Thrombospondin-1 stimulates platelet aggregation by blocking the antithrombotic activity of nitric oxide/cGMP signaling

Abstract

Platelet alpha-granules constitute the major rapidly releasable reservoir of thrombospondin-1 in higher animals. Although some fragments and peptides derived from thrombospondin-1 stimulate or inhibit platelet aggregation, its physiologic function in platelets has remained elusive. We now show that endogenous thrombospondin-1 is necessary for platelet aggregation in vitro in the presence of physiologic levels of nitric oxide (NO). Exogenous NO or elevation of cGMP delays thrombin-induced platelet aggregation under high shear and static conditions, and exogenous thrombospondin-1 reverses this delay. Thrombospondin-1-null murine platelets fail to aggregate in response to thrombin in the presence of exogenous NO or 8Br-cGMP. At physiologic concentrations of the NO synthase substrate arginine, thrombospondin-1-null platelets have elevated basal cGMP. Ligation of CD36 or CD47 is sufficient to block NO-induced cGMP accumulation and mimic the effect of thrombospondin-1 on aggregation. Exogenous thrombospondin-1 also reverses the suppression by NO of alphaIIb/beta3 integrin-mediated platelet adhesion on immobilized fibrinogen, mediated in part by increased GTP loading of Rap1. Thrombospondin-1 also inhibits cGMP-mediated activation of cGMP-dependent protein kinase and thereby prevents phosphorylation of VASP. Thus, release of thrombospondin-1 from alpha-granules during activation provides positive feedback to promote efficient platelet aggregation and adhesion by overcoming the antithrombotic activity of physiologic NO.

Figure 1

Exogenous TSP1 reverses the delay of platelet aggregation by NO. Washed human platelets in Tyrode buffer (2 × 10⁵ platelets/μL) were incubated in the presence of thrombin (0.2 U/mL) and exogenous NO (DEA/NO 10 μM) for 5 minutes under high shear (1200 rpm, A,B) or static (C) conditions and absorbance was recorded. In other experiments, fresh, washed human platelets in Tyrode buffer (500 μL) were treated with TSP1 (2.2 nM) (D) or the indicated concentrations of TSP1 or fibronectin (E) and DEA/NO (10 μM) for 5 minutes and lysed, and cGMP was determined via immunoassay. Platelets in PRP were treated with TSP1 (2.2 nM) for 15 minutes followed by NO (DEA/NO 10 μM) for 5 minutes and lysed, and cGMP was determined via immunoassay (F). Data presented are representative of at least 3 experiments (A-C). Results are the mean (± SD) of at least 3 experiments (D,F).

Figure 2

Endogenous TSP1 limits NO/cGMP signaling in murine platelets. Equal numbers of murine C57BL/6 WT and TSP1-null platelets were incubated in Tyrode buffer in the presence of 10 μM DEA/NO for 5 minutes, and cGMP was determined by immunoassay (A). In other experiments WT and TSP1-null platelets were preincubated with exogenous TSP1 (2.2 nM) for 15 minutes and then treated with NO (10 μM DEA/NO) (B) or treated with l-arginine at the indicated doses for 20 minutes and cGMP levels determined via immunoassay (C). Results are the mean (± SD) of at least 3 experiments.

Figure 3

Endogenous TSP1 is necessary for platelet aggregation in the presence of NO. Equal numbers of murine C57BL/6 WT and TSP1-null platelets were incubated under standard high shear (A,B) or static aggregation (C) conditions. Aggregation profiles were determined in the presence of a titrated dose of thrombin (0.1 U/mL added at time points indicated by arrows, A) or a fixed thrombin dose (0.2 U/mL, B,C) plus or minus DEA/NO (10 μM). Alternatively, equal numbers of WT and TSP1-null platelets were treated with a fixed dose of thrombin (0.2 U/mL) and 8-Br-cGMP (10 μM) under high shear conditions and aggregation was determined (D). Data presented are representative of at least 3 experiments.

Figure 4

Regulation of platelet adhesion by NO is blocked by TSP1. Fresh, washed human platelets were added to 35 × 10-mm plastic dishes precoated with either type I collagen (3 μg/mL) or fibrinogen (5 μg/mL) (A) and incubated in Tyrode buffer and the indicated treatment agents for 1 hour at 37°C. Wells were washed and platelets fixed, stained, and counted. Human platelets were incubated on collagen-coated plates in the presence of DEA/NO (10 μM) plus or minus ODQ (10 μM) (B) or collagen and fibrinogen coated plates plus or minus exogenous TSP1 (2.2 nM) (C), and adhesion was determined. Results are the mean (± SD) of at least 3 experiments.

