight regulation of the levels of these proteins.
⌬p85 Expression Dramatically Inhibits Phosphorylation of Nuclear Erks-The dramatic reduction in cellular proliferation resulting from ⌬p85 expression we described previously (10) suggests a failure to activate genes and proteins required for cell cycle progression. Nuclear Erk1 and -2 are responsible for phosphorylating and activating Ets family transcription factors, such as Elk-1, important for mitogenesis and regulation of cyclin D1. Therefore, we investigated whether expression of ⌬p85 affected the levels of activated Erks present in the nu-cleus following IL-3 stimulation. IL-3 dose-response analyses demonstrated that doses of 1 ng/ml IL-3 and greater induced activation of both Erk1 and Erk2 in cells incubated in the presence of Tet. Expression of ⌬p85 reduced the ability of IL-3 to activate Erk1 and -2 (see Fig. 4A(i)). Somewhat surprisingly, when the immunoblot was reprobed with anti-Erk antibodies, Erk1 and -2 were present in the nucleus under basal conditions, and their levels did not appear to be affected by IL-3 ( Fig.  4A(ii)).
Presence of Mek in Nuclear Fractions in BaF/3 Cells-The presence of Erk1 and -2 in the nucleus under basal conditions prompted us to investigate whether Mek was also present in the nucleus under the same conditions. Mek contains a nuclear export sequence, but recent data suggest that it can shuttle back and forth from the cytosol to the nucleus (38). To investigate whether Mek was also present in the nuclear fractions of BaF/3 cells, the immunoblots shown in Fig. 4A(i) were stripped and reprobed with either anti-phospho-Mek (␣-pMek, Fig.  4A(iii)) or anti-Mek antibodies (␣-Mek, Fig. 4A(iv)). It can be clearly observed that Mek is present in the nuclear fraction of BaF/3 cells, in both unstimulated and IL-3-treated cells (Fig.  4A(iv)). Probing with the anti-phospho-Mek antibody demonstrated Mek phosphorylation in cells treated with at least 1 ng/ml IL-3. This phosphorylation was inhibited in the presence of ⌬p85 expression.
The experiments described above examined concentrated nuclear extracts directly but give no indication as to the relative levels of Erks and Mek in the nucleus versus the cytosol under the different conditions used. To address this, cytosolic and nuclear extracts were prepared from the same cell sample, and the same number of cell equivalents of each fraction was immunoblotted with anti-Erk and Mek antibodies. The results are shown in Fig. 4B, (i) and (ii). In cells incubated in the presence FIG. 3. ⌬p85-mediated inhibition of Erk1 and Erk2 activation is enhanced at sub-optimal doses of IL-3. ⌬p85 BaF/3 cell clone 1D8 was incubated in the presence (ϩ) or absence (Ϫ) of 2 g/ml tetracycline (Tet) for 18 h. Cells were treated for the indicated times (in minutes) with 5 ng/ml IL-3 or left untreated (0). 40 g of cell protein extract was loaded per sample and separated through 10% acrylamide gels, and duplicate gels were prepared. Immunoblotting was performed with an antibody specific for active Erks (␣pErk) (A). This membrane was stripped and reprobed with an anti-Erk antibody (B) or an anti-p85 antibody (E). An antibody specific for phosphorylated Mek (␣-pMek) was used as the primary antibody (C), and the membrane was stripped and reprobed with an antibody specific for Mek (D). The positions of phosphorylated and unphosphorylated enzymes are indicated. or absence of Tet, the vast majority of Erk1, Erk2, and Mek are located in the cytosolic fraction, with very low levels (less than 1% of total) detected in the nucleus. These levels did not detectably alter upon IL-3 treatment or ⌬p85 expression. In addition, in these experiments it was very difficult to detect any Mek in the nucleus, suggesting that Mek is present at even lower levels than the Erks. To assess the purity of the cytosolic versus nuclear extracts, duplicate gels were prepared and immunoblotting was performed using antibodies recognizing the murine IL-3 receptor chain, Aic2A, and the largely cytosolic adaptor protein, Shc. The results shown in Fig. 4B, (iii) and (iv), demonstrate that neither Aic2A nor Shc could be detected in the nuclear extracts, suggesting the presence of Erk1 and Erk2 in the nuclear extracts is unlikely to be due to cytosolic contamination. In addition, immunofluorescence using an antipan Erk antibody detected low levels of Erk in the nucleus of these cells under basal conditions and when stimulated with IL-3 (data not shown).
