Phosphopleckstrin inhibits gbetagamma-activable platelet phosphatidylinositol-4,5-bisphosphate 3-kinase.

Pleckstrin, the prototypic protein containing two copies of the pleckstrin homology domain, is a prominent substrate of protein kinase C in platelets and neutrophils. Both cell types have p85 subunit-containing phosphoinositide 3-kinase (p85/PI3K) and non-p85-containing PI3K (PI3Kγ) that is activated by βγ subunits of heterotrimeric GTP-binding proteins. We have shown that a PI3K product, phosphatidylinositol (PI) 3,4,5-trisphosphate, promotes pleckstrin phosphorylation in platelets. Since pleckstrin homology domains are thought to interact with Gβγ heterodimers and/or PI(4,5)P2, we have examined the effects of recombinant pleckstrins on platelet PI3Kγ and p85/PI3K activities. Depending upon its phosphorylation/charged state, pleckstrin inhibits PI3Kγ, but not p85/PI3K. Pleckstrin-mediated inhibition of PI3Kγ is overcome by excess Gβγ and is restricted to PI(4,5)P2 as substrate, i.e. pleckstrin does not inhibit phosphorylation of PI(4)P or PI. Consistent with this, activation of protein kinase C by exposure of platelets to β-phorbol diester (to increase endogenous pleckstrin phosphorylation) prior to platelet lysis causes inhibition of Gβγ-stimulatable PI3K activity only with respect to PI(4,5)P2 substrate. This phosphopleckstrin-mediated inhibition is overcome by increasing concentrations of Gβγ. We propose that phosphorylation of pleckstrin may constitute an important inhibitory mechanism for PI3Kγ-mediated cell signaling.

To date, two major forms of phosphoinositide 3-kinase (PI3K) 1 have been reported to be activated by cell ligands that stimulate a diverse array of cellular responses, ranging from chemotaxis and aggregation to proliferation and transformation. The first and better known form (here designated p85/ PI3K) (reviewed in Refs. 1 and 2) is a heterodimer composed of 85-kDa (p85) and 110-kDa (p110) subunits, functioning as adaptor and catalytic entities, respectively. Some p85/PI3K is associated with the cytoskeleton of resting platelets. In stimulated platelets, however, more p85/PI3K is activated and recruited to the cytoskeleton and is partially dependent upon the small GTP-binding protein Rho (3)(4)(5)(6). This isoform appears to contribute to the conversion of integrin ␣ IIb ␤ 3 to a fibrinogenbinding conformation (6). The second form (here designated PI3K␥) has a 110-kDa catalytic entity that lacks a binding site for p85, but contains a potential pleckstrin homology (PH) domain (7). PI3K␥ is activated by ␤␥ subunits of heterotrimeric G-proteins (G␤␥) (7)(8)(9) and also associates with the cytoskeletal fraction of thrombin-stimulated platelets (10).
The phosphorylation of pleckstrin, a cytoplasmic protein that contains two different PH domains separated by a 150-residue peptide, is a well known marker of platelet activation (11). The biological functions of pleckstrin are just beginning to be explored. PH domains have been reported to interact both with PI(4,5)P 2 and G␤␥ (12)(13)(14)(15). We have shown that, when overexpressed in COS-1 cells, pleckstrin inhibits PI(4,5)P 2 hydrolysis mediated by phospholipase C activated either by G-protein or growth factor receptors (16,17), consistent with an interaction between pleckstrin and PI(4,5)P 2 . This inhibitory effect is dependent on the phosphorylation of Ser 113 , Thr 114 , and Ser 117 within pleckstrin. Substitution of glycine residues at these sites creates a pleckstrin variant (PLECK G ) that is relatively inactive, while insertion of charged residues (e.g. glutamate) creates a variant (PLECK E ) that is constitutively active. Recently, we reported that a product of PI(4,5)P 2 -directed PI3K activity, PI(3,4,5)P 3 , promotes the phosphorylation of pleckstrin in platelets (18). We have investigated in this study whether pleckstrin affects p85/PI3K and/or PI3K␥ activities and whether any such effects are specific to phosphopleckstrin.

