Regulation of Outside-in Signaling in Platelets by Integrin-associated Protein Kinase C (cid:1) *

Studies with inhibitors have implicated protein kinase C (PKC) in the adhesive functions of integrin (cid:2) IIb (cid:1) 3 in platelets, but the responsible PKC isoforms and mechanisms are unknown. (cid:2) IIb (cid:1) 3 interacts directly with tyro- sine kinases c-Src and Syk. Therefore, we asked whether (cid:2) IIb (cid:1) 3 might also interact with PKC. Of the several PKC isoforms expressed in platelets, only PKC (cid:1) co-immuno-precipitated with (cid:2) IIb (cid:1) 3 in response to the interaction of platelets with soluble or immobilized fibrinogen. PKC (cid:1) recruitment to (cid:2) IIb (cid:1) 3 was accompanied by a 9-fold in- crease in PKC activity in (cid:2) IIb (cid:1) 3 immunoprecipitates. RACK1, an adapter for activated PKC (cid:1) , also

In addition to their roles in cell adhesion, integrins transmit signals in both directions across the plasma membrane to regulate cytoskeletal organization, motility, and other anchoragedependent cellular responses (1). In platelets, for example, ␣ IIb ␤ 3 responds to "inside-out" signals with an increase in affinity for cognate ligands such as fibrinogen that bridge platelets to each other and mediate platelet adhesion to sites of vascular damage. In turn, ligand binding to ␣ IIb ␤ 3 triggers outside-in signals that promote cytoskeletal changes necessary for full platelet aggregation and spreading (2,3). Bidirectional ␣ IIb ␤ 3 signaling is controlled, in part, by specific intracellular proteins that interact with the relatively short cytoplasmic tails of ␣ IIb or ␤ 3 . For example, binding of talin to ␤ 3 is a final common step in the cellular modulation of ␣ IIb ␤ 3 affinity (4), and binding of c-Src and Syk protein-tyrosine kinases to ␤ 3 is required for platelet spreading on fibrinogen (5,6). Several other intracellular proteins, for example, CIB and ␤ 3 -endonexin, can also bind to ␣ IIb or ␤ 3 tails, respectively, and may influence ␣ IIb ␤ 3 functions (7)(8)(9). However, the full complement of intracellular proteins that are capable of interacting directly or indirectly with ␣ IIb ␤ 3 is unknown.
The protein kinase C (PKC) 1 subfamily of AGC serine/threonine kinases has been implicated in integrin function or dynamics in many cell types (10). In platelets, PKC is thought to regulate ␣ IIb ␤ 3 affinity, based on the stimulatory effects of phorbol esters, which bind to PKC C1 domains, and the blocking effects of broad spectrum PKC inhibitors (3). However, the lack of specificity of these compounds limits data interpretation and does not permit conclusions about the roles of specific PKC isoforms (11). PKCs have been categorized as classical (diacylglycerol-and Ca 2ϩ -regulated through C1 and C2 domains, respectively), novel (Ca 2ϩ -independent but diacylglycerol-regulated), or atypical (Ca 2ϩ -and diacylglycerol-independent) (12,13). Platelets are reported to contain members of all three classes of PKC isozymes (14 -21), as well as the related protein kinase D (22). Recently, experimental tools have become available to study specific PKC isoforms in cells, including overexpression, gene targeting and gene knock-down strategies, and molecular imaging (23)(24)(25). Some of these tools are potentially relevant to platelets.
PKC function depends on the maturation of catalytic activity of the enzyme through phosphorylation and PKC binding to membranes or specific proteins. The latter interactions place PKC in proximity to substrates and relieve autoinhibitory restraints imposed by the binding of a pseudosubstrate sequence to the active site (12,13,26). One group of PKC targeting proteins has been termed RACK, which binds selectively to activated PKCs (27). The best characterized protein of this group is RACK1, a 36-kDa protein composed of seven WD40 repeats. RACK1 was originally identified based on its interaction with activated PKC␤ and subsequently shown to interact with certain other PKC isoforms and with several other pro-teins, most notably integrin ␤ cytoplasmic tails and c-Src (see Fig. 1A) (28 -30).
Given the potential for the cytoplasmic tails of ␣ IIb ␤ 3 to serve as binding sites for signaling molecules and the apparent functional relationships between PKC and ␣ IIb ␤ 3 , the present studies were carried out to determine whether specific PKCs associate with ␣ IIb ␤ 3 and if so to determine what the functional relevance of the association is. By using human and mouse platelets and a CHO cell model system, the results show that one particular PKC isoform, PKC␤, inducibly associates with ␣ IIb ␤ 3 in response to fibrinogen binding to cells. The PKC␤/ ␣ IIb ␤ 3 interaction appears to be mediated by RACK1 and is required for cytoskeletal reorganization and platelet spreading on fibrinogen, but it is dispensable for the affinity modulation of ␣ IIb ␤ 3 .

