Arterial thrombosis may initiate after the rupture of an unstable atherosclerotic plaque, and it involves multiple platelet adhesion and agonist receptors (1) as well as activation of clotting with fibrin deposition (2- 3). Two adenosine 5′-diphosphate (ADP) receptors, P2Y₁ and P2Y₁₂, mediate platelet stimulation induced not only by exogenous ADP (4- 6), but also by shear forces or interactions with extracellular matrixes that cause ADP release from storage granules (7- 9). In particular, P2Y₁₂ concurs to the stability of platelet aggregates (10) and may exert a similar effect in developing arterial thrombi (11). In experimental studies, platelets that adhere and aggregate onto immobilized von Willebrand factor (VWF) exhibit cyclic oscillations in intracytoplasmic Ca⁺⁺ concentration ([Ca²⁺]_i) (12). These have been shown to depend on the concurrent function of P2Y₁₂ and glycoprotein (GP) IIb/IIIa (integrin α_IIbβ₃), and are linked to the recruitment and activation of flowing platelets into growing thrombi (12). Such findings, and the results obtained with a thrombosis model in P2Y₁₂-deficient mice (13), suggest that continuing stimulation of ADP receptors may be required to initiate as well as propagate thrombus growth on damaged vascular surfaces. Experiments with selective inhibitors have confirmed the importance of P2Y₁₂ in this regard (12), while P2Y₁ may be key for the initial activation of ADP-stimulated platelets (14) but have no role in thrombus propagation. In the present study, we have used a fibrillar type I collagen surface exposed to flowing blood and specific antagonists of the ADP receptors and GP IIb/IIIa to evaluate how aggregating platelets are incorporated irreversibly into a thrombus. Our findings may contribute to clarifying the mechanism of action of antithrombotic agents.

Venous blood was obtained from medication-free volunteers (6 men, 2 women; age 28 to 43 years) with their informed consent, and transferred into plastic tubes containing 1/10 volume of the thrombin inhibitor Argatroban (Mitsubishi Kagaku, Tokyo, Japan) to yield a final concentration of 100 μmol/l (15- 16), which does not decrease the plasma divalent cation concentration. Platelet-rich plasma (PRP) was separated by centrifugation at 100 g for 15 min and platelet-poor plasma (PPP) by further centrifugation at 800 g for 10 min. The platelet count in PRP was adjusted to 3 × 10⁵/μl.

AR-C69931MX (17) was from AstraZeneca (Loughborough, Leicestershire, United Kingdom). MRS2179 (18) was obtained from Dr. Savi (Sanofi-Synthelabo Recherche, Toulouse, France). Lanthanum chloride (19), acid insoluble fibrillar collagen type I from bovine Achilles tendon, mepacrine (quinacrine hydrochloride), acetyl salicylic acid, ADP, and epinephrine were from Sigma Chemical Co. (St. Louis, Missouri). Tirofiban (Aggrastate) was from Merck & Co. (Allentown, Pennsylvania). Fluo-3 acetoxymethyl ester (Fluo-3AM) was from Molecular Probe (Eugene, Oregon).

Platelets were rendered fluorescent by adding 10 μmol/l mepacrine (16,20) or 1 μg/ml of the fluorescein isothiocyanate (FITC)-labeled Fab fragment of the anti-GP IIb/IIIa monoclonal antibody, YM337 (Yamanouchi Pharmaceutical Co. Ltd., Tokyo, Japan). Thrombi of similar size were obtained in either case (not shown). A rectangular flow chamber with type I collagen fibrils coated on the glass bottom (15- 16,20) was assembled onto the stage of an inverted epifluorescence microscope (Leica, Germany). Blood was aspirated through the chamber with a syringe pump (Harvard Apparatus, Holliston, Massachusetts) at a constant flow rate to yield a wall shear rate of 1,500 s⁻¹. Images were digitized on-line with a color CCD video camera (L-600, Leica, Germany). Thrombus growth was evaluated in two dimensions by measuring the surface area covered by platelets (16) and in three dimensions by confocal microscopy (Figure 1) as previously reported (15). The effect of inhibitors of platelet function on the stability of platelet thrombi was tested in two-stage experiments, consisting in the perfusion of untreated blood for 6 min followed by blood containing or not a test substance for an additional 6 min.

