Abstract
Bacterial LPS induces rapid thrombocytopenia, hypotension, and sepsis. Although growing evidence suggests that platelet activation plays a critical role in LPS-induced thrombocytopenia and tissue damage, the mechanism of LPS-mediated platelet activation is unclear. In this study, we show that LPS stimulates platelet secretion of dense and α granules as indicated by ATP release and P-selectin expression, and thus enhances platelet activation induced by low concentrations of platelet agonists. Platelets express components of the LPS receptor-signaling complex, including TLR (TLR4), CD14, MD2, and MyD88, and the effect of LPS on platelet activation was abolished by an anti-TLR4-blocking Ab or TLR4 knockout, suggesting that the effect of LPS on platelet aggregation requires the TLR4 pathway. Furthermore, LPS-potentiated thrombin- and collagen-induced platelet aggregation and FeCl3-induced thrombus formation were abolished in MyD88 knockout mice. LPS also induced cGMP elevation and the stimulatory effect of LPS on platelet aggregation was abolished by inhibitors of NO synthase and the cGMP-dependent protein kinase (PKG). LPS-induced cGMP elevation was inhibited by an anti-TLR4 Ab or by TLR4 deficiency, suggesting that activation of the cGMP/protein kinase G pathway by LPS involves the TLR4 pathway. Taken together, our data indicate that LPS stimulates platelet secretion and potentiates platelet aggregation through a TLR4/MyD88- and cGMP/PKG-dependent pathway.
Bacteria-derived LPS plays a fundamental role in sepsis. Following its release into the bloodstream, LPS forms a complex with LPS-binding protein (1, 2). This complex binds to CD14, a high-affinity LPS receptor present on the surface of several types of cells (3), and induces cellular responses through TLR4, the first-discovered mammalian homolog of Drosophila Toll (4, 5). Recognition of LPS by TLR4 requires an extracellular adaptor protein, MD2. TLR4-induced intracellular signaling requires multiple adaptor proteins, including MyD88, the MyD88 adaptor-like protein, Toll/IL-1R-containing adaptor molecule (TRIF), and TRIF-related adaptor molecule (6). Although TLR4 is the principal signal transducer for most types of LPS and TLR2 is a major receptor for lipoteichoic acid from Gram-positive bacteria, TLR2 is also a signal transducer for at least some Gram-negative bacteria (7, 8, 9, 10). TLR2 is expressed in platelets (11, 12). Signal transduction by TLR2 also requires the MyD88 pathway. Whether the entire LPS receptor-signaling complex is physically and functionally present in blood platelets remains unclear.
Patients with sepsis are often thrombocytopenic and i.v. injection of LPS in mice also induces rapid thrombocytopenia (13, 14). Under these conditions, platelet aggregates are found in lung and liver microvasculature (15, 16). Recently, TLR4 has been found to be expressed in platelets and play important roles in LPS-induced thrombocytopenia (17, 18). LPS enhances microvascular thrombosis in wild-type mice, but not TLR4- deficient mice (19). Furthermore, infused platelets from wild-type but not from TLR4 knockout mice accumulate in the lungs of LPS-treated wild-type mice (17). Despite these in vivo data, several studies suggest that LPS does not affect human platelet function, while other studies report that LPS inhibits human platelets in vitro (19, 20, 21, 22). Stahl et al. (23) recently reported that LPS activates the ligand binding function of integrin αIIbβ3. Thus, it remains controversial whether LPS directly induces platelet activation. Furthermore, it is unclear how TLR4 transmits LPS signals leading to platelet activation and whether platelets express the necessary components of the TLR4 signaling complex. In this study, our experimental data suggest that LPS primarily stimulates platelet secretion of granule contents and thus enhances integrin-dependent platelet aggregation induced by multiple stimuli. We show that the components of the TLR4-MyD88 receptor-signaling complex required for LPS signaling are present in platelets. Importantly, we demonstrate that LPS-mediated platelet activation requires TLR4/MyD88-dependent activation of the NO and cGMP-dependent protein kinase pathway.
