Protein Kinase C (cid:100) Specifically Associates with Phosphatidylinositol 3-Kinase Following Cytokine Stimulation*

Phosphatidylinositol (PI) 3-kinase is activated as a result of cytokine-induced association of the enzyme with specific tyrosine-phosphorylated proteins. PI 3-ki- nase lipid products, PI 3,4-P 2 and PI 3,4,5-P 3 , have been shown, in vitro , to directly activate novel and atypical protein kinase C (PKC) isozymes. However, the mechanism by which PI 3-kinase may be involved in regulation of PKC isoforms in vivo is presently unknown. We investigated a possible relationship by looking for associa- tions between these enzymes. We found that in a human erythroleukemia cell line, as well as in rabbit platelets, PI 3-kinase and PKC (cid:100) associate in a specific manner that is modulated by cell activation. Granulocyte-mac- rophage colony-stimulating factor treatment of cells caused increased association of PKC (cid:100) and PI 3-kinase as did treatment of platelets with platelet-activating fac- tor. Results using two PI 3-kinase inhibitors, wortmannin and LY-294002, showed that the former inhibited this association, while the latter did not, suggesting that PI 3-kinase lipid products may not be a prerequisite for the PI 3-kinase/PKC (cid:100) association. Our results also suggest that tyrosine phosphorylation of PKC (cid:100) is not involved in its association with PI 3-kinase.

events have yet to be worked out. Tyrosine phosphorylation of Shc, followed by SH2-mediated association with the adaptor protein Grb-2 and the p21 ras activator, m-SOS1, have been shown to be involved in the activation of p21 ras (5,6,10), and activation of PI 3-kinase in hemopoietic cells is a result of cytokine-induced association of PI 3-kinase with specific tyrosine-phosphorylated proteins (11)(12)(13)(14). A tyrosine-phosphorylated 70-kDa substrate was shown to associate with PI 3-kinase following treatment of cells with IL-3 or GM-CSF (14), although this substrate was subsequently shown to be the tyrosine phosphatase, SH-PTP2, and its association may be indirect in a complex with Grb-2 (9).
The involvement of p21 ras in mitogenesis is well understood, but the role of PI 3-kinase, which is thought to be essential as well (15), is less well characterized. Recent evidence suggests that two serine/threonine kinases, p70 S6 kinase (16,17) and PKB (18), are thought to be activated downstream of PI 3-kinase. In hemopoietic cells, we showed previously that IL-3, IL-5, and GM-CSF as well as IL-4 and Steel factor all activate PI 3 kinase (14), and indeed, this signal is important in the ability of cytokines to inhibit apoptosis (19).
An intriguing set of observations made in the past few years has suggested that PI 3,4-P 2 and PI 3,4,5-P 3 can activate several novel and atypical PKC isoforms (PKC⑀, -␦, -, and -) (20,21). These results were obtained from in vitro assays, but the mechanism whereby PKC isoforms may be activated by PI 3-kinase in vivo is not known. The activation of PKC activity by cytokines has been reported in a few publications (22,23), but it is not clear how this may occur. In the case of cytokines such as IL-3, IL-5, and GM-CSF, there is no evidence that a classical PI-phospholipase C pathway is operating, although evidence for increased production of diacylglycerol derived from phosphatidylcholine has been reported (24 -26).
In an effort to further delineate pathways operating downstream of the related receptors for IL-3, IL-5, and GM-CSF, we have examined the role of PKC isoforms. This is a complex task, as there are at least 11 distinct family members that may be regulated by several independent mechanisms. In this report, we focus on the potential interaction of one specific PKC isoform, PKC␦, with the PI 3-kinase enzyme. We find that the two enzymes can be co-immunoprecipitated as shown both by immunoblotting and enzyme activities. Furthermore, the association is modulated by cytokines, particularly by GM-CSF, in a human hemopoietic cell line. In another system, platelets activated by platelet-activating factor (PAF), we can also demonstrate an increase in the association between the two enzymes following activation of the cells. In addition, increased tyrosine phosphorylation of PKC␦ in response to Fc⑀RI receptor activation in another cell type does not lead to any change in association between PKC␦ and PI 3-kinase. Therefore, we have discovered an association between PI 3-kinase and one specific PKC isoform that is probably independent of tyrosine phospho-* This work was supported by a grant from the Medical Research Council of Canada. 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 1 The abbreviations used are: PI, phosphatidylinositol; IL, interleukin; GM-CSF, granulocyte-macrophage colony-stimulating factor; PAF, platelet-activating factor; PKC, protein kinase C; WM, wortmannin. rylation of PKC␦ and is somehow modulated by specific receptor-mediated events.