Figure 5

TSP-1 enhances platelet adhesion by antagonizing NO and 8Br-cGMP signaling via Rap1 and blocks VASP-Ser²³⁹ phosphorylation by inhibiting cGK. Platelets preincubated in the presence or absence of TSP-1 (2.2 nM) were treated with DEA/NO (10 μM) or 8Br-cGMP (100 μM) 2 minutes prior to stimulation with 0.5 U/mL thrombin (A). Rap1 activation was analyzed by affinity purification using GST-RalGDS-RBD fusion protein immobilized on Glutathione-Sepharose beads. Platelets incubated at 37°C in the presence or absence of GGTI-298 (10 μM for 30 minutes) before addition of 1 U/mL thrombin for 1 minute, or incubated with TSP1 (2.2 nM for 15 minutes), were lysed and subjected to a Rap activation assay (B). Fresh, washed human platelets were used directly or preincubated in the presence of GGTI-298 for 30 minutes, added to 35-mm plastic dishes precoated with either type I collagen (3 μg/mL) or fibrinogen (5 μg/mL) (C,D), and incubated in Tyrode buffer and the indicated treatment agents for 1 hour at 37°C. Wells were washed and platelets fixed, stained, and counted. Washed human platelets in Tyrode buffer (2 × 10⁵ platelets/μL) were incubated in the presence of thrombin (0.2 U/mL) and exogenous NO (DEA/NO 10 μM) plus or minus TSP1 (2.2 nM) for 5 minutes or preincubated with GGTI-298 for 30 minutes and treated as described earlier for 5 minutes; aggregation was determined under high shear (E). Washed human platelets, either untreated or treated with thrombin (0.1 U/mL), 8Br-cGMP (100 μM for 2 minutes), or 2.2 nM TSP1 followed by 8Br-cGMP, were lysed, resolved on SDS gels, blotted, and probed with a polyclonal antiserum against Ser²³⁹-phosphorylated VASP (F). Washed human platelets were preincubated with TSP1 (2.2 nM) or Rp-8pCPT-cGMP (5 μM) for 15 minutes prior to treatment with NO (DEA-NO 10 μM) for 5 minutes (G) or 8Br-cGMP (100 μM) for 1 minute (H). The platelets were chilled to terminate the reaction, washed, lysed, and centrifuged. Lysate supernatants containing equal amounts of protein (100 μg) were assayed for phosphorylation of the cGK-I–selective substrate Arg-Lys-Arg-Ser-Arg-Ala-Glu. Data are representative of at least 3 experiments (A,B, E-H). Results are the mean (± SD) of at least 3 experiments (C,D).

Figure 6

CD36- and CD47-binding domains of TSP1 block NO-driven delay of platelet aggregation. Washed human platelets (2 × 10⁵ platelets/μL) were incubated in the presence of thrombin (0.2 U/mL) and exogenous NO (DEA/NO 10 μM) for 5 minutes in the presence of recombinant TSP1 constructs 3TSR and E123CaG-1 (2.2 nM) or NoC1 (2.2–22 nM) and aggregation was determined under high shear (A,B) or low shear (C) conditions. In other experiments, washed platelet were preincubated with the indicated concentrations of recombinant fragments and treated with DEA/NO (1 μM) for 60 seconds and lysed, and cGMP levels were determined by immunoassay (D). Data are representative of at least 3 experiments (A-C). Results are the mean (± SD) of at least 3 experiments (D).

Figure 7

CD47- and CD36-binding peptides antagonize the NO delay in platelet aggregation. Washed human platelets (2 × 10⁵ platelets/μL) were incubated in Tyrode buffer in the presence of thrombin (0.2 U/mL) and exogenous NO (DEA/NO 10 μM) for 5 minutes. Peptide sequences were derived from relevant domains of TSP1 including 7N3 (FIRVVMYEGKK, 10 μM) and control peptides p604 (FIRGGMYEGKK, 10 μM) and p605 (FIRVAIYEGKK, 10 μM, A), and p7N3 (0.1-10 μM, B), and absorbance was determined under high shear, or p459 (4N1–1, RFYVVMWK, 10 μM, C) and 4N1K (KRFYVVMWKK, 10 μM, D), and absorbance was determined under low shear conditions. Washed human platelets in Tyrode buffer (2 × 10⁵ platelets/μL) were incubated in the presence of thrombin (0.2 U/mL) and exogenous NO (DEA/NO 10 μM) for 5 minutes. Peptide sequences were derived from TSP1, including p7N3 and p907 (GDGV[D-I]TRIR, E) (10 μM) and aggregation was determined under high shear, and p906 (VTAGGGVQKRSRL, F; 10 μM) and p246 (KRFKQDGGWSHWSPWSS, G; 10 μM), and absorbance was measured under low shear. Human platelets were pretreated with TSP1-based peptides p907 and p7N3 (10 μM) before adding DEA/NO (10 μM), and cGMP levels were determined (H). Data are representative of at least 3 experiments (A-G). Results are the mean (± SD) of at least 3 experiments (H).

Figure 8

Proposed mechanism for TSP1 antagonism of NO/cGMP signaling in platelets. Using recombinant domains and peptides of TSP1, we show that ligation of CD36 or CD47 is sufficient to block an NO-mediated delay in platelet aggregation. TSP1 blocks a delay mediated by either exogenous NO or NO synthesized by endogenous eNOS using Arg as substrate. The ability of TSP1 to prevent cGMP synthesis stimulated by exogenous NO identifies sGC as one target of TSP1 signaling. The ability of TSP1 to inhibit cGK-I–mediated phosphorylation of VASP and a cGK-I–selective peptide (RKRSRAE) stimulated by a cell-permeable cGMP analog (8Br-cGMP) identifies cGK-I as a second target of TSP1 signaling in platelets. VASP is required for NO/cGMP-mediated inhibition of agonist-induced platelet aggregation as well as platelet adhesion. TSP1 prevents cGK-I–mediated phosphorylation of VASP at Ser²³⁹. NO also stimulates phosphorylation of the cGK-I target Rap1GAP2, so TSP1 inhibition of sGC and cGK-I also controls GTP loading of Rap1, which is required for thrombin-stimulated activation of the adhesion receptor α_IIb/β₃.