Materials and Methods
Reagents
LPS (Escherichia coli 0111:B4 and 055:B5) and the protein kinase G (PKG)3 inhibitor Rp-pCPT-cGMPS were purchased from Calbiochem. FeCl3, LPS (E. coli 0127:B8), the purified LPS (E. coli 0111:B4, catalog no. L4391), a mAb against β-actin (AC74), and Nω-nitro-l-arginine methyl ester (L-NAME), were from Sigma-Aldrich. Kdo2-Lipid A was a generous gift from Dr. A. J. Morris (University of Kentucky, Lexington, KY). Polyclonal Abs against human TLR4 (H-80) or MyD88 (HFL-296) and mAbs against CD14 (UCH-M1) or a complex of TLR4-MD2 (HTA125) were purchased from Santa Cruz Biotechnology. Blocking mAbs against human TLR4 (HTA125) or TLR2 (clone T2.5) were from eBioscience. Anti-thrombin was from Enzyme Research Laboratories. Collagen and luciferin-luciferase reagent were purchased from Chronolog. MyD88 knockout mice were obtained from S. Akira (Research Institute for Microbial Diseases, Osaka University, Osaka, Japan) (24, 25). MyD88 knockout and wild-type mice obtained from heterozygous breeding were used for the experiments. The TLR4-deficient mouse strain C57BL/10ScCr and wild-type control C57BL/10J were from The Jackson laboratory. Mice were bred and maintained in the University of Illinois Animal Care Facility following institutional and National Institutes of Health guidelines after approval by the Animal Care Committee.
Preparation of washed platelets
Fresh blood from healthy volunteers was anticoagulated with one-seventh volume of acid-citrate dextrose (ACD; 85 mM trisodium citrate, 110 mM dextrose, and 78 mM citric acid) as described previously (26). For the preparation of mouse platelets, 6- to 8-wk-old mice of either sex were anesthetized with an i.p. injection of pentobarbital and blood was drawn from the inferior vena cava. Blood from five to six mice of either genotype was pooled using one-seventh volume of ACD (85 mM trisodium citrate, 83 mM dextrose, and 21 mM citric acid) as anticoagulant and platelets were isolated by differential centrifugation as previously described (27). Platelets were washed twice with CGS buffer (0.12 M sodium chloride, 0.0129 M trisodium citrate, and 0.03 d-glucose (pH 6.5)), resuspended in freshly made but not sterile Tyrode’s buffer, and allowed to rest for at least 1 h at 37°C before use (28). In platelet aggregation experiments using platelet-rich plasma (PRP), one-tenth volume of 3.8% trisodium citrate was used as anticoagulant.
Platelet aggregation and secretion
Platelet aggregation was measured in a turbidometric platelet aggregometer (Chronolog) at 37°C with stirring (1000 rpm) (26). To examine the effect of LPS on platelet activation, various concentrations of LPS were added simultaneously with or without low concentrations of platelet agonists to induce platelet aggregation. ATP release was monitored in parallel with platelet aggregation. To examine the effect of LPS on ATP release, washed human platelets were incubated with LPS at room temperature for 10 min; ATP in the supernatant was measured by addition of luciferin-luciferase reagent. Platelets were also incubated with 0.025 U/ml thrombin at room temperature for 10 min as a positive control. Quantification was performed using the ATP standard. Statistical analysis was performed using the t test.
P-selectin expression
Washed human platelets were incubated with LPS or thrombin in the presence of a monoclonal anti-human P-selectin Ab, SZ51 (29), or control mouse IgG for 30 min at 22°C. After washed once with PBS, the platelets were incubated with a FITC-conjugated rabbit anti-mouse IgG Ab. P-selectin expression was analyzed using a FACSCalibur flow cytometer (BD Biosciences).
Detection of TLR4 and MyD88 expression in platelets by Western blot
Washed human platelets or leukocytes were solubilized in SDS-PAGE sample buffer. Western blot was performed as described previously (27) with polyclonal anti-TLR4 or MyD88 Abs.
Detection of TLR4, MD2, and CD14 on the platelet surface by flow cytometry
Human platelets were incubated with a polyclonal Ab against human TLR4 (H-80), or a mAb (HTA125) that recognizes the TLR4-MD2 complex, or a mAb against human CD14 (UCH-M1) for 30 min at 22°C. Platelets were also incubated with control IgG for 30 min. After washed once with PBS, platelets were incubated with a FITC-conjugated anti-rabbit or mouse IgG Ab. TLR4, TLR4/MD2, or CD14 expression was analyzed using a FACSCalibur flow cytometer.