EXPERIMENTAL PROCEDURES
Cells-TF-1 cells, a human erythroleukemia cell line (a kind gift from Dr. T. Kitamura, DNAX, Palo Alto, CA), were maintained in RPMI 1640 (Life Technologies, Inc.) with 10% fetal bovine serum (Intergen) plus 5 ng/ml IL-5 (purified recombinant human IL-5 prepared in a baculovirus expression system was kindly provided by Dr. D. Nicholson, Merck-Frosst, Kirkland, Quebec). In experiments in which cells were stimulated with cytokines, human recombinant cytokines were used; IL-3 and GM-CSF were from R&D Systems and IL-5 as described above. RBL-2H3 cells (a kind gift from Dr. John Rivera, NIH) were maintained in RPMI 1640 with 10% fetal bovine serum. Cells were trypsinized and passaged every 3 days.
PI 3-Kinase Activity-Following immunoprecipitation as described above, beads were used to assay for PI 3-kinase activity as described (14,27). Briefly, the beads were washed twice with solubilization buffer and three times with 10 mM Tris-Cl buffer, pH 7.4. Phosphatidylinositol (10 g/sample) in 30 mM Hepes, pH 7.4, and 40 l of kinase buffer (30 mM Hepes, pH 7.4, 30 mM MgCl 2 , 50 M ATP, 200 M adenosine, [ 32 P]ATP 10 Ci/l) were added to the beads. The reaction was stopped after 15 min with 1 N HCl. Chloroform:methanol (1:1) was used to extract the lipid, and the labeled PI-3-P was separated by thin layer chromatography using oxalate-treated aluminum-backed silica gels (EM Scientific) and solvent comprised of NH 4 OH:water:methanol:chloroform. Rabbit platelets were prepared as described (28).
Fc⑀RI Cross-linking in RBL-2H3 Cells-RBL-2H3 cells were trypsinized and washed in RPMI with serum. Cells were passively sensitized by incubation in the same medium with supernatant from a monoclonal antibody line producing IgE specific for dinitrophenol (a kind gift from Dr. John Rivera, NIH) at a dilution of 1:25. After 1 h at 37°C, cells were washed twice in Tyrode's buffer, pH 7.4, with Ca 2ϩ and Mg 2ϩ added, and resuspended in the same buffer at 5 ϫ 10 6 cells/ml. Cross-linking of the Fc⑀RI receptor was achieved by addition of dinitrophenol-human serum albumin (Sigma) at a final concentration of 400 ng/ml. At various times, cells were centrifuged, then solubilized, and PKC␦-immunoprecipitated as described above.

RESULTS AND DISCUSSION
The association of PKC isoforms ␦ and ⑀ with PI 3-kinase was examined in the erythroleukemia cell line TF-1. Cell lysates from TF-1 cells, immunoprecipitated with antibodies to either PKC␦ or PKC⑀ and analyzed by immunoblotting with antibodies to the p85 or the p110 subunits of PI 3-kinase, showed that PI 3-kinase was being co-immunoprecipitated (Fig. 1, A and B,  and Fig. 2). In reciprocal experiments, immunoprecipitation with antibody to p85 co-immunoprecipitated PKC␦ and -⑀, as detected by immunoblotting. Other PKC isoforms found in TF-1 cells include ␣, ␤, , , and . 2 None of the latter PKC isoforms were found to co-immunoprecipitate with PI 3-kinase, as determined by immunoblotting following immunoprecipitation with anti-p85 antibody, either in untreated cells or following stimulation with GM-CSF, IL-3, or IL-5 (data not shown).
