Leukocytes Can Enhance Platelet-mediated Aggregation and Thromboxane Release via  Interaction of P-selectin Glycoprotein Ligand 1 with P-selectin

After obtaining institutional review board approval (Joint Commission on Clinical Investigation at the Johns Hopkins Medical Institutions, Baltimore, MD) and informed consent, venous blood was drawn from healthy male and female volunteers between the ages of 20 and 47 yr. After discarding the initial 3 ml, blood was anticoagulated with heparin (1 U/ml for aggregation studies or 10 U/ml for granulocyte isolation) or citrate (9:1 for aggregation, cytometric, and enzyme-linked immunosorbent assays). Preliminary studies demonstrated similar results for aggregation experiments performed in heparin and citrate anticoagulated blood. All volunteers were free of aspirin and nonsteroidal anti-inflammatory drugs for 10 days before blood drawing.

Blocking antibodies to P-selectin (Endogen, Woburn, MA), PSGL1 (Immunotech, Miami, FL), L-selectin (Endogen), CD11b (Endogen), and immunoglobulin (Ig)G isotype control (Pharmingen, San Diego, CA) were all azide-free and certified as no–low endotoxin contamination. The following fluorochrome-labeled antibodies were used in flow cytometry: phycoerythrin–CD15 antibody (Immunotech, Marseilles, France) and perCP-CD61 antibody (Becton Dickinson Immunocytometry Systems, Bedford, MA) were used to label granulocytes and platelets, respectively; and phycoerythrin–IgG (Sigma Chemical Co, St. Louis, MO) and perCP-IgG (Immunotech) antibodies were used as controls. All antibodies were of the IgG1 isotype. Thrombin receptor activating peptide (TRAP) was obtained from Bachem Biosciences (King of Prussia, PA). Adenosine diphosphate and luciferase were obtained from Chrono-log Corporation (Havertown, PA). Arginine-glycine-aspartate-serine peptide, Hank’s balanced salt solution (HBSS), and indomethacin were obtained from Sigma. c-18 sepPak cartridges (Waters Corporation, Milford, MA) were used for thromboxane extraction. β-thromboglobulin and thromboxane B2 enzyme immunoassay kits were obtained from Diagnostica Stago (Asneires-Sur-Seine, France) and Neogen (Marseilles, France), respectively.

Whole blood was centrifuged at 180 g  for 15 min to obtain platelet-rich plasma (PRP). PRP was diluted with platelet-poor plasma to adjust platelet counts to 100,000/μl (final concentration in aggregation samples). Polymorphonuclear leukocytes (PMNs) were isolated from whole blood by Histopaque gradient centrifugation (Histopaque 1077/1119, Sigma Chemical Co.), washed once in HBSS, and resuspended in HBSS to achieve a final concentration of 1500 ± 300 cells/μl (final PMN:platelet of 1:67 in mixed-cell aggregation samples). PMN isolation by Histopaque gradient centrifugation, with minimization of washing steps and avoidance of hypotonic erythryocyte lysis, has been suggested as a means of limiting in vitro  leukocyte activation. 16Leukocytes prepared in this way were found to be more than 98% neutrophils (Wright-Giemsa staining) and more than 98% viable (Trypan blue exclusion). All samples were ready for in vitro  analysis within 1.5 h of blood drawing.

Impedance aggregometry was performed using a Chrono-log Lumi-Aggregometer (Havertown, PA) as described by the manufacturer with modification. Briefly, anticoagulated whole blood was diluted in an equal volume of HBSS and incubated for 20 min (37°C) in the presence and absence of blocking or IgG isotype control antibodies (20 μg/ml). Aggregation in response to TRAP (5–20 μm) or adenosine diphosphate (10 μm) was measured as the change in impedance units over 5 min and peak aggregation responses recorded. Aggregation was performed at 37°C, and samples were stirred at 1,200 min^-1to simulate arterial shear conditions. Simultaneous measurement of dense granule secretion was determined by agonist-induced adenosine triphosphate release and measured as change in luminescence units after addition of luciferase (0.1 ml/ml blood). Luminescence units were converted to adenosine triphosphate concentration (nanomoles/ml) using an adenosine triphosphate standard.