MATERIALS AND METHODS
Assay Components-Human platelets were isolated, and cytoskeletal fractions (CSK R and CSK A ) were prepared from resting or thrombinactivated (2 units/ml, 60 s) washed platelets, respectively, as described (6,10), for assay of p85/PI3K and PI3K␥ activities. We have reported (10) that cytoskeletal preparations from resting platelets (CSK R ) contain p85/PI3K activity that can be stimulated (as is true for immunoprecipitated p85/PI3K) by GTP␥S, but no PI3K␥ activity; however, thrombin-activated platelet cytoskeletal fractions (CSK A ) contain both p85/PI3K and PI3K␥. Approximately 60% of the PI3K activity of CSK A is inhibitable by removing endogenous G␤␥ with ␤ARK-PH, i.e. is due to activated PI3K␥ (6,10). Antibodies to ␣and ␤-p85 isoforms were used to immunoprecipitate p85/PI3K from Triton-soluble fractions or cytosol of resting platelets (6, 10) for studies of p85/PI3K activity. p85/PI3K had a lower apparent K m for PI(4,5)P 2 than did recombinant PI3K␥ (7). 2 In some experiments, platelets were incubated for 60 s at 37°C with 200 nM ␣-PMA or ␤-PMA or buffer prior to lysis with ice-cold Triton buffer and removal from the Triton-soluble fractions of p85/PI3K (Ͼ96% by immunoprecipitation) and of Triton and PMA (by Extractigel) (6). The eluates obtained were used in PI kinase assays.
PI3K␥ was purified Ͼ2000-fold from fresh pig platelet cytosol by sequentially using polyethylene glycol precipitation, Q-Sepharose, gel filtration, and heparin-Sepharose column chromatography. 2 Separation of PI3K␥ from p85/PI3K was achieved using immunoabsorption resin containing coupled anti-p85 antibodies. 2 G␣ subunits had only minor effects on PI3K␥ in comparison with G␤␥, which was purified as published (19). The native size of G␤␥-responsive PI3K (PI3K␥) was estimated to be 210 kDa following gel filtration. No p85 protein was present, as monitored by Western blotting, but p110␥ immunoreactivity, detected using an anti-peptide antibody (the kind gift of Dr. R. Wetzker), copurified with G␤␥-stimulated PI3K activity.
The cDNAs encoding PLECKs (i.e. PLECK WT , PLECK E , and PLECK G ) have been described (16,17). The cDNA coding the interdomain region of PLECK WT was prepared by reverse transcription-polymerase chain reaction, and its sequence was confirmed. These cDNAs were subcloned into a derivative of pET-11b, a plasmid vector that expresses proteins utilizing T7 polymerase, modified to insert a sixhistidine peptide tag at the amino terminus. Expressed proteins were purified over a nickel-agarose column eluted with 20% glycerol, 20 mM Tris-HCl, pH 7.9, 100 mM KCl, 5 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride, and 80 mM imidazole. Elution buffer was substituted for PLECKs in control assays. Coomassie blue staining of recombinant PLECKs confirmed protein concentration and Ͼ95% purity. Furthermore, peptides corresponding to a portion of the pseudo-phosphorylated or non-phosphorylatable interdomain regions (residues 107-117) of PLECK E (QKFARKEERRE; PEPT E ) and PLECK G (QK-FARKGGRRG; PEPT G ), respectively, near the N-terminal PH domain and the highly conserved Trp 92 , were synthesized at the central facility of the University of Pennsylvania.

