A Sequence within the Cytoplasmic Tail of GpIIb Independently Activates Platelet Aggregation and Thromboxane Synthesis*

All integrin α subunits contain a highly conserved KXGFFKR motif in their cytoplasmic domains that plays a crucial role in the regulation of integrin affinity for their ligands. We show that a lipid-modified peptide corresponding to the cytoplasmic region, 989–995, of the platelet integrin subunit glycoprotein GpIIb (αIIb), palmitoyl-KVGFFKR (Ppep; 10 μm), but not a similarly modified scrambled peptide (palmitoyl-FKFVRGK), can specifically induce platelet activation and aggregation equivalent to that of strong agonists such as thrombin. Ppep-induced aggregation is also associated with indices of platelet activation including thromboxane A2 (TXA2) synthesis (EC50 = 45 ± 5 μm), secretion of α-granules detected as enhanced surface expression of P-selectin (EC50 = 52 ± 8 μm), and conformational changes in GpIIb/IIIa measured by the monoclonal antibody, PAC-1 (EC50 = 3.7 ± 1 μm). The TXA2 receptor antagonist, SQ29548, PGE1, and the ADP scavenger, apyrase, differentially inhibit the aggregation response and TXA2 synthesis in response to Ppep. Similarly, GpIIb/IIIa antagonists (RO-449883 and integrelin), which inhibit aggregation by greater than 90%, have little effect on peptide-induced TXA2 synthesis, suggesting that this event is independent of fibrinogen binding to GpIIb/IIIa. Alanine-stepping of the Ppep sequence identifies GFFK(991–994) as the critical residues in all peptide-mediated events. We conclude that this peptide can imitate the cytoplasmic domain of GpIIb and initiate parallel but independent signaling pathways, one leading to ligand binding and platelet aggregation and the other to intracellular signaling events such as TXA2 synthesis and secretion.

Integrins are a family of cell adhesion molecules composed of two subunits, ␣ and ␤, which form a complex on the cell surface. Ligand recognition by integrins may be modulated by intracellular signals that interact with the cytoplasmic tails of the subunits. This has been demonstrated most clearly for the platelet glycoprotein (Gp) 1 IIb/IIIa (␣IIb␤3), the most abundant platelet integrin, which acts as a receptor for fibrinogen, fibronectin, and other RGD-containing macromolecules (1). Under resting conditions, this receptor has a low affinity for its ligands (2). However, when platelets are stimulated by agonists such as thrombin or ADP, GpIIb/IIIa undergoes a conformational change (3) detected by the appearance of neoepitopes for monoclonal antibodies such as PAC-1 (4) and acquires a high affinity binding for its ligands, principally fibrinogen (5,6). The binding of fibrinogen results in platelet aggregation, an early step in the generation of a thrombus. Deletion or mutation of the cytoplasmic domains of the integrin subunits can produce a constitutively active or inactive receptor (7)(8)(9), suggesting that signals resulting from cell activation interact with the intracellular components of GpIIb/IIIa to modify ligand recognition (inside-out signaling).
The cytoplasmic domains are also important for events occurring as a consequence of ligand-integrin interactions, socalled "outside-in" signaling. In the case of GpIIb/IIIa, ligand binding and receptor clustering is followed by an array of intracellular signals including thromboxane A 2 (TXA 2 ) generation and tyrosine phosphorylation events (10,11). After ligand binding, GpIIb/IIIa becomes tightly associated with the membrane skeleton as focal contact structures form, composed of several cytoskeletal elements including talin, vinculin, and spectrin. Potential signaling molecules such as pp60c-src, pp62c-yes, phosphoinositide 3-kinase, and protein kinase C also associate with these focal contact structures (12,13). Several sites within the cytoplasmic tail of the two integrin subunits have been implicated in this process (14 -16). Phosphorylation of the cytoplasmic tail of GpIIIa occurs in parallel with ligand occupancy, permitting the binding of signaling proteins SHC and GRB2 (17,18). Naik and co-workers (19) showed that a 25-kDa calcium-and integrin-binding protein (CIB) interacts with the GpIIb cytoplasmic tail.
