Endothelium-mediated control of vascular tone: COX-1 and COX-2 products

Abstract

Endothelium-dependent contractions contribute to endothelial dysfunction in various animal models of aging, diabetes and cardiovascular diseases. In the spontaneously hypertensive rat, the archetypal model for endothelium-dependent contractions, the production of the endothelium-derived contractile factors (EDCF) involves an increase in endothelial intracellular calcium concentration, the production of reactive oxygen species, the predominant activation of cyclooxygenase-1 (COX-1) and to a lesser extent that of COX-2, the diffusion of EDCF towards the smooth muscle cells and the subsequent stimulation of their thromboxane A2-endoperoxide TP receptors. Endothelium-dependent contractions are also observed in various models of hypertension, aging and diabetes. They generally also involve the generation of COX-1- and/or COX-2-derived products and the activation of smooth muscle TP receptors. Depending on the model, thromboxane A₂, PGH₂, PGF_2α, PGE₂ and paradoxically PGI₂ can all act as EDCFs. In human, the production of COX-derived EDCF is a characteristic of the aging and diseased blood vessels, with essential hypertension causing an earlier onset and an acceleration of this endothelial dysfunction. As it has been observed in animal models, COX-1, COX-2 or both isoforms can contribute to these endothelial dysfunctions. Since in most cases, the activation of TP receptors is the common downstream effector, selective antagonists of this receptor should curtail endothelial dysfunction and be of therapeutic interest in the treatment of cardiovascular disorders.

LINKED ARTICLES

This article is part of a themed issue on Vascular Endothelium in Health and Disease. To view the other articles in this issue visit http://dx.doi.org/10.1111/bph.2011.164.issue-3

Introduction

In 1980, Furchgott and Zawadzki (1980) unequivocally demonstrated that the presence of the endothelium was required in order to observe relaxations of isolated arteries to acetylcholine. This seminal discovery not only led to the identification of the L-arginine nitric oxide (NO) synthase pathway and the overwhelming role of NO as an intercellular messenger but also led to the quest for other endothelium-derived vasoactive factors, in particular endothelium-derived hyperpolarizing factor (EDHF) and endothelium-derived contracting factors (EDCF) (for review, Félétou and Vanhoutte, 2006a,b; Félétou et al., 2009).

However, although the era of endothelium-derived relaxing factors truly began with the scientific breakthrough of Furchgott and Zawadzki (1980), prostaglandins (PG) were in fact the first endothelium-derived vasoactive paracrine substances to be identified (Moncada et al., 1976; 1977;). PGs and thromboxane A₂ are critical modulators of vascular tone and platelet activity under both physiological and pathophysiological conditions (Moncada and Vane, 1979; Félétou et al., 2010a). The fatty acid arachidonic acid, the most common precursor of PGs, is released from the cell membrane phospholipids primarily by phospholipase A₂ and can be metabolized by several enzymatic systems including prostaglandin H (PGH) synthases, lipoxygenases and cytochrome P450 monooxygenases or be transformed in a radical catalyzed non-enzymatic manner into isoprostanes (Morrow et al., 1980; Smith and Marnett, 1991). PGH synthase, the first and rate-limiting enzyme involved in the biosynthetic pathway of PGs, possesses both a cyclooxygenase (COX) catalytic activity leading to the formation of prostaglandin G₂ (PGG₂) and a peroxidase activity catalyzing the reduction of PGG₂ to prostaglandin H₂ (endoperoxide, PGH₂). Although this single protein is associated with both COX and peroxidase activities, PGH-synthases are usually termed COX (Vane et al., 1998).

COX- and endothelium-dependent contractions have been reported in arteries and veins of different species in response to various agonists and substances that increase the endothelial intracellular calcium concentration ([Ca²⁺]_i) in a receptor-independent manner as well as in response to physical stimuli such as stretch (Miller and Vanhoutte, 1985; Katusic et al., 1987; 1988; Ihara et al., 1999; Okon et al., 2002; Yang et al., 2004b; Tang et al., 2007). Endothelium-dependent contractions have been observed in healthy blood vessels, suggesting that they play a physiological role in the endothelium-dependent regulation of vascular tone. For instance, the endothelium may contribute to the autoregulation of cerebral blood flow during increases in transmural pressure by the increased production and release of PGs, which causes activation of the underlying vascular smooth muscle (Katusic et al., 1987). However, endothelium-dependent contractions are also frequently associated with cardiovascular disease in both animals and humans. These responses counterbalance the endothelium-dependent vasodilatations produced by NO and/or EDHF and contribute to endothelial dysfunction (Vanhoutte et al., 2005; Félétou et al., 2009; 2010b;).

This brief review will highlight the physiological and pathological role of endothelial COX-derived vasoactive factors, especially in aging, hypertension and diabetes.

Mobilization of arachidonic acid

The fatty acid arachidonic acid, the most common precursor of PGs, is released from the cell membrane phospholipids. Two major phospholipases are implicated in prostanoid formation, phospholipase A2 (PLA₂) acting on phosphatidyl-ethanolamine, phosphtidyl-choline or plasmalogens, as well as phospholipase C, which together with the diacylglycerol lipase acts sequentially on phosphatidyl-inositols derivatives (Smith and Marnett, 1991).

The phospholipase A₂ super family includes in mammals at least 25 enzymes identified with PLA₂ activity and is subdivided in five main groups: secreted PLA₂ (sPLA₂), cytosolic PLA₂ (cPLA₂), calcium-independent PLA₂ (iPLA₂), the platelet-activating factor acetylhydrolases and the lysosomal PLA₂. Although sPLA_2, cPLA₂ and iPLA₂ can generate arachidonic acid, cPLA₂α (or group IV-A) is the only PLA₂ enzyme that shows significant selectivity towards phospholipids containing arachidonic acid (Burke and Dennis, 2009). cPLA₂α is expressed ubiquitously and constitutively in most cells and tissues and is an essential component of the initiation of the metabolism of arachidonic acid. The translocation and activation of this intracellular enzyme is initiated by submicromolar calcium concentrations and by phosphorylations, both of which are critical events in the post-receptor signalling transduction (Kudo and Murakami, 2002). iPLA₂s, as their name indicates, do not require calcium for membrane association or enzymatic activity and have been implicated in a number of physiological and pathophysiological processes. In both endothelial and smooth muscle cells, iPLA₂β is an important effector of calcium signalling and contributes with STIM and ORAI to capacitative Ca²⁺ entry (Bolotina, 2008).

COXs

Two COXs (COX-1 and COX-2) encoded by two different genes have been cloned and characterized. Although COX-1 and COX-2 share a high level of homology (65%), the activity and expression of these enzymes are regulated differentially, and they can function independently within the same cell type (Davidge, 2001). An enzymatically active splice variant of COX-1, termed COX-3, is expressed in the heart and cerebral cortex, but the regulation of its transcription appears similar to that of COX-1 (Park et al., 2006). Fatty acids such as arachidonic acid are the preferential substrate of COX-1, while COX-2 used as substrates both and equally well fatty acids and 2-arachidonyl glycerol. Therefore, COX-2 can generate group of products that COX-1 cannot synthesize (Smith and Song, 2002). Both COXs depend on the presence of lipid peroxides for their activation, but the activation of COX-2 requires 10-fold lower concentrations of hydroperoxide than COX-1, suggesting that COX-2 can function in the presence of COX-1, without the latter being activated (Morita, 2002; Smith and Song, 2002). In most tissues, COX-1 is expressed constitutively but can also be overexpressed, for instance by shear stress (Doroudi et al., 2000), while COX-2 is often induced at sites of inflammation. However, COX-2 is also expressed constitutively in several organs and cell types, including endothelial cells where its expression is also up-regulated by shear stress (Funk and Fitzgerald, 2007).