EXPERIMENTAL PROCEDURES
Reagents-Monoclonal antibodies to PKC␣, -␤, -␦, -, -⑀, -, andwere from Transduction Laboratories (Lexington, KY), and antibodies specific for PKC␤I, PKC␤II, or cortactin were from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). The phospho-specific antibody to c-Src, tyrosine 418, was from BIOSOURCE, and anti-phosphotyrosine antibodies 4G10 and PY20 were from Upstate Biotechnology (Lake Placid, NY) and Transduction Laboratories, respectively. Monoclonal antibody 327 to c-Src was from Dr. Joan Brugge (Harvard Medical School), and monoclonal antibody D57 (specific for ␣ IIb ␤ 3 ) and polyclonal antibody Rb8053 (specific for integrin ␤ 3 ) were from Dr. Mark H. Ginsberg (University of California San Diego, La Jolla, CA). Horseradish peroxidase-conjugated secondary antibodies were from Bio-Rad, and Cy-5and TRITC-conjugated secondary antibodies were from Jackson Immu-noResearch Laboratories, Inc. (West Grove, PA). Rhodamine-phalloidin was from Molecular Probes (Eugene, OR), purified human fibrinogen was from Enzyme Research Laboratories, Inc. (South Bend, IN), and protein A-Sepharose was from Amersham Biosciences. Cell-permeable PKC antagonist peptide p99 was purchased from Stanford University (31). Src kinase inhibitor SU6656 was from SUGEN, Inc. (South San Francisco, CA). Platelet agonists were from Sigma, except for convulxin, which was a gift from Dr. Steve Watson (University of Birmingham, Birmingham, UK).
Cell Culture, Plasmids, and Transfections-A5 CHO cells stably expressing ␣ IIb ␤ 3 were maintained in culture as described previously (32). Transfections were performed at 70 -80% confluency using Lipofectamine (Invitrogen) according to the manufacturer's instructions. Twenty-four hours later, the concentration of fetal calf serum was decreased from 10 to 0.5%, and cells were cultured for another 24 h before being used in functional assays. Mammalian expression plasmids included pEGFP-PKC␤I (a chimera containing GFP fused to the N terminus of full-length PKC␤I (a gift from Dr. Stephen S. G. Ferguson, Robarts Research Institute, Ontario, Canada)), pRC-CMV/Src (encoding wild-type non-neuronal murine c-Src), and pcDNA/RACK1/ WD6/7/HA (a hemagglutinin-tagged chimera containing RACK1 WD repeats 6 and 7). To construct the latter, pTarget/WD6/7 (a gift from Dr. Arnaud Besson, Fred Hutchinson Cancer Research Center, Seattle, WA) was subjected to PCR using Platinum Pfx polymerase to introduce a 5Ј-KpnI site and a hemagglutinin tag-3Ј-XhoI site. After digestion and ligation into pcDNA3.1, transformed colonies were screened by colony PCR, and coding sequences were verified by DNA sequencing. Plasmids were amplified and purified before use (QIAfilter plasmid Maxi kit, Qiagen, Inc., Chatsworth, CA).
Interaction of Cells with Fibrinogen-Human platelets or mouse platelets from PKC␤ Ϫ/Ϫ , PKC␤ ϩ/Ϫ , and PKC␤ ϩ/ϩ littermates (33) were obtained from fresh, anticoagulated whole blood and were washed and resuspended to 3 ϫ 10 8 cells/ml in a platelet incubation buffer (34,35). The PKC␤ status of mouse platelets was established by genotyping the animals and by Western blotting of platelet lysates. CHO cells were harvested using 0.5 mM EDTA, washed once in Dulbecco's modified Eagle's medium, resuspended to 3 ϫ 10 6 cells/ml, and incubated for 45 min in the presence 20 M cycloheximide. Binding of fluorescein isothiocyanate-fibrinogen to mouse platelets was quantified by flow cytometry (35). To test the effects of soluble fibrinogen binding on the interaction of intracellular proteins with ␣ IIb ␤ 3 , platelets or CHO cells were incubated for 20 or 30 min, respectively, with 250 g/ml fibrinogen in the presence or absence of 0.5 mM MnCl 2 (to activate ␣ IIb ␤ 3 ) (36) and in some cases with 2 mM RGDS (to block fibrinogen binding). Cells were collected by centrifugation, washed once with phosphate-buffered saline, and solubilized for 10 min on ice in a lysis buffer containing 0.5% Nonidet P-40, 50 mM NaCl, 50 mM Tris, pH 7.4, and inhibitors (1 mM sodium vanadate, 0.5 mM sodium fluoride, and 1ϫ Complete protease inhibitor (Roche Applied Science)). Lysates were clarified at 4°C by sedimentation at 10,000 rpm in a microcentrifuge and subjected to immunoprecipitation and Western blotting. To test the effects of cell adhesion to fibrinogen, 100-mm bacterial culture dishes were precoated with 5 mg/ml bovine serum albumin (BSA) or 100 g/ml fibrinogen. After blocking with heat-denatured BSA, 4.5 ϫ 10 8 platelets in 1.5 ml or 3 ϫ 10 6 CHO cells in 2 ml were added to each dish, and incubations FIG. 1. RACK1 associates with ␣ IIb ␤ 3 in platelets. A, schematic diagram of the domain structure of RACK1 illustrating WD repeats and regions of the protein implicated in direct interactions with integrin ␤ tails, c-Src, and PKC␤II. B, washed human platelets were plated on fibrinogen (Fib) for 30 min or maintained in BSA suspension. Platelet lysates (Lys) were immunoprecipitated (IP) with an antibody to ␤ 3 or with normal rabbit serum (NRS) as control. Immunoprecipitates were subjected to SDS-PAGE and probed on Western blots with antibodies to RACK1 and ␤ 3 integrin. This experiment is representative of three so performed.

