Adenylic Dinucleotides Produced by CD38 Are Negative Endogenous Modulators of Platelet Aggregation*

Diadenosine 5′,5‴-P1,P2-diphosphate (Ap2A) is one of the adenylic dinucleotides stored in platelet granules. Along with proaggregant ADP, it is released upon platelet activation and is known to stimulate myocyte proliferation. We have previously demonstrated synthesis of Ap2A and of two isomers thereof, called P18 and P24, from their high pressure liquid chromatography retention time, by the ADP-ribosyl cyclase CD38 in mammalian cells. Here we show that Ap2A and its isomers are present in resting human platelets and are released during thrombin-induced platelet activation. The three adenylic dinucleotides were identified by high pressure liquid chromatography through a comparison with the retention times and the absorption spectra of purified standards. Ap2A, P18, and P24 had no direct effect on platelet aggregation, but they inhibited platelet aggregation induced by physiological agonists (thrombin, ADP, and collagen), with mean IC50 values ranging between 5 and 15 μm. Moreover, the three dinucleotides did not modify the intracellular calcium concentration in resting platelets, whereas they significantly reduced the thrombin-induced intracellular calcium increase. Through binding to the purinergic receptor P2Y11, exogenously applied Ap2A, P18, and P24 increased the intracellular cAMP concentration and stimulated platelet production of nitric oxide, the most important endogenous antiaggregant. The presence of Ap2A, P18, and P24 in resting platelets and their release during thrombin-induced platelet activation at concentrations equal to or higher than the respective IC50 value on platelet aggregation suggest a role of these dinucleotides as endogenous negative modulators of aggregation.

Diadenosine 5,5ٟ-P1,P2-diphosphate (Ap2A) is one of the adenylic dinucleotides stored in platelet granules. Along with proaggregant ADP, it is released upon platelet activation and is known to stimulate myocyte proliferation. We have previously demonstrated synthesis of Ap2A and of two isomers thereof, called P18 and P24, from their high pressure liquid chromatography retention time, by the ADP-ribosyl cyclase CD38 in mammalian cells. Here we show that Ap2A and its isomers are present in resting human platelets and are released during thrombin-induced platelet activation. The three adenylic dinucleotides were identified by high pressure liquid chromatography through a comparison with the retention times and the absorption spectra of purified standards. Ap2A, P18, and P24 had no direct effect on platelet aggregation, but they inhibited platelet aggregation induced by physiological agonists (thrombin, ADP, and collagen), with mean IC 50 values ranging between 5 and 15 M. Moreover, the three dinucleotides did not modify the intracellular calcium concentration in resting platelets, whereas they significantly reduced the thrombin-induced intracellular calcium increase. Through binding to the purinergic receptor P2Y 11 , exogenously applied Ap2A, P18, and P24 increased the intracellular cAMP concentration and stimulated platelet production of nitric oxide, the most important endogenous antiaggregant. The presence of Ap2A, P18, and P24 in resting platelets and their release during thrombin-induced platelet activation at concentrations equal to or higher than the respective IC 50 value on platelet aggregation suggest a role of these dinucleotides as endogenous negative modulators of aggregation.
The dinucleoside diphosphates diadenosine 5Ј,5ٞ-P1,P2diphosphate (Ap2A), 2 adenosine guanosine diphosphate (Ap2G), and diguanosine diphosphate (Gp2G) represent a new class of growth-promoting extracellular mediators present in platelet secretory granules (1) and in cardiac myocytes (2), capable of stimulating cardiac myocyte proliferation (1) and believed to play a role in the control of cardiovascular tone (3). The intraplatelet concentration of each one of these dinucleotides has been estimated to be in the range between 30 and 100 M (1). It has also been shown that the concentration of Ap2A, Ap2G, and Gp2G in the supernatant of platelets stimulated with 0.05 units/ml thrombin is ϳ60% of the total intraplatelet amount of each dinucleoside diphosphate, suggesting that their primary function is extracellular (1). The enzyme responsible for their synthesis has not been identified, and the effect of these dinucleotides on platelet function has not yet been investigated.
We have recently demonstrated that ADP-ribosyl cyclases from Axinella polypoides, (Porifera, Demospongiae), Aplysia californica (Molluscs), and human CD38 can synthesize three adenylic dinucleotides from cyclic ADP-ribose and adenine (Ade): Ap2A and two isomers thereof, called P18 and P24, which are characterized by an unusual N-glycosidic bond between one adenine and the ribose (C1Ј-N1 in P18 and C1Ј-N3 in P24) (4). CD38 is present in human platelets (5,6), and activation with thrombin induces an increase of CD38 activity associated with the platelet cytoskeleton (7). These results prompted us to investigate the presence of P18 and P24 in human platelets; the effect of Ap2A, P18, and P24 on platelet aggregation; and possible synergistic or antagonistic effects of these adenylic dinucleotides with thrombin, ADP, or collagen, the most potent physiological platelet agonists.