Various biologically active eicosanoids are formed from the short-lived but biologically active PGH₂, through the action of a set of PG synthases, PGD, PGE, PGF, PGI and thromboxane synthases. The five primary PGs formed, PGD₂, PGE₂, PGF_2α, PGI₂ and thromboxane A₂, interact with prostanoid (P) receptors, which belong to the G-protein-coupled seven transmembrane domains family, and are classified in five subtypes (DP, EP, FP, IP and TP receptors; Figure 1) in function of their preferential affinity towards those PGs (Tsuboi et al., 2002). Additionally, activation of COXs is a source of superoxide anions because of their ability to co-oxidize substances such as NADPH (Wolin, 2000; Tang et al., 2007). The oxidative modification of polyunsaturated fatty acids via a non-enzymatic, free radical–catalyzed mechanism can generate isoprostanes (Morrow et al., 1980). Isoprostanes are often used as markers of oxidative stress (Pratico et al., 2001) and have been associated with various cardiovascular diseases, including coronary stenosis (Wang et al., 2006a). However, these prostanoids are also biologically active compounds that stimulate PG receptors (Janssen, 2002; Yang et al., 2004a; Figure 2). Furthermore, a COX-dependent production of 8-isoprostanes can take place in endothelial cells (Watkins et al., 1999). For instance, the COX-2-dependent production of 8-iso-PGF_2α, which activates TP receptors, may contribute to the hypoxia-induced hyper-responsiveness of pulmonary arteries (Delannoy et al., 2010).

Figure 1.

Cyclooxygenases and arachidonic acid metabolism. Prostacyclin and thromboxane synthases belong to the cytochrome P-450 superfamily (in the human, CYP8A1 and CYP5 respectively). The preferential receptors for the five primary prostaglandins and their subtypes are indicated: IP, DPs, EPs, FP and TP for prostacyclin, prostaglandin D₂, prostaglandin E₂, prostaglandin F_2α and thromboxane A₂ respectively. PGG₂, prostaglandin G₂; PGH₂, prostaglandin H₂; PGI₂, prostacyclin; TXA₂, thromboxane A₂; PGD₂, prostaglandin D₂; PGE₂, prostaglandin E₂; PGF_2α, prostaglandin F_2α; COX, cyclooxygenase; PGHS, prostaglandin H synthase; PGIS, prostacyclin synthase; TXS, thromboxane synthase; PGDS, prostaglandin D synthases; cPGES, cytosolic prostaglandin E₂ synthase; mPGES, membrane prostaglandin E₂ synthase; PGFS, prostaglandin F₂ synthase.

Figure 2.

Prostanoids-induced contractions in SHR aorta and TP activation. In isolated aortic rings of SHR (in the presence of the NO synthase inhibitor L-nitro-arginine), PGD₂, PGE₂, PGF_2α, the analogue of thromboxane A₂, U 46619, 8-isoprostane, PGH₂ and prostacyclin, all produce concentration-dependent contractions, which are inhibited by the selective TP antagonist, S 18886 (100 nM). SHR, spontaneously hypertensive rat; PGD₂, prostaglandin D₂; PGE₂, prostaglandin E₂; PGF_2α, prostaglandin F_2α.

In healthy blood vessels, both endothelial and, to a lesser extent, vascular smooth muscle cells express the two COXs, COX-1 being the predominant isoform (De Witt et al., 1983; Tang and Vanhoutte, 2008). Similarly, in humans, although COX-2 appears to be the major contributor of the overall systemic generation of prostacyclin, in endothelial cells of both healthy and diseased blood vessels, COX-1 is also an important source of PGs (McAdam et al., 1999; Flavahan, 2007; Funk and FitzGerald, 2007; Rovati et al., 2010).

In the cardiovascular system, thromboxane A_2, a powerful aggregating factor and a vasoconstrictor, is derived predominantly from platelet COX-1 but can also be generated by endothelial cells. In most blood vessels, prostacyclin, first described as a potent anti-aggregating agent and as a vasodilator, is the principal metabolite of arachidonic acid, the endothelium being the major site of its synthesis (Moncada et al., 1976; 1977; Figure 3). However, in many cardiovascular diseases, the widespread vascular and organ inflammation and the associated oxidative stress enhance the production of eicosanoids and shift their production/effects from vasodilatation and anti-thrombosis to vasoconstriction, pro-thrombosis and further inflammation (Félétou et al., 2010a).

Figure 3.

Release of prostaglandins by WKY isolated aortic rings with and without endothelium under resting conditions and upon stimulation by the calcium ionophore A 23187 (1 µM), acetylcholine (10 µM) and ATP (300 µM). In WKY aorta, as in most blood vessels, endothelium-derived prostacyclin is the predominant metabolite of arachidonic acid. WKY, Wistar-Kyoto rats.

In mice, disruption of COX-1 is associated with a decrease in constitutive PG synthesis, but the inducible synthesis remains unaffected. The knockout mice show normal survival but with an impaired platelet aggregation and inflammatory responses and with an increased tolerance to pain and some degree of airway hyper-responsiveness (Loftin et al., 2002; Morita, 2002). They show a reduced hypertensive response to angiotensin II and, in isolated aortic rings, a complete inhibition of endothelium-dependent contractions (Tang et al., 2005; Smyth et al., 2009). COX-1 deletion prevents atherosclerotic lesion formation in ApoE null mice (McClelland et al., 2009).

The phenotype associated with COX-2 deletion is more severe. Thirty to 40% of the pups from COX-2 knockout mice die within 48 h from a patent ductus arteriosus (Loftin et al., 2001). Furthermore, the surviving animals have a shorter life span than the wild-type mice. The disruption of COX-2 is associated with a normal constitutive PG synthesis, but the inducible synthesis is impaired (Morham et al., 1995). Females have a deficient reproductive function (Lim et al., 1997), and in both sexes, the inflammatory responses are altered (Morita, 2002). COX-2 knockout mice have a compromised postnatal kidney development associated with a reduced renal blood flow, a decreased urine production and the development of nephropathy (Loftin et al., 2001). COX-2 disruption decreases plasma renin in mice subjected to a low-salt diet. The knockout mice are hypertensive. and although platelet aggregation is normal, since mature platelets only express COX-1, they are more susceptible to thrombosis (Loftin et al., 2002; Morita, 2002; Yu and Funk, 2007). They show an enhanced pressor effect in response to angiotensin II and no or minor inhibition of endothelium-dependent contractions. In mice with a specific deletion of COX-2 in the cardiomyocytes, heart failure and fibrosis are observed. However, in mouse models of abdominal aortic aneurysms, the disruption of COX-2 decreases their formation, and in various experimental models of cancer, it reduces tumour formation (Tang et al., 2005; Smyth et al., 2009).

The double deletion of COX-1 and COX-2 is lethal within hours after birth because of patent ductus arteriosus (Davidge, 2001).