FIG. 2. Inducible association of PKC␤ with ␣ IIb ␤ 3 in platelets.
A, lysates from human platelets were subjected to SDS-PAGE (10 g of protein/lane) and analyzed by Western blotting with antibodies to specific PKC isoforms. Lysate from rat cerebrum was included as a positive control. Chemiluminescence development time was 1 min. B, platelets were plated on fibrinogen (Fib) for 30 min or maintained in BSA suspension in the presence or absence of 200 nM PMA (top panel) or 2 M apyrase (bottom panel). Lysates (Lys) were immunoprecipitated (IP) with an antibody to ␤ 3 or normal rabbit serum (NRS), and immunoprecipitates were probed on Western blots with antibodies to PKC␤ and ␤ 3 integrin. This experiment is representative of five so performed. C, platelets were incubated for the indicated times in the presence or absence of 250 g/ml fibrinogen, 0.5 mM MnCl 2 , and 2 mM RGDS. Then ␤ 3 immunoprecipitates were probed on Western blots with antibodies to PKC␤ and ␤ 3 . This experiment is representative of two so performed.
were carried out in a CO 2 incubator for the indicated periods of time at 37°C. Non-adherent cells from the BSA plates were sedimented and lysed immediately. Cells adherent to fibrinogen were gently washed twice with phosphate-buffered saline, lysed on the plates, and subjected to immunoprecipitation and Western blotting.
Immunoprecipitation and Western Blotting-Five hundred-microli-ter aliquots of lysate containing equal amounts of protein (ranging from 500 to 800 g, depending on the experiment) were incubated with relevant antibodies for 2 h or overnight at 4°C with gentle agitation. Then, 50 l of protein A-Sepharose (50% v/v) were added for 2 h at 4°C. Immune complexes were washed twice with phosphate-buffered saline, SDS-PAGE sample buffer was added, and samples were electrophoresed in 7.5 or 10% SDS-polyacrylamide gels. After electrotransfer to nitrocellulose, Western blotting was carried out using enhanced chemiluminescence for detection (Supersignal West Pico substrate, Pierce). PKC Activity Measurements-␣ IIb ␤ 3 was immunoprecipitated from platelet lysates with antibody Rb8053 and protein A-Sepharose. Beads were washed twice with lysis buffer and resuspended in kinase buffer and [ 32 P]ATP, and PKC activity was measured using a substrate peptide (QKRPSQRSKYL) according to the manufacturer's instructions (PKC assay kit, Upstate Biotechnology, Inc., Lake Placid, NY).
Cell Spreading and Migration Assays-Glass coverslips were coated with 100 g/ml fibrinogen, washed cells were allowed to adhere for 45 min at room temperature, and cell morphology and spreading were assessed by confocal microscopy (6,34). To study CHO cell migration on fibrinogen, cells were serum-starved overnight in Dulbecco's modified Eagle's medium containing 0.5% fetal calf serum and resuspended in Dulbecco's modified Eagle's medium containing 10% fetal calf serum, and 1-ml aliquots were seeded at 30,000 cells/ml onto fibrinogen-coated coverslips. After 1 h, cells were microinjected with a 50 g/ml solution of RACK1 WD6/7 and/or GFP-PKC␤I or GFP. Microinjected cells were monitored in real time in a temperature-controlled environment chamber using a Nikon TE2000U microscope. Images were acquired every 10 min for a 6-h period with a CoolSnap HQ charge-coupled device camera (Roper Scientific, Tucson, AZ) running on a Linux work station using FIG. 3. Interaction of PKC␤ with ␣ IIb ␤ 3 correlates with an increase in PKC activity associated with the integrin. A, platelets were incubated as indicated with 250 g/ml fibrinogen (Fib), 0.5 mM MnCl 2 , and 2 mM RGDS. Then, PKC activity in ␤ 3 immunoprecipitates (IP) was measured by PKC in vitro kinase assay, as described under "Experimental Procedures." Data are expressed as a percentage of the maximal response observed when platelets were exposed to a combination of 200 nM PMA, 250 g/ml fibrinogen, and 0.5 mM MnCl 2 . Results represent means Ϯ S.E. of three experiments. B, platelets were preincubated with 1 M p99 peptide or 12 M bisindolylmaleimide I (Bis) for 10 min and plated on fibrinogen or maintained in BSA suspension for 30 min. Lysates were immunoprecipitated with an antibody to ␤ 3 and probed on Western blots with antibodies to PKC␤ and ␤ 3 . This experiment is representative of three so performed.