The PI3K Inhibitor LY294002 Also Inhibits Erk Activation in Response to IL-3 in the Cytosol and Nucleus-To complement our studies targeting p85 of class I A PI3Ks, we also investigated the effects of the PI3K inhibitor LY294002 on IL-3induced activation of Erk1 and Erk2 and Mek in BaF/3 cells. We have shown previously that incubation with LY294002 inhibits IL-3-induced proliferation of BaF/3 cells and also reduces IL-3-induced activation and phosphorylation of PKB (39), in a manner similar to expression of ⌬p85 (10). Untransfected BaF/3 were preincubated with 10 or 30 M LY294002 for 30 min prior to IL-3 stimulation. Cytosolic and nuclear extracts were prepared from the same cells. Preincubation with LY294002 reduced activation of Erk1, -2, and Mek in the cytosol and nucleus (Figs. 5, A and B, (i) and (iii)). As with ⌬p85 expression, this effect was most evident at early times following IL-3 treatment (1 and 2 min), although still apparent at 10 min. At later time points, inhibition by LY294002 was much less pronounced (data not shown).

PD98059 Inhibits IL-3-driven Proliferation of BaF/3 Cells-
The results above demonstrate that ⌬p85 expression and incubation with LY294002 consistently reduce activation of Erk1, -2, and Mek in both the cytosol and nucleus following IL-3 treatment. We were initially interested in identifying potential targets of PI3K action in regulating proliferation, and these results support a role of the Erk cascade in this process. To examine this on a cellular level, we investigated what effects the Mek activation inhibitor, PD98059, has on IL-3-induced proliferation in the absence and presence of ⌬p85 expression. When ⌬p85 is expressed we observe a reduction in IL-3-driven proliferation (Fig. 6A), as we have reported previously (10). Incubation with 50 M PD98059, which has been used previously in BaF/3 cells to reduce thrombopoietin-induced proliferation (40), reduced IL-3-induced proliferation, to a level similar to that observed with ⌬p85 expression alone. These results are consistent with the view that Mek activation is important for IL-3-driven proliferation of BaF/3 cells, implicating a physiological role for Mek downstream of PI3K in regulation of IL-3 growth. The combination of PD98059 and ⌬p85 expression reduced IL-3-driven proliferation further. There are two possible explanations for this as follows: (i) treatment with either inhibitor alone did not inhibit Mek activation completely, but both PD98059 treatment and ⌬p85 expression together did, or (ii) additional pathways controlled by PI3Ks are also involved in the regulation of IL-3-dependent proliferation. To address the first possibility, the effectiveness of PD98059 at inhibiting Mek was assessed. BaF/3 cells expressing ⌬p85 or left uninduced were preincubated with 10 or 50 M PD98059 for 30 min, prior to IL-3 treatment. In uninduced cells, 10 M PD98059 led to partial inhibition of Mek and Erks, and at 50 M PD98059 there was still some low but residual activation of Mek, see Fig.  6B. In cells expressing ⌬p85, 10 M PD98059 almost completely inhibited Mek and Erk activation by IL-3, and no activation of either Mek or Erks was detectable in the presence of 50 M PD98059. These results suggest that the presence of both PD98059 and expression of ⌬p85 are more effective at inhibiting Mek which correlate with the further decrease in IL-3induced proliferation observed. However, we cannot rule out the possibility that additional pathways are also regulated by PI3K, and we are currently investigating this.
