Identification of a nucleotide-binding site on glycoprotein IIb. Relationship to ADP-induced platelet activation.

Formalin-fixed platelets have been used to study the binding of adenine nucleotides in order to avoid the complications of nucleotide metabolism and to achieve steady-state binding. Sp-adenosine-5'-(1-thiotriphosphate) (Sp-ATP-alpha-S) binds to platelets at two sites (Kd1 3 nM; 31,000 sites/platelet; Kd2 200 nM; 300,000 sites/platelet) as compared with values for ADP under these conditions (Kd1 30 nM; 25,000 sites/platelet and Kd2 3 microM; 400,000 sites/platelet) (bound/total approximately 0.1). Competition binding experiments showed that both of the ATP-alpha-S sites were accessible to ADP and vice versa. [35S]ATP-alpha-S was photoaffinity cross-linked to unfixed platelets by direct irradiation with ultraviolet light. A single radiolabeled component (120 kDa) was identified and shown to be identical with the alpha subunit of GPIIb based on two-dimensional sodium dodecyl sulfate-polyacrylamide gel electrophoresis followed by Western blotting with anti-GPIIb monoclonal antibodies, by isoelectric focusing (pI 4.5-5.5), by immunoaffinity adsorption using monoclonal anti-GPIIb/IIIa antibodies coupled to Sepharose, and by crossed immunoelectrophoresis. Amino-terminal sequencing of a tryptic fragment labeled with [35S]ATP-alpha-S identified an 18-kDa domain beginning at Tyr-198 in the primary sequence of GPIIb alpha. These studies demonstrate the presence of an adenine nucleotide-binding site on GPIIb alpha.

Formalin-fixed platelets have been used to study the binding of adenine nucleotides in order to avoid the complications of nucleotide metabolism and to achieve steady-state binding. Sp-adenosine-5'-(l-thiotriphosphate) (S,-ATP-a-S) binds to platelets at two sites (&I 3 nM; 31,000 sites/platelet; Kdz 200 nM; 300,000 sites/ platelet) as compared with values for ADP under these conditions (& 30 nM; 25,000 sitedplatelet and Kd2 3 PM; 400,000 sites/platelet) (bound/total -0.1). Competition binding experiments showed that both of the ATP-a-S sites were accessible to ADP and vice versa.
[36S]ATP-a-S was photoaffinity cross-linked to unfixed platelets by direct irradiation with ultraviolet light. A single radiolabeled component (120 kDa) was identified and shown to be identical with the alpha subunit of GPIIb based on two-dimensional sodium dodecyl sulfate-polyacrylamide gel electrophoresis followed by Western blotting with anti-GPIIb monoclonal antibodies, by isoelectric focusing (PI 4.5-5.5), by immunoaffinity adsorption using monoclonal anti-GPIIb/ IIIa antibodies coupled to Sepharose, and by crossed immunoelectrophoresis. Amino-terminal sequencing of a tryptic fragment labeled with [35S]ATP-a-S identified an 18-kDa domain beginning at Tyr-198 in the primary sequence of GPIIb,. These studies demonstrate the presence of an adenine nucleotide-binding site on GPIIb,.
Despite the importance of ADP in platelet function, the mechanisms by which it activates platelets are imperfectly understood. For example, it has not been resolved whether platelet activation is mediated by a single type of receptor that is coupled to two pathways, one inducing activation and the other affecting adenylyl cyclase, or whether there are two different receptors each separately affecting one of the pathways. The first, or single receptor hypothesis, is supported by the constant ratio between the activities of a wide range of structurally diverse ADP analogues in their effects as agonists or antagonists of platelet activation and the inhibition of cAMP accumulation Hourani, 1982a, 1982b). The two-receptor hypothesis is supported by the fact that the adenosine analogue 5'-fluorosulfonylbenzoyl adenosine inhibits ADP-induced platelet aggregation but does not affect * The work was supported in part by United States Public Health Service Grant HL39438. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisernent" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. $ To whom correspondence should be addressed: Cell Biology Laboratory, American Red Cross, 15601 Crabbs Branch Way, Rockville, MD 20855. cAMP accumulation (Mills et al., 1985) while, conversely, pchloromercuribenzene sulfonate inhibits the effect of ADP on cAMP accumulation but does not inhibit aggregation (Mills and MacFarlane, 1977).