FIG. 4. Effect of Src inhibition on the interaction of PKC␤ with
␣ IIb ␤ 3 . Platelets were plated on fibrinogen (Fib) or maintained in BSA suspension for 30 min in the presence or absence of 2 M SU6656 or vehicle control (Me 2 SO). Lysates were immunoprecipitated with an antibody to ␤ 3 and probed on Western blots with antibodies to PKC␤ and ␤ 3 (A) or to c-Src, Tyr(P)-418, and ␤ 3 (B). Experiment is representative of two so performed.
ISee software (ISee Imaging Systems, Raleigh, NC). At the end of the filming period, cells were stained with an antibody to hemagglutinin tag to detect those cells expressing RACK1 WD6/7.

RESULTS
Interactions between ␣ IIb ␤ 3 and RACK1 in Platelets-The RACK1 adapter molecule contains binding sites for c-Src, certain isoforms of activated PKC, and integrin ␤ cytoplasmic tails ( Fig. 1A) (27,29,30). In platelets, ␣ IIb ␤ 3 is constitutively associated with c-Src, and fibrinogen binding leads to c-Src activation (5). Therefore, we considered whether RACK1 might also be associated with ␣ IIb ␤ 3 . Washed human platelets were allowed to adhere for 30 min to immobilized fibrinogen or incubated in suspension over BSA. During this time, adherent platelets reorganize their actin cytoskeletons and spread to varying degrees (34). After washing and cell lysis in buffer containing Nonidet P-40 detergent, ␣ IIb ␤ 3 immunoprecipitates were probed for the presence of RACK1 by Western blotting. ␣ IIb ␤ 3 associated with RACK1 whether platelets were in suspension or adherent to fibrinogen (Fig. 1B). Therefore, potential relationships among ␣ IIb ␤ 3 , RACK1, and PKC were assessed.
Interactions between ␣ IIb ␤ 3 and PKC in Platelets-As reported previously (14 -21), two classical PKC isoforms (PKC␣ and PKC␤) and two novel isoforms (PKC␦ and PKC) were detected readily in human platelets, whereas other isoforms were less prominent or undetectable with the antibodies used ( Fig. 2A). Of the PKC isoforms detected, only PCK␤ was inducibly associated with ␣ IIb ␤ 3 in response to platelet adhesion to fibrinogen (Fig. 2B). In addition, when platelets were incubated in suspension for 10 min with 200 nM phorbol 12-myristate 13-acetate (PMA) to activate PKC, PKC␤ became associated with ␣ IIb ␤ 3 (Fig. 2B). Similar results were obtained if platelets were stimulated instead with 50 M PAR1 thrombin receptor-activating peptide (SSFLRN) or if murine platelets were used instead of human platelets (data not shown). There are two splice variants of PKC␤, PCK␤I and PKC␤II, in which the sequences differ in the C-terminal V5 region (37). Human platelets are reported to contain both variants (16), and we could detect both in human and mouse platelets with variantspecific antibodies (data not shown). The platelet studies described below used an antibody that recognizes both variants of PKC␤.
The inducible association of PKC␤ with ␣ IIb ␤ 3 could be a direct consequence of fibrinogen binding and ␣ IIb ␤ 3 clustering (38) or the result of additional signaling events stimulated during platelet spreading caused by co-stimulation with endogenous ADP (3). However, the association of PKC␤ with ␣ IIb ␤ 3 that was induced by platelet adhesion to fibrinogen was not affected by the presence of 2 M apyrase, an amount of enzyme sufficient to remove any ADP that might be released from the platelets (Fig. 2B). In addition, soluble fibrinogen binding to platelets in suspension was induced with 0.5 mM MnCl 2 , which activates ␣ IIb ␤ 3 directly. Under these conditions, PKC␤ became associated with ␣ IIb ␤ 3 as early as 30 s after fibrinogen binding, the earliest time point tested, and the association was stable for up to 20 min (Fig. 2C). This association was maintained even if platelets were pretreated with 10 M latrunculin A to block actin polymerization (data not shown). However, the interaction was blocked by 2 mM RGDS, a competitive inhibitor peptide of fibrinogen binding to ␣ IIb ␤ 3 (Fig. 2C), and it was not observed whether fibrinogen was omitted from the incubation mixture. Thus, the recruitment of PKC␤ to ␣ IIb ␤ 3 is caused by fibrinogen binding and is not the result of additional signaling events induced by ADP co-stimulation.
Factors Regulating PKC␤ Association with ␣ IIb ␤ 3 -Recruitment of PKC to plasma membranes often coincides with PKC activation (12,13). To test PKC activity associated with ␣ IIb ␤ 3 , platelets were incubated with MnCl 2 with or without fibrinogen for up to 20 min, and PKC activity was quantified in ␣ IIb ␤ 3 immunoprecipitates. Immunoprecipitates from control platelets incubated with MnCl 2 or fibrinogen alone displayed relatively low PKC activity. However, ␣ IIb ␤ 3 immunoprecipitates from platelets incubated with MnCl 2 and fibrinogen displayed a significant increase in PKC activity, even at 30 s (p Յ 0.001). At 20 min, this increase was 9-fold over baseline, which was 25-30% of the response observed with maximal PKC activation by PMA, and it could be blocked by RGDS (Fig. 3A). In addition, the fibrinogen-dependent increase in the association of PKC␤ with ␣ IIb ␤ 3 was blocked partially by p99, a cell-permeable pseudosubstrate peptide inhibitor of classical and novel PKCs (31) and by bisindolylmaleimide I, a general PKC inhibitor (Fig. 3B). These results indicated that fibrinogen binding stimulates an increase in PKC activity associated with ␣ IIb ␤ 3 and suggest that the association of PKC␤ with ␣ IIb ␤ 3 requires catalytically active PKC␤.