A highly conserved amino acid sequence exists immediately proximal to the transmembrane-spanning region in the ␣ cytoplasmic domains of all integrins. This motif, KXGFFKR, has been shown to interact with calreticulin in an inducible manner in cells expressing the collagen receptor ␣ 2 ␤ 1 (20). Furthermore, studies in which this sequence is deleted or mutated, have strongly suggested that it is involved in the regulation of the integrin affinity state (9,16,(21)(22)(23). However, deletion or mutation of this region also modifies ligand recognition by the integrin and presumably, the conformation of the receptor, which will impact upon cellular events after ligand occupancy. Moreover, since modified integrins cannot be analyzed in native cells, it is necessary to study their functions in transfected cells that may not have a full complement of signaling molecules. As an alternative approach, we have examined the functional effects of a synthetic peptide sequence corresponding to the conserved amino acid motif on intact, human platelets. Our findings demonstrate that a lipid-soluble peptide, palmitoyl-KVGFFKR mediates platelet activation in a highly specific manner and identifies a component of GpIIb that is involved in thromboxane formation independent of fibrinogen binding or platelet aggregation. Vectashield mounting medium for fluorescence was obtained from Vector Laboratories Inc. (Burlingame, CA). CD62P (anti-GMP-140) was obtained as a phosphatidylethanolamine conjugate from Cymbus Corporation (Hants, UK). TXB 2 enzyme-linked immunoabsorbant assay kits were obtained from Assay Designs (Ann Arbor, MI)). The GpIIb/ IIIa antagonist Integrelin was obtained from COR Therapeutics, (San Francisco, CA). RO44 9883 (1-(N-(p-amidinobenzoyl) L-tyrosyl)-4-piperidinyl) oxy) acetic acid) was a gift from Dr. S. Roux of Hoffman LaRoche (Basle, Switzerland). All other materials were reagent grade or better. Deionized water (Purite, grade 1) was used throughout all experiments.

Materials-Prostaglandin
Peptide Synthesis-A peptide corresponding to the membrane proximal portion of the GpIIb cytoplasmic tail containing the amino acid sequence K989-R995 (KVGFFKR) and a scrambled version of this sequence (FKFVRGK) were synthesized on an Applied Biosystems automated peptide synthesizer (model 432A, Norwalk, CT) using a standard solid-phase Fmoc procedure. Corresponding peptides palmitoylated on the amino-terminal amino acid were also synthesized in an identical manner. Portions of both KVGFFKR and palmitoylated-KVGFFKR were subsequently labeled with carboxyfluorescein and used in immunofluorescent assays. Sequential alanine- were also synthesized. All peptides were purified after synthesis using reverse-phase high pressure liquid chromatography and confirmed using electrospray mass spectrometry.
Platelet Aggregation-Platelets obtained from volunteers free from medication were collected into 0.15 vol./vol. acid-citrate-dextrose (ACD: 38 mM citric acid, 75 mM Na citrate, 124 mM dextrose) anticoagulant and washed using a modification of the method of Mendelsohn et al., 1991 (24). Briefly, blood was centrifuged at 180 g for 10 min. at room temperature. Platelet-rich plasma was then acidified to pH 6.5 with ACD and PGE 1 (1 M) was added. The platelets were pelleted through plasma by centrifugation at 750 g for 10 min. at room temperature. The supernatant was removed and the platelet pellet resuspended in 130 mM NaCl, 10 mM trisodium citrate, 9 mM NaHCO 3 , 6 mM dextrose, 0.9 mM MgCl 2 , 0.81 mM KH 2 PO 4 , 10 mM Tris pH 7.4, adjusted to 2 ϫ 10 8 /ml and supplemented with 1.8 mM CaCl 2 . Platelet aggregations were performed at 37°C in a BioData PAP-4 aggregometer (Horsham, PA). Peptide was dissolved in deionized water at a concentration of 1 mg/ml and used at the concentrations indicated to induce aggregation. Thrombin (0.2units/ml) or thrombin receptor activating peptide (TRAP; 10 M; SFLLRN) were used as positive controls in all experiments. Aliquots for TXB 2 analysis were snap frozen and stored at Ϫ80°C until analysis.
Fluorescent Microscopy-Near-confluent human erythroleukemic (HEL) cells (25) were cultured overnight on Falcon culture Slides ™ in the presence of 10 nM phorbol 12-myristate 13-acetate in standard tissue culture medium (10% fetal calf serum in DMEM) in order to facilitate adhesion of these platelet-like cells which normally grow in suspension. Alternate slides were incubated for 1-2 min with 45 M carboxyfluorescein-labeled KVGFFKR or carboxyfluorescein-labeled, palmitoylated-KVGFFKR. The slides were washed three times by immersion in phosphate-buffered saline and then mounted in vectashieldmounting medium and analyzed on a Nikon diaphot 300 microscope with epifluorescent attachment using 400X magnification.
TXA 2 Analysis-Washed platelets were diluted to 2 ϫ 10 8 /ml and stimulated for 3 min whereas stirring in an aggregometer before being snap-frozen in liquid nitrogen and stored for analysis. Platelet TXA 2 was measured as its stable metabolite TXB 2 by ELISA (26).