PGD₂ and DP receptors

PGD₂ is a major PG in the central nervous system and in immune cells. It can also be synthesized in the vascular wall as both types of PGD synthases are expressed in endothelial cells, where they are up-regulated in response to an increase in fluid shear stress (Taba et al., 2000). However, PGD₂ is mainly involved in the regulation of sleep and in allergic responses (Smyth et al., 2009).

Two distinct types of PGD synthases have been identified, the lipophilic ligand carrier protein (lipocalin) type enzyme and the haematopoietic enzyme. PGD₂ is dehydrated to produce the J series of PG (PGJ₂ analogues). PGJ₂ acts as a ligand for the PPARγ nuclear receptor, which is involved in the differentiation of adipocytes (Urade and Eguchi, 2002). PPARγ is also expressed in endothelial and smooth muscle cells (Jackson et al., 1999; Law et al., 2000). Two PGD₂ receptors, DP1 and DP2, have been identified, the latter being also known as CRTH2, a member of the chemokine receptor family (Hirai et al., 2001; Figure 1). Activation of DP1 increases blood flow and vascular permeability (Smyth et al., 2009). PGD₂ can produce not only endothelium- and NO-dependent relaxations (Abran et al., 1997) but also endothelium-independent contractions, the latter effect involving activation of TP receptors (Gluais et al., 2005; Figure 2).

PGE₂ and EP receptors

PGE₂, the most abundant PG in the human body, is involved in multiple physiological effects including gastric secretion and motility, bone formation, inflammation, female reproductive function, body temperature, sleep–wake activity and the regulation of kidney functions and blood pressure (glomerular filtration rate, tubular salt and water transports, renin release, renal blood flow and vascular tone) (Park et al., 2006; Sugimoto and Narumiya, 2007; Legler et al., 2010). The diverse effects of PGE₂ (for instance, this PG can produce both relaxation and contraction of vascular smooth muscle) can be attributed to the existence of four receptor subtypes, namely EP1, EP2, EP3 (of which three splice variants exist in the mouse and at least eight in the human) and EP4, which are coupled to different signalling pathways (Kobayashi and Narumiya, 2002; Hata and Breyer, 2004; Sugimoto and Narumiya, 2007; Figure 1).

Three distinct PGE synthases, responsible for the synthesis of PGE₂, have been identified and characterized, one cytosolic, cPGES, and two membrane bound, mPGES-1 and mPGES-2 (Kudo and Murakami, 2005). cPGES is a constitutive enzyme expressed ubiquitously and in abundance in many tissues and cell types. This cytosolic form is preferentially associated with COX-1, suggesting that this isoform contributes physiologically to the production of PGE₂ (Murakami et al., 2002; Park et al., 2006), and its function is thought to overlap that of COX-1. Knockout mice studies demonstrate that the deletion of cPGES is lethal (Hara et al., 2010). The mPGES-1 is localized in perinuclear membranes, is up-regulated in response to stimuli that induce COX-2 expression and is associated preferentially with COX-2. In the vascular wall, mPGES-1 is also expressed by both endothelial and vascular smooth muscle cells (Murakami et al., 2002). Cells from mice deficient in mPGES-1 display impaired LPS-stimulated PGE₂ generation, but the basal formation, which is cPGES-dependent, is preserved. mPGES-1 plays a critical role in fever, inflammation, pain, gastrointestinal and renal homeostasis as well as in tumorigenesis (Hara et al., 2010). Additionally, the deletion of the mPGES-1 gene does not affect arterial blood pressure or thrombogenesis but delays atherogenesis, possibly by diverting the metabolism of PGH₂ towards prostacyclin (Cheng et al., 2006; Wang et al., 2006b) and prevents angiotensin-II-induced oxidative stress and aortic aneurysm formation (Wang et al., 2008). Conversely, mPGES-1 deficient mice show adverse left ventricular remodelling after myocardial infarction (Degousee et al., 2008) and can, depending on the model, develop severe hypertension when subjected to a high-salt diet (Jia et al., 2006; Francois et al., 2007). mPGES-2 is a constitutive, Golgi membrane–associated protein that can be coupled to both COX-1 and COX-2 (Kudo and Murakami, 2005). The real significance of its contribution in PGE₂ production under physiological or pathophysiological conditions is questionable since knockout of this enzyme does not result in a specific phenotype or a reduction in PGE₂ generation (Hara et al., 2010).

Although PGE₂ is the common ligand of the four receptor subtypes, the amino acid homology among these receptors is limited. The stimulation of the EP1 receptor is predominantly coupled with an increase in intracellular calcium concentration (calcium channel gating and phospholipase C–dependent IP₃ formation), that of EP2 and EP4 with an increase in cAMP (activation of adenylyl cyclase) and that to EP3 with a decrease in cAMP (inhibition of adenylate cyclase) (Sugimoto and Narumiya, 2007; Legler et al., 2010). In the rat aorta, the mRNAs of the four receptor subtypes are expressed, and in various rat arteries, EP1 activation is associated with contraction (Bolla et al., 2004; Tang et al., 2008). The studies performed on mice knockout for these various receptor subtypes as well as experiments with subtype selective agonists and antagonists indicate that the EP1 receptor is associated with stress responses and hyperalgesia; the EP2 receptor with ovulation and fertilization as well as pain and inflammation; the EP3 with fever, angiogenesis, duodenal secretion and pain; and the EP4 receptor with the closure of the ductus arteriosus, bone formation and possibly vascular headache in migraine as well as protection against inflammation (bowels) and generation of inflammation (arthritis) (Sugimoto and Narumiya, 2007; Jones et al., 2009).

As regards the cardiovascular role of EP receptor subtypes (Kobayashi and Narumiya, 2002), the deletion of the various EP receptor genes show that in the kidney, PGE₂ stimulates renin release via the stimulation of EP2 and EP4 receptors, but that the four subtypes are involved in the control of renal vascular tone, with EP1 and EP3 increasing and EP2 and EP4 decreasing it (Schweda et al., 2004). EP1-deficient mice have a urine concentration defect due to decrease in vasopressin release resulting in hypotension. EP1 receptor deletion reduces the elevated blood pressure and cardiac hypertrophy induced by angiotensin-II (Guan et al., 2007; Smyth et al., 2009). Paradoxically in EP1 knockout males, but not in female, the depressor response to exogenous administration of PGE₂ is reduced. However, in both genders, EP2 receptors contribute to the decrease in blood pressure produced by PGE₂ (Audoly et al., 1999), and EP2 knockout mice develop salt-sensitive hypertension (Ma et al., 2001). EP3 receptor–deficient mice have a bleeding tendency and are resistant to thromboembolism (Yuhki et al., 2010). In male mice, EP3 receptor activation opposes the vasodepressor response to PGE₂, while in female mice, EP4 receptors contribute also to the depressor action of PGE₂ (Audoly et al., 1999). Endothelium- and NO-dependent vasodilatation to PGE₂ involves EP4 activation (Hristovska et al., 2007). The germline disruption of the EP4 gene enhances infarct size in the heart following ischaemia–reperfusion (Xiao et al., 2004), and specific deletion of EP4 in cardiomyocytes exacerbates the decline in cardiac function observed after myocardial infarction (Smyth et al., 2009).