Expression of ⌬p85 Reduces Coupling of the Tyrosine Phosphatase SHP-2 to Grb2-The effects of expression of ⌬p85 and treatments with either LY294002 or PD98059 suggest that PI3Ks are important in regulation of Erk activation by IL-3, at some point upstream of Mek. The coupling of the IL-3R to activation of the Ras/MAPK cascade may be mediated by a number of signaling complexes. We have previously shown that both SHP2 (41) and Shc (42) bind the IL-3R␤ chain and both SHP2 and Shc are tyrosine-phosphorylated by IL-3 at sites that are then bound by the SH2 domain of Grb2, providing a link to Ras via Sos (34,43). In addition, the scaffolding protein, Gab2, can form complexes with SHP2 and PI3K (44 -46) and has been reported to play a role in activation of Erks (46). We had noticed in whole cell lysates a reduction in IL-3-induced tyrosine phosphorylation of a 70-kDa protein when ⌬p85 was expressed (see Fig. 7A, left panel). SHP2 is a 70-kDa protein, and so we hypothesized that if ⌬p85 expression could affect SHP2 tyrosine phosphorylation and/or its interactions with IL-3R, Grb2 or Gab2, this could provide a mechanism for the effects that expression of ⌬p85 has regulation of Erk activation by IL-3.
We first examined the level of tyrosine phosphorylation of SHP2 in cells expressing ⌬p85, and we found it to be reduced compared with control cells (see Fig. 7A), although the amount of SHP2 immunoprecipitated was equivalent (Fig. 7A, lower  panel). In addition to decreased SHP2 phosphorylation, we observed a decrease in the coprecipitation of a tyrosine-phosphorylated protein of molecular mass of 120 kDa when ⌬p85 was expressed, combined with increased precipitation of a 170-kDa phosphotyrosyl protein, which we have shown to be the scaffolding protein, IRS-2 (data not shown). Given that we observe decreased tyrosine phosphorylation of SHP2, we next examined whether this influenced the ability of Grb2 to interact with SHP2. Pull-down experiments were performed using a glutathione S-transferase fusion protein containing the SH2 domain of Grb2 from both cells induced to express ⌬p85 and control cells. Anti-phosphotyrosine immunoblotting demonstrated decreased levels of a 70-kDa tyrosine-phosphorylated protein precipitated by Grb2 in cells expressing ⌬p85 but not in control cells (see Fig. 7B, (i) and (ii), upper panels). Reprobing with anti-SHP2 antibodies demonstrated this 70-kDa protein corresponded to SHP2 (Fig. 7B, (i) and (ii), lower panels) and showed a decrease of ϳ50% in the amount of SHP2 precipitat- ing with the Grb2 fusion protein, upon ⌬p85 expression. Similar results were observed when anti-Grb2 immunoprecipitates were performed (Fig. 7B(iii)). These results confirmed that expression of ⌬p85 resulted in a reduction in SHP2 tyrosine phosphorylation and its ability to interact with Grb2. Levels of the 50-kDa phosphotyrosyl protein pulled down by the Grb2 fusion protein and in Grb2 precipitates, previously identified as Shc (34), were unchanged in the ⌬p85-expressing cells.
Expression of ⌬p85 Results in Increased Tyrosine Phosphorylation of Gab2 but Decreased Association between SHP2 and Gab2-We have reported previously that ⌬p85 expression results in increased tyrosine phosphorylation of a protein of 100 kDa (10). We were interested to determine if this protein was Gab2 and, if so, whether binding of SHP2 to Gab2 is influenced by ⌬p85 expression. Gab2 immunoprecipitates were prepared from cells induced to express ⌬p85 or left uninduced as controls. The results are shown in Fig. 8A and clearly demonstrate that Gab2 tyrosine phosphorylation induced by IL-3 is increased upon expression of ⌬p85. We next examined whether this increased Gab2 phosphorylation resulted in alterations of proteins able to interact with Gab2. Thus, precipitates were immunoblotted with antibodies specific for SHP2 and showed a consistent decrease in SHP2 associated with Gab2 when ⌬p85 was expressed (see Fig. 8B). When we performed the reciprocal experiment, less Gab2 could be detected in SHP2 immunoprecipitates when ⌬p85 was expressed (Fig. 8C). Interestingly, we also observed that in total cell lysates, in uninduced cells, Gab2 underwent a very dramatic shift in mobility, so that after 10 min of IL-3 treatment a triplet was detected. However, in the cells expressing ⌬p85, only a doublet could be detected, suggesting that not all phosphorylated forms were present, despite the overall increase in Gab2 tyrosine phosphorylation observed in these samples. Similar observations were noted in Gab2 precipitates.