FACS Analysis of P-selectin and PAC-1-Washed platelets were placed in the aggregometer at 37°C in the presence or absence of thrombin receptor activating peptide (TRAP) (10 M) or the indicated concentrations of Ppep and stirred for 3 min. 50 l aliquots were then placed on ice and incubated with CD62P-phosphatidylethanolamine as a marker for ␣-granule degranulation (27). Parallel samples were incu-bated with PAC-1 fluorescein isothiocyanate to determine the activation status of GpIIb/IIIa. Incubations were for 30 min. Samples were washed and resuspended in phosphate-buffered saline containing 0.1% bovine serum albumin (phosphate-buffered saline/bovine serum albumin) and analyzed on a Becton Dickinson FACS-Scan (San Jose, CA) at 488/510 nm. Controls were included in all assays and were obtained by incubating platelets as above in the presence of 20 M TRAP or 0.2 units/ml thrombin (positive control) or in the absence of any agonist (negative control). Nonspecific binding was determined in control and activated platelets using a fluorescein isothiocyanate-labeled ␥-1 mouse IgG. Data are expressed as percent positive cells for P-selectin assays and as mean fluorescence for PAC-1 assays.

RESULTS
The membrane-proximal seven amino acids (Lys-989 -Arg-995) of GpIIb were synthesized either as unmodified peptides or as palmitoylated isoforms and purified by high performance liquid chromatography. Peptide purity was ascertained by electrospray mass spectrometry and was routinely Ͼ99%.
Unmodified KVGFFKR had no effect on platelets. However, the addition of a palmitate group to the peptide increased its lipid permeability and permitted an interaction with intracellular components of the platelet. Palmitoylated peptide (Ppep45 M), but not equal concentrations of palmitate alone, KVGFFKR alone, or palmitate plus unconjugated KVGFFKR caused aggregation in washed human platelets (Fig. 1A). There was a lag time of between 10 and 20 s before the initiation of aggregation. No obvious shape change response was observed in this time period. Palmitoylated scrambled peptide (P-FK-FVRGK), even at higher concentrations (120 M), failed to cause significant platelet aggregation (Fig. 1B).
To establish if the peptide was gaining access to the platelet cytoplasmic milieu, carboxyfluorescein-labeled peptides were synthesized with and without the palmitate modification. In Fig. 2, we show that only the palmitoylated peptide can be observed in the cytoplasm of HEL cells, a platelet-like, human megakaryocyte cell line (25). No fluorescence was observed in cells incubated with carboxyfluorescein-labeled KVGFFKR that lacked the lipid modification. Similarly, flow cytometric analysis of either HEL cells or platelets shows that only the palmitoylated, fluoresceinated peptide had significant cellular association (Fig. 2, c and d). Access of labeled Ppep to cells is independent of the presence of GpIIb/IIIa, as excess of either Ppep or palmitoylated scrambled peptide compete equally for label uptake into cells. Furthermore, we have evidence of Ppep uptake into cells that do not express GpIIb/IIIa, including human umbilical vein endothelial cells (data not shown).
Ppep-induced aggregation was acutely dose-dependent and was maximal at 10 M (Fig. 3a). Platelet aggregation induced by Ppep, like that caused by thrombin, was accompanied by TXA 2 synthesis in a dose-dependent manner (Fig. 3b). The maximum response determined by Michaelis-Menton kinetics was 644.5 Ϯ 49.22 ng/ml TXB 2 , with half-maximal effect (EC 50 ) at 45 M peptide. Ppep-induced thromboxane formation was paralleled by secretion, which can be measured by expression of the ␣-granular marker, CD62P or P-selectin (27), on the surface of the platelet (Fig. 3b). The EC 50 for this effect was 52 M, which corresponded to 41.3 Ϯ 14% positive cells compared with 0.5 Ϯ 3.1% positive cells in untreated, control platelets. The disparity in the EC 50 values for the signaling pathways (45-52 M; thromboxane synthesis and secretion) compared with the dose of Ppep that produces a maximal aggregation response (10 M) would suggest a different mechanism of activation for these respective events. Finally, Ppep induced PAC-1 expression (Fig. 3c) in a time-and dose-dependent manner. This antibody recognizes an epitope on the GpIIb/IIIa complex that is exposed after platelet activation, permitting fibrinogen binding (4).