EXPERIMENTAL PROCEDURES
Materials-If not otherwise indicated, all chemicals were from Sigma and were of the highest purity grade available.
Blood Collection and Platelet Preparation-Venous blood, freshly drawn from healthy volunteers, was collected in a 130 mM aqueous trisodium citrate anticoagulant solution (9:1). The donors declared not to have taken drugs known to interfere with platelet function during 2 weeks prior to blood collection. Washed platelets (WP) were prepared as described (8). Briefly, platelet-rich plasma, obtained by centrifugation of the whole blood at 100 ϫ g for 25 min, supplemented with 2 g/ml apyrase and 1 M PGE 1 , was centrifuged at 1000 ϫ g for 15 min. The platelet pellet was washed once with ACD solution (75 mM trisodium citrate, 42 mM citric acid, and 136 mM glucose, pH 4.8), centrifuged at 1000 ϫ g for 15 min, and then resuspended in Ca 2ϩ -free Hepes buffer (145 mM NaCl, 5 mM KCl, 1 mM MgSO 4 , 10 mM glucose, 10 mM Hepes, pH 7.4) or in Ca 2ϩ -free Tyrode's-Hepes buffer (134 mM NaCl, 12 mM NaHCO 3 , 2.9 mM KCl, 1 mM MgCl 2 , 5 mM glucose, 5 mM Hepes, pH 7.4).
Cell Culture and Transfection-HeLa cells were grown on 24-well culture dishes and maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum and 4 mM L-glutamine at 37°C in a 5% CO 2 environment. Cells were transfected using FuGENE 6 (Roche Applied Science) transfection reagent 48 h prior to assay with either pcDNA 3.1 (empty vector), pcDNA3-hP2Y 1 , or pcDNA3-hP2Y 12 together with pcDNA1-G␣ q/i . pcDNA1-G␣ q/i directs expression of a chimera of G␣ q containing the last five amino acids of G␣ i , which couples G␣ i -coupled receptors to activation of phospholipase C (9).
1321N1 cells expressing the human P2Y 11 receptor were generated by retroviral infection as described in Ref. 10 and cultured in Dulbecco's modified Eagle's medium with 10% fetal bovine serum and antibiotics.
Production and Purification of P18, P24, and Ap2A-P18 and P24, to be used for functional studies and as standards in the HPLC purification of the platelet dinucleotides, were produced with the purified ADP-ribosyl cyclase from A. polypoides and HPLC-purified as described by Basile et al. (4). 14 C-Labeled P18, P24, and Ap2A were produced from [ 14 C]NAD ϩ with the purified ADP-ribosyl cyclase from A. polypoides and HPLCpurified (4). Ap2A was synthesized and HPLC-purified as described by Rossi et al. (11).
HPLC Determination of P18, P24, and Ap2A Content in Human Platelets-The presence of the three adenylic dinucleotides was investigated both in resting and in thrombin-stimulated human platelets. In a typical experiment, 500-l aliquots of WP suspension (5 ϫ 10 8 platelets/ml) were incubated for 20 min at 37°C without or with 0.05 units/ml thrombin. Thereafter, aliquots were centrifuged, each supernatant was recovered, and pellets were sonicated twice for 15 s with an interval of 1 min. Lysed platelets were centrifuged (2600 ϫ g for 5 min), and supernatants and pellets were recovered and deproteinized with trichloroacetic acid (5% final concentration). Trace amounts of 14 C-labeled P18, P24, and Ap2A were added to the acid extracts to calculate the percentages of recovery of the various dinucleotides at the end of the purification procedure. Excess trichloroacetic acid was removed by diethylether extraction, and samples were analyzed by HPLC for the presence of the adenylic dinucleotides. All HPLC analyses were performed on a Hewlett-Packard 1090 instrument, equipped with a diode array spectrophotometric detector (HP 1040). Aliquots of 250 l were subjected to the analytical phosphate HPLC (4). The column was a Delta Pak C18 column (150 ϫ 4.6 mm, 5 m; Waters). Solvent A was 0.1 M KH 2 PO 4 containing 5 mM tetran-butylammonium, pH 5.0; solvent B was solvent A with 30% (v/v) methanol. The solvent program was a linear gradient starting at 100% A and increasing to 100% B in 30 min and then remaining at 100% B for 10 min. HPLC fractions containing P18, P24, or Ap2A were pooled, lyophilized, redissolved in deionized water, and injected into an analytical PLSAX column (50 ϫ 4.6 mm, 8 m; CPS). Solvent A was 20 mM ammonium bicarbonate, pH 7.6; solvent B was 500 mM ammonium bicarbonate, pH 7.6. The solvent program was the same as the one described above for reverse phase HPLC. The HPLC fractions containing the three dinucleotides were collected, dried, redissolved in deionized water, and reinjected into the analytical phosphate HPLC for final identification and quantitation (see Fig. 2 for representative chromatograms of the final purification step for each dinucleotide); P18, P24, and Ap2A were identified by comparison of their absorption spectra with computer-stored standards and were quantified by comparison of the computer-integrated peak areas with those of known amounts of the relevant dinucleotide standard. The percentage of recovery of the three adenylic dinucleotides at the end of the purification procedure was calculated from the yield of the corresponding radioactive tracer and was between 15 and 25% in the various purifications. Identification was confirmed by HPLC analysis of a nucleotide pyrophosphatase-digested parallel sample, showing conversion of the dinucleotide into the corresponding iso-AMP (or AMP for Ap2A) moiety (4). For final calculation of the intraplatelet concentration of the various dinucleotides, a platelet volume of 7 fl was considered.
