Association Between Ligand-Induced Conformational Changes of Integrin _IIbβ₃ and _IIbβ₃-Mediated Intracellular Ca²⁺ Signaling

INTEGRINS ARE heterodimeric glycoproteins consisting of α and β subunits that are a family of cell surface molecules that mediate cellular attachment to the extracellular matrix and cell cohesion.1 Integrins are involved in many physiological functions such as development, immune response, tissue repair, and hemostasis, and they are now recognized as important signaling molecules that can mediate the bidirectional transfer of information from the outside to the inside of the cell and also from the inside to the outside of the cell.2-4 

α_IIbβ₃ (GPIIb-IIIa), a prototypic integrin, is expressed exclusively on platelets and megakaryocytes and acts as a receptor for fibrinogen, von Willebrand factor, vitronectin, and fibronectin. The interaction of this integrin with fibrinogen or von Willebrand factor appears to be mediated, at least in part, via an Arg-Gly-Asp (RGD) sequence, and α_IIbβ₃ is essential for platelet aggregation that leads to hemostatic plug formation and pathological thrombus formation.5 Recent studies have demonstrated that conformations of α_IIbβ₃ are dynamically regulated and that the following steps are necessary for maximal platelet aggregation6: (1) agonist-induced α_IIbβ₃ activation via a process termed inside-out signaling, (2) ligand (fibrinogen) binding, and (3) postreceptor occupancy events via a process termed outside-in signaling that involves change in intracellular Ca²⁺ level and pH, tyrosine phosphorylation, and cytoskeletal reorganization.7,8 Binding of fibrinogen, a multivalent ligand, to α_IIbβ₃ leads to expression of neo-epitopes on α_IIbβ₃, termed ligand-induced binding sites (LIBS), as well as clustering of α_IIbβ₃. LIBS expression has been well documented on both α_IIb and β₃ subunits and might explain the capacity of α_IIbβ₃ to initiate outside-in signaling.9-11 However, the association between LIBS expression and integrin outside-in signaling remains elusive.

In this report, using six unrelated α_IIbβ₃-specific peptidomimetic compounds as a monovalent ligand instead of the multivalent ligand, fibrinogen, we attempted to determine whether LIBS expression on α_IIbβ₃ may be associated with outside-in signaling such as α_IIbβ₃-mediated intracellular Ca²⁺ changes. Using a panel of monoclonal antibodies (MoAbs) against LIBS, we showed that α_IIbβ₃-specific peptidomimetic antagonists can be divided into two groups. In group I, antagonists can induce LIBS on both α_IIb and β₃ subunits. In group II, antagonists can induce LIBS on the α_IIb subunit, but not on the β₃ subunit. Interestingly, only group I antagonists dose-dependently augmented the [Ca²⁺]_i increase in thrombin-stimulated platelets in an aggregation-independent manner. Our data suggest that β₃ LIBS expression is associated with α_IIbβ₃-mediated intracellular Ca²⁺ changes.

OP-G2 is an MoAb specific for α_IIbβ₃-complex. OP-G2 behaves like a macromolecular RGD-containing ligand and has been shown to bind at or near the ligand recognition site.12,13 AP5 (anti-β₃ amino-terminus, residues 1-6) was kindly provided by Dr Thomas J. Kunicki (Scripps Research Institute, La Jolla, CA). PMI-1 (anti-α_IIb heavy chain, residues 844-859), anti-LIBS1 (anti-β₃), and anti-LIBS2 (anti-β₃, residues 602-690) were generously donated by Dr Mark H. Ginsberg (Scripps Research Institute).14-16 AP5, anti-LIBS1, and anti-LIBS2 recognize LIBS on the β₃subunit, and PMI-1 recognizes LIBS on the α_IIb subunit. Monoclonal IgG was purified from ascites fluid by affinity chromatography on Protein A Sepharose CL-4B (Pharmacia, Piscataway, NJ).

All α_IIbβ₃-specific peptidomimetic compounds were synthesized in Fujisawa Pharmaceutical Co (Osaka, Japan) and the chemical structures of these compounds are shown in Fig 1. FK633, MK383, Ro44-9883 and SC54701 have been reported previously.17-20 FR169824 and FR184764 were newly designed and synthesized by Fujisawa based on the chemical structure described previously.21 The high pressure liquid chromatography (HPLC) profile of each compound showed a single sharp peak with the expected molecular weight. In addition, nuclear magnetic resonance (NMR) studies of FK633 gave only a normal NMR spectrum (data not shown). These data indicate that each compound is monomeric in solution. The purity of each compound was more than 98%.