A major difficulty in obtaining kinetic binding data on the interaction of ADP with platelets has been their ability to rapidly metabolize the ligand and its analogues and the complicating effects of ADP secretion from the activated platelets. We have recently developed the use of fixed platelets to avoid these difficulties and to provide an indicator system for measuring the ability of ADP analogues to compete with ADP under steady-state conditions (Jefferson et al., 1988;Agrawal et al., 1989). Of 20 ADP analogues examined, the most effective competitor at the high affinity site was the S, diastereoisomer of adenosine 5'-O-(l-thiotriphosphate) (ATP-a-S) which bound about 10 times more strongly than did ADP itself (Agrawal et al., 1989). It has been clearly established that essentially any modification in the structure of ADP, whether in the purine, ribose, or phosphate moieties, results in the loss of platelet aggregating activity. The only exception to the strict structural requirements for ADP analogues in inducing aggregation is if substitution occurs in the C-2 position (for review, see Haslam and Cusack, 1981). Attempts to identify the platelet ADP receptor by affinity labeling techniques have been constrained by this structural limitation. 2-Azidoadenosine-5'-diphosphate labeled several components in intact platelets but none were competed by ADP suggesting that nonspecific labeling may have occurred (Macfarlane et al., 1982). We have synthesized 2-aminopropylthio-ADP but found it only about one-seventh as effective an agonist as ADP (Jefferson et al., 1987) and have subsequently found that its even larger ["Hlarylazide analogue was impractical as a photoaffinity probe due to low incorporation and low specific radioactivity.' The adenosine analogue 5'-fluorosulfonylbenzoyl adenosine, which has been used to identify adenine nucleotide-binding sites in over 40 isolated proteins (Colman, 1983), has been shown to label a 100-kDa membrane protein in intact platelets (Bennett et al., 1978), but the role of this protein in ADP-induced activation has not been clearly established (Figures et al., 1987;Colman, 1988).
Photoaffinity labeling of a specific site on a molecule usually requires a ligand substituted with a specific photolabile group (Chowdry and Westheimer, 1979). However, in a number of cases heterocyclic aromatic compounds have been photoactivated and directly coupled to cellular substrates by ultraviolet irradiation in the absence of photolabile substituents. This was first achieved in binding cyclic AMP to receptors in testicular homogenates (Antonoff and Ferguson, N. J. Greco, N. N. Tandon, and G. A. Jamieson, unpublished results. 1974;Obrig et al., 1975) and was subsequently extended to the specific binding of phencyclidine (Blaustein and Ickowicz, 1983) and benzodiazepines (Mohler et al., 1980) to synaptosomes, and of noncompetitive blockers to the acetylcholine receptor of Torpedo marmorata (Oswald and Changeux, 1981).
This approach of radiolabeling specific sites in cells by direct photoactivation suggested to us a way of circumventing the difficulties associated with the photoaffinity labeling of the ADP-binding sites. Moreover, since our studies in the fixed platelet system had shown that ATP-a-S bound avidly to the ADP-binding site (Agrawal et al., 1989) and since ATPa-S is a potent antagonist of ADP-induced platelet activation (Cusack and Hourani, 1982a), interpretation of labeling patterns would not be complicated by simultaneous platelet activation as might be the case in labeling with an agonist. Finally, "S-labeled ATP-a-S is readily available at high specific radioactivity which would compensate for low labeling efficiency expected in direct photoactivation.
In our present studies, we have combined the high binding affinity of ATP-a-S to platelets with direct photoactivation and have identified a nucleotide-binding site on GPIIb,.
