Role of Prostacyclin in the Cardiovascular Response to Thromboxane A2

See allHide authors and affiliations

Science  19 Apr 2002:
Vol. 296, Issue 5567, pp. 539-541
DOI: 10.1126/science.1068711


Thromboxane (Tx) A2 is a vasoconstrictor and platelet agonist. Aspirin affords cardioprotection through inhibition of TxA2 formation by platelet cyclooxygenase (COX-1). Prostacyclin (PGI2) is a vasodilator that inhibits platelet function. Here we show that injury-induced vascular proliferation and platelet activation are enhanced in mice that are genetically deficient in the PGI2receptor (IP) but are depressed in mice genetically deficient in the TxA2 receptor (TP) or treated with a TP antagonist. The augmented response to vascular injury was abolished in mice deficient in both receptors. Thus, PGI2 modulates platelet-vascular interactions in vivo and specifically limits the response to TxA2. This interplay may help explain the adverse cardiovascular effects associated with selective COX-2 inhibitors, which, unlike aspirin and nonsteroidal anti-inflammatory drugs (NSAIDs), inhibit PGI2 but not TxA2.

Prostacyclin (PGI2) is the major product of cyclooxygenase (COX) catalyzed metabolism of arachidonic acid in macrovascular endothelium (1, 2). It is a potent inhibitor of platelet activation by all recognized agonists and a vasodilator. Biosynthesis of PGI2 is increased in human syndromes of platelet activation (3), suggesting that it functions as a homeostatic response to accelerated platelet–vessel wall interactions. However, its importance in vivo has remained speculative. Recent studies of the coxibs, which are selective inhibitors of COX-2 (4) have suggested that this isozyme is a major source of PGI2 under physiological conditions in humans (5, 6), perhaps reflective of COX-2 induction in endothelial cells by hemodynamic shear (7). PGI2 synthase preferentially couples with COX-2 rather than COX-1 in coexpression systems (8).

It is possible that PGI2 modulates cardiovascular homeostasis in humans, as suggested by the outcome of a controlled comparison of a selective COX-2 inhibitor, rofecoxib, and an isoform nonselective inhibitor, naproxen (the VIGOR study). In this trial, rofecoxib was associated with a higher risk of myocardial infarction by a factor of 5 (9). Two mechanistic hypotheses have been advanced to explain these results: (i) A cardioprotective effect of naproxen, mediated by its inhibition of COX-1–dependent production of platelet TxA2, and (ii) enhancement of the deleterious cardiovascular effects of TxA2 in the patients on rofecoxib, which, in contrast to naproxen, inhibits only PGI2 production, leaving TxA2 unopposed (6).

To test the hypothesis that PGI2 modulates the cardiovascular effects of TxA2 in vivo, we generated mice with disordered expression of receptors for these eicosanoids and monitored the response to vascular injury. Both PGI2 and TxA2 activate G protein–coupled receptors, the prostacyclin receptor (the IP) and the thromboxane receptor (the TP), respectively (10, 11). Others have shown that IP knockout mice (IPKOs) do not develop spontaneous thrombosis but are more susceptible to thrombotic stimuli than their wild-type (WT) littermates (12). TPKOs, by contrast, exhibit a mild bleeding tendency (13). Both IPKOs and TPKOs are normotensive.