Given these alterations in SHP2-Gab2 associations in ⌬p85expressing cells, we also examined the effects of LY294002 on SHP2 and Gab2 tyrosine phosphorylation, the association of SHP2 and Gab2, and the association of SHP-2 and Grb2. Somewhat surprisingly, we consistently observed little effect of LY294002 treatment on phosphorylation of SHP2 or Gab2 or their association induced by IL-3 (see Fig. 9). In addition, LY294002 did not consistently affect the association of SHP2 with Grb2 (data not shown). In addition to examining effects of short term treatment of LY294002 (30 min), we repeated these experiments following long term treatment of BaF/3 with LY294002 (16 h). However, we again did not detect any consistent effects of LY294002 on SHP2 associations (data not shown).
Previously, we have demonstrated that ⌬p85 expression and LY294002 both inhibit proliferation of BaF/3 cells in response to IL-3, lead to greater than 90% inhibition of PKB phosphorylation and activation by IL-3, and inhibit IL-3-induced hyperphosphorylation of Bad (10,39). Here we have demonstrated that expression of ⌬p85 and LY294002 also both lead to decreases in the early phase of IL-3-induced activation of Erk1, Erk2, and Mek. However, whereas ⌬p85 expression influenced SHP2-Gab2 and SHP2-Grb2 associations, LY294002 appeared not to.
One possibility that could potentially account for these differences is if ⌬p85 and LY294002 have different effects on the in vivo levels of D3 phosphoinositides, and hence we examined this directly. The results obtained are shown in Tables I and II. In ⌬p85 BaF/3 transfectants (Table I), in the presence of Tet, IL-3 induced a 3-fold increase in PI(3,4)P 2 levels and a 3.9-fold increase in PI(3,4,5)P 3 levels. Expression of ⌬p85 abolished the ability of IL-3 to induce an increase in PI(3,4)P 2 or PI(3,4,5)P 3 levels. In fact, the levels of both lipids decreased following IL-3 stimulation, most likely due to the fact that SHIP and PTEN were still activated in these cells by IL-3. In BaF/3 cells (Table  II), IL-3 induced a 2-fold increase in PI(3,4)P 2 levels. However, a 30-min preincubation with 30 M LY294002 reduced the basal levels of PI(3,4)P 2 in unstimulated BaF/3 cells by greater than 50%, and this fell further following IL-3 treatment. Levels of PI(3,4,5)P 3 were also reduced upon incubation with LY294002 but to a lesser extent. These results demonstrate that there are in fact differences in the effects of these PI3K inhibitors on levels of D3 phosphoinositides in vivo. ⌬p85 expression appears to inhibit specifically the IL-3-induced increase in PI(3,4)P 2 and PI(3,4,5)P 3 levels, whereas LY294002, reduced the basal levels of these lipids in unstimulated cells. These differential effects on D3 phosphoinositides could explain the difference in the effects on SHP2-Gab2 and SHP2-Grb2 associations.