Optical aggregometry on mixed-cell suspensions was performed using a Chrono-log Lumi-aggregometer using the method of Schattner et al.  17with modification. Briefly, PRP and PMNs were isolated as previously described and incubated separately for 20 min at 37°C. During the 20-min incubation period, PRP was incubated in the presence and absence of P-selectin or IgG isotype control antibodies (20 μg/ml). PRP aggregation in response to TRAP (15 μm) was measured in the presence of PMNs or HBSS control, which were added to PRP immediately before agonist stimulation. In some experiments, PMNs were centrifuged at 10,000 g  for 5 min (after the 20-min incubation period), and the granulocyte supernatant was added to PRP instead of the PMN suspension. Aggregation was measured as percent light transmission relative to a platelet-poor reference and was recorded as peak change in optical density in 5 min. Aggregation was performed at 37°C, and samples were stirred at 1,000 min^-1to simulate arterial shear.

Samples were prepared as described previously for whole-blood impedance aggregometry, except arginine-glycine-aspartate-serine peptide (400 μm) was added just before TRAP activation to prevent platelet–platelet aggregate formation. Five minutes after agonist stimulation (or unstimulated control), 50-μl aliquots were removed from the aggregometer, and platelet–leukocyte conjugate formation was assessed by two-color flow cytometry as previously described 8with modification. Briefly, aliquots of blood were incubated with phycoerythrin–CD15 and perCP-CD61 antibodies to label granulocytes and platelets, respectively. After incubating samples for 20 min at 22°C, erythrocytes were lysed and samples fixed with FACS lysing solution (Beckton Dickinson, San Jose, CA). Samples were stored at 4°C and analyzed by flow cytometry within 24 h of fixation. Flow cytometry was performed using a FACScan cytometer (Beckton Dickinson) equipped with a 488-nm argon excitation laser. Scatter data were acquired in linear mode and fluorescence data in logarithmic mode. Data from 10,000 granulocyte events were collected and analyzed. Granulocytes were identified by their phycoerythrin–CD15 fluorescence and scatter profile, and perCP–fluorescence histograms were generated from the gated granulocyte population. The presence of platelet–leukocyte conjugates was expressed as the percent of granulocytes displaying perCP–fluorescence. Unstimulated whole blood incubated with perCP-IgG isotype control antibody served as the negative control, and positivity gates for the negative control were set at 2%.

Samples were prepared as previously described for whole-blood impedance aggregometry. Five minutes after TRAP stimulation, 50 μl of blood was removed from the aggregometer and incubated with an equal volume of ice-cold acid-citrate-dextrose quenching buffer. (Quenching buffer included 20 μm indomethacin for thromboxane assays.) Samples were incubated on ice for 5 min, centrifuged at 10,000 g  for 5 min, and plasma was removed for β-thromboglobulin and thromboxane assays. β-thromboglobulin and thromboxane were determined by commercially available enzyme-linked immunosorbent assays according to the manufacturers’ protocols.

Data were analyzed by paired t  test and repeated-measures analysis of variance with Bonferroni-corrected post hoc  analyses, as appropriate. Data are expressed as mean ± SD.

To determine the impact of platelet–leukocyte interactions on platelet agonist–induced aggregation, we inhibited platelet–leukocyte adhesion with blocking antibodies to P-selectin and PSGL1. P-selectin and PSGL1 antibodies inhibited TRAP-induced aggregation in whole blood, but IgG control antibodies did not (figs. 1A and 1B). Using an increasing dose response to TRAP (5–20 μm), we observed a similar inhibition of aggregation in response to 10 and 20 μm TRAP; however, aggregation was not reliably induced in response to 5 μm TRAP, regardless of the presence of blocking antibody (data not shown). Adenosine diphosphate–induced aggregation was also inhibited, although the magnitude of this inhibitory effect was less (11 ± 3, 11 ± 2, 8 ± 3*, and 7 ± 2* for no antibody, IgG control, P-selectin, and PSGL1, respectively; *P < 0.05 vs.  no antibody and IgG control by Bonferroni post hoc  test). In contrast, blocking antibodies to L-selectin, a constitutively expressed leukocyte selectin, had no impact on TRAP-induced aggregation (24 ± 2, 24 ± 2, and 22 ± 2 for no antibody, IgG control, and L-selectin, respectively). Similarly, blocking antibodies to CD11b, an activation-dependent leukocyte integrin capable of mediating platelet–leukocyte adhesion, 2did not significantly reduce aggregation in response to TRAP (26 ± 7, 26 ± 5, and 23 ± 4, for no antibody, IgG control, and CD11b, respectively).