Platelets in PRP were incubated for 30 min at 37°C with fluo-3AM (8 μmol/l), then mixed with erythrocytes separated from the same blood and washed three times by centrifugation and resuspension in a buffer composed of 10 mmol/l HEPES, 140 mmol/l NaCl, pH 7.4 (HEPES buffer). The washed cells were resuspended in homologous PPP containing Argatroban (100 μmol/l) at a 40% hematocrit. Unlike previously suggested (21) but in agreement with recent reports (22), we did not use probenecid to prevent the leakage of fluo-3 because of its effects on platelet function (23). The [Ca²⁺]_i of 10 randomly selected platelets incorporated at different positions within a thrombus was measured using confocal microscopy. Variations in the fluorescence intensity of fluo-3AM were converted into [Ca²⁺]_i using the equation: $[Ca2+]i=Kd(F−Fmin)/(Fmax−F)$ where Kd (495 nmol/l) is the dissociation constant of fluo-3AM for the interaction with Ca²⁺ (21); F is the measured fluorescent intensity of single platelets; Fmax is the fluorescence intensity of single platelets treated with the Ca²⁺ ionophore A23187 (10 μmol/l; Sigma) in the presence of 2 mmol/l Ca⁺⁺; and Fmin is the fluorescent intensity of unstimulated single platelets.

The platelet binding of FITC-conjugated PAC-1, a monoclonal antibody that selectively interacts with activated GP IIb/IIIa, was measured by flow cytometry (FACScan, Becton-Dickinson, San Jose, California). Platelets in PRP were activated with the combination of ADP and epinephrine (25 μmol/l each) or with the thrombin receptor activation peptide (1 mmol/l). Then, FITC-conjugated PAC-1 was added at a final concentration of 2.77 μg/ml, followed by HEPES buffer containing or not AR-C69931MX (100 nmol/l), MRS2179 (100 μmol/l), or lanthanum chloride (1 mmol/l). PAC-1 binding was measured 5, 10, 30, 45, and 60 min after addition of the last solution. All these experiments were performed under static conditions. The median fluorescence of 10,000 single platelets was calculated using the CellQuest software (Becton Dickinson Biosciences).

All numerical data are expressed as mean values ± SD unless otherwise specified. The effect of various concentrations of AR-C69931MX and MRS2179 on the surface coverage by platelets was tested by one-way analysis of variance. Differences between two groups of data were compared by Newman-Keuls test. A p value of <0.05 was considered to denote statistical significance.

In agreement with previous results (7- 8), ADP receptor antagonists inhibited thrombus growth on type I collagen fibrils exposed to blood flowing with a wall shear rate of 1,500 s⁻¹. Platelet surface coverage decreased from 38.0 ± 6.6% to 13.6 ± 3.9% after blocking P2Y₁₂ with 100 nmol/l AR-C69931MX, and to 19.4 ± 5.4% after blocking P2Y₁ with 100 μmol/l MRS2179 (p < 0.01). Both antagonists also inhibited thrombus volume (Figure 2). Untreated blood perfused for 6 min formed multilayered thrombi with a height of 13.2 ± 2.3 μm (n = 8), which was reduced to a single layer of platelets with a height of 3.2 ± 1.1 μm by 100 nmol/l AR-C69931MX (n = 8). With blood containing 100 μmol/l MRS2974, the thrombus height was 6.1 ± 3.5 μm (n = 8), less than with untreated blood (p < 0.01) but more than with the P2Y₁₂ antagonist (p < 0.01).

Blood perfused over collagen type I fibrils for 6 min at the wall shear rate of 1,500 s⁻¹ formed platelet thrombi with a height of 13.5 ± 3.3 μm (n = 8; Figure 3), increasing to 16.3 ± 4.5 μm after an additional 6 min. In contrast, when blood perfused in the second 6-min period contained 100 nmol/l AR-C69931MX, thrombus size progressively decreased until a single layer of adherent platelets with a height of 3.3 ± 1.3 μm remained. A similar effect, resulting in a 3.4 ± 1.2 μm high single layer of adherent platelets, was observed with the Ca²⁺ channel blocker, lanthanum chloride (1 mmol/l), whereas 100 μmol/l MRS2974 reduced thrombus height to 5.4 ± 2.2 μm. The latter value was less than with untreated blood but more than with the anti-P2Y₁₂ antagonist (n = 8; p < 0.01 for both comparisons; Figure 3). Addition of 0.5 μmol/l tirofiban, a GP II/IIIa antagonist (15), to the blood used in the second perfusion also reduced thrombi to a single layer of adherent platelets with a height of 3.1 ± 1.1 μm (n = 8; Figure 4), whereas 100 μmol/l aspirin allowed continued growth to a height of 15.8 ± 5.5 μm (n = 8; Figure 4).