Given the results above, we wanted to re-affirm that both  DISCUSSION We have previously demonstrated that specifically targeting class I A PI3Ks by regulated expression of dominant negative ⌬p85 results in a dramatic reduction in the ability of IL-3 to induce proliferation of the IL-3-dependent cell line, BaF/3 (10). In this report we have examined targets of PI3K action that may contribute to this decrease in IL-3 proliferation. We now report that expression of ⌬p85 leads to a reduction in the ability of IL-3 to induce activation of components of the Ras/ MAPK cascade, namely Mek, Erk1, and Erk2. This occurs in both cytosolic and nuclear locations and is most apparent at early time points following IL-3 treatment. The pharmacological PI3K inhibitor, LY294002, mimicked these effects. Inhibition of Mek activation by PD98059 reduced IL-3-driven proliferation, consistent with the view that the effect of ⌬p85 expression on Mek and Erk activation is physiologically relevant. We also demonstrate that tyrosine phosphorylation of SHP2 is decreased upon ⌬p85 expression, and this correlated with a decrease in the ability of Grb2 to interact with SHP2. SHP2 association with the scaffolding protein Gab2 was also reduced upon expression of ⌬p85, although interestingly, overall levels of Gab2 phosphorylation were increased. The decrease in coupling of SHP2 to both Grb2 and Gab2 are consistent with the observed reduction in Mek and Erk activation, since both of these pathways have been implicated in the regulation of MAPK activation by IL-3. Incubation with LY294002 did not appear to influence these protein interactions, potentially pointing toward a different mode of action, and we detected differences in the effects of these PI3K inhibitors on the in vivo levels of D3 phosphoinositides. This study highlights the complexity of signaling interactions occurring upon IL-3 stimulation and their relationship to functional responses.
Previous reports have implicated PI3Ks in the regulation of the Erk cascade (22)(23)(24)(25)(26)(27)47). In many cases, inhibition of PI3Ks results in a decrease in activation of Erks (22)(23)(24)(25)(26)(27)47) although in other situations, little effect has been reported (28,29). The fact that we observe inhibition of Erk1, Erk2, and Mek in cytosolic and nuclear locations in response to IL-3 following either expression of ⌬p85 or treatment with LY294002 suggests that in IL-3-dependent BaF/3 cells, PI3K is involved in the regulation of this cascade. It was of interest to note that at maximal doses of IL-3 (20 ng/ml) marked effects on activation of MAPKs were observed in the very early phase of Erk activation (Figs. 1, 3, and 5, 1-10 min). At sub-optimal doses of IL-3 (5 ng/ml, Fig. 3), the effects appeared to be more sustained, implying inhibition by ⌬p85 was more effective. These observations are consistent with the reports of others (30,31) that suggest the role of PI3K in regulation of Erk activity is related to receptor levels, cytokine dose, and cell type. Importantly, our data suggest that PI3K activity is involved in the initial phase of Erk activation by IL-3.
Our data with PD98059 support a role for Mek in IL-3-dependent proliferation, and this in turn supports the view that the decrease in Mek and Erk activation observed upon ⌬p85 expression contributes, at least partially, to the decrease in IL-3-induced proliferation we have reported previously (10). Analyses are under way to assess the effects of constitutively active and dominant negative versions of Mek in these cells, but previous reports have demonstrated inhibition of proliferation of BaF/3 cell transfectants by PD98059 (40,48), and Mek function has been shown to be necessary for optimal IL-3 stimulation of S-phase entry, again in BaF/3 (49). This accumulated evidence is consistent with the view that Mek is important for regulation of proliferation. Both ⌬p85 and PD98059 together were more effective at inhibiting Mek, which may have led to the enhancement of inhibition of IL-3-induced proliferation observed with both inhibitors. However, PI3K may have additional targets involved in regulation of proliferation, and recent evidence indicates cyclin and cyclin-dependent kinase inhibitor expression can be regulated by PI3K-dependent mechanisms (50 -57). We have made similar observations in our system, 2 but whether these events are Mek-dependent or -independent have yet to be fully determined, since Mek can regulate cyclin D1 expression (20).
