Platelets release preformed mediators and generate eicosanoids that regulate acute hemostasis and inflammation, but these anucleate cytoplasts are not thought to synthesize proteins or cytokines, or to influence inflammatory responses over time. Interrogation of an arrayed cDNA library demonstrated that quiescent platelets contain many messenger RNAs, one of which codes for interleukin 1β precursor (pro–IL-1β). Unexpectedly, the mRNA for IL-1β and many other transcripts are constitutively present in polysomes, providing a mechanism for rapid synthesis. Platelet activation induces rapid and sustained synthesis of pro–IL-1β protein, a response that is abolished by translational inhibitors. A portion of the IL-1β is shed in its mature form in membrane microvesicles, and induces adhesiveness of human endothelial cells for neutrophils. Signal-dependent synthesis of an active cytokine over several hours indicates that platelets may have previously unrecognized roles in inflammation and vascular injury. Inhibition of β3 integrin engagement markedly attenuated the synthesis of IL-1β, identifying a new link between the coagulation and inflammatory cascades, and suggesting that antithrombotic therapies may also have novel antiinflammatory effects.
Platelets are anucleate blood cells that are required for hemostasis, and also participate in pathologic thrombotic syndromes. Additionally, platelets have inflammatory actions and are a rich source of chemokines, cytokines, and growth factors that are preformed and packaged in storage granules (Klinger, 1997). After platelet activation, these mediators influence wound surveillance, repair, and vascular remodeling by signaling target cells and inducing leukocyte accumulation (Weyrich et al., 1996; Mannaioni et al., 1997; Barry and FitzGerald, 1999). Activated platelets also synthesize eicosanoids from arachidonic acid released from membrane phospholipids, a process that contributes to inflammation and thrombosis (Barry et al., 1997, 1999; Barry and FitzGerald, 1999). Secretion of granular proteins and eicosanoid generation are rapid processes that occur within minutes after platelet activation.
Platelets are thought to play a minor role, if any, in ongoing inflammation or vascular injury and repair beyond initial adhesion, secretion, and eicosanoid synthesis. However, activation of platelets does not appear to decrease their life span (Reimers et al., 1976; Michelson et al., 1996; Berger et al., 1998), and thrombogenic adhesive platelets remain functional (Berger et al., 1998). This observation suggests that activated platelets may contribute to thrombotic and inflammatory cascades for much longer periods than previously believed. Consistent with this notion, we recently demonstrated that activated platelets translate constitutive mRNAs into proteins, and that this process continues for hours after stimulation with thrombin- or integrin-mediated adhesion (Weyrich et al., 1998; Pabla et al., 1999). In this report, we use a model of platelet–fibrin clot formation to demonstrate for the first time that activated platelets synthesize an inflammatory cytokine, interleukin (IL)*-1β. Accumulation of IL-1β is prolonged, attenuated by blocking engagement of β3 integrins, and mediates signaling of endothelial cells with consequent neutrophil (polymorphonuclear leukocyte [PMN]) adhesion. It was reported earlier that stimulated platelets possess IL-1β activity, but they were not thought to actively synthesize this inflammatory protein (Hawrylowicz et al., 1989; Loppnow et al., 1998). Regulated cytokine synthesis establishes the potential for previously unrecognized biologic roles for platelets in inflammation and vascular injury.
Results And Discussion
The mRNA for IL-1β and other transcripts are associated with the polysomes of resting platelets
We extracted mRNA from resting platelets and identified constitutively present transcripts using an arrayed cDNA library. Interrogation of 588 cDNAs demonstrated that numerous mRNAs are present in resting platelets (Fig. 1 A). Table I is an abbreviated list of these mRNAs, highlighting transcripts for which the corresponding protein has been reported in platelets. The hybridization patterns were similar among different platelet donors (unpublished data) and identical when mRNA from resting platelets and platelets activated with thrombin were examined, indicating that as expected, there was no new transcription (unpublished data). The mRNA for IL-1β was one of the constitutive transcripts in unstimulated platelets (Fig. 1, A and B; Table I). Leukocyte-specific transcripts were not detected, indicating that platelets were the source of IL-1β mRNA and protein (see below) in our assays (Fig. 1 B; Table I).