Radioligand Binding Assay-For competition binding studies, platelets were fixed with paraformaldehyde and then washed with phosphate-buffered saline prior to resuspension in Tyrode's-Hepes, p H 7.4 (buffer B) a t 2 X lo9 platelets/ml (Jefferson et al., 1988Agrawal et al., 1989. Binding data were obtained using tracer concentrations of radiolabeled ligands and increasing concentrations of competitor (Agrawal et al., 1989). The ratio of bound/total ligand (B/ T) was set a t approximately 0.1 at the lowest ligand concentrations examined (Bylund and Yamamura, 1990). To achieve this ratio platelet numbers differed between assays: filtration, ["HIADP (1.5 X 10' platelets, 0.5 ml), [:"S]ATP-a-S (3 X lo7, 0.5 ml); centrifugation, ['HI ADP (5 X lo7 platelets, 0.1 ml), [%]ATP-a-S (1 X lo7 platelets, 0.5 ml). Following incubation a t 22 "C for 15 min, bound ligand was separated from free ligand by direct centrifugation (12,000 X g, 5 min) or by rapid filtration (5 s) through Whatman GF/B filters followed by two washes with 1.5 ml of buffer B using a Brandel MR-24 cell harvester (Brandel, Gaithersburg, MD). Eighteen point binding curves were performed in duplicate using platelet preparations from different donors and analyzed using the LIGAND program (Munson and Rodbard, 1980). Nonspecific binding for ["HIADP and [3sS]ATP-a-S assays were determined in the presence of 1 mM ADP and 20 p M ATP-a-S and The abbreviations used are: TPCK, L-1-tosylamido-2-phenylalanine chloromethyl ketone; SDS-PAGE, sodium dodecyl sulfate-polyacrylamide gel electrophoresis; PG, prostaglandin; HEL, human erythroleukemia; B/T, bound/total. were determined to be, respectively, -30 and 15% in the centrifugation assay technique and approximately 4 and 7% by the filtration assay technique.
Photoaffinity Labeling-Washed platelets (2 X 10") in modified Tyrode's buffer B (with 1 pg PGEJml but without bovine serum albumin) were mixed with [35S]ATP-a-S at a concentration of 85 nM, corresponding to 30 times the Kd of the high affinity binding site as determined on fixed platelets, in a total volume of 100 ~1 . Samples were immediately photolyzed on ice at 254 nm on white ceramic plates a t a distance of 4 cm with a UVS-54 Mineralight lamp (Ultraviolet Products, San Gabriel, CA) using a Woods-type filter. Optimal irradiation time was determined to be 5 min. A 4.5-mm thick quartz cuvette filled with buffer B was placed between the incubation well and the ultraviolet lamp. Incident energy was 360 microwatts/ cm' as determined by a UV meter (model 5255, UltraViolet Products).
To determine the ability of nucleotides to compete with ["S]ATPa-S in photoaffinity labeling requires special care to ensure that the same amount of incident energy is available both in the primary well and in the well containing the high concentration of competing nucleotide. T o ensure this, experiments were carried out in parallel: in one case the quartz cuvette placed between the source and the reaction well contained buffer B while the reaction well contained ["S]ATP-a-S plus the appropriate concentration of the competing nucleotide. In the other case, the quartz cuvette contained the same concentration of competing nucleotide in buffer B while the well contained [3'S]ATP-a-S alone. This experimental configuration ensured that any observed decrement in binding was due to competition and not to the absorption of ultraviolet radiation by the competing nucleotide. After photolysis, free ligand was removed by centrifugation (12,000 X g, 2 min) of the platelets through 10% sucrose in buffer B containing PGE,. The platelet pellet was washed without resuspension with the same buffer to remove unincorporated ["S]ATP-a-S, and the pellet was then solubilized for 15-20 h with solubilization buffer (1% Triton X-100 in 50 mM Tris, 2 mM phenylmethylsulfonyl fluoride, 2 mM benzamidine, and 50 pg of leupeptin/ml at pH 7.4). Triton X-100-soluble material was recovered by ultracentrifugation for 30 min at 217,000 X g.
Zmmunoaffinity Isolation-Purified AP-2 (1.86 mg) was coupled to 0.4 g of cyanogen bromide-activated Sepharose 4B (93% cross-linking) as per manufacturer's instructions (Pharmacia LKB Biotechnology Inc.). AP-2 immunoaffinity columns, stored in 50 mM Tris, 0.2 M NaC1,0.02% NaN3, pH 7.2, or in phosphate-buffered saline, pH 7.2, containing 0.02% NaN3, were washed before use with phosphatebuffered saline and equilibrated in solubilization buffer containing 5 mM Ca2+. Typically 1 X 10'oplatelets labeled with ["S]ATP-a-S were solubilized in Triton X-100 and then treated with 1 mM Ca" to ensure formation of the GPIIb/IIIa complex (Kunicki et al., 1981, Brass et al., 1985 before addition of the solution to 1 ml of AP-2-Sepharose. Sepharose-coupled AP-2 was calculated to be in 100-150fold excess of the maximum number of molecules of GPIIb/IIIa present. This mixture was rotated a t 22-25 "C end-over-end for 2-3 h before being transferred to a 10-ml Econo-column (Bio-Rad). After the breakthrough solution was collected, the column was washed with three column volumes of solubilization buffer containing 1 mM ca2+, 0.5 M NaCl, 0.1% Triton X-100 followed by five column volumes of solubilization buffer containing 1 mM Ca'+ and 0.1% Triton X-100. GPIIb/IIIa was then eluted with 0.1 M glycine, 0.1 M NaCI, 0.1% Triton X-100, pH 2.8; the eluted fractions were immediately neutralized by the addition of 0.025 volume of 1 M Tris.