Initially, we studied the impact of IP deletion on the proliferative response to catheter-induced carotid vascular injury. This model involves passage of a polyamide catheter containing a fine wire down the common carotid artery, causing endothelial denudation, neointimal formation, and vascular smooth muscle cell proliferation (14, 15). Neointimal proliferation was not observed 2 weeks after surgery in sham-treated mice, but it was induced by injury in both IPKOs and their WT littermates (15). Deletion of the IP resulted in an enhanced proliferative response to injury: the intima-to-media ratio increased significantly in the IPKOs (Fig. 1A). The percentage of luminal stenosis was also increased after injury in the IPKOs [Web table 1 (15)]. We also assessed intimal cellular proliferation by bromodeoxyuridine (BrdU) labeling. There was no significant labeling in the intima or media of uninjured WTs. Injury caused a marked increase in labeling of both intimal and medial cells in WTs. However, intimal and medial proliferation was enhanced in IPKOs (Fig. 1B). In humans, vascular angioplasty is associated with an increase in platelet activation, reflected by urinary excretion of the Tx metabolite (Tx-M), 2,3-dinor-TxB2 (16). Patients undergoing coronary angioplasty are given aspirin, which suppresses TxA2 formation and reduces the periprocedural rate of myocardial infarction by ∼50% (17). We found that catheter-induced carotid injury in the mouse also results in a significant increase in Tx biosynthesis (Fig. 1C), which was augmented in the IPKOs compared with IPWTs (Fig. 1D). These data support the hypothesis that PGI2 modulates platelet activation by TxA2 in vivo.

Figure 1

Morphometric analysis of carotid arteries and Tx-M biosynthesis after vascular injury in WT and KO mice. (A) Intima-to-media ratio; (B) the intimal BrdU-labeled cell number. Results are means ± SEM of values from eight or nine animals in each group. *P < 0.05, **P < 0.01 compared with corresponding WTs group, respectively by Student's t test. (C) Urinary excretion of 2,3-dinor-TxB2 (Tx-M) increased in response to vascular injury in WTs. The data are depicted at the midpoint of the timed collections. Tx-M increased significantly in the periprocedural aliquot of urine (*P < 0.05 versus preinjury by Kruskal-Wallis test). Tx-M did not change in the sham group. (D) The periprocedural increment in urinary Tx-M times baseline values in WTs versus KOs after vascular injury (*P < 0.05).

To address the functional importance of TxA2in this mouse model, we investigated the effect of S18886, a highly selective, long acting TP antagonist activity (18). We measured the impact of TP antagonism on the response to vascular injury in both WT mice and transgenic mice with vascular overexpression of the human TP (TPOEs) (15, 19). Oral administration of S18886 (10 mg/kg body weight, per os) completely inhibited platelet aggregation induced by the TP agonist, U46619, in both WT and TPOEs (20). Both 10 mg/kg and 100 mg/kg of S18886 significantly decreased the intima-to-media ratio, percentage stenosis, and the number of intimal cells labeled with BrdU in WT mice (Table 1). The response to injury was markedly exaggerated in TPOEs (Fig. 2). Thus, the intima-to-media ratio (1.2 ± 0.1 versus 0.6 ± 0.1; P < 0.01), the percentage stenosis (77.6 ± 4.3 versus 31.7 ± 5.2; P< 0.05), and the number of intimal BrdU-labeled cells (198.8 ± 26.4 versus 94.9 ± 19.5; P < 0.05) were all exaggerated in the transgenic animals. The increase in platelet activation evoked by vascular injury, as reflected by urinary Tx-M, was also more pronounced in TPOEs (6.2 ± 1.1 times basal) than in WT mice (3.2 ± 0.6 times basal; P < 0.05). TP antagonism also reduced the response to vascular injury in the TPOEs (Fig. 2), although the dose-response relationship was modified compared with that in WT mice. Thus, oral administration of S18886 (100 mg/kg per day) reduced the intima-to-media ratio (0.5 ± 0.1; P< 0.01), the percentage stenosis (39.2 ± 9.7; P< 0.001), and intimal BrdU-labeled cells (52.9 ± 17.9;P < 0.01) evoked by injury in TPOEs. Although S18886 does not inhibit enzymes in the Tx biosynthetic cascade in vitro (18), treatment with the antagonist reduced the procedure-related increment in Tx-M (1.6 ± 0.4; P< 0.01 versus untreated TPOEs). This is likely to reflect interruption of the continued stimulation of platelets by TxA2 released after TP-mediated platelet activation (21).