To define the membrane-proximal residues of the IIb subunit important for the aggregating activity of this peptide, we sequentially substituted each residue in the peptide with alanine (Ala). The Ala-stepped peptides were then assayed for their ability to induce platelet aggregation, and samples were analyzed in parallel for TXA 2 production (Fig. 4). Platelet aggregation activity was lost in Ppep G991A, Ppep F992A, Ppep F993A, and Ppep K994A. In contrast, activity was present but submaximal in Ppeps K989A, V990A, and R995A. A parallel profile of activity was observed for TXA 2 synthesis activity with the Ala-stepped peptides. Thus, TXA 2 synthesis and aggregation seem to be associated facets of Ppep stimulation.
To explore the mechanisms of Ppep-induced platelet activation, we examined the response to several inhibitors (Fig. 5). SQ29 548, a highly specific thromboxane receptor antagonist (10 M), reduced platelet aggregation by 63% but did not abolish it. Similarly, apyrase (10 units/ml), an ADP scavenger, reduced platelet aggregation by 73%. Aggregation was also only partially inhibited by PGE 1 (5 M), a concentration that abolished responses to thrombin (0.2 units/ml). These findings demonstrate that platelet aggregation was augmented by the formation of TXA 2 and the release of ADP but was not initiated by these events. Similarly, Ppep-induced platelet activation is not mediated via the activation of a G-protein-dependent mechanism as these events are inhibited by PGE 1 . Thus it would appear that Ppep directly induces a high affinity state in GpIIb/ IIIa and this is responsible for initiating the intracellular signaling events and platelet aggregation. Moreover, none of the inhibitors used altered TXA 2 formation, suggesting a dissociation between aggregation and TXA 2 synthesis. Consistent with this, two GpIIb/IIIa antagonists, RO 449883 (28) and integrelin

FIG. 2. Ppep but not unmodified peptide is cell-permeable.
HEL cells were adhered to tissue culture slides by incubating overnight in 10 nM phorbol 12-myristate 13-acetate as described under "Experimental Procedures." Cells were then incubated with carboxyfluorescein-labeled palmitoylated-KVGFFKR peptide (a) or carboxyfluorescein-KVGFFKR (b) and analyzed by fluorescent microscopy using a 400ϫ magnification. HEL cells (c) and platelets (d) were also analyzed by flow cytometry for the cellular association of the fluoresceinated, unmodified peptide (solid histograms) or fluoresceinated-Ppep (open histograms). (29), suppressed platelet aggregation by greater than 90% but had no effect on TXA 2 formation. DISCUSSION The KXGFFKR sequence is common to ␣ subunits of integrins and appears critical for receptor function. Deletion or mutation analysis directed against this motif suggests that this sequence or a factor binding to it is critical for maintaining the integrin in its low affinity binding state (15,16,22). Thus, deletion of the KVGFFKR sequence in GpIIb/IIIa results in a constitutively active receptor capable of recognizing its ligand and binding it with high affinity. This cytoplasmic segment also recognizes intracellular proteins that may play a role in signal transduction such as calreticulin (20,30,31) and the calcium and integrin binding protein, CIB (19). We synthesized the peptide KVGFFKR and added a palmitoyl group to facilitate its access to the intraplatelet milieu, an approach that has been successful with other peptides (32)(33)(34). Fluorescence microscopy and flow cytometry demonstrate, using carboxyfluorescein-labeled peptides, that only peptides modified with palmitate gained access to intracellular regions of HEL cells and platelets. Staining in HEL cells was homogeneously distributed within the entire cytoplasmic area and was notably absent from the nuclear region.
The palmitoylated peptide, at concentrations as low as 10 M, induced platelet activation and aggregation, whereas a scrambled peptide, similarly palmitoylated, had no effect even at concentrations as high as 120 M. Unconjugated peptide, in the presence or absence of equimolar amounts of free palmitate, had no effect on platelet function, affirming that the peptide must gain access to intracellular sites for biological effect. Neither palmitate alone or palmitoylated-scrambled peptide induced platelet aggregation, verifying that the effect on platelets was due to a specific Ppep-mediated response and not nonspecific effects such as to disruption of the cell membrane or cell lysis. Aggregation resulted from activation of GpIIb/IIIa as there was enhanced binding of the monoclonal antibody, PAC-1, which recognizes the active conformation of the integrin (4). Moreover, platelet aggregation was inhibited by integrelin and RO44 9883, highly specific GpIIb/IIIa antagonists, verifying that the aggregation response is dependent on activation of GpIIb/IIIa. The steep dose-response curve for Ppep-induced platelet aggregation reflects the inability of platelet aggregometry to detect small aggregates of platelets. Such aggregates can be observed by flow cytometry at doses as low as 2 M Ppep as an increase in both forward and sideward scatter within the platelet population. The appearance of these small aggregates is accompanied by PAC-1 expression in a parallel manner. The EC 50 for Ppep induction of PAC-1 binding was 3.7 Ϯ 1.1 M, and the maximum effect was achieved at 10 M.