BM13505, a thromboxane A₂ (TXA₂) receptor antagonist, was also synthesized in Fujisawa.22 

Fibrinogen (Kabi, Stockholm, Sweden) was labeled basically according to the method of Faraday et al.23 Briefly, fibrinogen at 17 mg/mL was incubated with FITC (20 μg/mg fibrinogen; Sigma, St Louis, MO) for 3 hours at 22°C. Excess FITC was removed by exhaustive dialysis against modified Tyrode-HEPES buffer (137 mmol/L NaCl, 2 mmol/L KCl, 12 mmol/L NaHCO₃, 0.3 mmol/L NaH₂PO₄, and 5 mmol/L HEPES, pH 7.4). FITC-fibrinogen was stored at 4°C and used within 1 week of preparation.

Platelet-rich plasma (PRP) was obtained by differential centrifugation of the acid-citrate-dextrose–anticoagulated blood as described previously.24 Prostaglandin E₁(PGE₁; Sigma) was added to a final concentration of 20 ng/mL. The PRP was then centrifuged at 750g for 10 minutes to sediment platelets. After three washes with Ringer’s citrate-dextrose containing PGE₁, pH 6.5, the platelet pellet was resuspended in an appropriate buffer. For loading of platelets with fura-2, the diluted PRP with modified Tyrode-HEPES buffer containing 1 mmol/L MgCl₂ (1:1 dilution) was incubated with 3.3 μmol/L acetoxymethyl esters (AM) of fura-2 (Dojin Chemical Co, Ltd., Kumamoto, Japan) for 15 minutes at 37°C in the dark, and then platelets were washed twice with Ringer’s citrate-dextrose containing PGE₁, pH 6.5, as described above.

Flow cytometry was performed as described previously, with slight modifications.14 Five-microliter aliquots of the washed platelets (1 × 10⁹/mL) suspended in 20 mmol/L HEPES, 137 mmol/L NaCl, 2 mmol/L CaCl₂, pH 7.4, plus 1% bovine serum albumin (BSA), and 20 ng/mL PGE₁ (test buffer) were added to tubes containing serial concentrations of synthetic antagonists in 40 μL test buffer. Five microliters of each biotinylated MoAb examined was then added to the mixture to make a final concentration of 5 μg/mL and incubated for 30 minutes at room temperature. The platelet suspensions were then incubated with a 1:320 final dilution of FITC-conjugated streptavidin (Sigma) for an additional 30 minutes without an intermittent washing step. The platelets were then diluted to 0.5 mL Tris-buffered saline (TBS; pH 7.4) and analyzed in a flow cytometer (FACScan; Becton Dickinson, Mountain View, CA).

For the analysis of FITC-fibrinogen binding to platelets, 40-μL aliquots of washed platelets (1.2 × 10⁸/mL) suspended in modified Tyrode-HEPES buffer plus 1 mmol/L CaCl₂ and 1% BSA were added to tubes containing 5 μL of serial concentrations of synthetic antagonists and 5 μL of 3 mg/mL FITC-fibrinogen. After adding 20 μmol/L ADP, the mixtures were incubated for 15 minutes at 37°C without stirring. The platelet suspensions were then diluted to 0.5 mL with modified Tyrode-HEPES buffer containing 1 mmol/L CaCl₂ and analyzed in the flow cytometer.

To examine the inhibitory effects of antagonists on ADP-induced platelet aggregation, a model PAP-4 NKK platelet aggregation tracer (Nikou Bioscience Inc, Tokyo, Japan) was used as described previously.24 

Change of intracellular free calcium concentrations ([Ca²⁺]_i ) and platelet aggregation were simultaneously measured using a Calcium Ion Analyzer FS-100 (Kowa, Osaka, Japan) that detects intensities of Fura-2 fluorescence at 380 nm (F380) and 320 nm (F320). Fura-2–loaded platelets were preincubated with each synthetic antagonist for 1 minute and then stimulated with 0.03 U/mL thrombin at 37°C with a stirring rate of 1,000/min. Changes in the fluorescence and the light transmittance were recorded. The [Ca²⁺]_i was automatically calculated from the ratio of F380 and F320 by a FS-100 computer program connected with a calcium-ion analyzer.