The activity and subcellular localization of classical PKCs are influenced by phosphorylation and products of phospholipid hydrolysis, e.g. Ca 2ϩ and diacylglycerol (12,13). Because fibrinogen binding to platelets leads to both the activation of integrin-associated c-Src (5) and the association of active PKC␤ with ␣ IIb ␤ 3 , functional relationships between PKC and c-Src were examined. In fibrinogen-adherent platelets, 2 M SU6656, a selective inhibitor of Src family kinases, failed to block the interaction of PKC␤ with ␣ IIb ␤ 3 (Fig. 4A), despite the fact that it adequately blocked c-Src activation, as assessed by tyrosine phosphorylation of c-Src activation loop Tyr-418 (Fig. 4B). Furthermore, the inhibition of PKC by 12 M bisindolylmaleimide I failed to block the activation of integrin-associated c-Src (data not shown). Platelet and ␣ IIb ␤ 3 Responses Dependent on PKC␤-Prominent responses following platelet adhesion to fibrinogen include reorganization of the actin cytoskeleton and spreading (34). Forty-five minutes after plating on fibrinogen, normal human platelets were generally well spread and exhibited prominent F-actin cables and a peripheral distribution of phosphotyrosine-containing proteins (Fig. 5B). In contrast, platelets incubated with the cell-permeable p99 peptide to inhibit classical and novel PKCs attached to fibrinogen and displayed filopodia but largely failed to reorganize their cytoskeletons, develop a peripheral distribution of phosphotyrosine-containing proteins, or spread (Fig. 5A). Similar results were obtained with bisindolylmaleimide I (Fig. 5C), whereas platelets incubated with the inactive congener, bisindolylmaleimide V, responded normally (Fig. 5D). The differences in spreading between PKC inhibitor-treated and control platelets were significant, as determined by computerized image analysis of cell surface areas (p Ͻ 0.001).
Because the inhibition of PKC by p99 or by bisindolylmaleimide I had also blocked the fibrinogen-dependent interaction of PKC␤ with ␣ IIb ␤ 3 (Fig. 3B), these results suggested that an integrin-associated pool of PKC␤ is required for full cytoskeletal and morphological responses of platelets to fibrinogen. However, because neither inhibitor is specific for PKC␤, this issue was also addressed by studying mouse platelets that were deficient in PKC␤ (PKC␤ Ϫ/Ϫ ). Similar to human platelets incubated with PKC inhibitors, PKC␤ Ϫ/Ϫ platelets failed to spread on fibrinogen, although platelets from PKC␤ ϩ/ϩ littermates spread normally (Fig. 6A). In addition, PKC␤ Ϫ/Ϫ platelets spread less well than PKC␤ ϩ/ϩ platelets in response to PMA. These differences in spreading between PKC␤ Ϫ/Ϫ and PKC␤ ϩ/ϩ platelets were statistically significant (p Ͻ 0.001). In contrast, PKC␤ Ϫ/Ϫ platelets spread as well as PKC␤ ϩ/ϩ platelets when co-stimulated with agonists to G protein-coupled receptors (PAR4 receptoractivating peptide, ADP) or an agonist to glycoprotein VI (convulxin) (Fig. 6A). Furthermore, PKC␤ Ϫ/Ϫ platelets bound soluble fibrinogen as well as PKC␤ ϩ/Ϫ platelets in response to G protein-coupled receptor agonists (Fig. 6B). Altogether, these results indicated that PKC␤ is required for normal outside-in signaling from ␣ IIb ␤ 3 to the platelet actin cytoskeleton. On the other hand, PKC␤ is dispensable for ADP-or thrombin-induced activation of ␣ IIb ␤ 3 . Because small amounts of endogenous platelet ADP are required for full spreading of platelets on fibrinogen (3), we cannot exclude a role for PKC␤ in this type of ADP co-stimulation.
