We have identified and purified a platelet integral membrane protein (140,900 mol wt), using the KC4 monoclonal antibody specific for activated platelets, that is internal in resting platelets but exposed on activated platelets (Hsu-Lin S.-C., C. L. Berman, B. C. Furie, D. August, and B. Furie, 1984, J. Biol. Chem. 259: 9121-9126.). The expression ofthe protein on the platelet surface is secretion-dependent. This protein has been named platelet activation-dependent granule-external membrane (PADGEM) protein. PADGEM protein is distinct from the surface glycoproteins of resting platelets, but identical to the S12 antigen, GMP-140. Using immunofluorescent staining, resting platelets failed to stain for PADGEM protein with the KC4 antibody, but after permeabilization showed a punctate staining of the cell interior. Thrombin-stimulated intact platelets stained with a peripheral rim pattern thus demonstrating the translocation of PADGEM protein from an internal location to the cell surface. PADGEM protein expression on the platelet surface at varying thrombin concentrations correlated with alpha granule release, as measured by the secretion of platelet factor 4. Further evidence for an alpha granule localization of PADGEM protein was provided by nitrogen cavitation of resting platelets followed by metrizamide density gradient centrifugation; PADGEM protein codistributed with platelet factor 4. Using immunoelectron microscopy, the protein was localized to the alpha granule in frozen ultrathin sections of resting platelets labeled using rabbit anti-PADGEM protein antibodies, whereas in thrombin-activated platelets, the plasma membrane was labeled. These studies indicate that PADGEM protein is a component of the alpha granule membrane of resting platelets and is incorporated into the plasma membrane upon activation and secretion.

We have measured the association of platelet surface membrane proteins with Triton X-100 (Triton) -insoluble residues in platelets surface labeled with ' 25 1 . In both concanavalin A (Con A)-stimulated and resting platelets, this fraction is composed largely of polypeptides with apparent molecular weights of 45,000, 200,000, and 250,000 which comigrate with authenic actin, myosin heavy chain, and actin binding protein, respectively, as judged by PAGE in SDS. Less than 10% of the two major ' 25 1-labeled surface glycoproteins, GPllb and GPIII, were associated with the Triton residue in resting platelets. Within 45 s after Con A addition, 80-95% of these two glycoproteins became associated with the Triton residue and the amount of sedimentable actin doubled . No cosedimentation of GPllb and III with the cytoskeletal protein-containing Triton residue was seen when Con A was added to a Triton extract of resting cells, indicating that the sedimentation of GPllb and III seen in Con A-stimulated platelets was not due to precipitation of the glycoproteins by Con A after detergent lysis. Treatment of Triton extracts of Con A-stimulated platelets with DNase I (deoxyribonucleate 5'-oligonucleotididohydrolase [EC 3.1 .4 .5]) inhibited the sedimentation of actin and the two surface glycoproteins in a dose-dependent manner . This inhibition of cosedimentation was not due to an effect of DNase I on Con A-glycoprotein interactions since these two glycoproteins could be quantitatively recovered by Con A-Sepharose affinity absorption in the presence of DNase I . When the Con A bound to the Triton residue was localized ultrastructurally, it was associated with cellsized structures containing filamentous material . In intact cells, there was simultaneous immunofluorescent cored istribution of surface-bound Con A and myosin under conditions which induced a redistribution of platelet myosin . These data suggest that Con A can, in the intact platelet, induce physical interactions between certain surface glycoproteins and the internal cytoskeleton .There is a considerable immunohistochemical and some biochemical evidence suggesting that cell surface proteins and glycoproteins interact in a transmembrane manner with the cell cytoskeleton, especially after surface proteins have been aggregated by multivalent ligands (1,2,5,6,(21)(22)(23)32) .Elucidation of the molecular architecture of this complex of surface proteins and the cytoskeleton requires cells in which the cytoskeleton with associated surface proteins can be isolated in bulk and in a relatively pure and intact state . The human platelet appears to meet these criteria. In addition, the glycoprotein composition of the platelet surface has been wellcharacterized (4,19,(26)(27)(28) . Phillips and his co-workers (29) have recently shown that two major surface glycoproteins, designated GPIlb and GPIII, become associated with platelet THE JOURNAL OF CELL BIOLOGY " VOLUME 92 FEBRUARY 1982 565-573©The Rockefeller University Press -0021-9525/82/02/0565/09 $1 .00 cytoskeletons i...

Human fibrinogen has an Arg-Gly-Asp-Ser (RGDS) sequence at residues 572- 575 of its A alpha-chain. Although RGDS-containing peptides inhibit fibrinogen binding to stimulated platelets, these peptides also inhibit platelet binding of human fibrinogen fragment X and rat fibrinogen, which lack RGDS sequences corresponding to A alpha 572–575. Thus competition between free RGD-containing peptides and internal RGDS sequence at A alpha 572–575 is not the basis for their inhibition of fibrinogen binding to platelets. Addition of a Thr to the carboxy- terminus and an Asn to the amino-terminus of the RGDS sequence, the amino acids corresponding to A alpha 576 and 571 respectively, reduced the inhibitory potency of RGDS-containing peptides by fourfold to tenfold. Arg-Gly-Asp-Phe (RGDF) corresponds to A alpha 95–98, and the RGDF peptide was an effective inhibitor of fibrinogen binding, fourfold to fivefold more potent than RGDS. Thus, local primary structure may play an important role in regulating the capacity of RGD sequences in proteins to interact with specific adhesion receptors.

Normal primary platelet aggregation requires agonist-mediated activation of membrane GPIIb-IIIa, binding of fibrinogen to GPIIb-IIIa, and cellular events after ligand binding. PAC1 monoclonal antibody distinguishes between resting and activated states of GPIIb-IIIa, and other antibodies preferentially recognize GPIIb (PMI-1) or IIIa (anti- LIBS1) after the binding of fibrinogen or fibrinogen-mimetic peptides, such as GRGDSP. Using these antibodies and platelet flow cytometry, we studied two distinct persistent platelet aggregation abnormalities. Platelets from a thrombasthenic variant, which contained near-normal amounts of GPIIb-IIIa, failed to aggregate or bind PAC1 in response to agonists. In addition, GRGDSP, which binds to normal GPIIb-IIIa without prior cell activation, failed to increase the binding of PMI-1 or anti- LIBS1 to the thrombasthenic platelets, suggesting a primary defect in ligand binding. Chromatography of detergent-solubilized platelets on a KYGRGDS affinity column confirmed that the patient's GPIIb-IIIa lacked the fibrinogen binding site. In another patient with myelofibrosis and defective aggregation, PAC1 failed to bind to adenosine diphosphate- stimulated platelets, but did bind when protein kinase C was directly activated with phorbol myristate acetate. Furthermore, the binding of PMI-1 and anti-LIBS1 increased in response to GRGDSP, confirming a defect in agonist-mediated fibrinogen receptor activation rather than in fibrinogen binding or events distal to binding. These studies indicate that this immunochemical approach is useful in classification of clinical abnormalities of platelet aggregation as defects in either (a) fibrinogen receptor activation, (b) fibrinogen binding, or (c) postoccupancy events.