P18, P24, and Ap2A Production and Degradation by Human Platelets-WP suspensions (1.0 ϫ 10 9 platelets/ml) were sonicated and incubated at 37°C with 0.65 mM NAD ϩ and 6.5 mM Ade in the presence of 2 mM MgCl 2 , for dinucleotide production, or with 0.02 mM Ap2A, P18, or P24, for dinucleotide degradation. At various times, aliquots of the suspension were withdrawn and deproteinized with 5% trichloroacetic acid, excess trichloroacetic acid was removed by diethylether extraction, and samples were injected into the analytical phosphate HPLC (4).
Measurement of Platelet Aggregation-Platelet aggregation, performed on a Menarini Aggrecoder PA-3210 aggregometer, was monitored according to Born's method (12) and quantified by light transmission for 3 min from the addition of the agonist. Platelet-rich plasma or WP (3.0 ϫ 10 8 platelets/ml) were preincubated without (control) or with P18, P24, and Ap2A for 2 min at 37°C before the addition of agonists (thrombin, ADP, or collagen).
Measurement of Intraplatelet Calcium Levels-WP (3.0 ϫ 10 8 platelets/ml), resuspended in Hepes buffer, were incubated with 1 M Fura-2/AM for 60 min at 37°C. Prostaglandin E 1 (2 M final concentration) and EGTA (1 mM final concentration) were added before centrifugation of the Fura-2-loaded platelets for 15 min at 1100 ϫ g. The platelet pellet was resuspended at 2.0 ϫ 10 8 platelets/ml in Hepes buffer and preincubated at 37°C for 5 min with saline before the addition of the adenylic dinucleotides to be tested. When the effect of Ap2A, P18, and P24 was tested on platelets stimulated with thrombin, WP were preincubated with the adenylic dinucleotides for 5 min at 37°C and then challenged with thrombin. Fluorescence of Fura-2loaded platelets was measured at 37°C in unstirred conditions in a PerkinElmer Life Sciences fluorescence spectrometer model LS50B with excitation at 340 and 380 nm and emission at 509 nm. The fluorescence of fully saturated Fura-2 (F max ) was obtained by lysing platelets with 50 M digitonin in the presence of 2 mM Ca 2ϩ , whereas F min was determined by the subsequent addition of 20 mM EGTA. The fluorescence was fully quenched with 5 mM MnCl 2 to yield the autofluorescence value. Dedicated software converted data into cytosolic Ca 2ϩ concentration applying a K d value for Fura-2 and Ca 2ϩ of 135 nM.
Inositol Phosphate Accumulation Assay in Transfected HeLa Cells-Cells were labeled overnight with 200 l of serum-free, inositol-free Dulbecco's modified Eagle's medium containing 0.4 Ci/well 3 H-labeled myo-inositol (American Radiolabeled Chemicals, St. Louis, MO). Assays were initiated with the addition of 5ϫ concentrations of the indicated compounds in 50 l of 50 mM LiCl, 250 mM Hepes, pH 7.25. Following incubation for 15 min at 37°C, the medium was removed by aspiration, and the reaction was terminated by the addition of 0.75 ml of boiling 10 mM EDTA, pH 8.0. [ 3 H]Inositol phosphates were isolated by Dowex column chromatography as described previously (13).
Measurement of Nitric Oxide (NO) Production-WP (0.5 ϫ 10 8 platelets/ml), resuspended in Hepes buffer containing 1 mM CaCl 2 and prewarmed at 37°C for 10 min with saline or additions, were incubated with 40 M L-arginine and the various dinucleoside diphosphates (10 M), alone or with thrombin (0.05 units/ml), for 15 min at 37°C under mild horizontal shaking. The incubation was stopped by sonicating samples on ice.