Functional Relationships between RACK1, PKC␤, and ␣ IIb ␤ 3 -A CHO cell model system was used to further explore relationships among RACK1, PKC␤, and ␣ IIb ␤ 3 . A5 CHO cells that express RACK1 endogenously and ␣ IIb ␤ 3 after stable transfection were transiently transfected with GFP-PKC␤I. Forty-eight hours later, cells were plated on fibrinogen or BSA for 45 min, and the presence of GFP-PKC␤ in ␣ IIb ␤ 3 immunoprecipitates was assessed. As had been observed with platelets, RACK1 was constitutively associated with ␣ IIb ␤ 3 (data not shown), and adhesion to fibrinogen (Fig. 7A) or binding of soluble fibrinogen (Fig. 7B) caused an increase in the amount of GFP-PKC␤I associated with ␣ IIb ␤ 3 . However, no PKC␤/␣ IIb ␤ 3 association was observed if cells had been simultaneously transfected with a RACK1 truncation mutant (WD6/7) consisting of WD repeats 6 and 7, which can interact with PKC but not with integrin ␤ cytoplasmic tails (39) (Fig. 7, A and B). RACK1 WD6/7 had no effect on the expression of RACK1, GFP-PKC␤I, or ␣ IIb ␤ 3 (Fig. 7A), nor did it affect the constitutive association of ␣ IIb ␤ 3 with c-Src (data not shown). Because RACK1 WD6/7 may sequester PKC␤ away from a RACK1/␣ IIb ␤ 3 complex, these results are consistent with the idea that RACK1 mediates the interaction of PKC␤ with ␣ IIb ␤ 3 , although it does not mediate the interaction of c-Src with ␣ IIb ␤ 3 . If RACK1 were the intermediary between PKC␤ and ␣ IIb ␤ 3 , this interaction would be dependent on the ␤ 3 cytoplasmic tail. Indeed, when fibrinogen binding was stimulated by MnCl 2 , GFP-PKC␤I co-immu-noprecipitated with ␣ IIb ␤ but not with ␣ IIb ␤ 3 ⌬724, a carboxylterminal truncation mutant lacking 39 of the 47 residues of the ␤ 3 tail (Fig. 7C).
RACKs mediate translocation of activated PKCs to specific subcellular locations (27). To evaluate whether RACK1 is capable of promoting GFP-PKC␤I localization to ␣ IIb ␤ 3 -based adhesion sites, A5 cells transfected with PKC␤I were plated on fibrinogen-coated coverslips for 45 min and evaluated by confocal microscopy. Adherent cells spread and exhibited prominent lamellipodia containing small peripheral focal complexes and larger somewhat more centralized focal adhesions containing ␣ IIb ␤ 3 (Fig. 8A). GFP-PKC␤I co-localized strongly with ␣ IIb ␤ 3 in these structures (Fig. 8, B and C). However, when cells were simultaneously transfected with GFP-PKC␤I and RACK1 WD6/7, GFP-PKC␤I appeared to be distributed more diffusely throughout the cells and exhibited less co-localization with ␣ IIb ␤ 3 (Fig. 8, A-C). To localize GFP-PKC␤I more precisely, fibrinogen-adherent cells were co-stained for cortactin, which co-localizes with ␣ IIb ␤ 3 primarily at lamellipodial edges, and paxillin, which localizes predominantly to focal adhesions. PKC␤1 co-localized with both cortactin and paxillin, although more completely with the former (78.9 Ϯ 1.3% co-localization) than the latter (54.2 Ϯ 2.5% co-localization) (Fig. 8D). Thus, RACK1 promotes the localization of PKC␤I to ␣ IIb ␤ 3 adhesion sites, especially at lamellipodial edges.
To assess the functional relevance of RACK1-mediated PKC␤I association with ␣ IIb ␤ 3 , the migration of A5 cells on fibrinogen was studied in the presence of fetal calf serum as a source of growth factor. After adhesion to fibrinogen for 1 h, FIG. 7. Association of PKC␤ and ␣ IIb ␤ 3 in CHO cells. A, A5 CHO cells expressing wild-type ␣ IIb ␤ 3 were transfected with cDNAs for GFP-PKC␤I or GFP-PKC␤I plus RACK1 WD6/7 (WD6/7), as indicated. Forty-eight hours later, cells were plated on fibrinogen (Fib) or maintained in BSA suspension for 45 min. Cells were lysed and immunoprecipitated (IP) with an antibody to ␤ 3 and probed on Western blots with antibodies to PKC␤ and ␤ 3 . Results are representative of three experiments. Lysates were analyzed to assess expression and gel loading of GFP-PKC␤I and RACK1. B, A5 CHO cells were transfected with cDNAs for GFP-PKC␤I or PKC␤I plus RACK1 WD6/7, as indicated. Forty-eight hours later, cells were incubated in the presence or absence of 250 g/ml fibrinogen and 0.5 mM MnCl 2 for 30 min. ␤ 3 immunoprecipitates were subjected to Western blotting with antibodies to PKC␤I and ␤ 3 . The amount of GFP-PKC␤I that co-immunoprecipitated with ␤ 3 was quantified by densitometry and normalized to the amount of ␤ 3 . Data represent the -fold change in the amount co-immunoprecipitated, with the MnCl 2 sample (without fibrinogen) assigned a value of 1 (means Ϯ S.E. of three experiments). C, A5 CHO cells or ␣ IIb ␤ 3 ⌬724 CHO cells were plated on fibrinogen or maintained in BSA suspension for 45 min. Cells were lysed and immunoprecipitated with an antibody to ␤ 3 and probed in Western blots with antibodies to PKC␤ and ␤ 3 . Results are representative of three experiments.