Prostaglandin F_2α and FP receptors

Prostaglandin F₂ isoforms are synthesized not only from PGH₂ by the membrane-associated 9,11-endoperoxide reductase but also from PGD₂ and PGE₂ by cytosolic PGD₂ 11-ketoreductase and PGE₂ 9-ketoreductase (or in the endometrium by the 20α-hydroxysteroid deshydrogenase) respectively (Watanabe, 2002; Helliwell et al., 2004).

Prostaglandin F_2α interacts with its preferential receptor the FP receptor (Figure 1), which generally generates an increase in the intracellular concentration of calcium. Prostaglandin F_2α is required in the female reproductive system for normal parturition. In the kidney, this PG is involved in water absorption, causing natriuresis and diuresis and, in the eye, is involved in the regulation of intraocular pressure. Prostaglandin F_2α is produced in the vascular wall, is a potent vasoconstrictor and may be involved in cardiac hypertrophy (Hata and Breyer, 2004; Jones et al., 2009; Smyth et al., 2009). Additionally, FP receptor can be expressed in endothelial cells and induce endothelium- and NO-dependent relaxations when stimulated (Chen et al., 1995). Deletion of FP receptors reduces arterial blood pressure and delays atherogenesis in hyperlipidemic mice (Yu et al., 2009).

Prostacyclin and IP receptor

Prostacyclin is an unstable substance formed by prostacyclin synthase, which belongs to the cytochrome P450 superfamily (in the human CYP8A1) of enzymes. In endothelial cells, prostacyclin synthase is highly expressed (Tang and Vanhoutte, 2008), and the enzyme is closely associated with COX-1 (Kawka et al., 2007). Prostacyclin synthase is also expressed in vascular smooth muscle cells, neurons, oviducts, intestinal epithelial cells and embryonic cells (Wu and Liou, 2005).

The deletion of this enzyme generates hypertensive mice with thickening and sclerosis of the arterial wall and with kidney infarction associated with interstitial fibrosis and nephrosclerosis (Smyth and FitzGerald, 2002; Wu and Liou, 2005). Conversely, overexpression of prostacyclin synthase, alone or in association with COX-1, prevents injury-induced intimal hyperplasia, pulmonary hypertension and vascular remodelling and protects brain tissue from ischaemia–reperfusion injury (Wu and Liou, 2005). Prostacyclin synthase is amongst the most sensitive targets of peroxynitrite and is inactivated by concentrations as low as 50 nM (Zou et al., 2002a,b; Schmidt et al., 2003). When PGI₂ synthase is inactivated, the excess PGH₂ is shunted towards other metabolic pathways leading to a variety of products that are, in general, deleterious to vascular function (Gluais et al., 2005).

Prostacyclin is not only a potent platelet inhibitor preventing their aggregation and their adhesion to the endothelial cell surface and an endothelium-derived vasodilator (Moncada et al., 1976; 1977; Moncada and Vane, 1979; Radomski et al., 1987a,b;) but is also involved in renal function by regulating renal blood flow, glomerular filtration rate and renin release. In addition, prostacyclin is involved the development, transport and implantation of the embryo, in pain tolerance, in gastric acid secretion, and in intracelullar signalling by interacting with nuclear receptors and regulating gene transcription (Smyth and FitzGerald, 2002; Wise, 2003; Wu and Liou, 2005; Smyth et al., 2009).

Prostacyclin is the preferential ligand of IP receptors (Figure 1), and its effects often, but not always, involve the activation of adenylyl cyclase and the subsequent elevation of intracellular cyclic AMP (Wise and Jones, 1996; Clapp et al., 2002; Wise, 2003; Orie et al., 2006). The genetic deletion of IP receptors is associated with increased injury-induced restenosis (Cheng et al., 2002), thrombotic events (Murata et al., 1997), atherosclerosis (Egan et al., 2004; Kobayashi et al., 2004) and reperfusion injury (Xiao et al., 2001). Some of the negative effects observed with IP receptor deletion, including restenosis and enhanced platelet activation, can be abrogated by coincidental TP receptor deletion indicating that prostacyclin regulates the cardiovascular effects of thromboxane A2 (Fetalvero et al., 2007). A cross-talk with the two prostanoid binding sites may involve the heterodimerization of these two receptors. TP receptor agonists stimulating these heterodimers would produce accumulation of cAMP. This and regulation of receptor endocytosis and trafficking would explain some of the braking effects of IP receptors on the cellular effects of TP receptor activation (Wilson et al., 2004; 2007;). In humans, a defect in prostacyclin receptor signalling, as observed in patients with a dysfunctional IP receptor mutation, leads to accelerated atherothrombosis (Arehart et al., 2008). The accelerated cardiovascular disease associated with this IP^R212C mutation is observed in individuals carrying only one copy of the variant allele and thus can be attributed to a dominant-negative action not only when dimerized with the wild-type IP receptor but also through dimerization with the TP receptor; activation of such dimers no longer leads to cAMP production (Ibrahim et al., 2010).

Selective and potent agonists and antagonists of the IP receptors have been synthesized. IP receptor antagonists could be of interest for the treatment of pain, inflammation and overactive bladder providing that cardiovascular side effects could be avoided (Jones et al., 2009). IP receptor agonists are prescribed for pulmonary hypertension but are currently administered intravenously or by inhalation and are associated with serious side effects. The recent synthesis of orally active and long-acting IP receptor agonists may prove beneficial not only in pulmonary hypertension but possibly also in atherosclerosis obliterans (Fetalvero et al., 2007; Kuwano et al., 2007).

Thromboxane A₂ and TP receptors

Thromboxane A₂ is a highly unstable intermediate enzymatically produced from PGH₂ as a substrate by thromboxane synthase, which belongs also to the cytochrome P450 superfamily (in the human, CYP5; Yokoyama et al., 1991). Thromboxane A₂ not only induces platelet aggregation and contraction of vascular smooth muscle but is also involved in allergies, modulation of acquired immunity, atherogenesis, neovascularization and metastasis of cancer cells (Nakahata, 2008).

Thromboxane A₂ is the preferential ligand of TP receptors (Figure 1), but PGH₂ and higher concentrations of other PGs, isoprostanes and hydroxyeicosatetraenoic acids (HETEs) can activate them with a various range of potency (Félétou et al., 2010b; Figure 2). By contrast, epoxyeicosatrienoic acids (EETs), which act as endothelium-derived hyperpolarizing factors in some vascular beds (Félétou and Vanhoutte, 2006b), and their dihydro-derivatives (DiHETs) are endogenous antagonists of TP receptors (Behm et al., 2009). The transduction signal associated with TP receptor activation involves mainly two types of G-proteins, the Gq and the G13 families, resulting in the stimulation of phospholipase C and RhoGEF respectively. In humans, but not in rodents, two alternatively spliced isoforms of TP, TPα and TPβ, have been described, which differ only in their C-terminal portions. Both isoforms are coupled to phospholipase C activation, but TPα stimulates adenylyl cyclise, whereas TPβ inhibits it, at least in transfected cells (Hirata et al., 1996). Furthermore, TP receptors can be engaged in cross-talks with receptor tyrosine kinases, such as the EGF receptor, to induce cell proliferation and differentiation (Nakahata, 2008). Reactive oxygen species enhance the stability and increase the density of functional TP receptors at the cell membrane (Valentin et al., 2004; Wilson et al., 2009), and, in endothelial cells, the activation of TP receptors inhibits NO production (Liu et al., 2009). Metabolically stable agonists of TP receptors with high affinity and potency have been synthesized and numerous synthetic TP antagonists have been designed including S 18886 (terutroban), which is an orally active, potent and selective antagonist of TP receptors (Simonet et al., 1997).