To measure the NO content, 0.55-ml aliquots of supernatant were added to equal volumes of glycine buffer (15 g/liter glycine NaOH, pH 9.7) containing activated cadmium beds (Fluka AG), which catalyze the chemical reduction of nitrate to nitrite (14), and incubated overnight at room temperature under horizontal shaking. Cadmium beds were activated immediately before each experiment by three subsequent washings with 0.2 N H 2 SO 4 , bidistilled water and glycine buffer, respectively. NO formation, determined by the Griess reagent (1% sulfanylamide in 2.5% H 3 PO 4 , 0.1% naphthylenediamine dihydrochloride), was measured at 540 nm using a sodium nitrite calibration curve (15).
Determination of cAMP Levels in 1321N1 Cells and Human Platelets-Control and hP2Y 11 -transfected 1321N1 cells were seeded in 35 ϫ 10-mm dishes (0.5 ϫ 10 6 cells/dish). After 24 h, the medium was replaced with 0.6 ml of HBSS. P24, Ap2A, or buffer (control) were added at the indicated concentrations, and incubations were stopped after 0, 2.5, 5, or 15 min by removal of HBSS and the addition of 200 l of ice-cold perchloric acid (PCA) (0.6 M). Cell extracts were collected and centrifuged to remove proteins; the cAMP content was measured on the neutralized cell extracts by radioimmunoassay ([ 3 H]cAMP assay system; GE Healthcare).
Aliquots (300 l, 1 ϫ 10 9 platelets/ml) of WP from different healthy donors were preincubated or not for 30 min at 37°C with 1 M NF157 prior to incubation with 10 M P18, P24, or Ap2A for 15 min. Incubations were stopped by the addition of 26 l of 9 M PCA, and then each sample was sonicated in ice at 2.5 watts for 10 s and centrifuged at 2600 ϫ g for 3 min at 4°C. The supernatant from each sample was collected, and the PCA was removed as described by Graeff and Lee (16). The cAMP content was measured on neutralized platelet extracts by radioimmunoassay ([ 3 H]cAMP assay system; GE Healthcare).
Determination of IP 3 Levels in 1321N1 Cells and Human Platelets-Control and hP2Y 11 -transfected 1321N1 cells were seeded in 35 ϫ 10-mm dishes (0.5 ϫ 10 6 cells/dish). After 24 h, the medium was replaced with 0.6 ml of HBSS. 10 M P24 or Ap2A or buffer (control) were then added, and at 0, 0.5, 1, and 2.5 min after the addition, the incubation was stopped by the removal of HBSS, the addition of 300 l of ice-cold 0.6 M PCA, and scraping of the cells. The IP 3 content was measured on the supernatants of the neutralized cell extracts by radioimmunoassay ([ 3 H]IP 3 Biotrak Assay System; Amersham Biosciences).
Aliquots (200 l, 1 ϫ 10 9 platelets/ml) of WP from different healthy donors resuspended in Hepes buffer were incubated at 37°C for 30 s in the presence of 10 M P24 or Ap2A or buffer (control). At the end of each incubation, the suspensions were withdrawn, and the reaction was stopped by adding 13 l of 9 M PCA at 4°C. PCA was removed as described (16). The intracellular IP 3 concentration was determined as described above.
Statistical Analyses-All parameters were tested by paired t test. p values of Ͻ0.05 were considered significant.

RESULTS
Production of P18, P24, and Ap2A by Human Platelets-Human CD38, along with the ADP-ribosyl cyclases from the sponge A. polypoides and from the mollusc A. californica, has been shown to produce P18, P24, and Ap2A from cyclic ADPribose (or NAD ϩ ) and Ade ( Fig. 1) (4). CD38 is expressed in human platelets (5,6), and its activity increases during platelet aggregation (7). Thus, we preliminarily tested whether platelets also produced these adenylic dinucleotides. Indeed, P18, P24, and Ap2A production was observed upon incubation of sonicated WP with NAD ϩ (precursor of cyclic ADP-ribose) and Ade (Table 1). Interestingly, dinucleotide production was higher in platelets stimulated with thrombin (0.05 units/ml) as compared with resting platelets (Table 1). In the presence of 20 mM nicotinamide, an inhibitor of ADP-ribosyl cyclase activity, production of the three adenylic dinucleotides was abrogated (Table 1). No degradation of P24 and Ap2A was detectable in lysed platelets, either resting or activated; conversely, P18 was hydrolyzed to ADPR and Ade, similarly to what was observed for recombinant human CD38 (4), and the P18-hydrolyzing activity decreased in thrombin-stimulated platelets ( Table 1).