We examined the monosome and polysome fractions from platelets and found that ribosomal profiles for resting and activated cells paralleled each other, and that ribosomal RNA is primarily located in monosomal fractions. In contrast, IL-1β transcripts were located in the polysomes of resting and activated platelets (Fig. 1 C), as were many other mRNAs (unpublished data). Thrombin activation did not alter the distribution of IL-1β mRNA in monosomes and polysomes (Fig. 1 C and unpublished data). These results were unexpected, as a shift from monosomes to polysomes is characteristic of translational control at the level of polypeptide initiation, an important posttranscriptional regulatory mechanism (Kaspar and Gehrke, 1994; Sachs et al., 1997). Thus, platelets have translational checkpoints and control mechanisms beyond chain initiation as reported for Nanos in Drosophila embryos (Clark et al., 2000), and for IL-1β in mononuclear leukocytes under certain conditions of stimulation (Kaspar and Gehrke, 1994). Such translational control mechanisms may be specialized for rapid synthesis of critical transcripts (Clark et al., 2000).
IL-1β mRNA is translated into protein by activated platelets during fibrin clot formation
Next, we examined expression of the IL-1β gene product in resting platelets by immunocytochemistry, and found neither the precursor nor the mature form (Fig. 2, A, top left, B, and Fig. 3). In contrast, IL-1β was robustly detected in complexes of platelets and fibrin after activation with thrombin in the presence of fibrinogen, a model of clot formation and retraction (Fig. 2 A, top right). The result is consistent with previous reports that IL-1β activity is present in stimulated platelets (Hawrylowicz et al., 1989; Loppnow et al., 1998). Pretreatment of platelets with puromycin, a translational inhibitor, completely inhibited IL-1β synthesis (Fig. 2, A, bottom left, B, and C). However, platelet–fibrin clumps formed in response to thrombin, indicating that puromycin was not nonspecifically toxic. Similarly, staining of parallel samples for P-selectin, which is constitutively stored in α-granules, was unchanged by pretreatment with puromycin, indicating that this translational inhibitor did not alter the detection of resident proteins (Fig. 2 A, bottom right). When platelets were stimulated with thrombin in the presence of fibrinogen, IL-1β and P-selectin were associated with the fibrin lattice (Fig. 2 A), consistent with a previous report for P-selectin distribution (Siljander et al., 1996) (also see below).
We confirmed that IL-1β is synthesized by activated platelets by incubating them with [35S]methionine, stimulating them with thrombin, and immunoprecipitating labeled proteins with an antibody that preferentially recognizes IL-1β precursor (pro–IL-1β). This process yielded a 32-kD band, which coincides with the expected mass of pro–IL-1β (Fig. 2 C). This labeled product was absent in thrombin-stimulated platelets that were pretreated with puromycin. Puromycin also completely inhibited accumulation of pro– and mature IL-1β when measured by ELISA, indicating that it inhibits synthesis and not conversion of pro–IL-1β into its mature form (Fig. 2 B).
We next measured IL-1β levels in resting and activated platelets over an extended time period. Accumulation of pro–IL-1β was sustained over hours and was followed by processing of the precursor into its mature, active form (Fig. 3 A). Pro–IL-1β was detectable in thrombin-stimulated platelets within 30 min (unpublished data), a rapid synthetic response that may result from localization of IL-1β transcripts in polysomes in resting platelets (see above) (Clark et al., 2000). Platelets synthesized IL-1β in every experiment (n > 30), although the magnitude was variable among donors (unpublished data). This variability was not likely to be due to contaminating leukocytes, as we did not detect macrophage colony stimulating factor (M-CSF) (60 ± 13 vs. 69 ± 1 pg/ml in control vs. activated cells [18 h], respectively), a monocyte-derived cytokine. In addition, we detected no increase in IL-1β protein levels when lymphocytes or monocytes were added to isolated platelets, concentrations of which were 10-fold greater than the trace numbers of leukocytes that were sometimes detected in our preparations (unpublished data). Platelet-activating factor (PAF) also induced IL-1β synthesis (Fig. 3 B) as did ADP, collagen, and epinephrine (Fig. 3 C). Together, Figs. 1–3 demonstrate that platelets synthesize an important cytokine, IL-1β, and have the potential to influence inflammatory events over several hours by activation-induced translation of a cytokine from constitutive mRNA.