RESULTS AND DISCUSSION
Phosphopleckstrin Selectively Inhibits PI3K␥ Activity-PI3K␥, purified from platelets, was able to phosphorylate PI and PI(4)P as well as PI(4,5)P 2 in a G␤␥-activated manner, but showed preferential activity with PI(4,5)P 2 (Fig. 1). G␤␥ was also more stimulatory for PI3K␥ acting on PI(4,5)P 2 than on the other substrates. Of especial interest, PI(4,5)P 2 inhibited the activity of purified PI3K␥ for PI or PI(4)P, implying that the same PI3K␥ utilizes all three substrates (Fig. 1). These studies were confirmed in experiments in which equimolar amounts of all three phosphoinositide substrates were present in the same assay mixture, in comparison with individually presented substrates (data not shown).
Pleckstrin greatly inhibited PI3K␥ activity. Under the standard assay conditions (substrate vesicles containing 100 M each PI(4,5)P 2 and PS), PLECK WT , PLECK E (the constitutively activated pseudo-phosphorylated variant), and PLECK G (the phosphorylation-deficient variant) all were equally effective inhibitors of G␤␥-stimulated purified PI3K␥ (Fig. 2) and PI3K␥ in CSK A (data not shown). Increasing the PI(4,5)P 2 concentra-2 X.-w. Tang and C. P. Downes, submitted for publication. tion to 1 mM (a concentration roughly comparable with that estimated to be present at the plasma membrane of cells) (2) by increasing the total substrate vesicle concentration and maintaining the PI(4,5)P 2 /PS ratio at 1:1 overcame inhibition of purified PI3K␥ by PLECK WT and PLECK G , but not inhibition by PLECK E . This diminished inhibition may have resulted in part from the decreased effective concentration of PLECK, assuming that it is bound to PI(4,5)P 2 , thereby decreasing the encounters between PLECKs and PI3K␥, and/or from competition by PI(4,5)P 2 for a PLECK-binding site on PI3K␥. In this scenario, PLECK E would compete most efficiently for PI3K␥. It appears likely (based on studies with other PH domains; see below) that PLECKs bind to PI(4,5)P 2 more efficiently when G␤␥ is present. In any event, these results suggest that all three variants have at least weak inhibitory effects, but that PLECK E , and hence phosphopleckstrin, has a greater affinity for G␤␥/PI(4,5)P 2 than does PLECK WT or PLECK G , or that PLECK E (phosphopleckstrin) has a greater affinity for PI3K␥ than does PLECK WT or PLECK G . Given the molar excess of PI(4,5)P 2 over PLECKs under all conditions (25:1 to 250:1), it seems unlikely that PLECKs (2 M) are acting simply as competitors for PI(4,5)P 2 , somehow sequestering this substrate. In contrast, G␤␥ (0.5 M) would be stoichiometrically limited.
At the 1 mM substrate concentration noted above, PLECK E and PLECK WT , but not PLECK G , were effective inhibitors of PI3K␥ in CSK A (Fig. 3A), whereas only PLECK E inhibited purified PI3K␥ (Figs. 2 and 3A). The apparent discrepancy between the inhibitory activity of PLECK WT on CSK A versus purified PI3K␥ is readily attributable to the ability of CSK A to phosphorylate PLECK WT (see Fig. 6), whereas the purified preparation lacks protein kinase activity capable of phosphorylating pleckstrin (data not shown). In contrast, neither PLECK WT nor any of the variants affected basal or GTP␥Sstimulated p85/PI3K activity. These results were observed using either cytoskeleton from unstimulated platelets (CSK R ) or immunoprecipitated p85/PI3K (Fig. 3B).