Figure 2

The response to vascular injury in WT and TPOE carotid arteries and the effect of the TP antagonist, S18886. Injured artery in WTs (A), TPOEs (B), and TPOEs treated with S18886 (C) are representative segments from hematoxylin and eosin (H&E)–stained sections. The sections of injured artery in WTs (D), TPOEs (E), and TPOEs treated with S18886 (F) stained with BrdU. The arrows and the arrowheads indicate the internal elastic laminae and the external elastic laminae, respectively. Scale bar, 50 μm.

Table 1

The effects of TP antagonism with S18886 on the proliferative response to vascular injury in WTs. Results are means ± SEM of values from eight animals in each group. *P < 0.05, **P < 0.01 compared with vehicle-treated group (one-way ANOVA analysis and Newman-Keuls multiple-comparison test).

View this table:

To exclude the possibility that properties unrelated to TP antagonism might account for the effects of S18886, we determined that TP deletion reduced the changes evoked in the intima-to-media ratio (Fig. 1A), percentage stenosis (Web table 1), and the number of BrdU-labeled cells (Fig. 1B) compared with WT littermates (15). Consistent with our observations with TP antagonism, TP deletion also reduced the procedure-related increment in urinary Tx-M (Fig. 1D). To address the contribution of exaggerated Tx biosynthesis to the vascular proliferative response in IPKO mice, we studied coincidental deletion of the TP and the IP [TPIP double knockouts (DKOs)], which abolished the increased response to injury observed in the IPKOs (15) (Fig. 1, A, B, and D, and Web table 1). All three indices of proliferation were similar in TPIP DKOs and their WT littermates. Furthermore, the augmented, procedure-related increment in platelet activation, observed in IPKOs, was also abolished by concomitant TP and IP deletion (Fig. 1D).

Biosynthesis of PGI2 is elevated along with TxA2in human syndromes of platelet activation (22,23), and the formation and activity of the two prostaglandins may interact via several mechanisms. TP agonists evoke PGI2 release from endothelial cells in vitro (24), and platelet-derived endoperoxide precursors of TxA2 may also be utilized by endothelial COX to generate PGI2 in vitro (25). Pharmacological inhibition of Tx synthase enhances rediversion of endoperoxides to platelet inhibitory prostanoids, such as PGI2, in vivo (26). Both the arachidonic acid in microparticles shed by activated platelets (27) and platelet-derived TxA2 evoke COX-2–dependent PGI2 formation by endothelial cells (28). Considerable evidence also exists for cross talk between IP- and TP-dependent signaling pathways. Thus, TP desensitization evokes sensitization to IP agonists (29), and IP activation evokes TP desensitization (30).

Our data establish that endogenous PGI2 modulates the cardiovascular actions of TxA2 in vivo. This interplay may be relevant to the cardiovascular effects of selective COX-2 inhibitors, which, unlike aspirin and NSAIDs, depress PGI2without coincidental inhibition of TxA2 formed by COX-1 in platelets. Although our results suggest that the enhancement of the deleterious effects of TxA2 is a plausible explanation for the cardiovascular outcome in the VIGOR trial, they do not exclude a contributory role for naproxen's cardioprotective effect. Indeed, both mechanisms may be important. If the estimate of the difference in incidence rates of myocardial infarction between the groups is accurate, it is roughly twice that expected if naproxen was affording cardioprotection via sustained inhibition of platelet COX-1 (31).

The response to carotid vascular injury was modified by both TP antagonism and TP deletion. Although our mouse models do not perfectly simulate percutaneous angioplasty on an atherosclerotic background in humans, it prompts reconsideration of the potential utility of TP antagonists in this setting. Two clinical trials have compared TP antagonists with aspirin for the prevention of restenosis after angioplasty (32, 33). TP antagonists, in contrast to aspirin, do not depress PGI2 formation. Patients received periprocedural aspirin before randomization to continuing aspirin or a TP antagonist for the following 6 months. However, the aspirin regimen used depresses the procedure-related increment in PGI2 biosynthesis (16). Our results suggest that this may have conditioned the proliferative response to injury and may have undermined the comparison of the two treatments.

  • * To whom correspondence should be addressed. E-mail: garret{at}


View Abstract

Navigate This Article