IL-1β synthesized by activated platelets signals endothelial cells and induces PMN adhesion
Approximately 60% of mature IL-1β is retained within the platelets or the fibrin mesh that surrounds aggregated platelets (Fig. 2 A). The remaining IL-1β is found in the extracellular milieu. Fractionation of the supernatant revealed that half of the IL-1β was associated with platelet microvesicles, a finding confirmed by assay of β3 integrins and P-selectin in the same samples (Fig. 4 A and unpublished data). The remaining IL-1β in the supernatant fraction was completely soluble (Fig. 4 A). Soluble and microvesicle-associated IL-1β delivered outside-in signals to human endothelial cells and induced their adhesiveness for neutrophils (Fig. 4 B and unpublished data), a response dependent on synthesis of E-selectin, intercellular adhesion molecule 1, and chemokines (McEver et al., 1998). However, IL-1β exported in microvesicles was at least fivefold more potent than soluble IL-1β when equivalent volumes of sample were tested (unpublished data), suggesting that other inflammatory mediators packaged in microvesicles (Barry et al., 1999) may act in concert with IL-1β to induce endothelial adhesiveness. In this regard, we recently found that IL-1β acts in concert with P-selectin to induce expression of cyclooxygenase (COX)-2 when human monocytes interact with activated platelets (unpublished data). Recombinant IL-1β receptor antagonist inhibited the response to platelet-derived IL-1β and purified IL-1β, but did not inhibit adhesiveness induced by tumor necrosis factor α (Fig. 4 B and unpublished data). These data indicate that IL-1β synthesized by activated platelets induces inflammatory responses of target endothelial cells.
β3 integrins regulate IL-1β synthesis by platelets
In this model of clot formation and maturation, IL-1β is synthesized by platelets as they associate with the fibrin network (Fig. 2 A). Therefore, we pretreated platelets with mAb c7E3, an antibody that interrupts engagement of integrins αIIbβ3 and αVβ3 and prevents the retraction of the fibrin mesh (Coller, 1985; Pabla et al., 1999). We found that pretreatment with c7E3 inhibited clot retraction, whereas a panel of control antibodies had no effect (Fig. 4 C, inset, and unpublished data). In addition, IL-1β levels were markedly reduced when platelets were pretreated with c7E3, compared with those treated with a control antibody in parallel (Fig. 4 C). These results indicate that β3 integrins transmit outside-in signals to translational mechanisms that regulate IL-1β synthesis, and suggest that some of the therapeutic benefits of blocking β3 integrin engagement may result from interrupting consequent inflammatory responses.
Although the exact mechanisms are not yet clear, integrins may regulate translation in platelets at checkpoints independent of initial ribosomal mRNA interactions, as IL-1β transcripts are constitutively present in platelet polysomes (see above). In parallel experiments we have found that translational control proteins are physically separate from mRNAs in resting platelets, and that integrin engagement redistributes these proteins to mRNA-rich areas (Lindemann et al., 2001). Dissection of mechanisms of signal-dependent synthesis of specific gene products such as IL-1β by platelets may yield new insights into translational control in specialized cells (Weyrich et al., 1998; Clark et al., 2000) and reveal new activities of platelets in atherogenesis, tissue remodeling, and tumor angiogenesis (Weyrich et al., 1996; Barry et al., 1997; Mannaioni et al., 1997; Pinedo et al., 1998).
Materials And Methods
Washed platelets were isolated from acid-citrate-dextrose–anticoagulated human blood as described previously (Weyrich et al., 1996). Platelets (5 × 108 total) were resuspended in M199 medium containing 100 μg/ml of fibrinogen, and were left quiescent or stimulated with thrombin, PAF, ADP, collagen, or epinephrine for 0.5–18 h. In selected experiments, platelets were pretreated with mAb c7E3 (ReoPro [Abciximab]; Lilly), control antibodies, or puromycin. After each time point, the platelets were pelleted by centrifugation and the supernatant was collected. The pellet was lysed in 0.5 ml of RIPA buffer. The lysate and supernatant were stored in aliquots at −20°C until further use.
Polysomes were isolated from platelets using minor modifications of published protocols (Kaspar and Gehrke, 1994). In brief, platelets (7.5 × 109) were lysed in a high-salt buffer to remove polysomes from the cytoskeleton (200 mM Tris, 520 mM KCl, 30 mM MgCl2, 4% Triton X-100, and 200 mM sucrose, pH 9.5). The cells were passed three times through a 21-gauge tuberculin syringe, followed by a brief centrifugation to remove all insoluble material and heparin (10 mg/ml), and NaCl (150 mM) was added to the lysate. Mitochondria were removed by further centrifugation (5 min, 14,000 g) and the resulting supernatant was placed on a 4-ml sucrose gradient (0.5–2 M) and centrifuged at 4°C for 2 h (43,700 rpm) using an SW55 Ti swinging bucket rotor (Beckman Coulter).