Platelets that have been analyzed for [Ca²⁺]_iwere incubated with 10 mmol/L EGTA and 100 μmol/L indomethacin at 4°C to stop the reaction and then pelleted by centrifugation at 1,600g for 5 minutes. TXB₂ in the supernatant was measured with Thromboxane B₂ [¹²⁵I] RIA Kit (Dupont, NEN Research Products).

FK633, MK383, Ro44-9883, and SC54701 have been characterized as a compound that selectively inhibits α_IIbβ₃.17-20 FR169824 and FR184764 were newly synthesized at Fujisawa Pharmaceutical Co and inhibited fibrinogen binding to activated α_IIbβ₃ (Table1). None of these antagonists inhibited the adhesion of human umbilical vein endothelial cells (HUVEC) to vitronectin-coated plates or the adhesion of Chinese hamster ovary (CHO) cells stably expressing recombinant human α_vβ₃ to fibrinogen-coated plates even at 10 μmol/L, suggesting that these compounds do not inhibit α_vβ₃ or other integrins (data not shown). Fifty percent inhibitory concentrations (IC₅₀) of these compounds for platelet aggregation induced by ADP and fibrinogen binding to ADP-stimulated platelets were summarized in Table 1. These antagonists were highly active against platelet aggregation (∼160- to 900-fold potency as compared with RGDW) and their IC₅₀values were in a similar range (19 to 110 nmol/L). As expected, these antagonists showed larger IC₅₀ values for platelet aggregation than for fibrinogen binding to activated α_IIbβ₃.

AP5 and anti-LIBS2 MoAb recognize residues 1-6 and residues 602-690 on the β₃ subunit, respectively.14,16 Anti-LIBS1 MoAb recognizes different regions from those for AP5 and anti-LIBS2.14 PMI-1 MoAb recognizes residues 844-859 on the α_IIb heavy chain.15 Using these MoAbs, we examined the effects of antagonists on LIBS expression on α_IIbβ₃. A typical set of results using AP5 and PMI-1 is shown in Fig 2. Ro44-9883 and MK383 had little effect on the induction of AP5 (Fig 2), anti-LIBS1, and anti-LIBS2 epitopes (not shown) even at 100 μmol/L, whereas FR184764, SC54701, FR169824, and FK633 markedly induced these LIBS on the β₃ subunit. However, all antagonists induced PMI-1 epitope on the α_IIb subunit. In this study, we designated antagonists inducing LIBS on both α_IIb and β₃ as group I and those not inducing LIBS on β₃ as group II. RGDW and fibrinogen γ-chain peptides [HHLGGAKQAGDV (H12)] at 1 mmol/L induced both AP5 and PMI-1 epitopes, although the effects of H12 on LIBS expression were weaker than RGDW, probably due to a low affinity of H12 for α_IIbβ₃.12 Thus, RGDW and H12 belong to group I. We then compared the extent of LIBS expression induced by fibrinogen bound to ADP-stimulated platelets with those induced by antagonists. Maximal LIBS expression was obtained at 4 μmol/L of fibrinogen. When LIBS expression induced by FK633 was taken as 100%, fibrinogen induced only 33.7% ± 15.0% and 6.1% ± 3.7% of AP5 and PMI-1 expression, respectively (n = 3).

Fifty percent effective doses (ED₅₀) for AP5, anti-LIBS1, anti-LIBS2, and PMI-1 expression are summarized in Table 2. As compared with IC₅₀s for fibrinogen binding, ED₅₀s of these antagonists for LIBS expression were much higher. ED₅₀s for the induction of each LIBS on the β₃ subunit were anti-LIBS1 < AP5 < anti-LIBS2, indicating that anti-LIBS1 epitope is most sensitive for ligand binding among these LIBS. The group I antagonists (Ro44-9883 and MK383) were more potent in the induction of PMI-1 epitope than group II antagonists (FR184764, SC54701, FR169824, and FK633) ([ED₅₀ for PMI-1 expression/IC₅₀ for fibrinogen binding] ratio; group I, 29.8 ± 19.1; group II, 1.6 ± 0.8; P < .05; Table 2). We also examined inhibitory effects of antagonists on the binding of a ligand-mimic MoAb, OP-G2, to platelets. Although OP-G2 recognizes at or near the ligand recognition site, OP-G2, like small RGD-containing peptides, binds to nonactivated platelets.12 Interestingly, group II antagonists were much more potent in the inhibition of OP-G2 binding than group I antagonists ([IC₅₀ for OP-G2 binding/IC₅₀ for fibrinogen binding] ratio; group I, 102.3 ± 32.2; group II, 6.7 ± 4.9;P < .01; Table 2). The apparent differences in the (ED₅₀ for PMI-1 expression/IC₅₀ for fibrinogen binding) ratio and the inhibitory effects on OP-G2 binding suggest that the binding sites of antagonists are distinct between the two groups.