Measurement of platelet cGMP levels
Washed human platelets, resuspended in modified Tyrode’s buffer, were incubated with various concentrations of LPS for 5 min in a platelet aggregometer with stirring (1000 rpm) at 37°C. Washed platelets were incubated with thrombin (0.02 U/ml), LPS, or thrombin plus LPS at 37°C in a platelet aggregometer for 5 min. Washed human platelets were also incubated with various concentrations of LPS for 2 min at 22°C. The reaction was stopped by addition of ice-cold 12% (w/v) trichloroacetic acid. The samples were then centrifuged at 2000 × g for 15 min at 4°C and the supernatant was extracted four times with 5 vol of water-saturated diethyl ether. The samples were lyophilized and cGMP concentrations were determined using a cGMP enzyme immunoassay kit from Amersham-Pharmacia Biotech (27). Results are expressed as mean ± SD. Statistical significance between groups was determined by the t test.
In vivo thrombosis
FeCl3-injured carotid artery thrombus formation was performed as described previously (30). Briefly, 0.5 μl of 4% FeCl3 was applied to a filter paper disc (1-mm diameter) that was immediately placed on top of the carotid artery for 3 min to induce vessel damage. LPS (0.5 mg/kg weight in 0.1 ml of saline) or saline was injected into the fundus oculi of the mice 1 min after the initiation of carotid artery injury. Time to occlusion was calculated as a difference in time between the removal of the filter paper and stable occlusion (no blood flow for 2 min). Statistical analysis was performed using the Mann-Whitney U test for the evaluation of differences in median occlusion time.
Results
LPS enhances platelet aggregation induced by low concentrations of platelet agonists
To determine whether LPS directly induces platelet activation, we examined the effect of LPS on platelet aggregation using washed human platelets. When used alone, LPS (0111:B4, from Calbiochem, contaminated ≤2% proteins) up to 100 μg/ml failed to induce aggregation (Fig. 1⇓A). However, when added simultaneously with subthreshold concentrations of the platelet agonists thrombin or collagen, LPS significantly enhanced platelet aggregation (Fig. 1⇓, B–D). These results were confirmed with purified LPS (0111:B4, Sigma-Aldrich catalog no. L4391, contaminated proteins <1%) (data not shown). Recently, a well-defined and highly pure LPS, Kdo2-Lipid A, has been shown to activate the TLR4 pathway with a similar bioactivity as LPS (31, 32). To further verify the stimulatory effect of LPS on platelet activation and exclude the possibility that the stimulatory effect of LPS on platelet activation is caused by contaminated proteins, we examined the effect of Kdo2-Lipid A on platelet activation. Similar to LPS, Kdo2-Lipid A enhanced platelet aggregation induced by low-dose thrombin (Fig. 1⇓E). Thus, LPS can directly interact with platelets and plays a stimulatory role that synergizes with low concentrations of platelet agonists to induce platelet aggregation.
LPS potentiates platelet aggregation. A, LPS (0111:B4, 100 μg/ml) was added to washed human platelets and incubated in the aggregometer. B, A subthreshold concentration of thrombin (0.0185 U/ml) was added to washed human platelets immediately followed by addition of various concentrations of LPS (0111:B4) or buffer. C, A subthreshold concentration of collagen (0.3 μg/ml) was added to the washed human platelets followed by addition of LPS (0111:B4, 5 μg/ml) or buffer. D, Quantitative data (mean ± SD) from four experiments described in B (LPS 10 μg/ml) and C. E, A subthreshold concentration of thrombin (0.0185 U/ml) was added to washed human platelets immediately followed by addition of Kdo2-Lipid A (10 μg/ml) or buffer. F, PRP from a healthy human donor was added with a collagen (0.6 μg/ml) immediately followed by addition of different sources of LPS (1 μg/ml) or buffer to induce platelet aggregation.
To determine whether LPS is also able to promote platelet aggregation in the presence of plasma, we examined the effect of LPS on platelet aggregation in human PRP. We found that LPS also significantly enhanced collagen-induced platelet aggregation in PRP (Fig. 1⇑F). The concentration of LPS required for potentiating platelet aggregation in plasma (1 μg/ml) is significantly lower than that required for washed platelets (5–10 μg/ml). We also compared LPS from different sources and found that their potency in potentiating collagen-induced platelet aggregation varies with the order of 0127:B8>0111:B4>055:B5 (Fig. 1⇑F).