To address which residues of the peptide were responsible for the biological activity, we sequentially replaced each amino acid with an alanine residue. The results showed that the response was highly specific and that the activity resided predominantly in the region GFFK corresponding to amino acids 991-994 of the GpIIb cytoplasmic tail. Mutation of the valine 990 resulted in a reduction in platelet aggregation, although the peptide was still active. This is interesting as this is the residue that varies most between different ␣ integrins, suggesting that cell activation and signaling via this mechanism may be common to all integrins. In studies with the alanine-stepped palmitoylated peptides, the peptides that gave a poor reaction in platelet aggregation assays were also poor inducers of TXA 2 synthesis. These data are in agreement with the mutational studies by Hughes et al. in which similar alanine substitution revealed identical critical residues for constitutive expression of PAC-1 binding (16).
In addition to platelet aggregation, Ppep also induced platelet activation as measured by TXA 2 formation and secretion of granular contents. However, the products of these positive feedback pathways are not the mediators of Ppep-induced platelet activation. This is proven by two separate observations. First, the EC 50 values for Ppep-induced TXA 2 synthesis and P-selectin expression are approximately 50 M, whereas maximal aggregation occurs at 10 M Ppep. Second, inhibitor studies show that agents that scavenge ADP or that prevent the actions of TXA 2 only partially inhibited Ppep-induced aggregation. This finding indicates that secretion of ADP or synthesis of TXA 2 augment but are not direct initiators of Ppepinduced aggregation. Conversely, fibrinogen binding or aggregation induced by Ppep do not directly cause TXA 2 synthesis, as this pathway is unaltered by GpIIb/IIIa antagonists. Thus, it would appear that fibrinogen binding is not necessary for thromboxane synthesis to occur. In our studies, there is a possibility that the synthetic GpIIb/IIIa antagonists used are themselves acting as ligands and permitting outside-in signaling events to occur. In recognition of this possibility, we used two antagonists, integrelin and RO 449883, that differed in their abilities to expose ligand-induced binding sites and, hence, to act as partial ligands. Integrelin, a cyclic heptapeptide RGD mimetic (35,36), causes marked conformational changes in GpIIb/IIIa, exposing ligand-induced binding site epitopes (37), whereas RO 449883 does not (38). Both antago-nists inhibited platelet aggregation but not TXA 2 synthesis. Identical results were also obtained using F(ab)Ј fragments of the inhibiting monoclonal antibody, 7E3 (39, 40) (data not shown). Thus, integrin occupancy by its ligand and integrininitiated signaling may be distinguished as independent events. We conclude, therefore, that the interaction of Pal-KVGFFKR with GpIIb/IIIa represents a branch point in these two pathways such that both pathways require interaction with the peptide sequence and have the same intolerance to sequence mutation. Structure-function analysis pinpoints the region Gly-991-Lys-994 as the important region for these events. However, once the peptide has bound to its target, the processes of integrin ligation, clustering, and platelet aggregation proceed independently from the signaling pathways leading to TXA 2 synthesis.
Our findings demonstrate that KVGFFKR can both activate GpIIb/IIIa and induce intracellular signals independently. We propose a model where the sequence KVGFFKR in the cytoplasmic region of GpIIb, adjacent to the transmembrane region, provides a site for tethering the integrin and stabilizing it in a low affinity state. This is consistent with the deletion and mutational studies of O'Toole et al. (9) and Hughes et al. (41) and may be mediated through an interaction with signaling protein(s). Competition by the peptide displaces the stabilizing signaling factor, simultaneously activating the receptor by releasing it from its tether and allowing signaling to proceed via the tethering protein. An alternative possibility is that the KVGFFKR sequentially promotes receptor clustering and li- gand binding. The occurrence of integrin clustering may lead to cell signaling, whereas ligand binding is necessary for aggregation. Consistent with this possibility is the recent observation that integrin clustering can mediate cell signaling independent of ligand binding (23).
In conclusion, structure-function analysis of the intracellular segment of the GpIIb subunit (Lys-989 -Arg-995) using the cell-permeable peptide, Palmitoyl-KVGFFKR, pinpoints this motif and, in particular, the region Gly-991-Lys-994, as an important region in integrin signaling. This peptide substitutes for the endogenous GpIIb cytoplasmic tail and independently stimulates integrin ligation, leading to aggregation and platelet signaling events such as TXA 2 synthesis and secretion.