The binding characteristics of group I (FR184764, SC54701, FR169824, and FK633) and group II antagonists (Ro44-9883 and MK383) were further examined in an inhibition assay. The binding of group I antagonists such as SC54701 was monitored by the binding of AP5 MoAb. As shown in Fig 3, the binding of SC54701 was markedly inhibited by all of the group II antagonists. Similarly, the binding of FR184764, FR169824, and FK633 was also markedly inhibited by all of group II antagonists (data not shown). These results indicate that the binding of a group I antagonist is inhibited by the binding of a group II antagonist.

Group I and group II antagonists induced LIBS on α_IIbβ₃ differently, especially on the β₃ subunit. Using these antagonists as a monovalent ligand, we examined whether the difference in LIBS expression of α_IIbβ₃ induced by antagonists might affect outside-in signaling via α_IIbβ₃. None of these antagonists affected the [Ca²⁺]_i in nonactivated platelets, even at a high concentration of 10 μmol/L, indicating that LIBS expression alone is not sufficient to cause this outside-in signaling (data not shown). As previously demonstrated by Yamaguchi et al,25 26 a low concentration of thrombin (0.03 U/mL) induces a two-peaked [Ca²⁺]_i increase (Fig 4). The latter peak has been shown to be dependent on both fibrinogen binding to α_IIbβ₃ and platelet aggregation. Each antagonist, irrespective of the group, abolished the latter [Ca²⁺]_i peak as well as platelet aggregation. However, when the concentrations of FK633 were increased up to 10 μmol/L to induce full expression of LIBS, another second [Ca²⁺]_i peak was induced even in the absence of platelet aggregation (Fig 4). All group I antagonists showed essentially the same effects on the [Ca²⁺]_ichange. In addition, 1 mmol/L RGDW peptide also induced the second [Ca²⁺]_i peak, indicating that this phenomenon is not specific for peptidomimetic antagonist (data not shown). In contrast, none of the group II antagonists showed such effects even at 10 μmol/L (Fig 4). When ADP or epinephrine was used as an agonist instead of thrombin, none of these antagonists had effects on the [Ca²⁺]_i change (data not shown).

Using platelets derived from a patient with Glanzmann thrombasthenia who has no detectable α_IIbβ₃,27we further examined whether the second [Ca²⁺]_i peak induced by group I antagonists may be specifically mediated by α_IIbβ₃ on platelets. Thrombin induced only first [Ca²⁺]_i peak in thrombasthenic platelets. Neither FK633 nor FR169824 induced the additional second [Ca²⁺]_i peak, even at 10 μmol/L (Fig 5). This patient possesses a molecular genetic defect in the α_IIb gene27 and expresses normal level of α_vβ₃ on platelets (data not shown). Therefore, the second [Ca²⁺]_i peak induced by group I antagonists is specifically mediated by α_IIbβ₃.

To elucidate the nature of the second peak that was induced by group I antagonists, the effects of aspirin and BM135052 (TXA₂receptor antagonist) were examined. Aspirin as well as BM13505 markedly inhibited the second [Ca²⁺]_i peak induced by FK633 (Fig 6), FR184764, SC54701, or FR169824 (data not shown). These data suggest that the second [Ca²⁺]_i peak is caused by the production of TXA₂. In addition, apyrase also abolished the second [Ca²⁺]_i peak induced by FK633, suggesting that endogenous ADP played some role in the induction of the second peak.