Mice genetically deficient in TP receptors are normotensive but have abnormal vascular responses to thromboxane A₂ and show a tendency to bleeding (Thomas et al., 1998). The deletion of TP receptors decreases vascular proliferation and platelet activation in response to intimal lesions (Cheng et al., 2002), delays atherogenesis in apoE^-/- mice (Kobayashi et al., 2004), prevents angiotensin-II- and L-NAME-induced hypertension and the associated cardiac hypertrophy but reduces the extent of kidney injury only in the former hypertensive model (Francois et al., 2004; 2008;). TP receptor knockout mice are also protected against various lipopolysaccharide-induced responses such as the increase in iNOS expression (Yamada et al., 2003), acute renal failure (Boffa et al., 2004) and inflammatory tachycardia (Takayama et al., 2005).

The phenotype of thromboxane A₂ synthase knockout mice is much less pronounced (Yu et al., 2004) most likely because thromboxane A₂ is only one of the endogenous agonists of TP receptor and also because the deletion of this enzyme may redirect the arachidonic cascade towards less pathogenic synthases.

COX products and endothelium-dependent relaxations

Since inhibitors of COX abolish the basal and stimulated generation of PGI₂, and potent and selective antagonists of the IP receptor block vasodilator responses that it causes (Gluais et al., 2005; Gomez et al., 2008), the contribution of PGI₂ in endothelium-dependent responses can be assessed. This PG plays a role in flow-mediated vasodilatation (Koller et al., 1993; Duffy et al., 1998), but its contribution to acute endothelium-dependent relaxations in response to neurohumoral mediators is often considered as minimal, because COX inhibitors, in particular indomethacin, do not affect these responses. However, the role of prostacyclin as an endogenous mediator of endothelium-derived relaxation has been generally overlooked since in some vascular beds, a major vasodilator effect of COX derivatives can only be observed when the other pathways leading to endothelium-dependent relaxations have been inhibited (Corriu et al., 1996; Zygmunt et al., 1998). The contribution of PGI₂ to endothelium-dependent responses is increased in eNOS knockout mice (Chataigneau et al., 1999; Sun et al., 1999). Similarly, in human with cardiovascular diseases, COX-2-derived PGs can play a compensatory role for the decreased NO bioavailability (Bulut et al., 2003; Szerafin et al., 2006) possibly explaining some of the detrimental cardiovascular effects associated with COX-2 inhibitors (Andersohn et al., 2006).

The vascular relaxation to PGI₂, or its synthetic analogues, is often associated with a concomitant hyperpolarization of the smooth muscle cells, which, depending on the blood vessels and the species, can involve the opening of various populations of potassium channels (Corriu et al., 2001; Félétou and Vanhoutte, 2007). Therefore, in numerous vascular beds, PGI₂ can act also as an endothelium-derived hyperpolarizing substance (Parkington et al., 2004).

Although PGE₂ can be a potent vasodilator and is acutely released by endothelial cells (Gluais et al., 2005), there is little evidence for a role of this PG in endothelium-dependent relaxations, at the possible exception of some rabbit veins (Rouaud et al., 1999). The concomitant activation of contractile EP1 and/or EP3 receptors could mask a potential relaxing effect of endothelium-derived PGE₂ in other blood vessels.

COX products and endothelium-dependent contractions

COX-1

Hypertension

Endothelium-dependent contractions, associated with endothelial dysfunctions, were observed first in the isolated aorta of spontaneously hypertensive rats (SHR; Luscher and Vanhoutte, 1986). They have been extensively characterized thereafter in that blood vessel. In the arteries of this hypertensive model, the generation of a diffusible EDCF opposes the relaxing effect of NO. Endothelium-dependent contractions are positively correlated with the severity of hypertension and the aging process, are delayed in female SHR and also occur in aging normotensive Wistar–Kyoto control rats (WKY; Félétou et al., 2009).

In SHR aorta, endothelium-dependent contractions are associated with multiple dysfunctions in both the endothelial and the smooth muscle cells. In the endothelial cells, they include (i) abnormal calcium handling, (ii) an increased expression of COX-1, (iii) the associated enhanced production of reactive oxygen species, (iv) a major increase in prostacyclin synthase expression, (v) the enhanced release of prostacyclin, thromboxane A₂, and possibly PGH₂. In vascular smooth muscle cells, they include (i) an exacerbated response of the TP receptor to prostacyclin and PGH₂, (ii) a deficient IP receptor function and (iii) an early dysfunction in the adenylyl cyclase pathway (Félétou et al., 2009; 2010a,b;).

When compared with WKY aorta and in response to receptor-mediated stimuli (acetylcholine), the amplitude of the endothelium-dependent contractions and the increase in intracellular calcium ([Ca²⁺]_i) in SHR endothelial cells are exacerbated while in response to receptor-independent stimuli (calcium ionophore, A 23187) the maximal amplitude of the endothelium-dependent contractions and the changes in [Ca²⁺]_i in both strains are similar (Gluais et al., 2005; 2006; Tang et al., 2007). Any event leading to an increase in endothelial [Ca²⁺]_i, activates the calcium-dependent phospholipase A₂ (cPLA₂) and provokes the mobilization of arachidonic acid. However, in response to receptor-dependent stimuli, the activation of the calcium-independent phospholipase A₂ (iPLA₂) allows the store-operated calcium channels (SOC)–dependent influx of extracellular calcium and the subsequent activation of cPLA₂. It mediates the initial part of the signalling cascade leading to endothelium-dependent contractions of the SHR aorta in response to acetylcholine. Substances, such as calcium ionophores, that bypass the cell membrane receptors causes an increase in [Ca²⁺]_i, and a direct activation of cPLA₂ (Wong et al., 2010b). Therefore, the iPLA₂ pathway associated with calcium mobilization is defective in SHR endothelial cells (Figure 4).

Figure 4.

Calcium signalling and the COX-1 production of endothelium-derived contracting factors (EDCF). Acetylcholine (ACh) activates muscarinic receptors (M) on the endothelial cell membrane and triggers the release of calcium from intracellular stores. The resulting calcium depletion process displaces the inhibitory calmodulin (CaM) from iPLA2. Activated iPLA2 produces lysophospholipids (LysoPL), which in turn open store-operated calcium channels (SOCs) leading to the influx of extracellular calcium into the endothelial cells. This large influx of calcium ions then activates cPLA2, which catalyses the production of arachidonic acids (AA). The later is then metabolized by cyclooxygenase-1 (COX-1) to prostanoids. cPLA2, calcium dependent phospholipase A2; EC, endothelial cells; iPLA2, calcium independent phospholipase A2 (modified from Wong and Vanhoutte, 2010).