Presence of P18, P24, and Ap2A in Resting and in Thrombinstimulated Platelets-It has been recently demonstrated that secretory granules of human platelets contain Ap2A, which is released during platelet aggregation evoked by thrombin (1). This observation, together with the ability of platelet CD38 to synthesize the Ap2A isomers P18 and P24, in addition to Ap2A (Table 1), prompted us to investigate whether P18 and P24 were also present in human platelets and whether they were released during aggregation. HPLC analysis of acid extracts of resting platelets confirmed the presence of Ap2A at a concentration (ϳ20 M) similar to that already reported (1). Both Ap2A isomers were indeed detected at a similar intraplatelet concentration (ϳ40 and 60 M for P18 and P24, respectively). Neither Ap2A nor its isomers were detectable in the supernatant from resting platelets ( Table 2). The intraplatelet concentration of all three adenylic dinucleotides increased in thrombin-stimulated, as compared with resting, platelets (1.5-, 1.3-, and 3-fold for P18, P24, and Ap2A, respectively; see Fig. 2 for representative chromatograms of the final purification step). All adenylic dinucleotides were released into the supernatant, where they were present at ϳ40, 50, and 15 M for P18, P24, and Ap2A, respectively (Table 2), when platelets were suspended at a concentration similar to that present in whole blood.
P18, P24, and Ap2A Inhibit Agonist-induced Platelet Aggregation-Production of P18, P24, and Ap2A from NAD ϩ and Ade by human platelet lysates, the presence of the dinucleotides in resting platelets, and their release upon platelet activation by thrombin prompted us to explore the possibility that P18, P24, and Ap2A may affect platelet aggregation. Pretreatment of platelets for 2 min at 37°C with increasing concentrations of P18, P24, or Ap2A (1, 5, and 10 M) had no effect on platelet aggregation, as observed during the following 3 min at 37°C under stirring. Conversely, down-regulation of platelet aggregation was observed when platelets were pretreated with the adenylic dinucleotides and then challenged with ADP, thrombin, or collagen. As shown in Fig. 3, ADP-induced platelet aggregation was significantly reduced by all three adenylic dinucleotides, with IC 50 values of 7.2, 7.9, and 6.3 M for P18, P24, and Ap2A, respectively (Table 3). In platelets treated with suboptimal concentrations of thrombin (0.05 units/ml) or collagen (1.5 g/ml), aggregation was also inhibited in a concentration-dependent way by each of the three dinucleotides, although IC 50 values were slightly higher compared with those observed upon ADP-induced aggregation (Table 3).
P18, P24, and Ap2A Reduce the Thrombin-induced [Ca 2ϩ ] i Rise in Intact Platelets-During platelet aggregation, an increase of the intracellular calcium concentration ([Ca 2ϩ ] i ) is known to occur downstream of the activation by ADP-, thrombin-, or collagen-specific receptors (17,18). The inhibitory effect of P18, P24, and Ap2A on agonist-induced platelet aggregation suggested that we should investigate the effect of these dinucleotides on the [Ca 2ϩ ] i . To this purpose, platelets were preincubated with Ap2A, P18, or P24 and then stimulated with thrombin. In platelets challenged with thrombin (0.05 units/ ml), the [Ca 2ϩ ] i increased rapidly from a resting value of 65 Ϯ 11 nM to a maximal value of 172 Ϯ 35 nM. At the same time point, the [Ca 2ϩ ] i increase observed in platelets pretreated for 5 min with the adenylic dinucleotides (10 M) before the addition of thrombin was significantly reduced (Table 4). P18 exerted the highest percentage of inhibition (54%), in line with the lower IC 50 value observed upon thrombin-induced aggregation (Table 3). In the absence of thrombin, all three adenylic dinucleotides evoked a very limited increase of the [Ca 2ϩ ] i (Յ8% of the basal value), which was apparently insufficient to

TABLE 1 P18, P24, and Ap2A production by human platelets