We next determined if this association was inducible by cytokine stimulation. After 15 min of stimulation with GM-CSF, there was an increased association of PI 3-kinase in anti-PKC␦ immunoprecipitates. This increased association was marginal in IL-5-treated cells but was not evident in IL-3treated cells (Fig. 1, A and B). There was no detectable increase in PI 3-kinase protein in any of the PKC⑀ immunoprecipitates (Fig. 2). The difference in the effect using GM-CSF compared with IL-3 was very reproducible at this time point, but further 2 S. L. Ettinger and V. Duronio, unpublished observations. studies should be done to test additional times.
To investigate whether the PI 3-kinase co-immunoprecipitated with PKC␦ or PKC⑀ remains active, anti-PKC␦ and anti-PKC⑀ immunoprecipitates were analyzed for PI 3-kinase activity in vitro. GM-CSF stimulation of the cells led to increased PI 3-kinase activity associated with PKC␦ compared with control samples or IL-5-or IL-3-stimulated cells (Fig. 3A), thereby correlating with the detection of PI 3-kinase in immunoblots. Anti-PKC⑀ immunoprecipitates also contained active PI 3-kinase; however, the PI 3-kinase activity was not affected by cytokine treatment of cells.
We next studied the role of PI 3-kinase activity in mediating the association with PKC␦, taking advantage of two potent inhibitors of PI 3-kinase, wortmannin (WM) and LY-294002. Cells were pretreated with 100 nM WM or 50 M LY-294002 prior to cytokine stimulation, conditions that are known to inhibit the majority of PI 3-kinase activity (29,30). 3 WM decreased the association of PKC␦ with PI 3-kinase in all cases (Fig. 3, A and B), but LY-294002 had no effect (data not shown). The relative decreases were consistently more pronounced in GM-CSF-and IL-5-treated cells than in IL-3-treated cells. WM is known to bind covalently to the active site of the p110 subunit, while LY-294002 is a competitive inhibitor. The contrasting results between WM and LY-294002 suggest that the PI 3-kinase/PKC␦ association is not modulated by the lipid products of PI 3-kinase activity but may be affected by alterations of the p110 subunit caused by WM. Alternatively, we cannot rule out the possibility that WM is having its effect on the PI 3-kinase/PKC␦ association independently of its effect on PI 3-kinase activity. Other inhibitory effects of WM have been described (31), and in our laboratory we have recently shown that WM has inhibitory effects on other kinases that are not seen with concentrations of LY-294002 that cause an equivalent inhibition of PI 3-kinase activity. 3 Recent reports have described tyrosine phosphorylation of PKC␦ following stimulation with 12-O-tetradecanoylphorbol-13-acetate, carbachol, substance P, or Fc⑀RI cross-linking (32)(33)(34). In TF-1 cells, a faint band is detected in anti-phosphotyrosine blots of anti-PKC␦ immunoprecipitates, and this band comigrates with PKC␦ (data not shown). However, we have been unable to detect any differences in the levels of tyrosine phosphorylation following cytokine stimulation. Phosphoamino acid analysis of PKC␦ will be necessary to confirm whether PKC␦ is being tyrosine-phosphorylated to any significant extent. In another independent system, we have confirmed the results of Haleem-Smith et al. (34), showing that cross-linking of the Fc⑀RI receptor in RBL-2H3 cells leads to a large increase in tyrosine phosphorylation of PKC␦ (Fig. 4A), but interestingly, no PI 3-kinase was present in PKC␦ immunoprecipitates, either before or after receptor activation (Fig. 4B). Together, these results strongly suggest that the association between PI 3-kinase and PKC␦ that we have observed is independent of PKC␦ tyrosine phosphorylation and is therefore not likely to be mediated by the SH2 domains of the PI 3-kinase p85 subunit.
To determine if PI 3-kinase associates with PKC isoforms in any other signaling systems we next investigated rabbit platelets treated with the potent agonist, PAF. PAF-treated platelets also showed an increase in PI 3-kinase protein co-immunoprecipitated with PKC␦, compared with controls, as 3 M. P. Scheid and V. Duronio, submitted for publication. determined on immunoblots using anti-p85 antibodies (Fig.  5A). This association was again inhibited by WM. The PI 3-kinase was active in the PKC␦ immunoprecipitates as determined by an in vitro PI 3-kinase assay (Fig. 5B).