To confirm that TRAP stimulation induced platelet–leukocyte conjugate formation in whole blood, and that blocking antibodies inhibited platelet–leukocyte adherence, we analyzed aggregation samples by flow cytometry. Flow cytometric experiments demonstrated that TRAP activation induced the formation of platelet–leukocyte conjugates and that blockade of P-selectin or PSGL1 nearly eliminated conjugate formation (table 1). However, blocking antibodies to L-selectin did not reduce the number of platelet–leukocyte conjugates formed in response to TRAP; and blocking antibodies to CD11b reduced conjugate formation by only 10% (data not shown).

Data from blocking antibody experiments suggested that platelet–leukocyte adhesion contributed to TRAP-induced aggregate formation in whole blood, and that the interaction of P-selectin with PSGL1 contributed substantially to this effect. To confirm that leukocytes promoted aggregation and that P-selectin-antibody–mediated inhibition was specific to a reduction in platelet–leukocyte adhesion, we repeated aggregation experi-ments in PRP in the presence and absence of isolated granulocytes. Results of these experiments are shown in figure 2and demonstrate that isolated granulocytes moderately increased TRAP-induced aggregation in PRP, and that P-selectin blockade attenuated this effect. Furthermore, P-selectin antibody had no effect on platelet aggregation in the absence of granulocytes (fig. 2), and TRAP-stimulated granulocytes had no ability to aggregate in the absence of platelets (data not shown). To determine if the physical presence of leukocytes was necessary to augment PRP aggregation, we performed aggregation experiments in the presence and absence of leukocyte supernatants instead of leukocyte suspensions. There was no difference in PRP aggregation between samples in which measurements were made in the presence and absence of leukocyte supernatants (79%± 23%vs.  75%± 22% for supernatant vs.  HBSS, respectively;P = nonsignificant). Thus, under the conditions of this study, the physical presence of leukocytes appeared necessary to augment TRAP-induced aggregate formation in PRP.

Whole-blood and PRP experiments indicated that platelet–leukocyte adhesion contributed to aggregate formation induced by platelet agonists; however, the direct effect of conjugate formation on platelet function was unclear. To determine the impact of platelet–leukocyte adhesion on platelet function, we measured platelet secretion responses in TRAP-stimulated blood incubated with P-selectin, PSGL1, or control antibodies. Blocking antibodies reduced thromboxane release from TRAP-stimulated whole-blood samples (fig. 3). However, P-selectin and PSGL1 antibodies had no effect on TRAP-induced platelet dense granule (as measured by adenosine triphosphate release) or α-granule (as measured by β-thromboglobulin) secretion (table 2).

We hypothesized that leukocytes modulate platelet agonist–induced aggregation and secretion and that platelet–leukocyte adherence via  interaction of P-selectin with PSGL1 contributes to these responses. Results of our experiments indicate that leukocytes contribute to platelet agonist–induced aggregate formation and thromboxane release under shear stress in whole blood, and that interaction of platelet P-selectin with leukocyte PSGL1 contributes to these responses. These findings suggest that leukocytes may participate in hemostasis by contributing to aggregate bulk formation in response to platelet activation and by promoting platelet activation and vasoconstriction through increased thromboxane production.

The physiologic importance of platelet–leukocyte interactions has received considerable attention over the past 10 yr, with substantial focus on the proinflammatory consequences of these interactions. 18–20However, the impact of leukocytes on platelet-mediated hemostatic functions remains controversial because both augmented 10,11,21,22and inhibited 12,13,17,23platelet activation are reported in vitro . Leukocytes are capable of modulating platelet functions through release of a variety of stimulatory and inhibitory substances, including cathepsin G, 10,22reactive oxygen species, 11and nitric oxide 12; however, release requires leukocyte activation, which may or may not occur, before platelet activation in vivo . Furthermore, the net affect of these stimulatory and inhibitory substances on platelet functions in the blood milieu is unclear because plasma proteases can neutralize cathepsin G,¹⁴and erythrocytes can scavenge nitric oxide and increase thromboxane production. 24 

In contrast to many previous studies, we assessed aggregation and secretion responses in live cells in whole blood and plasma to approximate physiologic conditions. Under these in vitro  conditions, leukocytes enhanced platelet agonist–induced aggregation through a process that involved platelet P-selectin and leukocyte PSGL1. Enhanced responses required cell–cell contact because blocking antibodies inhibited aggregation in whole blood and PRP, but cell-free leukocyte supernatants had no effect. The ability of P-selectin and PSGL1 blocking antibodies to inhibit platelet agonist–induced aggregation was greater for TRAP than adenosine diphosphate (approximately 50%vs.  30% inhibition), which is consistent with greater induction of P-selectin expression (i.e. , α-granule secretion) known to occur with TRAP stimulation. However, in neither case was antibody blockade completely effective at suppressing aggregation. This observation is consistent with the fact that platelet aggregation involves multiple mechanisms, the most important of which is the agonist-induced binding of fibrinogen to GPIIb-IIIa receptors on activated platelets. 25 