It is not clear whether PI3K acts at some point upstream or downstream of Ras in regulation of MAPK activation, and this is still an area of some controversy (30,37,58,59). We have shown previously that IL-3 activates Ras and MAPKs in a rapid and transient manner (4,5), and it has been reported  10. Expression of ⌬p85 or treatment with LY294002 reduce IL-3-induced activation of PKB to similar extents. A, BaF/3 ⌬p85 transfectants were incubated in the presence (ϩTet) or absence (ϪTet) of 2 g/ml tetracycline for 18 h. B, BaF/3 cells were incubated with 10 M LY294002 (LY) or vehicle alone (CON) for 30 min. Cells were treated for the indicated times (in minutes) with 20 ng/ml IL-3 or left untreated (0). Extracts equivalent to 4 ϫ 10 5 cells were immunoblotted with an antibody that specifically recognizes phosphorylated serine 473 of PKB (␣-pPKB). that IL-3 regulates activation of A-Raf in a PI3K-dependent manner (47). Three documented pathways have been implicated in IL-3 regulation of the Ras-MAPK cascade as follows: (i) the Shc-Grb2-Sos pathway (34), (ii) the SHP2-Grb2 pathway (43), and (iii) the SHP2-Gab2 pathway (46,60,61). Therefore, to gain some insight into the mechanism of action of ⌬p85, we examined whether ⌬p85 expression had any effect on the regulatory protein interactions upstream of Ras. No effects were observed on the coupling of Shc to Grb2 when ⌬p85 was expressed (see Fig. 7B and not shown). However, we consistently observed a reduction of ϳ50% in SHP2 tyrosine phosphorylation in response to IL-3 and decreased interaction with Grb2, suggesting that phosphorylation of Tyr-304 or Tyr-542, the two potential Grb2 SH2 domain recognition motifs in SHP2 (43), was phosphorylated to a lesser extent when ⌬p85 was expressed. The scaffolding/adaptor protein, Gab2, has also been implicated in regulation of SHP2 action, and we and others (44 -46) have shown that it associates with both SHP2 and PI3K upon IL-3 stimulation. SHP2 association with Gab2 or Gab1 appears to be important for the ability of SHP2 to activate Erk1 and Erk2 in a number of cell systems (46,(62)(63)(64)(65). Therefore, the decreased association between SHP2 and Gab2 when ⌬p85 is expressed could also contribute to the reduction in IL-3-induced activation of Mek and Erk that we observe. A recently published report supports our observations. Yart et al. (66) demonstrated that in EGF-treated Vero cells, inhibition of PI3K activity reduced Erk activation and also reduced association of SHP2 with Grb2 and association of SHP2 with Gab1. This group implicated PI3K activity upstream of Ras, with Gab1 downstream of PI3K. Interestingly, in our experiments LY294002 treatment did not consistently lead to clearly detectable alterations in Gab2-SHP2 or SHP2-Grb2 associations, despite the fact that both expression of ⌬p85 and LY294002 treatment reduced IL-3-induced PKB activation to similar extents ( Fig. 10; see Refs. 10 and 39). However, it must be remembered that ⌬p85 will specifically target class I A PI3K, known to be activated by IL-3 (6), whereas LY294002 inhibits virtually all classes of PI3Ks (67). Therefore, the apparent discrepancies in their influences on SHP2 interactions could be due to differences in the effects of ⌬p85 expression and LY294002 treatment on in vivo D3 phosphoinositide levels. Indeed, we have shown that whereas ⌬p85 expression only abolished IL-3-induced increases in both PI(3,4)P 2 and PI(3,4,5)P 3 (Table I), pretreatment with LY294002 reduced levels of PI(3,4)P 2 in unstimulated cells, as well as abolishing IL-3-induced increases in PI(3,4)P 2 and PI(3,4,5)P 3 (Table II). These findings argue for greater specificity of inhibiting class I A PI3Ks by expression of ⌬p85, compared with LY294002. Therefore, in the LY294002-treated cells, the overall balance of membrane phospholipids will be altered which could perturb the presence of PH-domain containing proteins, certain of which may be important for regulation of tyrosine phosphorylation/dephosphorylation events that may regulate SHP2 interactions with Grb2 and Gab2. Clearly, further detailed investigations are needed to clarify the modes of action of LY294002 and ⌬p85 in regulation of Mek and Erk activation.
It is also apparent that Gab1 and Gab2 may be regulated differently. The fact that we observe no effect of short or long term LY294002 treatment on Gab2 tyrosine phosphorylation is consistent with a recent report (68) showing that treatment of BaF/3 with wortmannin has no effect on overall levels of Gab2 tyrosine phosphorylation. This is in contrast to Gab1, where treatment of cells with wortmannin reduced tyrosine phosphorylation of Gab1 stimulated by EGF treatment, suggesting membrane localization via D3-phosphorylated lipids is impor-tant for its phosphorylation (66,69). It is not clear why this difference is observed between Gab1 and Gab2.