The subsequent steps involve the activation of COX and the production of reactive oxygen species along with that of prostanoids. Aortic endothelial cells express preferentially COX-1 versus COX-2 (Kawka et al., 2007; Tang and Vanhoutte, 2008). In SHR endothelial cells, the mRNA and protein expression of COX-1 are enhanced when compared with that of WKY, and in the two strains, both are augmented by aging (Ge et al., 1995; Tang and Vanhoutte, 2008). Conversely, in SHR, the decrease expression of COX-1 produced by a chronic treatment with vitamin D reduces the endothelium-dependent contractions (Wong et al., 2010a). In response to acetylcholine, endothelium-dependent contractions and the associated generation of PGs are blocked consistently by selective inhibitors of COX-1 and partially inhibited, although to various extent depending on the experimental conditions, by selective inhibitors of COX-2 (Ge et al., 1995; Yang et al., 2003a; Gluais et al., 2005; 2006;). However, if the endothelium-dependent contractions and the release of PGs by A 23187 are also fully blocked by COX-1 inhibitors, these responses are less sensitive to COX-2 inhibition (Figure 5). This could possibly be explained by the fact that low concentrations of arachidonic acid are preferentially oxygenated by COX-2, while higher ones are preferentially metabolized by COX-1 (Morita, 2002). Alternatively, the effects observed with the COX-2 inhibitors could nevertheless be attributed to COX-1 inhibition. Indeed, the ability of COX-2 inhibitors to inhibit COX-1 depends obviously not only on the degree of selectivity of any given inhibitor but also on other factors such as substrate availability, endogenous lipid peroxide levels and plasma protein concentration, explaining why COX-2 inhibitors are systematically more potent in preventing the endothelial production of PGI₂ than the platelet production of thromboxane A₂ (Mitchell et al., 2006; Warner et al., 2006). In agreement with a preponderant role for COX-1 in endothelium-dependent contractions, these responses are abolished in aortae taken from COX-1 knockout mice, while they are maintained in aortic rings of COX-2 knockout animals (Tang et al., 2005).

Figure 5.

Effects of inhibitors of COX-1 and COX-2 on endothelium-dependent contractions and prostaglandins production in SRH aortic rings. Top panels: effects of acetylcholine. Lower panels: effects of A 23187. The effects of A23187 are less sensitive to the COX-2 inhibitor, NS 398, than those produced by acetylcholine. Data are shown as means ± SEM. The asterisk indicates a statistically significant effect of a COX inhibitor. The sharp sign indicates that the response in presence of the COX-1 inhibitor, SC 560, is significantly different from the response observed in the presence of NS 398. COX, cyclooxygenase; SHR, spontaneously hypertensive rat.

Additionally, COX is also involved in the endothelial generation of reactive oxygen species, a key factor in the generation of endothelium-dependent contractions (Yang et al., 2002; Tang et al., 2007). Reactive oxygen species decrease NO bioavailability (Gryglewski et al., 1986; Rubanyi and Vanhoutte, 1986) and, as a positive feedback loop, the formation of hydroperoxides further activates COX (Morita, 2002). In addition, since reactive oxygen species diffuse towards the vascular smooth muscle cells, they can stimulate COX in these cells and produce more contractile prostanoids.

The generated PGs diffuse towards the vascular smooth muscle cells and directly activate TP receptors (Luscher and Vanhoutte, 1986; Auch-Schwelk et al., 1990; Yang et al., 2003a). In the rat aorta, the five major PGs and 8-isoprostane produce contractions that predominantly involve TP receptor activation (Figure 2). However, the involvement of PGD₂ and 8-isoprostane in endothelium-dependent contractions can be ruled out since their generation is not affected by acetylcholine (Gluais et al., 2005).

In SHR aortic endothelial cells, the expression of thromboxane synthase is enhanced when compared with that in WKY endothelium (Tang and Vanhoutte, 2008). In response to ATP or the calcium ionophore A 23187, this is associated with an increase generation of thromboxane A₂, and the endothelium-dependent contractions are partially inhibited by dazoxiben, a selective inhibitor of thromboxane synthase that abrogates the production of thromboxane A₂ (Gluais et al., 2006; 2007;). By contrast, acetylcholine produces only a minor dazoxiben-sensitive increase in thromboxane A₂ production, and the endothelium-dependent contractions that it evokes are not affected by the presence of the thromboxane synthase inhibitor, indicating that thromboxane A₂ is only one of the EDCFs that can be released from SHR aortic endothelial cells (Koga et al., 1989; Kato et al., 1990; Ge et al., 1995; Gluais et al., 2005; 2006; 2007;).

Paradoxically, prostacyclin is likely to be a major EDCF in SHR aorta. In SHR endothelial cells, prostacyclin is by far the most abundant PG released in response not only to receptor-dependent stimuli but also to calcium ionophores (Gluais et al., 2005; 2006; 2007;). This may come as a surprise since prostacyclin synthase is rapidly nitrosylated and inactivated by peroxynitrite (Zou et al., 2002a,b; Schmidt et al., 2003). However, in the SHR aorta, the massive increase in the expression of prostacyclin synthase (Tang and Vanhoutte, 2008) may compensate the loss of activity due to peroxynitrite-dependent tyrosine nitration. Furthermore, in that preparation, prostacyclin does not produce relaxations but evokes TP receptor–dependent contractions (Rapoport and Williams, 1996; Gluais et al., 2005; Figure 2). In fact, prostacyclin, like PGH₂, is also more potent in producing contraction in SHR than in WKY aortae (Ge et al., 1995; Gluais et al., 2005). The absence of relaxation in response to prostacyclin is attributed to an early (as young as 12 weeks old) dysfunction of the IP receptors of vascular smooth muscle. This dysfunction is tissue specific since the platelet response to prostacyclin (or its analogues) is unaffected or even enhanced (Anand-Srivastava, 1993; Gomez et al., 2008). In order to explain this specific smooth muscle cell dysfunction, a decrease in the aortic expression of IP receptors (Numaguchi et al., 1999) and an early impairment of adenylyl cyclase signalling have been evoked (Anand-Srivastava, 1988; Masuzawa et al., 1989). However, these two hypotheses can only, at best, partially explain the total disappearance of IP receptor–mediated relaxations in SHR aorta. Indeed, the decrease expression of the IP receptor has not been confirmed in latter experiments (Tang and Vanhoutte, 2008), and when compared with WKY, the relaxations to prostacyclin in SHR aorta are much more severely affected than those produced by other agents that stimulate adenylyl cyclase, such as isoproterenol and forskolin (Gomez et al., 2008). A potential additional/alternative hypothesis, which requires proper validation, could be the oxidative damage of the IP receptor itself, which contain redox-sensitive cysteines that play an essential role in determining its structure, addressing and function (Stitham et al., 2006).

Prostacyclin has also been identified as a major contributing factor accounting for the endothelial dysfunction in the aorta and mesenteric artery of WKY and SHR treated with aldosterone (Blanco-Rivero et al., 2005; Xavier et al., 2008). Thus, although as a rule prostacyclin is a vasodilator and an anti-aggregating agent, depending on the circumstances, the prostanoid can also act as an EDCF.

Any levels of prostacyclin synthase inactivation would theoretically lead to an excess of free PGH₂. Since PGH₂ is the second most potent agonist at TP receptors and is more effective in activating TP receptors in vascular smooth muscle from SHR than in that of WKY, the endoperoxide is also a suitable candidate as EDCF (Kato et al., 1990; Ge et al., 1995; Gluais et al., 2005; Figure 2). Finally, the shunting of PGH₂ metabolism towards other metabolic pathways can lead to a variety of products, including PGE₂ and/or PGF_2α, which also produce contractions by activating TP receptors (Figure 2). Therefore, thromboxane A₂, PGH₂, PGI₂, PGE₂ and PGF_2α can all act theoretically as EDCF (Gluais et al., 2005; Félétou et al., 2010a,b;).