Inhibition of G␤␥-stimulated PI3K␥ by Phosphopleckstrin Is Overcome by Increasing the Concentration of G␤␥-To evaluate further the role of pleckstrin in regulating PI(3,4,5)P 3 synthesis by PI3K␥, we tested whether varying the concentration of G␤␥ would affect the inhibitory activity of PLECK E . Inhibition by PLECK E decreased with increasing [G␤␥] at low or high [PI(4,5)P 2 /PS] (Fig. 4). This implies that phosphopleckstrin might be inhibiting PI3K␥ activity by binding to G␤␥, thereby making it unavailable to activate PI3K␥. Alternatively, phosphopleckstrin, while binding PI(4,5)P 2 and G␤␥ near PI3K␥, might impair the activity of PI3K␥ by a more direct route, e.g. one dependent upon a phosphopleckstrin-susceptible site on PI3K␥, and overcome by G␤␥. Consistent with either of these possibilities is the finding that the basal activity of purified PI3K␥, i.e. in the absence of G␤␥, was not inhibited by PLECK E (Fig. 4) or PLECK WT (data not shown) at 0.1 or 1 mM PI(4,5)P 2 . The elucidation in detail of the inhibitory mechanism that is at work is obviously impaired by the fact that three PH domains (two on pleckstrin and one on PI3K␥), each potentially vying for G␤␥/PI(4,5)P 2 , are present in this system.
To rule out a direct inhibitory interaction between the phosphorylated interdomain region and PI(4,5)P 2 , G␤␥, or PI3K␥, two peptides and the expressed interdomain region of PLECK WT were utilized in some assays. None of these affected PI3K␥ activity (purified or in CSK A ) (data not shown).
Inhibition of PI3K␥ by Phosphopleckstrin Is Specific for where Control is the G␤␥-stimulated activity. Only PLECK WT was observed to be phosphorylated and only by cytoskeletal fractions. About 60% of the PI3K activity in activated cytoskeleton is due to PI3K␥, as assayed by inhibition by ␤ARK-PH (10), and the remaining activity is presumed to consist of Rho/GTP␥S-activable p85/PI3K (4,10). B, the activity of p85/PI3K in CSK R and immunoprecipitated (I.P.) p85/PI3K was assayed as described under "Materials and Methods," without G␤␥, in the presence and absence of GTP␥S (10 M). There was no effect of PLECKs on basal activity (data not shown) or GTP␥Sstimulated activity. Results are the means Ϯ S.D. of two separate experiments performed in duplicate, utilizing different preparations of CSK R and immunoprecipitated p85/PI3K, normalized to controls. Control values ranged from 1.5 ϫ 10 3 to 10 4 dpm. PI(4,5)P 2 Phosphorylation-Since PI3K␥ can phosphorylate all three phosphoinositides (Fig. 1), we determined whether the inhibitory effects of pleckstrin were restricted to PI(4,5)P 2 . In contrast to their effects on the phosphorylation of PI(4,5)P 2 , neither PLECK E nor PLECK WT altered the phosphorylation at the 3-OH position of PI or PI(4)P by PI3K␥ in CSK A (Fig. 5). Yet, the fragment of ␤-adrenergic receptor kinase that contains a PH domain and binds G␤␥ (␤ARK-PH) (10) was inhibitory for all three substrates (Fig. 5). Clearly, PLECKs do not act, like ␤ARK-PH, to "chelate" G␤␥ in the absence of PI(4,5)P 2 . Notably also, the phosphorylation of PLECK WT that occurred in the presence of PI(4,5)P 2 was more than three times that observed with PI or PI(4)P (Fig. 6). Thus, PI(4,5)P 2 strongly augments pleckstrin phosphorylation by activated cytoskeletal fractions, which may be related to the ability of the PI(4,5)P 2 3-kinase product, PI(3,4,5)P 3 , to stimulate local protein kinase activity and thereby phosphorylate pleckstrin (18). The absence of pleckstrin phosphorylation was not the sole reason for the lack of effect of PLECK WT on PI3K␥ acting on PI or PI(4)P, however, since the pseudo-phosphorylated variant, PLECK E , also did not inhibit PI3K␥ when PI or PI(4)P was employed. It appears, therefore, that inhibitory effects of pleckstrin are specific for PI(4,5)P 2 substrate, consistent with a recent report that the bisphosphate group is necessary for binding of the inositol ring to PH domains (20). The data also imply that binding of PI(4,5)P 2 by phosphopleckstrin may be necessary either for phosphopleckstrin to bind G␤␥ optimally or to hinder binding of G␤␥ by PI3K␥, or otherwise to allow an inhibitory interaction between phosphopleckstrin and PI3K␥. The first interpretation would be in keeping with the findings of Lefkowitz and coworkers (14,15) that suggest a cooperativity between PI(4,5)P 2 and G␤␥ for binding to PH domains.