In the case of ⌬p85 expression one possible mechanism through which it could act relates to the Btk family tyrosine kinase, Tec, which can be transiently tyrosine-phosphorylated and activated in response to IL-3 in myeloid and BaF/3 cells (70,71). Also in BaF/3 cells, it has been reported that p85 can associate with Tec in response to IL-3, possibly via tyrosine 594 of Tec and p85 SH2 domains (72). Therefore, ⌬p85 may bind to Tec potentially inhibiting its activity. If Tec is important for phosphorylation of substrates downstream of IL-3, then less active Tec could lead to reduced phosphorylation of SHP2 at Grb2-binding sites and less tyrosine phosphorylation of Gab2 at SHP2-binding sites. This would result in less SHP2 associating with Gab2 and, given that Gab2 is a potential SHP2 substrate, could lead to increased tyrosine phosphorylation of Gab2. We are currently investigating these alternatives.
It is possible that the effect of ⌬p85 expression on elevation of Gab2 tyrosine phosphorylation is not related to the effects we observe on SHP2-Gab2 or SHP2-Grb2 associations. It is also formally possible that expression of ⌬p85 in BaF/3 cells protects Gab2 from dephosphorylation, via SH2 domain interactions, and as with all such approaches, it is difficult to rule this possibility out entirely. However, it should be noted that we do not see increased phosphorylation of all IL-3 substrate proteins in ⌬p85-expressing cells, and although ⌬p85 is associated with Gab2 in unstimulated cells (see Fig. 8D), upon IL-3 treatment, tyrosine phosphorylation of Gab2 rises considerably, but associated ⌬p85 only increases by 2-3-fold (Fig. 8, A and D).
Although LY294002 did not affect Gab2 tyrosine phosphorylation, it reduced the characteristic electrophoretic mobility shift of Gab2 observed following IL-3 stimulation (Fig. 9), as did expression of ⌬p85 (Fig. 8), suggesting a feedback mechanism. Similar results were observed using PD98059, 3 implicating PI3Ks and Erks. Similar findings have been reported recently by Gu et al. (68), and in addition, they observe a shift in Gab2 mobility in the absence of detectable Gab2 tyrosine phosphorylation, suggesting that the shift may not be due to tyrosine phosphorylation but to threonine or serine phosphorylation.
The demonstration that we detect effects on activation of Erks in the nucleus adds further to the evidence supporting a role for MAPKs in regulating PI3K-dependent IL-3-induced proliferation. Reports have suggested that sustained activation of Erks is required for proliferation (73) and that nuclear localization is important for Erk action (74,75). Our observation that Erk1, Erk2, and Mek are present in nuclear extracts under non-stimulated conditions is surprising, particularly as other reports, using fibroblast cell systems, demonstrate a clear exclusion of Mek from the nucleus but the appearance of Erks (74,75). When we examined the relative distribution between cytosol and nucleus of Mek and Erks, the overwhelming majority of both Mek and Erks was retained in the cytosol. Thus, although detectable levels were present in the nucleus, these represented a very small proportion of the total protein. The recent characterization of a Mek-dependent export of Erks from the nucleus (38), is consistent with a view that Mek can constantly shuttle to and from the nucleus. Hence, at any given time, a small proportion would be in the nucleus, which is consistent with our observations.
Overall, the evidence presented here suggests that expression of ⌬p85, or treatment with LY294002, results in a decrease in activation of Erk1, -2, and Mek by IL-3 which may contribute to the decrease in proliferation of these cells observed in response to IL-3. In addition, the mechanism of this regulation by the class I A PI3Ks may be via influences on SHP2 and Gab2, which are consistent with the roles of these molecules in IL-3 signaling via the Ras/Erk pathway. Further detailed investigations are under way that should enhance our understanding of these important interactions and regulation of cytokine signaling.