Platelets within thrombi exhibited cyclic [Ca²⁺]_i oscillations (Figure 5). The lowest average [Ca²⁺]_i was 212 ± 38 (SEM) nmol/l and the highest 924 ± 458 nmol/l, with a cycle length from peak to peak of 4.2 ± 2.8 s (Figure 5). These [Ca²⁺]_i variations appeared to depend on specific ion channels because they were blocked by the Ca²⁺ channel antagonist lanthanum chloride (not shown). Addition of the P2Y₁₂ antagonist to the perfused blood rapidly decreased the platelet [Ca²⁺]_i within formed thrombi; after 2 min the value ranged between 182 ± 22 nmol/l and 244 ± 96 nmol/l without detectable cycles in most platelets (Figure 5). After addition of the P2Y₁ antagonist, the low and high [Ca²⁺]_i values ranged between 192 ± 34 nmol/l and 558 ± 211 nmol/l, respectively, and some but not all platelets showed measurable cycles of 5.8 ± 2.5 s (Figure 5). Neither tirofiban, in spite of the pronounced effect on platelet thrombus size, nor aspirin had any influence on the cyclic [Ca²⁺]_i of aggregated platelets (not shown).

Fluorescein isothiocyanate-conjugated PAC-1 binding to platelets, measured as the median fluorescence intensity of 10,000 platelets, increased when platelets were activated by the combination of ADP and epinephrine or the thrombin receptor activating peptide (the results were similar and only the former are shown). Bound PAC-1 slowly but significantly decreased in time when no additional exogenous agonist was added after the initial activation (Figure 6). The reduction was more marked after adding the P2Y₁₂ inhibitor (Figure 6), suggesting that continuous stimulation is necessary to maintain the active state of GP IIb/IIIa. A similar effect was observed with the P2Y₁ inhibitor or the Ca²⁺ channel blocker (Figure 7), but not with aspirin (data not shown).

Our findings show that platelet thrombi growing on collagen may disperse several minutes after the initial aggregation when exposed to blood containing an ADP receptor antagonist or a putative Ca²⁺ channel blocker or a GP IIb/IIIa inhibitor. Thrombus stability, therefore, may depend on sustained ligand binding to activated GP IIb/IIIa, which in turn may be mediated by [Ca²⁺]_i elevations induced by the continuous stimulation of specific signaling pathways (Figure 8). Our findings add to the concept of intercellular calcium communication, suggesting that it operates bidirectionally from platelets at the growing edge of a thrombus not only to activate newly recruited platelets (12) but also to ensure stability throughout the entire aggregate. Such a conclusion is based on the observation that perfusion of blood containing ADP receptor antagonists reduced the cyclic [Ca²⁺]_i increases within platelet thrombi even several minutes after the initial adhesive contacts had been established. Our experiments, however, could not define whether platelets at the center of thrombi differed in their Ca²⁺ responses from those at the outer edges. It has been proposed that cooperation between P2Y₁₂-mediated activation and GP IIb/IIIa engagement by ligand promotes the activation of newly recruited platelets in a thrombus (12). We observed, however, that blocking GP IIb/IIIa on the perfused platelets had no effect on the [Ca²⁺]_i oscillations within the thrombus even though it caused progressive disaggregation. Thus, the effect of GP IIb/IIIa inhibition on Ca²⁺ signals may reflect the need of maintaining newly recruited platelets in contact with adherent ones until activation is induced. In the case of platelets already aggregated into a thrombus, the replacement of an adhesive ligand by a GP IIb/IIIa antagonist may result in the detachment of platelets at the edge without effects on the activation of those still in the thrombus.