The effects of LPS on platelet granule secretion
Platelet secretion plays a critical role in potentiating platelet activation induced by low-dose agonists. To determine whether platelet secretion accounted for the potentiating effect of LPS on platelet aggregation, we examined LPS-induced ATP release in the human platelets, which indicates the secretion of dense granules. LPS alone is sufficient to induce the release of ATP (ATP concentrations were 573.33 ± 62.22 (basal level) vs 986.67 ± 97.78 for 0111:B4 and 1186.67 ± 62.22 for 0127:B8 nmol/L ATP in the supernatant of solvent-treated platelets vs LPS-stimulated platelets) (Fig. 2⇓A), although the amount of ATP release induced by LPS stimulation is much lower than that induced by platelet agonists, such as thrombin (thrombin, at 0.025 U/ml, induced an ∼38-fold increase of ATP release than the basal level) (supplemental Fig. 14). The low level secretion induced by LPS stimulation may explain why LPS alone is insufficient to induce platelet aggregation. Similarly, LPS alone induced P-selectin expression in human platelets, indicating that LPS also stimulates α granule secretion (Fig. 2⇓, B and C). Thus, LPS stimulates platelet secretion of both dense and α granules.
LPS stimulates platelet secretion. A, Washed human platelets were incubated with 1 μg/ml LPS at 22°C for 10 min. The release of ATP into the platelet supernatant was determined by a luciferin-luciferase assay. B and C, Washed human platelets were incubated with LPS (1 μg/ml) or thrombin (0.025 U/ml) in the presence of a monoclonal anti-human P-selectin Ab, SZ51, for 30 min at room temperature. After washing, platelets were further incubated with a FITC-conjugated anti-mouse IgG. Expression of P-selectin was analyzed by flow cytometry. Quantitative results from three experiments are expressed as the P-selectin expression index (fluorescence intensity (mean) of platelets stimulated with an agonist/fluorescence intensity (mean) of unstimulated platelets) (B). Data from a representative experiment are shown in C.
Platelets express components of the LPS receptor-signal complex, TLR4, MD2, CD14, and MyD88
Platelets have been previously reported to express TLR4 (12, 17, 18). Consistently, we detected TLR4 in human platelets by Western blot analysis with a rabbit anti-TLR4 polyclonal Ab (Fig. 3⇓A). The expression of TLR4 on human platelets was further confirmed by flow cytometry using the anti-TLR4 polyclonal Ab (Fig. 3⇓B). In addition, we found that other members of the TLR4 receptor complex, MD2 (Fig. 3⇓C) and CD14 (Fig. 3⇓D), are also present on the surface of platelet as analyzed by flow cytometry. Furthermore, the intracellular adaptor protein MyD88 was detected in human platelet lysates. It is important to note that the MyD88 detected in our platelet preparation is unlikely to be from contaminated white blood cells, since the MyD88 expression level in the platelet preparation was much higher than that detected in 10 times as much of the leukocytes as the contamination level (0.016% as estimated by HEMAVET HV950FS multispecies hematology analyzer) (Fig. 3⇓E). Thus, platelets express important components of the TLR4 signaling complex, including TLR4, CD14, MD2, and the adaptor protein MyD88.
Expression of LPS receptor components in platelets. A, Washed human platelets (1 × 109/ml) were solubilized in 1× SDS-PAGE sample buffer. TLR4 was detected by immunoblotting with a rabbit (Rb) anti-human TLR4 polyclonal Ab. β-Actin was detected by immunoblotting with a mouse mAb against β-actin as a positive control. B–D, Washed human platelets were incubated with a rabbit anti-human TLR4 polyclonal Ab (B) or incubated with a mouse mAb, which recognizes the human TLR4-MD2 complex (C), or with a mouse anti-human CD14 mAb (D). The platelets were alternatively incubated with rabbit IgG or mouse IgG as controls. Expression of TLR4, TLR4-MD2 complex, and CD14 was analyzed by flow cytometry. E, Washed human platelets or various concentrations of WBCs from the same donor were solubilized in 1× SDS-PAGE sample buffer. MyD88 was detected by immunoblotting with a rabbit anti-human MyD88 polyclonal Ab. β-Actin was detected by immunoblotting with a mouse mAb against β-actin as a positive control.