Isolation and Enzymatic Digestion of GPZZb-For sequence analysis, ["S]ATP-a-S-labeled platelets from 25 irradiated wells (1 X 10'" platelets) were pooled and solubilized, and labeled GPIIb/IIIa was isolated as described above. The labeled fraction eluted from AP-2-Sepharose was simultaneously concentrated and exchanged with phosphate-buffered saline using a Centricon 30 microconcentrator (Amicon, Danvers, MA). The concentrate (50 pl) was diluted with 10 volumes of phosphate-buffered saline, pH 7.2, containing 0.1% Triton X-100 and reconcentrated before proteolysis. Digestion of ["SIATPa-S-labeled GPIIb/IIIa was carried out using TPCK-treated trypsin in phosphate-buffered saline containing 0.1% SDS and 100 p M dithiothreitol for 2 h a t 23 "C with a 1:20 (w/w) enzyme to substrate ratio (D'Souza et al., 1990). T o inactivate the enzyme after digestion, samples were heated at 70 "C for 10 min after the addition of onefourth volume of sample buffer containing 5% (v/v) P-mercaptoethanol. Proteolyzed samples were electrophoresed through an 8 M urea, 4% acrylamide stacking gel then into a 8 M urea, 12.5% acrylamide gel (13 X 18 cm) to optimize separation between oligopeptides (Swank and Munkres, 1971), electrophoretically transferred (50 mA) to Im-mobilon (Millipore Corp.), subjected to autoradiography to detect radioactive fragments of [3sS]ATP-a-S-labeled GPIIb which were then excised from Immobilon and examined by gas-phase sequencing for NH2-terrninal analysis (Moos et al., 1988;Tempst and Riviere, 1989) using an Applied Biosystems 477A pulsed liquid-phase sequencer and on-line model 120A phenylthiohydantoin analyzer. The initial yield for the 18-kDa fragment was 11.2 pmol.

RESULTS
Binding Studies-In our previous studies with the fixed platelet system, we used centrifugation through silicone oil (Jefferson et al., 1988) or direct centrifugation (Agrawal et al., 1989) to separate platelets with their bound, radiolabeled nucleotide from unbound ligand but B/T ratios were 0.2-0.3. Since, in the present work, we have maintained a B/T ratio -0.1 we have repeated our studies on the binding of [3H]ADP to platelets and on the ability of various nucleotides to compete in this binding using both direct centrifugation and a semiautomated filtration system to separate bound from unbound ligands. Competition binding isotherms are shown in Fig. 1. Under both sets of conditions, essentially identical binding parameters were obtained corresponding to two-site  (Table I): ADP bound to high (Kdl, 30 nM) and low (Kdp, -3 p M ) affinity sites while ATP-a-S bound with about 10-fold greater affinity to both high (&I, 3 nM) and low (Kd2, -0.2 p~) affinity sites. There was also good agreement between the two techniques in Ki values obtained from heterologous competition binding studies measuring the ability of ATP-a-S to compete with ADP and vice versa.
There was also good agreement in the total number of high affinity binding sites/platelet (-25,000) from both the heterologous and homologous competition binding studies using both the centrifugation and filtration techniques. There was, however, considerable variation seen in the number of low affinity sites/platelet as reflected both in the standard errors within one series of experiments and the different values obtained in different experiments. These differences arise from wide biological variations in the number of low affinity sites between different donors.
Several experiments were carried out to determine whether the various manipulations associated with the photoactivation experiments (for example, exposure to 4 "C and irradiation at 254 nm) altered the binding of [3H]ADP to platelets. These experiments were done by subjecting platelets to these manipulations and then fixing them before carrying out binding assays. No significant differences were seen in binding parameters using platelets that had been processed in the presence or absence of PGEl prior to fixation, platelets that had been activated with 1 nM human a-thrombin for 5 min at 37 "C without stirring in the absence of PGE1, or using platelets that had been irradiated under conditions of the photoaffinity labeling experiments prior to or after fixation (data not shown).