WP suspensions (1 ϫ 10 9 platelets/ml), preincubated or not with 0.05 units/ml thrombin in the presence or absence of 20 mM nicotinamide, were sonicated and incubated at 37°C with 0.65 mM NAD ϩ and 6.5 mM Ade for dinucleotide production or with 0.02 mM Ap2A, P18, or P24 for dinucleotide degradation. At various times, aliquots of the suspensions were withdrawn, deproteinized, and analyzed by HPLC. Production and degradation of Ap2A, P18, and P24 by human platelets are expressed as pmol/min/10 9 platelets. Results are the mean of three experiments, performed on platelets from different healthy donors. ND, not detectable. induce platelet aggregation, since neither dinucleotide per se had any proaggregant effect (see above). Are P18, P24, and Ap2A ADP Antagonists on P2Y 1 and P2Y 12 Receptors?-Platelet granules are known to contain high concentrations of adenylic nucleotides, most notably ADP and ATP, which are released during platelet aggregation and con-tribute to a positive feedback mechanism, by acting through specific purinergic receptors on the platelet membrane (19 -21). Thus, platelet aggregation induced by collagen, thrombin, or endothelial injury induces release of ADP, which in turn binds to P2Y 1 and P2Y 12 receptors, further stimulating aggregation (20). The structural homology between the adenylic dinucleotides (Ap2A, P18, and P24) and ADP prompted us to investigate whether they behaved as ADP antagonists on P2Y 1 or P2Y 12 receptors. HeLa cells were transfected with human P2Y 1 or co-transfected with human P2Y 12 and G␣ q/i to generate a cell system responsive to ADP with an increase of the inositol phosphate concentration. Incubation of P2Y 1 -or P2Y 12 /G q/i -transfected cells with 30 M P24 did not increase the inositol phosphate concentration in either cell type (Fig. 4). Conversely, 1 M 2MeSADP, a stable analogue of ADP, strongly increased the inositol phosphate concentration both in P2Y 1and P2Y 12 /G q/i -transfected cells, and this increase was not inhibited by the simultaneous addition of P24 (Fig. 4). The inability of P24 to bind to P2Y 1 receptors was also confirmed by Scatchard analysis. P24 failed to displace [ 32 P]MRS2500, a specific P2Y 1 receptor radioligand (22), from the P2Y 1 receptor (not shown). The fact that P24 did not interact with P2Y 1 or with P2Y 12 suggested that we should explore other mechanisms of action, resulting in an inhibition of platelet aggregation by Ap2A, P18, and P24. P18, P24, and Ap2A Stimulate Platelet Nitric Oxide Production in the Presence of Thrombin-NO, a potent antagonist of platelet aggregation (23)(24)(25), is synthesized in human platelets by a constitutive NO synthase (26 -28). The inhibitory effect of Ap2A, P18, and P24 on platelet aggregation induced by the physiological agonists ADP, thrombin, or collagen prompted us to investigate whether these adenylic dinucleotides could stimulate platelet NO production. WP (0.5 ϫ 10 8 platelets/ml) resuspended in Hepes buffer containing 1 mM CaCl 2 and prewarmed at 37°C for 10 min were incubated with 40 M L-arginine and the various dinucleoside diphosphates (10 M) alone or with thrombin (0.05 units/ml) for 15 min at 37°C under mild shaking. In platelets treated with thrombin alone, NO production was not increased over control values, measured in untreated platelets (Fig. 5, white bars), as expected due to the suboptimal thrombin concentration used for platelet activation (29).  Neither adenylic dinucleotide, in the absence of thrombin, modified basal NO production (Fig. 5, striped bars). Conversely, in platelets incubated with thrombin and either of the three adenylic dinucleotides, NO production increased over control values, 3.1-, 2.1-, and 1.3-fold in P18-, P24-, and Ap2Atreated-platelets, respectively (Fig. 5, black bars). P18, P24, and Ap2A Increase the [cAMP] i and the [IP 3 ] i in Human Platelets through P2Y 11 Receptor Stimulation-Stimulation of NO production by P18, P24, and Ap2A suggested that we should explore the effect of the dinucleotides on the [cAMP] i , since NOS activity is known to be stimulated by PKA (30). Moreover, an increase of [cAMP] i is known to induce a complete inhibition of platelet function (31,32). Platelet incu-bation for 15 min at 37°C with 10 M P18, P24, or Ap2A indeed increased the [cAMP] i 2.1-, 1.8-, or 1.5-fold, respectively, over values measured in untreated platelets ( Fig. 6A; control, n ϭ 5, p Ͻ 0.05). The only purinergic receptor known to stimulate adenylic cyclase is P2Y 11 (10,33), which is also expressed in human platelets (34). Preincubation of platelets for 30 min with 1 M NF157, a specific inhibitor of P2Y 11 (35), reduced cAMP overproduction induced by P18, P24, and Ap2A by ϳ70, 75, and 100%, respectively (Fig. 6A).