A5 cells were microinjected with either GFP, GFP-PKC␤I, or GFP-PKC␤I plus RACK1 WD6/7. Two hours later, individual cell movement was monitored for a period of 6 h (Fig. 9A). Cells expressing GFP-PKC␤I exhibited rates of migration similar to the control cells injected with GFP. However, cells expressing GFP-PKC␤I and RACK1 WD6/7 displayed significantly less migration (p Ͻ 0.01) (Fig. 9Bi). The effect of RACK1 WD6/7 on the cell migration rate was enhanced over time (Fig. 9Bii) presumably because of a progressive increase in expression of the mutant protein. Thus, RACK1 WD6/7 affects ␣ IIb ␤ 3 -dependent cell migration. Although there is little evidence that platelets migrate, these results in CHO cells indicated that RACK1 has the potential to influence ␣ IIb ␤ 3 function by forming a bridge between activated PKC␤ and the integrin. DISCUSSION Important functional relationships exist between integrins and PKCs in many cell types (10). These relationships include PKC regulation of integrin affinity, avidity, and trafficking (3,40), PKC involvement in cellular responses that are dependent on integrins (41)(42)(43), and the modulation of PKC activity by integrin-mediated cell adhesion (44). In platelets, PKC has been implicated in the agonist regulation of ␣ IIb ␤ 3 affinity and initiation of platelet aggregation (3,45). This conclusion is based largely on the stimulatory effects of phorbol esters and the inhibitory effects of PKC inhibitors. However, because these compounds can target signaling molecules in addition to PKC, observations derived solely from their use are ambiguous (11). Furthermore, platelets express several PKC isoforms, and with rare exception (46) the relationships of individual isoforms to ␣ IIb ␤ 3 function have not been evaluated. In addition, the role of specific PKC isoforms in outside-in ␣ IIb ␤ 3 signaling is unknown, and the mechanisms by which PKCs and ␣ IIb ␤ 3 regulate each other's functions remain poorly characterized.
Therefore, several experimental systems were used here to better understand the relationship between ␣ IIb ␤ 3 and PKC. The main conclusions are: 1) of all the PKC isoforms expressed in platelets, only PKC␤ becomes associated with ␣ IIb ␤ 3 in response to fibrinogen binding; 2) the interaction of PKC␤ with ␣ IIb ␤ 3 is dependent on the cytoplasmic tail of ␤ 3 , mediated likely by RACK1, and results in an increase in PKC activity associated with ␣ IIb ␤ 3 ; 3) the recruitment of PKC␤ to ␣ IIb ␤ 3 localizes the enzyme to lamellipodial edges and focal complexes, precisely the subcellular locations where the early phases of outside-in ␣ IIb ␤ 3 signaling take place (3,47); 4) PKC␤ is required for cytoskeletal reorganization and spreading of platelets on fibrinogen; and 5) the identification of PKC␤ within ␣ IIb ␤ 3 -based signaling complexes provides new support for a scaffold function of ␣ IIb ␤ 3 in adhesive signaling.
Role of RACK1 in Bridging ␣ IIb ␤ 3 and PKC␤-In cells other than platelets, some PKC isoforms may be able to interact with ␤ 1 integrins, perhaps directly (e.g. PKC␣) (48) or indirectly through tetraspanins (e.g. PKC␤II) (49) or RACK1 (28). Several observations suggest that RACK1 is physically interposed between ␣ IIb ␤ 3 and activated PKC␤ in platelets. First, RACK1 can bind directly to integrin ␤ 1 , ␤ 2 , and ␤ 3 tails in vitro through interactions that require conserved, membrane-proximal residues in the integrin tails and the WD5-7 repeats in RACK1 (28,50). Second, PKC␤ is one of the PKC isoforms that has been shown to bind directly to RACK1 through interactions involving the C2 domain of PKC␤ and repeats WD3 and WD6 of RACK1 (27,51). Interestingly, the V5 region of PKC␤II but not PKC␤I contains a second binding site for RACK1 (52). Third, RACK1 is constitutively associated with ␣ IIb ␤ 3 in platelets, as assessed by co-immunoprecipitation (Fig. 1B), thus placing a pool of this adapter molecule in the proper location to bridge ␣ IIb ␤ 3 and activated PKC␤. The constitutive association of RACK1 with ␣ IIb ␤ 3 contrasts with the reported PMA-inducible association of recombinant glutathione S-transferase (GST)-RACK1 with ␤ 1 integrin in 293T cells (28). Fourth, fibrinogen binding to platelets was associated with a marked increase in PKC activity in ␣ IIb ␤ 3 immunoprecipitates (Fig. 3A). This would be expected if the interaction was mediated by RACK1 because it binds selectively to activated PKCs (27,51). Finally in ␣ IIb ␤ 3 CHO cells, expression of the RACK1 WD6/7 truncation mutant, which can bind PKC but not integrins (28,39), abrogated fibrinogen-dependent interaction of GFP-PKC␤ with ␣ IIb ␤ 3 (Figs. 7 and 8) and blocked cell migration (Fig. 9).