Inhibition of Activation-We have confirmed (Cusack and Hourani, 1982a) the inhibitory effects of ATP-a-S on ADPinduced platelet aggregation with an approximate ECso of 2 p~ (Fig. 2 A ) . In the absence of added fibrinogen, ADP induced only shape change which could also be inhibited in a dosedependent fashion by ATP-a-S (Fig. 2B). In the presence of 50 nM PG12, basal cAMP levels increased from 3.1 f 0.6 pmol/ 2 X 10' platelets to 97 f 26 pmol/2 X 10' platelets (mean k S.D., n = 3 assayed in duplicate or triplicate). ADP (2.5 p~) reduced the PGIp-stimulated cAMP levels by 77 f 4% to 24 k 18 pmol/2 X lo8 platelets. When ADP and ATP-a-S were added simultaneously to PGIp-stimulated platelet suspensions, only 54 f 17% and 18 f 6% reductions in stimulated cAMP levels were observed using 2.5 and 5.0 p~ ATP-a-S, respectively, demonstrating that ATP-a-S antagonized the ADP-dependent inhibition of the PG12-stimulated elevation of CAMP.
Photoaffinity Labeling-By using isolated platelet membranes (Barber andJamieson, 1970, Harmon et al., 1991), [35S]ATP-a-S photolabeled three components of 120,108, and 39 kDa together with several additional minor bands (data not shown). In contrast, photoaffinity labeling of intact platelets with [35S]ATP-a-S yielded a single band of radioactivity which had a molecular mass of 120 kDa (red)/135 kDa (unred) (Fig. 3, lunes 1 and 3 ) . Photoincorporation of [35S]ATP-a-S into this component was completely blocked in the presence of 850 p M ADP, ATP, or ATP-a-S (Fig. 3, lane 2 ) taking the precautions described under "Materials and Methods" to ensure that there was equal ultraviolet illumination in both the control sample and in the sample containing the competing nucleotide. Photoincorporation was not affected by the presence or absence of PGEl (1 pg/ml), Cap+ (1 mM), Mg2' (1 mM), or 0.35% bovine serum albumin but was decreased 60% if irradiations were carried out at 22 "C as compared with those a t 4 "C, and was reduced approximately 90% in the tivity was obtained in Triton X-lOf)-soluhle form and was shown to be identical with the 120-kDa component obtained in the absence of EDTA. Approximately 0.02"; of total atldetl [.'"SJATP-n-S was photoincorporated into Triton-soluble GPIIb as determined by immunoprecipitation with AI'-2 and an approximately equal amount was hound to CIPIIb in the Triton X-100-insoluble fraction. In approximately "Or; of the photoaffinity labeling experiments, a faint band o f 108 kI)a was occasionally detectable.
Characterization of Gfllhtr-The 120-kDa component was identified as GPIIbn by two-dimensional electrophoresis (Fig.  4), isoelectric focusing (PI 4.5-5.5) (Fig. 5 ) , and crossed immunoelectrophoresis (Fig.  6 ) . T h e [ "S]ATI'-tr-S-Iaheled GPIIb/IIIa complex was immunopurified on a column of AI'-2-Sepharose (Fig. 7 ) , and digested with trypsin (120) yielding a single radiolabeled band (-18 kDa) as well as three silver staining bands which were also found in tc-psin autolysates in the absence of GPIIb (Fig. 8). Cas-phase sequencing of two independently photolabeled and digested samples of the labeled 18-kDa peptide were the 11 -mer \r'AQA(;FFSVVS and the homologous 16-mer YAEAGFSSVVT(K/Q)AGEL. This sequence is homologous to the sequence YCEAGFSSVVTQA-GEL beginning at Tyr-198 as determined from the cDNA sequence for GPIIb#, in human erythroleukemia (HEL) cells (Poncz et al., 1987) and is preceded by arginine as expected for a trypsin-generated peptide. Based on a molecular mass GPIIh/llla isolated from [ "'S]ATI'-n-S-laheled platelets was proteolyzed for 2 h at 22 "C with a 20:l (w/w) ratio of laheled Gl'llh/llIa to TPCK-treated trypsin. Proteolvsis mixtures were separated hy X M urea, 12.57 acrvlamide gel. transferred to Immohilon. then subjected to autoradiography. The radiolaheled hand was excised and sequenced directly. Pond A, silver-stained gel. /'and N. autorntliography of Immohilon strip. The o w n orrows indicate silver-stained components derived from trypsin in the ahsence of (;PlIh/llIa. of 18 kDa the peptide would he expected to extend to one of the several Lys or Arg residues occurring in the region of amino acids 350-360 in the GPIIh,. protein sequence. The identified amino-terminal sequence showed no homology to sequences in GPIIIa (Fitzgerald et al., 19871, GPIIh,, (Poncz et al., 1987), trypsin, or platelet actin (Intelligenetics Suite, Release 5.37, 1990. In these experiments, undigested ["SI ATP-n-S-labeled GPIIh,. was included as a positive control and gave the expected NH2-terminal sequence of LNLDP-VQLXF (Poncz et al., 1987).