LPS-mediated platelet activation requires TLR4/MyD88
To assess the role of TLR4 in LPS-induced platelet responses, human platelets were preincubated with an anti-human TLR4-blocking Ab or control IgG and then exposed to a subthreshold concentration of thrombin, in the presence of LPS, to induce platelet aggregation and secretion. LPS significantly enhanced low-dose thrombin-induced platelet aggregation and ATP release in control IgG-treated platelets, but not in anti-TLR4 Ab-treated platelets (Fig. 4⇓A). The stimulatory effect of LPS on platelet aggregation and secretion was also abolished in TLR4-deficient platelets (Fig. 4⇓, B and C). These results suggest that the stimulatory effect of LPS on platelet activation is mainly TLR4 dependent. Also, LPS enhanced aggregation and secretion of wild-type mouse platelets, but not MyD88-deficient platelets (Fig. 5⇓, A and B). MyD88 plays important roles not only in TLR receptor signaling, but also in IL-1 signaling (33). It has been recently reported that LPS stimulation induces IL-1 synthesis in platelets (34). Thus, to exclude the possibility that the effect of MyD88 deficiency on LPS-promoted platelet activation results from its role in IL-1 signaling, we examined whether IL-1 is involved in LPS-potentiated platelet aggregation. LPS-potentiated platelet aggregation was not affected by a recombinant human IL-1 receptor antagonist, IL-1ra (from Imgenex) (supplemental Fig. 2), suggesting that IL-1 is not required for LPS-promoted platelet activation. Thus, these results suggest that the stimulatory role of LPS in platelet aggregation requires TLR4/MyD88 signaling.
LPS potentiates platelet activation via a TLR4-dependent signaling pathway. A, Washed human platelets were preincubated with a mouse monoclonal anti-human TLR4-blocking Ab (10 μg/ml) or control mouse IgG at 37°C for 5 min and then exposed to a subthreshold concentration of thrombin in the presence of LPS (0111:B4, 10 μg/ml) or buffer. B and C, Washed platelets from wild-type mice (C57BL/10J) (B) or TLR4-deficient mice (C) were exposed to low-dose thrombin in the presence or absence of LPS (0111:B4, 10 μg/ml) to induce platelet secretion and aggregation.
The role of MyD88 in LPS-promoted platelet activation and thrombus formation. A and B, Washed platelets (resuspended in Tyrode’s buffer at 3 × 108/ml) from wild-type (A) or MyD88-deficient mice (B) were added with low-dose thrombin, followed by addition of LPS (0111:B4, 5 μg/ml), and incubated in the aggregometer. C, FeCl3-induced carotid artery injury was performed and time to occlusive thrombosis was recorded as described in Materials and Methods. LPS or saline was injected into fundus oculi of the C57BL/6- or MyD88-deficient mice 1 min after the initiation of carotid artery injury. The occlusion time of each mouse is shown as circles. The bars represent the median occlusion time.
LPS promotes thrombus formation in vivo
To investigate the role of LPS in platelet-dependent thrombosis in vivo, we determined the effect of LPS on FeCl3-injured carotid artery thrombus formation. Fig. 5⇑C shows that the median time from injury to formation of stable occlusive thrombus in LPS-treated mice (184.0 s) is significantly shortened compared with that of control mice (775.0 s; p < 0.001). Thus, LPS significantly aggravates thrombus formation in vivo. In contrast, LPS failed to accelerate the FeCl3-induced in vivo thrombosis in MyD88 knockout mice (Fig. 5⇑C). Therefore, the TLR-associated adaptor protein MyD88 is critical to the role of LPS in accelerating FeCl3-induced arterial thrombosis.