To confirm that a group I antagonist induces TXA₂production with the costimulation by thrombin, TXB₂, a major metabolite of TXA₂, was measured. As shown in Fig 7, TXB₂ was initially produced by thrombin stimulation. All antagonists at low concentrations markedly inhibited TXB₂ production dose-dependently, indicating that the greater part of the TXB₂ production under these conditions is dependent on both fibrinogen binding and platelet-aggregation. However, at higher concentrations, group I antagonists induced TXB₂ production in a concentration-dependent manner. In contrast to group I antagonists, the group II antagonists did not induce TXB₂ production even at high concentrations. In addition, the levels of TXB₂production in the absence of platelet aggregation correlated with the induction of AP5 epitope (compare Figs 2A and 7). These data confirm that the TXA₂ production, induced by a high concentration of group I antagonists, is responsible for the second [Ca²⁺]_i peak.

In the present study, we demonstrated that α_IIbβ₃ antagonists can be divided into two groups, group I and group II, according to the effects on LIBS expression on the α_IIb and the β₃ subunits. We designated antagonists inducing LIBS on the β₃ subunit as group I (FR184764, SC54701, FR169824, and FK633) and those not inducing as group II (Ro44 9883 and MK383). However, in contrast to the data reported by Steiner et al,28 29 using the PMI-1 MoAb as a probe, we demonstrated that all six antagonists can induce LIBS on the α_IIb subunit. A group II antagonist was apparently more potent in the induction of PMI-1 epitope than a group I antagonist. Using these antagonists as a monovalent ligand, we have readily demonstrated that only group I antagonists can induce α_IIbβ₃-mediated Ca²⁺ signaling in platelets stimulated with thrombin in an aggregation-independent manner. These data suggest that LIBS expression on the β₃subunit is a prerequisite for outside-in signaling through α_IIbβ₃.

Group I and group II antagonists induce distinct conformational changes on α_IIbβ₃. The ratio of ED₅₀for PMI-1 expression/IC₅₀ for fibrinogen binding clearly showed that the difference in the ability of PMI-1 expression between the two groups is not due to the difference in the affinities to α_IIbβ₃. In addition, group II antagonists are apparently more active against the binding of OP-G2 MoAb than group I antagonists. These data suggest that the binding sites of antagonists are distinct between the two groups. However, the binding of group I antagonists to α_IIbβ₃ that was monitored by AP5 binding was abolished by the binding of group II antagonists. These findings are consistent with the data reported by Diaz-González et al.30 Taken together, our data suggest that group I and group II antagonists interact with distinct but mutually exclusive sites on α_IIbβ₃. Although it has been demonstrated that RGD and H12 peptides interact with distinct but mutually exclusive sites,31 these peptides induced LIBS on both α_IIb and β₃ subunits. Therefore, the difference in the binding sites between group I and group II antagonists does not simply reflect the difference between RGD and H12 peptides.

Miyamoto et al32 demonstrated that, in fibroblasts, direct ligand occupancy by a monovalent ligand was not a sufficient signal for cytoskeletal protein organization. In contrast, integrin clustering without ligand occupancy induced intracellular accumulation of FAK and tensin, but not of other cytoskeletal proteins such as talin. Both ligand occupancy and integrin clustering were necessary for accumulation of talin, α-actinin, paxillin, vinculin, F-actin, and filamin.33 Although ligand occupancy would lead to LIBS expression on integrin receptors, how LIBS expression contributes to integrin outside-in signaling is not well understood. In platelets, fibrinogen binding to α_IIbβ₃ activated by the activating MoAb PT25-2 per se does not induce [Ca²⁺]_i increase, even in the presence of platelet aggregation.34 In contrast, a low concentration of thrombin (0.03 U/mL) induces the two-peaked [Ca²⁺]_i increase in platelets. The first [Ca²⁺]_i peak is generated by the thrombin receptor, whereas the latter [Ca²⁺]_i peak is not observed in thrombasthenic platelets and is dependent on both fibrinogen-binding to α_IIbβ₃ and platelet aggregation (this study and Yamaguchi et al26). Monovalent ligands such as RGDS peptide abolish the latter [Ca²⁺]_i peak. The latter peak can be also abolished by aspirin, BM13505 (TXA₂ receptor antagonist), or apyrase.35 Accordingly, the latter peak represents post-α_IIbβ₃ occupancy events by multivalent ligands such as fibrinogen. Both endogenous ADP release and platelet aggregation are needed for TXA₂ production via α_IIbβ₃, which is responsible for the latter peak. As expected, both group I and group II antagonists at low concentrations abolished the latter [Ca²⁺]_ipeak. However, group I antagonists at high concentrations induced the new second peak in thrombin-stimulated platelets, even in the absence of platelet aggregation, whereas group II did not induce it, despite LIBS expression on the α_IIb subunit. Group I antagonists did not induce the second peak in thrombasthenic platelets expressing normal level of α_vβ₃, even at 10 μmol/L, indicating that this signal is specifically mediated by α_IIbβ₃. The second [Ca²⁺]_i peak was dependent on the production of TXA₂ and was abolished by aspirin, BM13505, or apyrase. Accordingly, the pathway involved in the group I antagonist-induced second [Ca²⁺]_i peak is essentially the same as that for the fibrinogen-induced and aggregation-dependent [Ca²⁺]_i peak. In other words, these data suggest that, in thrombin-stimulated platelets, LIBS expression induced by group I antagonists is associated with the cyclooxgenese pathway possibly through activation of phospholipase A₂.36 Interestingly, in a canine coronary thrombolysis model, Murphy et al37 demonstrated that administration of a group I antagonist, Ro43-5054, increased the level of urinary 2,3-dinor-TXB₂, similar to controls, whereas a group II antagonist, Ro44-9883, markedly inhibited the increase. Our in vitro data presented here may explain their in vivo data.