Dinucleotide production
Since P2Y 11 is also coupled to a phospholipase C, we investigated whether the dinucleotides induced intraplatelet inositol 1,4,5-trisphospate ([IP 3 ] i ) rise. As shown in Fig. 6B 3 ] i rise produced by the two adenylic dinucleotides was significantly less than that observed with 0.1 units/ml thrombin, in agreement with the inability of the three adenylic nucleotides to

TABLE 4
Effect of P18, P24, and Ap2A on the thrombin-induced ͓Ca 2؉ ͔ i rise in human platelets FURA2-loaded WP (2 ϫ 10 8 platelets/ml) were preincubated in the absence (control) or in the presence of the adenylic dinucleotides for 5 min at 37°C and then challenged with 0.05 units/ml thrombin. The percentage of inhibition (I) on thrombin-induced ͓Ca 2ϩ ͔ i rise is reported. Results are the mean Ϯ S.D. of three different experiments performed on platelets from different healthy donors.
36 Ϯ 3 evoke aggregation. Finally, we investigated whether stimulation of NO production by Ap2A, P18, and P24 was downstream of P2Y 11 activation. Pretreatment of platelets with 1 M NF157 before stimulation with 10 M P24 or Ap2A completely abrogated the release of NO by the two adenylic dinucleotides (Fig. 5). 11 -transfected 1321N1-To confirm that the observed increase of the [cAMP] i and of the [IP 3 ] i in platelets was a consequence of binding of the adenylic dinucleotides to the P2Y 11 receptor, we measured the concentration of these two intracellular second messengers in control and hP2Y 11 -transfected 1321N1. Cells stimulated with 10 M P24 or Ap2A showed a time-dependent increase in [cAMP] i , with the maximum value reached for both adenylic dinucleotides after 15 min of incubation (Fig. 7A) (Fig. 7B).

DISCUSSION
Here we show that platelets produce and release three adenylic dinucleotides, Ap2A and two isomers thereof, characterized by an unusual N-glycosidic bond between one of the adenines and the ribose: C1Ј-N1 and C1Ј-N3 in P18 and P24, respectively ( Fig. 1) (4). Ap2A, P18, and P24 behave as negative modulators of platelet aggregation induced by the physiological agonists ADP, thrombin, or collagen.
The presence of Ap2A in human platelets was already reported, although the enzyme responsible for its synthesis and the effect of the dinucleotide on platelet aggregation were not investigated (1). Human CD38 has been recently demonstrated to synthesize Ap2A, P18, and P24 ( Fig. 1) (4). The fact that synthesis of the adenylic dinucleotides by the platelet lysate is inhibited by nicotinamide (Table 1), a known inhibitor of ADP-ribosyl cyclase activity (36), strongly suggests that platelet CD38 is responsible for synthesis of Ap2A and of its isomers.
P18 and P24 are present in resting platelets at micromolar concentrations, similar to those already reported for Ap2A (1) ( Table 2). Platelet activation by thrombin stimulates production of Ap2A, P18, and P24 (Table 2), suggesting the presence of a feedback mechanism, limiting platelet aggregation through generation of the three adenylic dinucleotides. Release of the adenylic dinucleotides from thrombin-activated platelets, at a  platelet density similar to that present in human plasma, results in extracellular concentrations of each dinucleotide equal to or higher than their IC 50 value upon agonist-induced platelet aggregation (Table 3). These results suggest a role for Ap2A, P18, and P24 as negative endogenous modulators of platelet aggregation. Release of antiaggregant molecules by agoniststimulated platelets is not unprecedented; platelet activation by thrombin, ADP, or collagen also results in the production of the most potent endogenous inhibitor of platelet activation, adhesion, and aggregation (i.e. NO) (25,23,37). Although it was recently reported that subthreshold concentrations of thrombin (Ͻ0.02 units/ml) induce release of very low (nanomolar) concentrations of platelet-derived NO, which behaves as a proaggregant (38), there is general agreement that proaggregant agonist concentrations induce release of higher (micromolar) concentrations of NO, which then behaves as a potent endogenous antithrombotic (30).
In nucleated cells, the biological activity is markedly different among Ap2A and its isomers. On human hemopoietic progenitors (CD34 ϩ cells), Ap2A stimulates proliferation, whereas P18 and, particularly, P24 induce apoptosis, with LD 50 values on colony growth of 1.0 and 0.18 M, respectively (4). P24 cytotoxicity, demonstrated on a wide range of different cell types (4), is due to its mitochondrial effects, which include inhibition of complex I of the respiratory chain and dissipation of the proton gradient (⌬⌿ m ); neither P18 nor Ap2A show any effect on respiration or on ⌬⌿ m (39). In contrast to the diverse effects observed on nucleated cells, all three adenylic dinucleotides share the same inhibitory effect on agonist-induced platelet aggregation, with P18 showing the lowest IC 50 values (Table 3). Since Ap2A, P18, and P24 all share a common ADP moiety, inhibition of platelet aggregation could be due to binding of these dinucleotides to the purinergic receptors involved in endogenous ADP-induced aggregation (i.e. P2Y 1 and P2Y 12 ). However, the following experimental findings rule out the possibility that the adenylic dinucleotides behave as ADP antagonists on P2Y 1 and on P2Y 12 . P24 neither induced an increase of the [IP 3 ] i by itself nor antagonized the increase of the [IP 3 ] i induced by the synthetic purinergic agonist 2MeSADP on P2Y 1 -transfected and on P2Y 12 /G i -transfected HeLa cells (Fig.  4). Moreover, P24 failed to displace [ 32 P]MRS2500, a specific P2Y 1 receptor radioligand (22), from the P2Y 1 receptor (not shown), ruling out a direct effect of this dinucleotide on P2Y 1 .