DISCUSSION
This report demonstrates that GPIIh,. contains a nucleotide-binding site which meets many of the requirements expected for a platelet ADP receptor. This binding site was recognized by direct photoaffinity labeling with ['"%]ATP-n-S, and this laheling was competahle hy the agonist ADP, and the antagonists ATP-a-S and ATP hut not hy adenosine which does not antagonize ADP-induced platelet activation. This pattern of competition is consistent with our hinding studies using the fixed platelet system. ATP-n-S competed at all sites accessible to ADP and, conversely, ADP competed at all sites accessihle to ATP-n-S. This implies that the photolabeling of GPIIh,, seen with [:'%]ATP-n-S represents a hinding site that is also completely accessible to ADP and that sites that are not accessihle to ADP would not photolabel with [:"'S]ATP-n-S. Our failure to detect similar photolaheling in experiments using ["]ADP and [''C]AI)P prohahlv reflects poor incorporation due to the fact that its hinding affinity (& 30 nM) is much weaker than that of ATP-n-S (K,, 3 nM), that it is rapidly metabolized to even more weakly bound products and that the availahle laheled forms of ADP are of much lower specific radioactivity than that available for [""SIATP-n-S.
The 18-kDa ATP-n-S-binding peptide isolated from human GPIIh,, has the amino-terminal sequence YAEAGFSSVVT (K/Q)AGEL beginning at Tyr-198. A corresponding sequence can be identified in the cDNA-deduced amino acid sequence from HEL cells (YCEAGFSSVVTQAGEL;Poncz et al., 1987;Heidenreich et al., 1990) which differs, however, from the platelet sequence in having Cys rather than Ala at position 199. In addition, we have consistently found an equimolar mixture of Lys and Gln at position 209 rather that Gln alone, as in the HEL cell clone: this may represent a polymorphic variation in platelet GPIIb', at this position. These variations between the primary (platelet) and deduced (HEL) amino acid sequences are unlikely to be due to a related integrin asubunit which complexes with GPIIIa in view of the fact that the NH2-terminal sequence for GPIIb, determined in control experiments is identical with that deduced from HEL cells.
The present steady-state binding studies were carried out under conditions such that the ratio (B/T) of bound/total ligand was -0.1 at the lowest concentration of ligand examined (Bylund and Yamamura, 1990). Under these conditions the affinity constants for ADP were 5-10-fold less than those previously determined at higher B/T ratios, while the number of high affinity sites was correspondingly fewer (Jefferson et al., 1988;Agrawal et al., 1989). These conclusions were confirmed by redetermining the binding parameters under the original conditions of B/T = 0.2-0.3 (data not shown).
The number of high affinity sites calculated for the steadystate binding of ADP and ATP-a-S in the present studies range from 18,500 f 2,500 to 32,000 & 4,800 and agree well with the number of GPIIb/IIIa complexes found from the binding of fibrinogen to stimulated platelets (14,000-38,000 sites/platelets) (Marguerie et al., 1980;Peerschke and Zucker, 1981;Plow and Marguerie, 1982;Plow et al., 1985) although higher numbers of binding sites have been reported for the binding of monoclonal antibodies ranging from about 40,000/ platelet for antibodies against GPIIb (McEver et al., 1980;Gulino et al., 1990) to about 50,00O/platelet for antibodies against the GPIIb/GPIIIa complex (Kunicki et al., 1981;Bennett et al., 1982;McEver et al., 1983).