LPS induces cGMP elevation in human platelets
The characteristics of LPS-mediated potentiation of platelet activation is similar to the recently identified stimulatory roles of NO and cGMP in platelet activation (29, 30, 35), in that both LPS and cGMP primarily stimulate secretion of platelet granules but not platelet aggregation. LPS stimulation of NO production has been shown in other cell types, resulting in the activation of soluble guanylyl cyclase and production of cGMP. To investigate whether the stimulatory effect of LPS on platelet activation involves LPS-mediated activation of the cGMP pathway, washed human platelets were stimulated with LPS and intracellular cGMP levels were measured by a solid-phase enzyme-linked immunoassay. LPS dose-dependently enhanced cGMP levels in human platelets (Fig. 6⇓, A and B). LPS-induced cGMP elevation was partially but significantly reduced in TLR4-deficient platelets (Fig. 6⇓C), suggesting the existence of both TLR4-dependent and TLR4-independent pathways mediating LPS-induced cGMP elevation.
LPS-induced platelet activation involves the cGMP/PKG pathway. A, Washed human platelets were incubated with various concentrations of LPS (0111:B4) at 22°C for 2 min. cGMP concentrations were determined using a cGMP enzyme immunoassay kit. The reaction was repeated three times using the platelets from the same donor. Results are expressed as mean ± SE. Statistical significance between groups was determined by the t test (p < 0.05 for 1 μg/ml LPS and <0.001 for 10 or 100 μg/ml LPS). The data shown here are representatives of at least three experiments from different donors. B, Washed human platelets were incubated with different sources of 100 μg/ml LPS (0111:B4 (B4), 055:B5 (B5), or 0127:B8 (B8)) at 22°C for 2 min. C, Washed platelets from TLR4-deficient mice or wild-type controls were incubated with 100 μg/ml LPS (0127:B8) for 0.5 or 2 min. D, Washed human platelets were preincubated with mouse IgG or mouse anti-TLR4 or TLR2 mAbs (10 μg/ml) for 5 min at room temperature and then incubated with 100 μg/ml LPS (0111:B4) for 2 min. Statistical significance between groups was determined by the t test (p < 0.05 for TLR4- or TLR2 Ab-treated platelets compared with IgG-treated platelets). E, Washed human platelets were preincubated with mouse IgG or an anti-TLR2-blocking Ab for 5 min at 37°C and then exposed to a low concentration of thrombin in the presence of LPS (0111:B4, 10 μg/ml) or buffer. F, Washed human platelets were preincubated with the NOS inhibitor L-NAME (1 mM) or the PKG inhibitor Rp-pCPT-cGMPS (200 μM) at 37°C for 5 min and then exposed to a subthreshold concentration of thrombin in the presence of LPS (0111:B4, 10 μg/ml) or buffer to induce platelet aggregation.
TLR2 has been shown to be expressed in platelets. To determine whether TLR2 is responsible for LPS-induced, TLR4-independent, cGMP elevation and platelet activation, we examined the effect of an anti-TLR2-blocking Ab on LPS-induced cGMP production and platelet aggregation. Fig. 6⇑D shows that LPS-induced cGMP elevation was inhibited by both anti-TLR4 and anti-TLR2 Abs. Furthermore, the effect of LPS on platelet aggregation was also blocked by the anti-TLR2 Ab (Fig. 6⇑E). Thus, these data indicate that TLR2 is also involved in LPS signaling in platelets.
PKG plays an important role in LPS-induced platelet activation
To determine whether the NO/cGMP/PKG pathway is important in LPS stimulation of platelet activation, we examined the effect of the inhibitors of NO synthase (NOS) and PKG on LPS-mediated potentiation of platelet aggregation induced by low doses of thrombin or collagen. The enhancing effect of LPS on platelet aggregation induced by low concentrations of thrombin (Fig. 6⇑F) was abolished by the NOS inhibitor l-NAME or the PKG inhibitor Rp-pCPT-cGMPS, indicating that the PKG pathway is important in LPS-mediated potentiation of platelet activation.
Discussion
In this study, we show that platelets express the necessary components of the LPS receptor-signaling complex, including CD14, TLR4/MD2, and MyD88. We have provided direct evidence that LPS stimulates platelet secretion and amplifies platelet aggregation mainly via the TLR4/MyD88- dependent mechanisms. Our data also reveal a novel signaling pathway mediating LPS-induced platelet activation, in which LPS binding to its platelet receptor induces cGMP production and activates PKG, leading to platelet secretion and thus amplification of aggregation.