Phosphorylation of Endogenous Pleckstrin in Vivo Mimics the Inhibitory Effect of PLECK E in Vitro-To begin to ascertain the relevance of pleckstrin phosphorylation by intact platelets specifically with respect to PI3K␥ activity, we assessed the effects of prior treatment of platelets with protein kinase Cactivating or -nonactivating isomers of PMA (␤-PMA and ␣-PMA, respectively) or dimethyl sulfoxide (vehicle)/buffer on several PI kinase activities. All Triton-soluble fractions contained traces of G␤␥, as detected by Western blotting (data not shown). As illustrated in Fig. 7, although all three PI3K activities were enhanced by additional exogenous G␤␥, only PI(4,5)P 2 -directed PI3K activity was inhibited significantly by prior treatment of platelets with ␤-PMA to stimulate pleckstrin phosphorylation in vivo. ␣-PMA was without effect (data not shown). Consistent with the data shown in Fig. 4, the inhibitory effect was overcome by inclusion of increasing concentrations of G␤␥, thereby decreasing inhibition from 55% (without exogenous G␤␥) to 15% (with 2 M exogenous G␤␥). The selectivity of the ␤-PMA is also consistent with the formation of phosphopleckstrin in response to ␤-PMA, but not ␣-PMA (Fig.  8) (6, 10), and the inhibitory effect with respect to PI(4,5)P 2 substrate is in keeping with our observation that inhibition by phosphopleckstrin is restricted to PI3K␥ acting on PI(4,5)P 2 (Fig. 5).
One would expect that there should be a negative regulatory mechanism for p85/PI3K as well, even though pleckstrin/phosphopleckstrin is not involved directly. A potential route for inhibition of p85/PI3K in vivo might involve the ability of PI(3,4,5)P 3 to compete with phosphotyrosine for binding to SH2 domains of p85 (24), thereby perhaps removing platelet p85/ PI3K from an important membrane locale. Binding of PI(3,4,5)P 3 to SH2 domains may even be directly inhibitory for p85/PI3K activity, as well as for localization, although it has not been demonstrated that such occupancy by PI(3,4,5)P 3 is not as stimulatory for p85/PI3K as is occupancy by phosphotyrosines. It is also noteworthy that PI(3,4)P 2 is significantly less effective than PI(3,4,5)P 3 in binding to SH2 domains (24). This would gradually favor PI(3,4)P 2 generation by p85/PI3K acting on PI(4)P, leading to a delayed increase in PI(3,4)P 2 , as observed (21)(22)(23). PI(3,4,5)P 3 , formed by PI3K␥, may also compete with phosphotyrosines for p85/PI3K, and in this setting, inhibition of PI3K␥ by phosphopleckstrin could relieve the  Fig. 7 were resolved on SDS gels as described under "Materials and Methods" and immunoblotted for pleckstrin (18). Lane 1, dimethyl sulfoxide/buffer; lane 2, ␣-PMA; lane 3, ␤-PMA. The upper band comigrated with 32 Plabeled pleckstrin in phosphorylation assays. PI3K␥-linked inhibition by PI(3,4,5)P 3 of p85/PI3K, favoring p85/PI3K activity. Our preliminary studies indicate that this is likely. 3 Based on the data that we have presented above, it appears that phosphopleckstrin, in amounts generated in vivo, may bind G␤␥ avidly when associated with PI(4,5)P 2 at the plasma membrane, more avidly than does pleckstrin, and may also interact locally with PI3K␥. The result is that the formation of PI(3,4,5)P 3 by PI3K␥ is inhibited. Studies are currently underway to define the roles of the two pleckstrin PH domains and the phosphorylation of the interdomain in achieving this inhibitory effect.