A Type II Phosphoinositide 3-Kinase Is Stimulated via Activated Integrin in Platelets

We have observed that aggregation of human platelets, caused by activation of integrin αIIbβ3 and its consequent binding of fibrinogen, stimulates a novel pathway for synthesis of phosphatidylinositol 3,4bisphosphate, thereby activating protein kinase B/Akt. Such synthesis depends upon both the generation of phosphatidylinositol 3-phosphate (PtdIns3P), which is sensitive to wortmannin (IC50 7 nm) and calpain inhibitors, and the phosphorylation of PtdIns3P by PtdIns3P 4-kinase. We now report that a recently characterized C2 domain-containing phosphoinositide 3-kinase isoform (HsC2-PI3K) is present in platelets and a leukemic cell line (CHRF-288) derived from megakaryoblasts, and is likely to be responsible for the stimulated synthesis of PtdIns3P observed in platelets. HsC2-PI3K, identifiable by Western blotting and immunoprecipitatable activity, is sensitive to wortmannin (IC50 6–10 nm), requires Mg2+, and shows strong preference for PtdIns over PtdIns4P or phosphatidylinositol 4,5-bisphosphate as substrate. HsC2-PI3K is activated severalfold when platelets aggregate in an αIIbβ3-dependent manner or when platelet or CHRF-288 lysates are incubated with Ca2+. Activation is prevented by calpain inhibitors. CHRF-288, which cannot undergo activation of αIIbβ3 and thereby aggregate in response to platelet agonists, do not generate PtdIns3P or activate HsC2-PI3K under conditions that stimulate other phosphoinositide 3-kinases. HsC2-PI3K may thus be an important effector for integrin-dependent signaling.

variety of physiologic changes, including proliferative responses to growth factors (1), differentiation (2), anti-apoptosis (3), cytoskeletal rearrangements and integrin activation (4,5), and integrin-mediated cell motility and carcinoma invasion (6,7). The second messengers involved in these events are thought to be PtdInsP 3 and/or PtdIns(3,4)P 2 , which are capable of stimulating the activity of protein kinases such as PKB/Akt (8), and some protein kinase C isoforms (9 -11), but may act in additional ways. Most of the PI3Ks studied in these contexts are, or have been assumed to be, of the "Type I" class (see Ref. 12 for review), where classification is based upon structural homology, substrate specificity, and mode of regulation. The Type I PI3Ks are heterodimers, containing 110 -130-kDa catalytic subunits associated with 50 -85-or 101-kDa adaptor entities that regulate localization and function. Type I PI3Ks phosphorylate PtdIns, PtdIns4P, and PtdIns(4,5)P 2 , with preference for PtdIns(4,5)P 2 , at the 3-OH site of the myoinositol ring in vitro and have been found to be stimulated by numerous growth factors and heterotrimeric GTP-binding protein-coupled receptors. They can be activated, depending upon the subtype, cell, and receptor that has been stimulated, by tyrosine phosphorylation, association with tyrosine-phosphorylated or proline-rich domains, small GTPases, or ␤␥ subunits of heterotrimeric GTPbinding proteins (12). Type II PI3Ks are catalytic entities about twice the size of Type I catalytic subunits, and it is unknown whether they, like Type I and Type III, have adaptor proteins or whether their increased size provides a built-in adaptor. They contain a defining C2 domain at their C termini, as well as an N-terminal extension. Importantly, as well, they cannot utilize, or utilize poorly, PtdIns(4,5)P 2 as a substrate, and their means of activation, or even whether they can be activated in a signal-transduction setting, has been unknown. The cloning and characterization of two new PI3Ks of the Type II class from human cells have been described recently (13,14). These PI3Ks have been designated PI3KC2␣ (13) and HsC2-PI3K (14). PI3KC2␣ (190 kDa), which has been expressed, is resistant to wortmannin (IC 50 420 nM) and resembles (90% homology) mouse m-cpk (15) and p170 (16). It preferentially phosphorylates PtdIns and PtdIns4P, but can, albeit poorly, phosphorylate PtdIns(4,5)P 2 . HsC2-PI3K is a closely related 185-kDa protein, except for its divergent 350 N-terminal residues, which encompass two proline-rich sequences appropriate for interaction with src-homology 3 domains (14). As described here, expressed HsC2-PI3K phosphorylates PtdIns in strong preference to the other two known PI3K substrates, PtdIns4P and PtdIns(4,5)P 2 , and is very sensitive to wortmannin (IC 50 ϳ10 nM), in contrast to PI3KC2␣. Mammalian Type III PI3K (HsVPS34) contains a catalytic subunit similar in size to those of Type I and an associated protein of 150 kDa. HsVPS34 is inhibited by nanomolar wortmannin, phosphorylates only PtdIns lipid substrate, and, in contrast to Type I and Type II kinases, is stimulated by Mn 2ϩ versus Mg 2ϩ . Type III PI3Ks, based upon findings with the yeast homologue, VPSp, are thought to participate constitutively in secretory protein sorting to vacuoles, rather than being stimulated in response to cellular agonists.