We have discovered here that LPS promotes platelet activation by inducing secretion of contents of both α and dense granules and thus amplifying secretion-dependent platelet aggregation. Previously, it was known that platelets are involved in the pathogenesis of severe sepsis (36, 37, 38) and that LPS stimulates thrombosis and formation of platelet microaggregates in vivo. It is also known that LPS stimulates the interaction between platelets and neutrophils, leading to robust neutrophil activation via TLR4-dependent mechanisms (18, 39, 40). However, it is unclear whether the in vivo effect of LPS on platelet activation is due to the direct response of platelets to LPS, and the mechanisms of platelet activation and activation of leukocyte-platelet interactions in patients with sepsis is poorly understood. P-selectin has been shown to play an important role in LPS binding to platelets (23) and LPS-mediated platelet-leukocyte interaction (41). However, it is controversial whether LPS is able to induce P-selectin expression in platelets. Although P-selectin was not detected by LPS stimulation in some previous reports (12, 17, 19, 39), recent studies suggest that LPS is able to induce platelet secretion of P-selectin (34, 42). It is not clear what caused this discrepancy. In this study, we confirmed that LPS is capable of inducing P-selectin expression on platelet surfaces. Although the potency to induce P-selectin expression varies between different sources of LPS, all of the tested LPS preparations from different sources induced P-selectin expression in human platelets in our hands (data not shown). We found that the resting state of platelets after preparation is an important factor affecting the detection of LPS-induced P-selectin expression, as partially activated platelets already express substantial levels of P-selectin. Under these conditions, the LPS-induced increase in P-selectin expression is no longer detectable. Our results suggest that LPS not only induces platelet secretion from α granules, but also induces platelet secretion from dense granules. ADP, secreted from dense granules, plays an important role in inducing integrin activation and platelet aggregation mainly through its P2Y1 and P2Y12 receptors. Thus, our findings that LPS stimulates platelet secretion and synergizes with platelet agonists in platelet activation may be a mechanism that explains the role of platelets in exacerbating LPS-induced sepsis. A recent report indicates that coagulation initiated by cytokines plays an important role in amplifying the inflammatory response in sepsis (43). It has also been reported that stimulation of platelets with LPS induces release of cytokines such as IL 8, epidermal growth factor, and TGF-β (42). Thus, platelet activation induced by LPS during sepsis may not only contribute to coagulation, but also contribute to inflammation during sepsis.
We conclude that LPS stimulates platelet secretion and aggregation via the TLR4/MyD88 receptor-signaling complex. This conclusion is supported by our data that TLR4, MD2, CD14, and MyD88 are all present in platelets and that the stimulatory effect of LPS is abolished by an anti-TLR4 Ab and in TLR4- or MyD88-deficient platelets. Our results are consistent with the findings that TLR4 is expressed on platelets (12, 17, 18), but further show that other necessary components of the TLR4 receptor-signaling complex are also present in platelets. We found that the high-affinity LPS receptor CD14 is present on platelet surfaces, which contradicts previous reports that CD14 is not expressed in platelets (34). Although the reason for this discrepancy is not clear, we cannot exclude the possibility that the detected CD14 may be from plasma CD14 since plasma contains microgram levels of soluble CD14. Our results agree with previous findings that platelet TLR4 plays an important role in LPS-induced thrombocytopenia and thrombus formation in vivo (17). Furthermore, the concentration of LPS required for promoting platelet aggregation in the presence of plasma is only 1 μg/ml in our studies, which is known to be achievable in patients with severe sepsis (1–10 μg/ml) (44), in whom disseminated intravascular coagulation and severe thrombocytopenia are more often seen (LPS is released from the local infected area, thus the local LPS concentration could be even higher than detected in systemic concentrations). Our results, along with previous findings, indicate that the TLR4/MyD88-signaling complex in platelets is important in the development of thrombotic complications in sepsis and in the pathogenesis of sepsis. We found that the concentration of LPS required for potentiating platelet aggregation in plasma (1 μg/ml) is significantly lower than that required for washed platelets (5–10 μg/ml) (Figs. 1⇑ and 2⇑A), suggesting the presence of a plasma cofactor sensitizing platelets to low concentrations of LPS.