The mechanism through which Ap2A and its isomers inhibit agonist-induced platelet aggregation seems to depend on (i) reduction of the agonist-triggered [Ca 2ϩ ] i rise (Table 4), which is causally related to platelet aggregation (17,18); (ii) increased platelet production of NO (Fig. 5); and (iii) increased intraplatelet concentration of cAMP ([cAMP] i ) (Fig. 6). The following results indicate that the receptor mediating the antiaggregant effects of Ap2A, P18, and P24 is the purinergic receptor P2Y 11 : (i) the P2Y 11 -specific inhibitor NF157 (35) (Fig. 7), in agreement with the fact that P2Y 11 activates both adenylyl cyclase and phospholipase C (10, 33). The increase of the intraplatelet [IP 3 ] i induced by the adenylic dinucleotides is significantly less than that observed with thrombin (Fig. 6B), in agreement with the absence of a significant [Ca 2ϩ ] i rise and of a proaggregant effect by the adenylic nucleotides per se.
P2Y 11 mRNA levels in platelets were reported to be much lower than those of P2Y 1 , P2Y 12 , and P2X 1 , although the level of mRNA depends on both its synthesis and degradation rates and may not linearly correlate with its protein expression (34). Also in human granulocytes, P2Y 11 mRNA is present at lower levels compared with mRNA of other purinergic receptors (40). In fact, expression of the P2Y 11 protein in granulocytes was found to be even lower than that observed in platelets (41). Nonetheless, ATP, NAD ϩ , and NAADP ϩ have been shown to induce significant biochemical and functional effects in granulocytes through P2Y 11 ligation (41)(42)(43). At present, binding of Ap2A, P18, and P24 to other purinergic and/or adenosine receptors, in addition to P2Y 11 , cannot be ruled out. Adenosine is known to induce an increase of the [cAMP] i and to exert antiaggregant effects on platelets by binding to A-type receptors (44), and Ap2A has been shown to be an agonist of A1 and A2 adenosine receptors in rat kidney vasculature (45). However, the fact that the P2Y 11 -specific antagonist NF157 inhibits by 100, 70, and 75% the [cAMP] i increase induced by Ap2A, P18, and P24, respectively, suggests that an effect of these dinucleotides on adenosine receptors may account for Յ20 -25% of the [cAMP] i increase and only for P18 and P24.
Both endogenous NO production and an intraplatelet [cAMP] i rise are known to inhibit platelet activation (25,31,32). NO increases intraplatelet cGMP, which inhibits cAMP phophodiesterase (46), contributing to the elevation of the [cAMP] i . Moreover, PKA is known to phosphorylate and activate platelet NOS (30). Thus, a positive feedback mechanism maintains elevated levels of [cAMP] i via NO. A high [cAMP] i inhibits IP 3 -induced Ca 2ϩ release in human platelets and megakaryocytes (47,48). The fact that Ap2A, P18, and P24 induce an increase of the [cAMP] i could result both in activation of NOS via PKA and in a reduction of the [Ca 2ϩ ] i in thrombin-stimulated platelets. NO release from platelets incubated with the adenylic dinucleotides occurred only in the presence of thrombin (Fig. 5), at a concentration (0.05 units/ml) that did not induce NO release per se (Fig. 5), as already reported (29,49). A possible synergism between the pathways triggered by thrombin and by P2Y 11 and leading to NOS activation could explain this observation. A similar interpretation of their results was proposed by Radomski et al. (49), who reported that the antiaggregant effect of L-arginine on platelets stimulated with thrombin (0.03 units/ml) became evident only upon the addition of subthreshold concentrations of antiaggregant prostacyclin (a known stimulator of [cAMP] i increase).
A deficiency of bioactive NO is known to be associated with arterial thrombosis both in animal models and in humans (50,51), and new NO donors with antithrombotic and vasodilating activities are being developed for treatment of arterial thrombosis (52). The discovery of new endogenous negative modulators of platelet aggregation that act through stimulation of platelet NO production could possibly lead to the development of a new family of antithrombotic drugs.