The components labeled with [%]ATP-a-S using isolated membranes were not extensively examined because their multiplicity indicated a much lower specificity of labeling than that seen with intact platelets. On the basis of their electrophoretic mobility the bands of 120 and 108 kDa were taken t o be GPIIb and GPIIIa, respectively. The band of 39 kDa was identified as actin by immunoprecipitation using a monoclonal antibody against platelet actin. Even with intact platelets an additional faint spot corresponding to GPIIIa was identified by two-dimensional SDS-PAGE in a small number of the photoaffinity labeling experiments. These results show that the major site for photoincorporation of ["'SIATP-a-S is GPIIb,, but that deformation of the membrane during platelet isolation or during membrane preparation can lead to limited labeling of GPIIIa. This is consistent with the fact that labeling of GPIIb,, with ATP-a-S appears to occur at the distal end of the molecule where GPIIb and GPIIIa come into proximity to form the fibrinogen-binding site.
The nucleotide-binding site identified in GPIIb, is located within the 18-kDa domain extending from Tyr-198. This region also contains a fibrinogen-binding site localized to residues 294-314 (D'Souza et al., 1990). Furthermore, two putative Ca"-binding domains are located in the regions defined by residues 274-285 and 328-339 based on their homologies to calcium-binding sites in troponin C and calmodulin, respectively (Poncz et al., 1987;D'Souza et al., 1990).
Under the present conditions of irradiation a high degree of selectivity in photolabeling was obtained, and platelet activation did not occur during irradiation as shown by the following studies. 1) Irradiation prior to fixation did not affect the calculated Kd and number of binding sites in competition binding experiments using either ["S]ATP-a-S uersus ATPa-S or [3H]ADP uersus ADP. 2) Release of serotonin from irradiated platelets (3 f 2%) was not significantly greater than in non-irradiated platelets. 3) Binding of lZ5I-fibrinogen did not differ between irradiated and non-irradiated platelets. 4) There were no observable changes in the crossed immunoelectrophoresis or SDS-PAGE patterns following irradiation (data not shown). A very low incorporation of radiolabel was, however, observed (-0.04% including both Triton-soluble and Triton-insoluble fractions) probably reflecting the fact that the dose-rate of incident ultraviolet light was 5-10fold less than the intensity of radiation required for the photoactivation of platelets (Dickson et al., 1971;Doery et al., 1973) or that used in previous studies on direct photolabeling of isolated enzymes (Eriksson et al., 1982;Modak and Gillerman-Cox, 1982;Biswa and Kornberg, 1984;Biswa and Biswa, 1987).
In competitive photoaffinity labeling studies using compounds with strong ultraviolet absorption, such as nucleotides, it is mandatory to use an experimental configuration similar to that described under "Materials and Methods" in order to compensate for the decreased incident energy at the cell surface due to absorption by high concentrations of competing ligand. In many cases this practice has not been followed and the reported competition data are actually artifacts due to decreased incident energy.
ADP receptors on platelets are distinct from P,, purinergic receptors on other cells since, in the former case, ADP is an agonist and ATP is an antagonist while, in the latter case, both ADP and ATP are agonists and there are no known antagonists (Haslam and Cusack, 1981). Moreover, comparison of the amino acid sequences within the nucleotide-binding region of GPIIb,, defined by the 18-kDa fragment shows that it does not possess any homology to the highly conserved glycine-rich consensus sequence GXGXXGXV found in many nucleotide-binding proteins (Wierenga and Hol, 1983;Hanks et al., 1988), or the DAVGIAK sequence identified in yeast plasma membranes using azido-nucleotides (Davies et al., 1990).
We have confirmed earlier observations (Cusack and Hourani, 1982b) that ATP-a-S antagonizes ADP-induced platelet aggregation and have shown that it decreases ADP-induced inhibition of PG1,-stimulated adenylyl cyclase. We have extended these results to show that ATP-a-S also inhibits ADPinduced shape change and serotonin release. We have also demonstrated that ATP-a-S competes in the binding of ADP to platelets and that it photolabels a specific domain of GPIIb,, in intact platelets.
While the present studies have identified an ADP-binding site on GPIIb,,, much remains to be done to determine whether this binding site functions as a receptor, occupancy of which is necessary for ADP-induced platelet activation. In particular, binding and photolabeling studies with Glanzmann's platelets, which show normal ADP-induced Ca2+ influx (Powling and Hardisty, 1985), are required as well as the localization of the binding domain to a precise amino acid sequence and the demonstration that domain-specific antibodies or peptide segments inhibit ADP-induced platelet activation.