We also characterized the serine/threonine kinase activity in anti-p85 immunoprecipitates and to date have found Ca 2ϩindependent protein serine/threonine kinase activity as evidenced by phosphorylation of a Ser-containing peptide substrate corresponding to the PKC⑀ pseudo-substrate site. There were no increases in kinase activity in response to calcium, and the activity was at least partially lipid-dependent. These features are consistent with the characteristics of PKC␦ and PKC⑀ activities. We are investigating this further to determine the role PI 3-kinase may play in regulation of the PKC activity, particularly that of PKC␦.
The importance of PI 3-kinase in signal transduction has been pointed out in numerous publications over the past few years. There have been many proteins shown to associate with PI 3-kinase following stimulation of cells, generally as a result of tyrosine-phosphorylated proteins associating via the PI 3-kinase p85 SH2 domains. This type of association is known to result in PI 3-kinase activation, independent of the tyrosine phosphorylation of the p85 subunit itself (35). We demonstrate here for the first time that at least two members of the PKC family of enzymes, PKC␦ and PKC⑀, also associate with PI 3-kinase, and in the case of PKC␦, the association is increased upon stimulation of hemopoietic cells with GM-CSF or upon activation of platelets. In our system, PKC␦ appears to be constitutively tyrosine-phosphorylated to a small extent, as seen on immunoblots. However, we cannot discern differences in levels of tyrosine phosphorylation with cytokine stimulation. Several PKC isoforms contain the consensus sequence recognized by PI 3-kinase SH2 domains in the p85 subunit including PKC␦, which has YNYM (36), and PKC⑀, which has YEMM (37). However, the lack of increased tyrosine phosphorylation of PKC␦ following cytokine stimulation, along with the lack of PKC␦/PI 3-kinase association in a system in which PKC␦ is heavily tyrosine-phosphorylated suggests that we are observing an association that is unlike previously demonstrated interactions with PI 3-kinase. However, we do not know whether we are observing a direct interaction or one that also involves other proteins.
In the TF-1 cells, IL-3, IL-5, and GM-CSF are each able to provide a complete mitogenic stimulus, while only GM-CSF and, to a lesser extent, IL-5 are able to cause increased PI 3-kinase/PKC␦ association. Coupled with the finding that PAF activation of platelets is able to increase the association, the results suggest that the functional role of this interaction may be unrelated to the mitogenic effect of cytokines. There have been conflicting reports regarding the role of PKC␦ in mitogenesis (38,39). Furthermore, as suggested by Myers et al. (40), FIG. 4. Tyrosine phosphorylation of PKC␦ does not result in its association with PI 3-kinase. RBL-2H3 cells were passively sensitized with specific IgE and then either left untreated or stimulated by addition of antigen for 1 min or 10 min. A, anti-phosphotyrosine immunoblots of PKC␦ immunoprecipitates show that cross-linking of the Fc⑀RI receptor results in increased tyrosine phosphorylation in RBL-2H3 cells and increased tyrosine phosphorylation of PKC␦. B, in the same immunoprecipitates, there was no evidence for PI 3-kinase being associated, although the p85 subunit of PI 3-kinase could be easily detected in the whole cell extracts. The whole cell extracts represent protein from 1 ϫ 10 5 cells, and the immunoprecipitatesw ere from 5 ϫ 10 6 cells. Ippt, immunoprecipitate; Pre-IP, cell extract prior to immunoprecipitation.

FIG. 5. Platelet lysates immunoprecipitated with anti-PKC␦ antibody also contain PI 3-kinase.
A, immunoblot analysis of anti-PKC␦ immunoprecipitates of platelet lysates revealed the presence of PI 3-kinase. Treatment of platelets with platelet-activating factor increased the amount of PKC␦ associated with PI 3-kinase. Pretreatment of platelets for 10 min with 100 nM WM decreases this association. The identity of the band below the PI 3-kinase p85 band is not known but may represent a degradation product. B, PI 3-kinase activity is present in the immunoprecipitates, and this activity is reduced by pretreatment of platelets with WM. Ippt, immunoprecipitate. the association of PI 3-kinase with various proteins may be unrelated to a role for that enzyme in mitogenesis. One might also speculate that the PI 3-kinase/PKC␦ association could be related to the effects of GM-CSF and perhaps IL-5 on inducing differentiation, effects that are distinct from those of IL-3. Further studies will be required to delineate the exact function of this novel association as well as other systems in which this type of regulation may be occurring.