Previous work has demonstrated that RACK1 can bind to a yeast two-hybrid bait representing the membrane-proximal 20 amino acid residues of the ␤ 2 cytoplasmic tail (28). In the current study, fibrinogen-dependent association of PKC␤ with ␣ IIb ␤ 3 in CHO cells no longer occurred when the ␤ 3 cytoplasmic tail was truncated at residue 724 (Fig. 7C). By eliminating 39 carboxyl-terminal residues from ␤ 3 , this truncation also eliminates many of the conserved residues that correspond to those sufficient for binding of ␤ 2 to RACK1 (28). Therefore, the weight of evidence suggests that RACK1 mediates the interaction of PKC␤ with ␣ IIb ␤ 3 in platelets. However, additional modes of PKC␤ interaction with ␣ IIb ␤ 3 are possible.
Role of PKC␤ in ␣ IIb ␤ 3 Signaling-PKC␤ co-immunoprecipitated with ␣ IIb ␤ 3 after fibrinogen binding to platelets, and it localized to lamellipodial edges and nascent adhesion sites in fibrinogen-adherent ␣ IIb ␤ 3 CHO cells. Thus, a stable association of PKC␤ with ␣ IIb ␤ 3 may be required for specific aspects of outside-in signaling. This formulation is supported by the inability of platelets from PKC␤ Ϫ/Ϫ mice to reorganize their cytoskeletons or to spread on fibrinogen (Fig. 6A) and by the migration defect of ␣ IIb ␤ 3 CHO cells expressing RACK1 WD6/7, which prevented PKC␤ association with ␣ IIb ␤ 3 (Figs.  8 and 9). On the other hand, PKC␤ Ϫ/Ϫ platelets retained inside-out signaling, as manifested by normal agonistinduced fibrinogen binding (Fig. 6B). Therefore, PKC␤ is not essential for ␣ IIb ␤ 3 activation, either because other PKC isoforms can compensate for the loss of PKC␤ in PKC␤ Ϫ/Ϫ platelets or because proteins that play key roles in inside-out signaling, such as talin, are substrates for PKC isoforms other than PKC␤ (3,4). There is precedent in platelets for individual PKC isozymes fulfilling unique roles by virtue of their targeting to specific substrates. For example, PKC interacts specifically with Bruton's tyrosine kinase downstream of glycoprotein Ib-IX-V and glycoprotein VI receptors, resulting in the reciprocal modulation of each other's activity (20). In contrast, PKC␦ selectively interacts with Fyn downstream from these same receptors, resulting in translocation of both kinases to the platelet membrane (21).
Several proteins that are required for normal outside-in signaling are known to bind directly or indirectly to ␣ IIb ␤ 3 and may, therefore, be considered potential substrates for PKC␤. These include c-Src, Syk, Vav, SLP-76, and PTP-1B (3,53). Some of these proteins contain potential phosphorylation sites for classical PKCs (motif scan analysis, medium stringency); however, the substrates of PKC␤ that participate in outside-in ␣ IIb ␤ 3 signaling remain to be identified. Interestingly, the ␤ 3 cytoplasmic tail becomes phosphorylated on Thr-753, an event proposed to be implicated in outside-in signaling; however, that site is said to be recognized preferentially by PDK1 and Akt/ PKB (54,55). In principle, PKC␤ might phosphorylate relevant substrates directly or influence the action of another protein kinase or phosphatase. In this context, both PKC␤ and c-Src can bind to RACK1, and in NIH 3T3 cells the c-Src interaction is enhanced by PKC, resulting in down-regulation of c-Src signaling, cytoskeletal rearrangements, and cell growth (29,56,57). However, in platelets, c-Src can bind directly to the ␤ 3 integrin cytoplasmic tail, suggesting that the c-Src interaction with RACK1 or PKC␤ may not be essential in the platelet system (5,6). In support of this, the inhibition of PKC with bisindolylmaleimide I had no effect on c-Src activation in fibrinogen-adherent platelets, and the inhibition of Src with SU6656 had no effect on ␣ IIb ␤ 3 interaction with RACK1 or PKC␤. Therefore in platelets, PKC␤ and c-Src may function in parallel rather than by direct cross-talk, at least during the initial phases of outside-in ␣ IIb ␤ 3 signaling.
In the future, advances in proteomics may allow unbiased detection of the range of phospho-proteins associated with ␣ IIb ␤ 3 in platelets, providing a more complete picture of integrin signaling (58). Nonetheless, by uncovering a functionally significant interaction between PKC␤ and ␣ IIb ␤ 3 , the present studies provide new support for the idea that the ␣ IIb ␤ 3 cytoplasmic tails function as a scaffold for the assembly of a protein complex that initiates and propagates signaling to the platelet actin cytoskeleton (3). These results also lead to new questions.
How does fibrinogen binding recruit PKC␤ to ␣ IIb ␤ 3 ? Is the catalytic activity of PKC␤ regulated by this interaction? Do PKC␤I and PKC␤II have different roles in outside-in signaling? What are the stoichiometries of RACK1 interaction with ␣ IIb ␤ 3 and PKC␤ interaction with RACK1? Is PKC␤ required for normal platelet function during hemostasis?