We propose that cyclic [Ca²⁺]_i increases within platelet thrombi reflect a mechanism that maintains GP IIb/IIIa activation and, thus, thrombus stability. The experiments with lanthanum, a putative Ca²⁺ channel blocker, support such a concept, but in this study we could not identify the ion channel involved or the signaling pathway regulating its function. Lanthanum, therefore, could have other effects, such as altering the ion binding sites of adhesive receptors and/or ligands (24) resulting in thrombus dissolution independently of Ca²⁺ channel blocking. Moreover, the cyclic [Ca²⁺]_i increases of aggregated platelets within thrombi could be a consequence, not the cause of, maintaining GP IIb/IIIa activation. In the alternative, the observation that [Ca²⁺]_i oscillations are linked to ADP receptor function may suggest an involvement of store-dependent Ca²⁺ entry (25- 27) mediated by P2Y₁₂ simulation. From a technical viewpoint, changes in platelet volume could have affected the results obtained with fluo-3AM, a single-wave Ca²⁺ indicator, thus influencing our conclusions. It should be noted that we have attempted to minimize such possible effects by using ultra-fast confocal microscopy, and no cyclic platelet volume changes have been reported during thrombus growth. In the end, sustained platelet activation necessary for thrombus stability may involve various signaling molecules, including ephrin/eph kinases (28), regulated by platelet-platelet contacts at the upper edge of the growing thrombus.

Although there is agreement that P2Y₁₂ is involved in the calcium signaling that sustains activation, conclusions on the role of P2Y₁ are not univocal, perhaps because previous studies were focused on the activation of newly recruited platelets (12) while we analyzed the activation of platelets already incorporated within thrombi. The two processes may be distinct, or the discrepant results may be caused by methodological differences. For example, continuous rather than periodic monitoring of [Ca²⁺]_i cycles may better reveal the relatively weak effect of blocking P2Y₁. Moreover, others have reported that only the combined blockade of P2Y₁ and P2Y₁₂ can inhibit platelet thrombus formation on the surface of collagen (7), whereas we have shown previously (8) and confirm here that inhibition of either receptor is almost equally effective in doing so. This discrepancy may result from differences in experimental observation times, because we measured thrombus volume after several minutes when the stability may be more influenced by interplatelet communication and less by the effects of the platelet-collagen interactions at the base of the thrombus.

Perhaps the main limitation of an ex vivo blood perfusion model is the need to use an anticoagulant to preserve blood fluidity, which in turn prevents the generation of thrombin and, consequently, of fibrin, both likely to have a role in providing thrombus stability (3,29). Nonetheless, ADP-induced signaling pathways may contribute to the overall effects of other agonists, including thrombin, that lead to nucleotide release from storage granules after primary platelet stimulation. Moreover, the mechanism highlighted here may provide initial stability to aggregated platelets exposed to flowing blood and, through the procoagulant function of activated platelets (30), contribute to fibrin formation. Our results, therefore, provide new insights into the mechanism of action of antiplatelet agents used in clinical practice and currently targeted against the P2Y₁₂ receptor. In situations such as acute coronary syndromes or coronary interventions with elevated thrombotic risk, P2Y₁₂ inhibitors may be used in combination with anticoagulants such as heparin, suggesting that an experimental model in which thrombin activity and fibrin formation are inhibited may be representative of at least some clinical conditions.

Our studies, although directly relevant only for thrombus formation on collagen fibrils under artificial blood flow conditions, suggest a possible role for distinct ADP receptors and their operating calcium signals in sustaining the long-term GP IIb/IIIa activation required to maintain platelet aggregate stability before the occurrence of fibrin generation. Such a mechanism may provide a more comprehensive understanding of the pharmacological effects of anti-P2Y₁₂ drugs, and suggest novel strategies to achieve the dispersion of platelet thrombi after they are formed.

The authors gratefully acknowledge the contribution of the staff of Tokai University Educational and Research Support Center.

For the supplemental videos, please see the online version of this article.

Supplemental movie for the three left panels of (Figure 2) (Control).

Supplemental movie for the three middle panels of (Figure 2) (ARC: AR-C69931MXl).

Supplemental movie for the three right panels of (Figure 2) (MRS: MRS2179l).

Supplemental movie for the left panel of (Figure 3)A (before starting second perfusion).

Supplemental movie for the right panel of (Figure 3)A (after finishing the second perfusion of control blood).

Supplemental movie for the left panel of (Figure 3)B (before starting second perfusion).

Supplemental movie for the right panel of (Figure 3)B (after finishing the second perfusion of blood containing AR-C69931MX).

Supplemental movie corresponding to the upper panel of (Figure 4)A.

Supplemental movie corresponding to the lower panel of (Figure 4)A.