In the presence of thrombin stimulation, the monovalent ligand to α_IIbβ₃ could induce the second [Ca²⁺]_i peak without platelet aggregation. Our data demonstrate that neither multivalency of the ligand nor platelet aggregation is essential to induce the α_IIbβ₃-dependent [Ca²⁺]_i increase. It is noteworthy that the extent of LIBS expression on the β₃ subunit, but not on the α_IIb subunit, correlated with the level of TXB₂ production. In contrast, the extent of AP5 and PMI-1 expression induced by fibrinogen was 33.7% ± 15.0% and 6.1% ± 3.7% (n = 3) of that induced by group I antagonists, respectively. Thus, group I antagonists can fully induce LIBS on α_IIbβ₃. These data suggest that group I antagonists-induced conformational change is needed for the second [Ca²⁺]_i peak as an outside-in signal via α_IIbβ₃.

There is increasing evidence that occupancy of one integrin can also suppress the functions of other integrins (trans-dominant inhibition).38-40 Recently, Diaz-González et al30 demonstrated that Ro43-5054, but not Ro44-9883, induces this trans-dominant inhibition. Similarly to our findings, they demonstrated that the inhibitory effect on the function of the target integrin α₅β₁ correlated with LIBS expression in the suppressive integrin α_IIbβ₃.

It has been well documented that the cytoplasmic domain of the β₃ subunit plays a critical role in α_IIbβ₃ outside-in signaling.41Truncation of the β₃ subunit cytoplasmic domain as well as a certain point mutation (S752P) abolished cell spreading mediated by α_IIbβ₃ and fibrin clot retraction, whereas truncation of the α_IIb subunit did not inhibit cell spreading.42,43 The trans-dominant inhibition of α_IIbβ₃ is also mediated by β₃cytoplasmic domain.30 As shown by three different MoAbs against β₃ LIBS here, β₃ extracellular domain dramatically changed its conformation by ligand binding. It is noteworthy that anti-LIBS2 recognizes the membrane proximal region of β₃. Therefore, it is likely that the conformational change detected by anti-LIBS MoAb may represent conformational changes of the β₃ cytoplasmic domain.

Our data presented here demonstrate that LIBS expression of the β₃ subunit could participate in outside-in signaling through this integrin during thrombogenesis and hemostasis. Antagonists for α_IIbβ₃ are likely to be the first anti-integrins to be widely used.44 From a therapeutic viewpoint, LIBS expression may facilitate to produce antibodies against the LIBS as a neo-epitope and cause subsequent immune thrombocytopenia.45 In addition, TXA₂ is one of platelets agonists and a potent constrictor of vascular smooth muscles.46 Although LIBS expression on both subunits and the stimulation of TXA₂ production by group I antagonists needs much higher concentrations than therapeutic range, we could not rule out the possibility that group I antagonists might have an adverse effect. Further study will be required to determine whether some adverse effects would be induced by group I antagonists in vivo experiment using therapeutic range, and better understanding of physiological roles of LIBS expression could contribute to the successful development of α_IIbβ₃antagonists.

The authors thank Dr Thomas J. Kunicki (Scripps Research Institute) for the MoAb AP5 and Dr Mark H. Ginsberg (Scripps Research Institute) for the MoAbs anti-LIBS1, anti-LIBS2, and PMI-1.