To determine if platelet–leukocyte conjugate formation could directly modify platelet function, we measured platelet-specific secretion responses in the presence and absence of blocking antibodies. These experiments demonstrated that leukocytes modulated thromboxane secretion through interaction of PSGL1 with P-selectin in TRAP-stimulated whole blood, but had no impact on platelet-dense granule and α granule secretion. Other investigators demonstrated previously that activated leukocytes (formyl-methionine-leucine-phenylalanine–stimulated leukocytes in platelet–leukocyte suspension) could transfer arachidonic acid to P-selectin–conjugated platelets and enhance thromboxane production. 26,27Our observations extend these findings by demonstrating the importance of PSGL1 as the leukocyte counter-receptor in mediating mixed-cell thromboxane production, and by suggesting that transcellular arachidonic acid metabolism may proceed in whole blood and when initiated by platelet activation instead of leukocyte activation.

We found no evidence that leukocytes inhibited platelet-mediated functions using whole blood and PRP, and when cell activation was initiated by platelet-specific agonists. Our observation that leukocytes enhanced platelet-mediated aggregation is similar to that of another study in which aggregation was assessed by the whole-blood impedance technique. 11Methodologic differences between this and other studies may account for discordant findings, with many previous studies performed in samples washed free of erythrocytes or plasma 10,12,13,21and using leukocyte activators in aggregation studies. 11,21,23In addition to removing elements capable of modulating platelet responses, washing steps may introduce artefactual platelet or leukocyte activation, because both types of cells are exquisitely sensitive to in vitro  manipulation. 16By reducing in vitro  cell preparation steps (e.g. , whole-blood aggregation, avoidance of platelet washing, etc.) we attempted to limit artefactual cell activation before initiating aggregation with platelet-specific agonists. Thus, although leukocytes augmented platelet agonist–induced aggregation and secretion in our studies, it remains possible that different effects could occur in vitro  or in vivo  if cell activation were induced by different agonists or if leukocyte activation preceded platelet activation. The importance of the sequence of cell activation is underscored by the studies of Lorant et al.  28that demonstrated a rapid decrease in leukocyte adherence to P-selectin after leukocyte activation.

The ability of leukocytes to modulate platelet responses has implications for in vitro  studies of platelet function and in vivo  hemostasis. Data from this study support the conclusion of other investigators that hemostasis is regulated by multiple blood cellular elements 29and suggest that whole-blood impedance aggregometry as described by Cardinal and Flower 30may reflect in vivo  aggregation events better than PRP aggregometry as described by Born and Hume. 31More importantly, the ability of leukocytes to augment platelet agonist–induced aggregation and thromboxane release suggests a prothrombotic potential for platelet–leukocyte interactions in vivo . Despite conflicting reports regarding the impact of leukocytes on platelet responses, other investigators have also suggested a prothrombotic role for P-selectin–PSGL1 interactions. 13The inhibitory effect of blocking antibodies on aggregation and thromboxane release in our studies varied from 20% to 50% depending on the experimental conditions, and it is difficult to know how these differences would affect in vivo  hemostasis. Although it is difficult to extrapolate from in vitro  to in vivo  settings, experimental conditions used in our studies may parallel in vivo  arterial thrombosis when initiated by platelet agonists and suggest that platelet–leukocyte adhesion may have a prothrombotic role in such circumstances. Support for such speculation is derived from a porcine model of arterial thrombosis and thrombolysis in which PSGL1 blockade accelerated clot lysis and decreased the rate of arterial reocclusion. 32 

Extrapolation of our results to the setting of cardiopulmonary bypass is more uncertain because platelet–leukocyte interactions are more likely to occur in association with prior or simultaneous leukocyte activation, 9,33and the experiments performed were not designed to determine the impact of leukocyte activation on platelet responses. However, other investigators have reported a reduction in TRAP-induced platelet P-selectin expression and platelet–leukocyte conjugate formation after cardiopulmonary bypass 34,35and have suggested that reduced platelet–leukocyte adhesion contributes to abnormal hemostasis after cardiac surgery. 34Additional studies are needed to further characterize the hemostatic implications of platelet–leukocyte interactions under a variety of physiologic and pathophysiologic conditions.