We have observed (17,18) that activation of platelet integrin ␣ IIb ␤ 3 , via its consequent fibrinogen-dependent aggregation or clustering, stimulates a novel pathway for the generation of PtdIns(3,4)P 2 in platelets that is a function of PtdIns3P production and phosphorylation of PtdIns3P by a 4-kinase. No PtdInsP 3 is generated during this integrin-dependent event. In the present study, we have examined the nature of the PI3K involved in the integrin-activated pathway, which results in the activation of platelet PKB␣/Akt (17,18). Our data indicate that a PI3K of the Type II subclass, HsC2-PI3K, is most likely to be responsible for the generation of PtdIns3P that occurs.

EXPERIMENTAL PROCEDURES
Most reagents were obtained from sources described (17,18). Calpain I and II inhibitors were purchased from Boehringer Mannheim. HsC2-PI3K cDNA was assembled using a combination of reverse transcription-polymerase chain reaction and standard cDNA screening, by using the sequence of HsC2-PI3K (14) cloned in the mammalian expression vector pBKCMV for transient expression in HEK 293 cells. Its activity after immunoprecipitation was assayed as described below. PI3K isoform-discriminant polyclonal antisera against the first 350 amino acid portion of HsC2-PI3K (14), expressed in Escherichia coli as an Nterminally fused glutathione S-transferase protein, were raised in rabbits. These antisera were used for all immunoprecipitations and Western blots directed at HsC2-PI3K and do not detect or immunoprecipitate Type I PI3Ks, HsVPS34, or PI3KC2␣. Antibodies to p85␣/␤ subunits and PI3KC2␣ were the generous gifts of Drs. Ivan Gout and Jan Domin (Ludwig Institute, London). Antibody to HsVPS34 was prepared by Dr. Volinia (19).
Western Blotting-Platelet and CHRF-288 Triton-soluble and insol-uble ("CSK") fractions and HsC2-PI3K immunoprecipitates were dissolved in SDS-reducing buffer and proteins resolved by one-dimensional SDS-polyacrylamide gel electrophoresis on 7.5% gels, prior to transfer to nitrocellulose for Western blotting and enhanced chemiluminescence detection, as described previously (5). CSK from Triton lysates were suspended in buffer at 4 -11 times their concentrations in Triton lysates before mixing with sample buffer, such that the same amount of CSK protein would be applied per lane.

RESULTS AND DISCUSSION
Exposure of platelets to a variety of agonists, under conditions that promoted ␣ IIb ␤ 3 ϩ FIB-dependent aggregation, led to the activation of HsC2-PI3K and, in 32 P-labeled platelets, transient accumulation of [ 32 P]PtdIns3P (Fig. 1). Both effects were inhibited by wortmannin (IC 50 7 nM, for [ 32 P]PtdIns3P; IC 50 6 nM for HsC2-PI3K), in keeping with the wortmannin sensitivity of expressed HsC2-PI3K (IC 50 ϳ10 nM). As shown in Fig. 1A, the increased activity of HsC2-PI3K in immunoprecipitates from Triton-soluble fractions of platelets incubated with SFLLRN (directed to the thrombin receptor) ϩ FIB was sustained for up to 14 min and slightly preceded the accumulation of [ 32 P]PtdIns3P in stimulated platelets. The transient accumulation of [ 32 P]PtdIns3P is probably attributable to the increased activity of PtdIns3P 4-kinase under these conditions (17,18). In the absence of FIB (not shown), or in the presence of maximally effective concentrations of calpeptin (IC 50 1 M) or calpain I inhibitor (IC 50 0.3 M), the increase in PtdIns3P levels and activation of HsC2-PI3K were abolished. Calpain I inhibitor (90% inhibition at 1 M) was more effective than calpain II inhibitor (18% inhibition at 1 M). Similarly, increases in HsC2-PI3K activity (Fig. 1B) and [ 32 P]PtdIns3P (Fig. 1C) in response to PMA (activating protein kinase C) ϩ FIB or LIBS (directly activating ␣ IIb ␤ 3 ) ϩ FIB were blocked by calpain inhibition and did not occur in the absence of FIB. Both PMA and SFLLRN are known to activate p85/PI3K, whether or not FIB is present, and thereby contribute to "inside-out" signaling leading to the sustained activation of ␣ IIb ␤ 3 (5). SFLLRN also activates another Type I enzyme, PI3K␥, which is dependent upon ␤␥ subunits of GTP-binding proteins, and apparently is not involved in signaling leading to ␣ IIb ␤ 3 activation (5). LIBS by-passes such a pathway, promoting the FIB binding conformation of ␣ IIb ␤ 3 by interaction with the ␤ 3 subunit (21) in a wortmannin-insensitive manner, i.e. without the prerequisite of p85/PI3K activation or stimulated accumulation of 3-OHphosphorylated phosphoinositides (5,17). Once FIB binds to ␣ IIb ␤ 3 and aggregation occurs, however, "outside-in" signaling that promotes calpain activation is triggered, leading to formation of PtdIns3P and PtdIns(3,4)P 2 (17,18). Calpain, a Ca 2ϩdependent thiol protease, is known to be activated intracellularly when FIB binds to ␣ IIb ␤ 3 and aggregates platelets that have been stimulated in the presence of Ca 2ϩ (22). To simulate these calpain-activating conditions, in the absence of integrin activation, unstimulated platelet lysates were incubated at room temperature in the presence of millimolar Ca 2ϩ , and the activity of immunoprecipitated HsC2-PI3K was then assayed. It was found that activity rose transiently (2.49 Ϯ 0.12-fold after 10 min, 4.08 Ϯ 0.18-fold after 20 min, and 3.24 Ϯ 0.17-fold after 60 min) and that the increase was blocked by omitting Ca 2ϩ or by including calpain inhibitors.