Sucrose gradients were fractionated using an ISCO UA-6 254-nm continuous flow chamber into 760-μl aliquots. Total RNA in each fraction was isolated for analysis of ribosomal RNA and mRNA for IL-1β and glutaraldehyde 3-phosphate dehydrogenase (GAPDH) as described previously (Weyrich et al., 1998). The total mRNA pool in platelets was also determined by interrogation of cDNA libraries according to the manufacturer's instructions (array 7742-1; CLONTECH Laboratories, Inc.). In selected experiments, sucrose cushions were used to isolate mRNAs associated with polysomes of resting and thrombin-activated platelets as described previously (Davies and Abe, 1995). The sucrose cushion was based on ribosomal profiles from complete sucrose gradients that were fractionated with the ISCO continuous flow chamber (see above).
ELISAs for pro–IL-1β, mature IL-1β, and M-CSF were performed according to the manufacturer (R&D Systems; Bender Med Systems). The pro– and mature IL-1β ELISAs did not cross react with one another.
Immunoprecipitation of [35S]methionine–labeled proteins
Immunoprecipitation of metabolically labeled IL-1β was conducted as described previously (Weyrich et al., 1998) using an antibody that preferentially recognizes pro–IL-1β (Research Diagnostics, Inc.).
Immunolocalization of platelet-derived IL-1β
Isolation of platelet microvesicles
Platelets were activated with PAF or thrombin as indicated above. After 8 h, the platelet suspensions were centrifuged for 4 min at 15,500 g. The pellet was lysed in RIPA buffer and the supernatant was collected and recentrifuged at 100,000 g in a Ti80 rotor (Beckman Coulter) for 90 min at 4°C to collect platelet microvesicles (George et al., 1982). The resulting pellet from high-speed centrifugation of the supernatant pool, which contained microvesicles, was lysed in RIPA buffer. Mature IL-1β was assayed in the pellet lysates, the microvesicle fraction, and the soluble supernatant by ELISA. The same samples were also examined by Western analysis using mAb S12 to detect P-selectin in the microvesicle fraction.
PMN adhesion to activated endothelial cells
Human umbilical endothelial cells were incubated with platelet supernatants or microvesicle fractions for 4 h in the presence or absence of a recombinant IL-1β receptor antagonist (R&D Systems). In parallel, human umbilical endothelial cells were also stimulated with tumor necrosis factor α (100 U/ml), recombinant active IL-1β, or recombinant pro–IL-1β. After 4 h, the human umbilical endothelial cells were thoroughly washed and PMN adhesion to the cell surface was quantitated as described previously in detail (Lorant et al., 1991; Lindemann et al., 2000).
Online supplemental material
Fig. S1(available at http://www.jcb.org/cgi/content/full/154/3/485/DC1) demonstrates that immunostaining of platelet-derived IL-1β is quenched when the antibodies are preincubated with purified IL-1β protein. Platelets were prepared for immunocytochemical detection of IL-1β protein as described in Fig. 2 A, the exception being that the anti–IL-1β antibody was quenched with excess IL-1β protein. The left panel depicts staining for IL-1β with quenched antibody. The middle panel shows the same cells stained for actin. The right panel is an overlay of the two panels. Note that staining for IL-1β is absent in the left and right panels.
We thank Donnie Benson, Jennifer Eyre, and Margaret Vogel for excellent technical assistance, Michelle Bills for preparation of the manuscript, Diana Lim for preparation of the figures, and our colleagues for helpful comments and critical reading of the manuscript.
This work was supported by grants from the National Institutes of Health (HL56713 to A.S. Weyrich, HL44525 to G.A. Zimmerman), the Atorvastatin Research Awards (A.S. Weyrich), the Western State Affiliate of the American Heart Association (S. Lindemann, AHA Grant 0020030Y), the American Heart Association (D.A. Dixon, AHA Grant 9930102N), and a National Institutes of Health Special Center of Research in Adult Respiratory Distress Syndrome (P50 HL50153).
The online version of this article contains supplemental material.
Abbreviations used in this paper: COX, cyclooxygenase; GAPDH, glutaraldehyde 3-phosphate dehydrogenase; IL, interleukin; M-CSF, macrophage colony stimulating factor; PAF, platelet-activating factor; PMN, polymorphonuclear leukocyte; pro–IL-1β, IL-1β precursor.