We conclude that LPS stimulates platelets via the cGMP-dependent signaling pathway. This is supported by the data that LPS induces cGMP elevation in platelets and that the stimulatory effect of LPS on platelets is inhibited by either the NOS inhibitor l-NAME or the PKG inhibitor Rp-pCPT-cGMPS. We have recently reported that the NO/cGMP/PKG pathway plays biphasic roles in platelet activation. At high concentrations, NO and cGMP inhibit platelets via a mechanism that involves cGMP-dependent protein kinase I (45, 46) and cAMP (47). However, during platelet activation, endogenously produced NO at low concentrations and cGMP can promote platelet activation (27, 30, 48). Importantly, the NO/cGMP/PKG pathway is similar to the TLR4-MyD88 complex in that both the NO/cGMP pathway and LPS stimulate secretion of platelet granules and potentiate platelet aggregation in the presence of low concentrations of agonists, but alone are not sufficient to cause platelet aggregation (29). Therefore, we have discovered a novel pathway of LPS-mediated platelet activation. In this pathway, LPS, by interacting with the TLR4-MyD88 receptor-signaling complex, activates the NO/cGMP pathway, stimulating platelet granule secretion, leading to potentiation and amplification of platelet activation and aggregation. Although the potency to induce cGMP elevation is different between different sources of LPS, all tested LPS preparations increased cGMP concentration in platelets. Consistently, 0127:B8 is the most potent in the promotion of platelet aggregation and induction of maximal cGMP production. In contrast, 055:B5 was the least potent in the promotion of platelet aggregation and induced the least amount of cGMP. Our data that LPS-induced cGMP elevation is only partially inhibited in TLR4-deficient platelets or anti-TLR4 Ab- treated human platelets suggest that there is a TLR4-independent signaling pathway mediating LPS-induced activation of the cGMP/PKG pathway. Platelets express functional TLR2 (11, 12). In addition to TLR4, TLR2 has been shown to be a signal transducer for some sources of LPS (7, 8, 9, 10). We found that LPS-induced cGMP elevation is significantly inhibited by an anti-TLR2-blocking Ab and that the potentiation of platelet aggregation by LPS is inhibited by anti-TLR2 Ab (Fig. 6⇑). These results indicate that TLR2 is also involved in LPS-mediated cGMP elevation and platelet activation. TLR2 signaling also requires MyD88. Therefore, it is not surprising that LPS-promoted platelet activation and thrombosis is abolished in MyD88-deficient mice. It is interesting to note that our finding that LPS stimulates platelet activation via the cGMP/PKG pathway may provide an explanation of previous controversy whether LPS stimulates or inhibits platelet activation. Previous studies showing the inhibitory effect of LPS on platelets allowed prolonged preincubation of high doses of LPS with platelets (19, 22). Since LPS induces cGMP elevation, it is possible that prolonged preincubation of high doses of LPS inhibited platelet activation in a way similar to the inhibitory effect of preincubation of high concentrations of cGMP. We show that LPS promotes platelet aggregation when added immediately before or after the agonist stimulation, which is also similar to the way endogenous cGMP stimulates platelet activation.
Acknowledgments
We thank Dr. Feng Qian for excellent technical assistance.
Disclosures
The authors have no financial conflict of interest.
Footnotes
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
↵1 This work is supported by American Heart Association National Scientist Development Grant 0430095N and American Heart Association Midwest affiliate Grand-in-Aid 0855698G (to Z.L.), Grants HL68819, HL62350, and 080264 from the National Institutes of Health/National Heart Lung and Blood Institute (to X.D.), and in part by the Centers of Biomedical Research Excellence in Obesity and Cardiovascular Disease Grant P20 RR021954-01A1 from the National Institutes of Health/National Center for Research Resources.
↵2 Address correspondence and reprint requests to Dr. Zhenyu Li, Division of Cardiovascular Medicine, The Gill Heart Institute, 741 South Limestone Street, Biomedical Biological Sciences Research Building, Room B251, University of Kentucky, Lexington, KY 40536-0200. E-mail address: zhenyuli08{at}uky.edu
↵3 Abbreviations used in this paper: PKG, cGMP-dependent protein kinase; L-NAME, Nω-nitro-l-arginine methyl ester; PRP, platelet-rich plasma.
↵4 The online version of this article contains supplemental material.
- Received September 2, 2008.
- Accepted April 3, 2009.
- Copyright © 2009 by The American Association of Immunologists, Inc.