As we have reported (20,23), CHRF-288, a leukemic cell line derived from a platelet precursor cell, the megakaryoblast (24), can be stimulated by a variety of physiological agonists to accumulate PtdInsP 3 and PtdIns(3,4)P 2 . These increases are sensitive to wortmannin, and CHRF-288 cells contain p85/ PI3K and PI3K␥, Type I enzymes whose activities in lysates can be stimulated by guanosine 5Ј-O-(thiotriphosphate) and ␤␥ subunits of GTP-binding proteins, respectively. Despite displaying apparently normal ␣ IIb ␤ 3 at their surface, however, CHRF-288 cells cannot undergo the activation of this integrin that leads to the binding of FIB (25) and aggregation. We found that CHRF-288 cells, after incubation with SFLLRN ϩ FIB (or PMA ϩ FIB, not shown) accumulated no [ 32 P]PtdIns3P, whereas [ 32 P]PtdInsP 3 and [ 32 P]PtdIns(3,4)P 2 were formed rapidly (Fig. 3A), a pattern similar to the "pre-integrin" accumulation of 3-OH-phosphorylated phosphoinositides in stimulated platelets (18). Furthermore, although much immunoprecipitatable HsC2-PI3K activity was present in Triton or RIPA lysates of CHRF-288 cells, no increased activity was observed when CHRF-288 were activated with SFLLRN ϩ FIB or PMA ϩ FIB (Fig. 3B). In contrast, lysates of CHRF-288 incubated in the presence of Ca 2ϩ showed a time-dependent increase in immunoprecipitatable HsC2-PI3K activity (2.3 Ϯ 0.2-fold in 10 min, 3.6 Ϯ 0.1-fold in 20 min, and 4.2 Ϯ 0.4-fold in 60 min), which was prevented by omitting Ca 2ϩ or including calpain inhibitor. This indicated that HsC2-PI3K of CHRF-288 cells could be activated in an apparently calpain-dependent manner if the requirement for activated integrin were bypassed. Thus, failure to activate ␣ IIb ␤ 3 is associated in intact CHRF-288 with failure to activate HsC2-PI3K and stimulate accumulation of PtdIns3P.
Western blotting of CSK and Triton-soluble (TS) fractions of platelets ( Fig. 4) with antibody to HsC2-PI3K revealed a protein band at about 200 kDa in both fractions. This band in the Triton-soluble fraction could be decreased by immunoprecipitation with HsC2-PI3K antibody (PS), and a corresponding band was found for the Western blot of the immunoprecipitate (IP). The migration of the band did not shift detectably for blots from activated platelets (SFLLRN) or from platelet lysates incubated with Ca 2ϩ .
In conclusion, our data indicate that HsC2-PI3K is activated in platelets in a manner dependent upon the reorganization of integrin ␣ IIb ␤ 3 , FIB binding and aggregation, and, most probably, calpain I activation. Given its strong preference for Ptd-Ins as a substrate, its susceptibility to wortmannin, and its activation under the same conditions required for the accumulation of PtdIns3P in intact cells, HsC2-PI3K is the most reasonable choice for the enzyme responsible for generating PtdIns3P in stimulated platelets. It is possible that HsC2-PI3K is a substrate for calpain I and thereby activated by it, but, at this point, other mechanisms involving other calpain targets are equally likely to contribute to the stimulation observed in vivo. Inasmuch as the antibody used for immunoprecipitation and identification of resting and activated HsC2-PI3K is directed to the N-terminal 350-amino acid region of HsC2-PI3K, and the catalytic domain lies approximately between amino acids 1037 and 1320 (14), any activating cleavage target for calpain would have to be C-terminal to these regions. If the cleavage site were in the C2 domain of HsC2-PI3K, for example, a decrease in size might not be easily detectable under our conditions. Further studies with tagged, expressed HsC2-PI3K and purified calpain I may be needed to identify possible activating cleavage sites for HsC2-PI3K in stimulated platelets.