In addition, in the SHR aorta, PGE₂-mediated relaxations are impaired, which could contribute to the observed endothelial dysfunction (Tang et al., 2008) and, in the femoral artery of diabetic rats, activation of the EP1 receptor contributes to the endothelium-dependent contractions (Shi et al., 2007).

Furthermore, some alterations at the level of the TP receptors should also be considered. Hydrogen peroxide prevents the translocation and degradation of TP receptors, increasing their density at the cell membrane and TP activation enhances TP stability through a reactive oxygen species–dependent post-transcriptional mechanism (Valentin et al., 2004; Wilson et al., 2009). This may explain the enhanced TP receptor–dependent contractions in response to PGH₂, prostacyclin and exogenously generated reactive oxygen species observed in SHR aorta (Auch-Schwelk et al., 1989; Ge et al., 1995; Yang et al., 2002; 2003b;, Gluais et al., 2005; García-Redondo et al., 2009). In addition, TP receptors are also expressed in endothelial cells and their stimulation induces the Rho kinase–dependent inhibition of NO production (Liu et al., 2009). Conversely, the isoform α of the human TP receptor is negatively and independently regulated by either NO or prostacyclin, following the phosphorylation of serine residues by protein kinase G and A respectively (Reid and Kinsella, 2003). Additionally, NO can inhibit the activity of thromboxane synthase (Wade and Fitzpatrick, 1997), indicating that a decrease in NO bioavailability may facilitate the TP receptor–dependent signalling pathway. Finally, EDCF- and TP-mediated responses, first observed in the aorta of the SHR, are not ubiquitous in SHR arteries but have been reported in other vascular territories such as the mesenteric, skeletal muscle and renal vascular beds (Félétou et al., 2009). In these peripheral arteries, the endothelial dysfunction additionally includes a marked attenuation of the EDHF-mediated component of the endothelium-dependent relaxations (Félétou and Vanhoutte, 2006b). TP receptor stimulation induces a loss in the activity of endothelial small conductance calcium-activated potassium channels (Crane and Garland, 2004; McNeish and Garland, 2007), an essential component of EDHF-mediated responses (Félétou and Vanhoutte, 2006b). Conversely, the impairment of EDHF-mediated responses can favour the development of endothelium-dependent contractions (Michel et al., 2008).

COX-2

COX-2 is traditionally believed to be the major generator of the vasodilator PGI₂, which functions especially when NO bioavailability is diminishing. Endothelial dysfunction is reversed by a complementary up-regulation of COX-2 expression and activity in the mesenteric vascular bed of mice with streptozotocin-induced diabetes (Nacci et al., 2009), while attenuated NO-dependent vasodilatations in diabetic patients may be compensated for by the emerging vasodilator effect of prostacyclin (Meeking et al., 2000). Compared with patients with non-documented diabetes, coronary arterioles from diabetic patients show a significant up-regulation of COX-2 expression, which contributes to enhanced bradykinin-induced vasodilatations (Szerafin et al., 2006). The availability of selective COX-2 inhibitors and genetically engineered mice allow a more in-depth investigation on the emerging role of COX-2 as an inflammatory mediator releasing vasoconstrictors in hypertension, diabetes and aging.

Hypertension and diabetes

Vasculopathies are the leading causes of morbidity and mortality in hypertensive and diabetic patients. In addition to COX-1, COX-2 also can generate vasoconstrictor prostanoids in the SHR endothelial cells. In both WKY and SHR endothelial cells, the induction of COX-2, not only in the aorta but also in resistance arteries, is accelerated by aging and can be associated with the generation of endothelium-derived contractile prostanoids. (Heymes et al., 2000; Alvarez et al., 2005; Blanco-Rivero et al., 2005; Shi et al., 2008; Virdis et al., 2009). For instance, in WKY rats, the impairment of the aortic endothelium-dependent relaxations, observed after a chronic treatment with fenofibrate, has been attributed to the endothelial release of COX-2-derived PGE₂ acting on smooth muscle TP receptors (Blanco-Rivero et al., 2007). In L-NAME-hypertensive rats, the up-regulation of COX-2 enhances the release of EDCFs (Qu et al., 2010). Additionally, the release of COX-2-derived PGF_2α and 8-isoprostane augments α-adrenoceptor-induced contractions in SHR arteries (Alvarez et al., 2005), and COX-2 also mediates vasoconstrictions in response to tert-butyl hydroperoxide, a product of lipid peroxidation (Garcia-Cohen et al., 2000). Elevated COX-2 expression contributes to deoxycorticosterone acetate salt–induced hypertension, which is reversed by intra-peritoneal administration of the preferential COX-2 inhibitor NS-398 (Adeagbo et al., 2005). Recent findings of Wong et al. (2010c) further support the pathological role of COX-2 in endothelial dysfunction in hypertension. Bone morphogenic protein 4 (BMP4) and COX-2 are elevated in the renal arteries of hypertensive patients. BMP4, by increasing NADPH oxidase–derived reactive oxygen species and activating p38 mitogen-activated protein kinase, up-regulates COX-2 in the murine aorta, resulting in the impairment of endothelium-dependent relaxations and appearance of endothelium-dependent contractions (Figure 6). A similar association is observed in SHR intrarenal arteries and renal arteries from hypertensive patients, in which both noggin (BMP4 antagonist) and celecoxib (selective COX-2 inhibitor) normalize the relaxations and/or abolish endothelium-dependent contractions.

Figure 6.

Exaggerated role of COX-2 in hypertension and aging. In hypertension, BMP-4 expression is elevated, resulting in the up-regulation of COX-2 expression and activity via NAPDH oxidase-mediated generation of reactive oxygen species and the subsequent activation of p38 MAPK. During aging, COX-2 expression, release of and vascular contractility to the COX-2-derived PGF_2α are augmented. Under these conditions, NO bioavailability is diminished, thus favouring the emergence of the exaggerated endothelium-dependent contractions. AA, arachidonic acid; ACh, acetylcholine; BMP4, bone morphogenic protein 4; COX-2, cyclooxygnease-2; EC, endothelial cells; EDC, endothelium-dependent contraction; EDR, endothelium-dependent relaxation; eNOS, endothelial nitric oxide synthase; NO, nitric oxide; PG, prostaglandin; PLA₂, phospholipase A₂.

While ROS up-regulate the expression and activity of COX-2, the latter in turn can increase oxidative stress in the vascular wall. Thus, COX-2-derived PGF_2α and superoxide anions underpin the diminished acetylcholine-induced relaxations in perfused mesenteric arteries of female Wistar rats with alloxan-induced diabetes (Akamine et al., 2006). Mesenteric arteries of type 2-diabetic Otsuka Long-Evans Tokushima Fatty (OLETF) rats exhibit attenuated endothelium-dependent relaxations and enhanced acetylcholine-induced contractions, attributed to the elevated expression of COX-1 and COX-2 and the release of thromboxane A₂ and PGE₂ (Matsumoto et al., 2007); these abnormal responses are reversed in part by chronic oral treatment with eicosapentaenoic acid (omega-3 fatty acid), metformin (oral anti-diabetic drug), pyrrolidine dithiocarbamate (thiol antioxidant) or ozagrel (thromboxane synthase inhibitor). The reversals are due to the suppression of extracellular signal-regulated kinase (ERK) and nuclear factor-κB (NF-κB)–mediated COX-2 expression (Matsumoto et al., 2008; Matsumoto et al., 2009a,b,c;).

Nephropathies and renovascular abnormalities are common in diabetic patients (Mogensen and Schmitz, 1988; Kamgar et al., 2006). Both type 1 and type 2 diabetic rats exhibit exaggerated renal cortical expression and activity of COX-2, accompanied by pathological haemodynamic alterations (Komers et al., 2001; 2005;). Preliminary data in human renal arteries from diabetic subjects reveal acetylcholine- and PGF_2α-induced vasoconstrictions, whereby the muscarinic agonist stimulates the celecoxib-sensitive liberation of PGF_2α (Wong et al., 2009), indicating a pivotal role of COX-2 in diabetic renovascular pathologies in humans.

Aging

COX-2 can play a constitutive role in the regulation of vascular tone. Thus, L-NAME-treated aortic rings from young hamsters show COX-2- and TP receptor–mediated endothelium-dependent contractions. In that preparation, endothelial COX-2 is constitutively expressed and generates PGF_2α (Wong et al., 2009). With aging, diminished NO-mediated endothelium-dependent relaxations favour the appearance of COX-dependent vasoconstrictions. Studies on aortae from old hamsters show that aging not only enhances COX-2-dependent endothelium-dependent contractions but also the expression of the isoform as well as the release of and the vascular responsiveness to PGF_2α (Wong et al., 2009). Due to a reduced NO bioavailability (as reflected by attenuated endothelium-dependent relaxations) in the aorta of the aged hamster, endothelium-dependent contractions can be observed under physiological conditions, that is in the absence of pharmacological inhibition of eNOS (Wong et al., 2009). The preferential COX-2 inhibitor NS-398 also restores normal endothelium-dependent relaxations and abolishes acetylcholine-induced contractions in small mesenteric arteries from aged rats (Alvarez de Sotomayor et al., 2007). Apparently, both COX-1 and COX-2 contribute to the exaggerated endothelium-dependent contractions to the calcium ionophore A23187 in femoral arteries from aged rats (Shi et al., 2008). Indeed, mRNA levels of COX-1, COX-2, thromboxane synthase, PGF synthase, haematopoietic-type PGD synthase and membrane PGE synthase-2 are all augmented in endothelial cells from aged rats (Tang et al., 2008), further indicating an increasing importance of the arachidonic acid–COX cascade during aging.

COX-2 derived PGF2α acting on TP receptor

Although PGF_2α is the natural agonist of the FP receptor, TP receptors appear to be the preferential cellular target of PGF_2α in inducing vasoconstrictions, possibly because of different relative expression levels of these receptors in the vasculature. Indeed, while TP receptors are well expressed in the hamster aorta and human renal arteries, FP receptors are nearly undetectable (Wong et al., 2009). As a consequence and because of the chemical similarity between various prostanoids, when PGF_2α is released in sufficient quantities, TP receptor activation ensues, leading to vasoconstriction.

Conclusions and perspectives

Aging and cardiovascular diseases are associated with multiple endothelial dysfunctions, which often involve COX activation and TP receptor stimulation. Generally, the production of EDCFs does not significantly influence systemic arterial blood pressure, but TP activation amplifies the endothelial dysfunction (Félétou et al., 2009).

In humans, the production of COX-derived EDCF is a characteristic of the aging blood vessels, with essential hypertension merely causing an earlier onset and an acceleration of this endothelial dysfunction (Taddei et al., 1993; 1995; 1997a,b;). In primary hypertensive patients, selective inhibition of COX-1 partially reverses the impairment of vasodilator responses to acetylcholine, while the selective inhibition of COX-2, which does not produce adverse effects in the forearm of healthy subjects (Verma et al., 2001), further reduces the increase in forearm blood flow produced by the muscarinic agonist (Bulut et al., 2003). These results indicate not only that COX-1-derived vasoconstrictor PGs contribute to the endothelial dysfunction, but that the production of vasodilator PGs by COX-2 is of minor importance in subjects with normal endothelial function but becomes relatively more important in hypertensive patients with endothelial dysfunction, presumably playing a beneficial compensatory role. By contrast, in patients with coronary artery disease, the impaired acetylcholine- and flow-induced forearm vasodilatation is restored by the administration of a TP receptor antagonist (Belhassen et al., 2003). The fact that the patients in this latter study were already treated with aspirin suggests that COX-2 activity, rather than COX-1, should be the main source of the vasoconstrictor prostanoids involved in this endothelial dysfunction. Indeed, in patients with severe coronary artery disease, COX-2 inhibition improved flow-mediated dilatation (Chenevard et al., 2003). These results indicate that, as it has been observed in animal models, COX-1, COX-2 or both isoforms can contribute to endothelial dysfunctions. Since in most cases, the activation of TP receptors is the common downstream effector, selective antagonists of this receptor could be of therapeutic interest in the treatment of cardiovascular disorders.

Glossary

Abbreviations

ATP

adenosine triphosphate

BMP4

bone morphogenic protein-4

iPLA₂

calcium-independent phospholipase A₂

cAMP

cyclic adenosine monophosphate

COX-1

cyclooxygenase-1

COX-2

cyclooxygenase-2

cPLA₂

cytosolic phospholipase A₂

DiHET

dihydroxyeicosatrienoic acids

EDCF

endothelium-derived contracting factor

EDHF

endothelium-derived hyperpolarizing factor

EETs

epoxyeicosatrienoic acids

EP

prostaglandin E₂ receptor

eNOS

endothelial nitric oxide synthase

ERK

extracellular signal-regulated kinase

FP

prostaglandin F_2α receptor

iNOS

inducible nitric oxide synthase

LPS

lipopolysaccharides

NADPH

nicotinamide adenine dinucleotide phosphate

NF-κB

nuclear factor κB

NO

nitric oxide

OLEFT

Otsuka Long-Evans Tokushima fatty rats

PPAR

peroxisome proliferator-activated receptor

PLA₂

phospholipase A₂

IP

prostacyclin receptor

PG

prostaglandin

PGD₂

prostaglandin D₂

DP

prostaglandin D₂ receptor

PGE₂

prostaglandin E₂

PGES

prostaglandin E synthase

cPGES

cytosolic PGE synthase

mPGES

membrane-bound

PGE-synthase

PGH₂, prostaglandin H₂, endoperoxide

PGF_2α

prostaglandin F_2α

PGG₂

prostaglandin G₂

PGI₂

prostaglandin I₂ or prostacyclin

PGIS

prostacyclin synthase

PGJ₂

prostaglandin J₂

sPLA₂

secreted phospholipase A₂

SHR

spontaneously hypertensive rat

SOC

store-operated calcium channels

TP

thromboxane/endoperoxide receptor

WKY

Wistar-Kyoto rats

Conflict of interest

Michel Félétou is an employee of a pharmaceutical company, the ‘Institut de Recherches Servier’, currently developing a TP receptor antagonist. Paul M. Vanhoutte is a former employee and is a consultant for this company.