Prostanoid function and cardiovascular disease

Prostanoids, including prostaglandins (PGs) and thromboxanes (TXs) are synthesized from arachidonic acid by the combined action of cyclooxygenases (COXs) and PG and TX synthases. Finally after their synthesis, prostanoids are quickly released to the extracellular medium exerting their effects upon interaction with prostanoid receptors present in the neighbouring cells. These agents exert important actions in the cardiovascular system, modulating vascular homeostasis and participating in the pathogenesis of vascular diseases as thrombosis and atherosclerosis. Among prostanoids, Tromboxane (TX)A₂, a potent platelet activator and vasoconstrictor and prostacyclin (PGI2), a platelet inhibitor and vasodilator, are the most important in controlling vascular homeostasis. Although multiple studies using pharmacological inhibitors and genetically deficient mice have demonstrated the importance of prostanoid-mediated actions on cardiovascular physiology, further analysis on the prostanoid mediated actions in the vascular system are required to better understand the benefits and risks for the use of COX inhibitors in cardiovascular diseases.

Keywords: Prostaglandins; thromboxanes; cyclooxygenases; prostanoid receptors; NSAIDs; cardiovascular physiopathology

Abbreviations COX Cyclooxygenase DP PGD receptor EP PGE receptor FP PGF receptor IL Interleukin IP PGI receptor NSAIDs non-steroidal anti-inflammatory drugs PG prostaglandin PGES PGE synthase PGDS PGD synthase PGFS PGF synthase PGIS PGI synthase TP Thromboxane receptor TX Thromboxane TXS Thromboxane synthase

Arachidonic acid, liberated from membrane phospholipids by several phospholipases, is metabolized by the sequential action of cyclooxygenases (COX) and prostaglandin or thromboxane synthases to produce the diverse classes of prostanoids (Simmons et al., [64]; Smith et al., [65]). These agents are a group of bioactive lipids including prostaglandins (PGs) and thromboxanes (TXs) that play a very important role in many physiological and pathological processes, including inflammation (Williams et al., [84]), cancer (Subbaramaiah & Dannenberg, [68]), angiogenesis (Iñiguez et al., 2003) and cardiovascular diseases (FitzGerald et al., [24]; Davidge, [20]; Alfranca et al., [1]; Belton & Fitzgerald, [4]). COX enzymes catalyse the formation of an unstable endoperoxide intermediate, PGH2, which in turn can be metabolized by cell-specific isomerases and synthases to a range of eicosanoids with potent and diverse biological effects, as PGD2, PGE2, PGF2α, PGI2, and TXA2. Prostanoids released to the extracellular medium exert their biological effects in an autocrine or paracrine fashion upon interaction with prostanoid receptors present in target cells (Figure 1).

Graph: Figure 1. Biosynthesis of prostanoids. Arachidonic acid is liberated from the membrane phospholipids by phospholipase A2 (PLA2) and converted to PGH2 and then to PGG2 by cyclooxygenases (COX-1 or COX-2). These enzymes are the target of NSAIDs that inhibit PG and TX synthesis. Glucocorticoids diminish prostanoid production through the inhibition of COX-2 transcriptional induction by inflammatory stimuli. Subsequent conversion of PGH2 to different prostanoids (PGs, prostaglandins and TXs, thromboxanes) is catalyzed by the respective synthases and isomerases. Prostanoids signal through G protein-coupled receptors with seven trans-membrane domains. PGE2 signals via four different receptors (EP1, EP2, EP3 and EP4). PGD2, PGF2α, PGI and TXA2 bind to the DP and CRTH2, FP, IP and TP receptors respectively.

In the cardiovascular system, prostanoids have been shown to modulate the pathogenesis of vascular diseases as thrombosis and atherosclerosis through a variety of processes, including platelet aggregation, vasorelaxation and vasoconstriction, local inflammatory response and leukocyte-endothelial cell adhesion (Vila, [78]; Kobayashi et al., [43]). Multiple studies using NSAIDs as pharmacological inhibitors of COXs and genetically deficient mice have revealed the importance of prostanoid-mediated actions on cardiovascular physiology (Belton & Fitzgerald, [4]; Burleigh et al., [10]; FitzGerald & Patrono, [25]; Fosslien, [27]; Ray et al., [56]; Martinez-Gonzalez & Badimon, [46]). However, recent withdrawal of COX-2 selective inhibitors from the clinic because of their adverse effects in patients with potential cardiovascular risk has opened a debate about the role of COX–derived prostanoids in vascular pathologies and the benefits and risks for the use of COX inhibitors in cardiovascular diseases.

Two main COX isoenzymes have been identified, named COX-1 and COX-2, encoded by two separate genes (Smith et al., [65]; Simmons et al., [64]; Warner & Mitchell, [82]). A COX-3 enzyme derived from alternative splicing of the COX-1 gene has been also described, although its role is still unclear (Chandrasekharan et al., [16]). COX-1 is constitutively expressed in most cell types and tissues being responsible for vascular, renal and gastric homeostasis (Morita, [48]). In contrast, COX-2 expression is generally induced at sites of inflammation by many stimuli that includes among others, pro-inflammatory cytokines as interleukin (IL)-1β, tumour necrosis factor (TNF)-α, growth factors, mechanical stress, oxidized lipids, free radicals and bacterial products (Smith et al., [65]; Tanabe & Tohnai, [69]). Those stimuli activate many transcription factors such as nuclear factor (NF)-κB, NF-IL6 (C/EBP), cAMP response element-binding protein (CREBP), Activator protein (AP)-1, interferon regulatory factors (IRFs) and nuclear factor of activated T cells (NFAT) (Tanabe & Tohnai, [69]; Blanco et al., [7]; Iniguez et al., [36]) that induce COX-2 transcription in different cell types. COX-2 transcripcional induction can be inhibited by anti-inflammatory and immunosuppressive drugs as glucocorticoids and Cyclosporin A (Iniguez et al., [37]; Kujubu & Herschman, [44]) as well as by anti-inflammatory cytokines such as IL-4, IL-10 and IL-13 (Niiro et al., [53]; Diaz-Cazorla et al., [21]). Increasing research on the physiological roles of COX-2 has revealed that, in addition to inflammation, this isoform also plays an important role in renal homeostasis, mucosal defence, reproduction and vascular homeostasis (Morita, [48]; Simmons et al., [64]; Warner & Mitchell, [82]).

Cyclooxygenases are the target of non-steroidal anti-inflammatory drugs (NSAIDs). According to their selectivity in the inhibition of both COX isoforms, these drugs can be divided into classic NSAIDs, such as aspirin and ibuprofen that inhibit both isoforms at standard doses, and highly selective COX-2 NSAIDs as those belonging to the family of "coxibs". The fact that COX-1 was mainly involved in house-keeping functions and COX-2 was associated to inflammation, led to the theory that inhibition of COX-1 may account for some of the unwanted side effects of these drugs such as gastrointestinal and renal toxicity whereas the ability of NSAIDs to inhibit COX-2 activity may explain their therapeutic effects as anti-inflammatory drugs. Therefore, most of the new research on anti-inflammatory drugs was aimed at targeting the COX-2 inducible production of PGs (Dannhardt & Kiefer, [19]; Simmons et al., [64]). As a consequence, new drugs with high selectivity against COX-2 such as Celecoxib and Rofecoxib and the newest coxibs Etoricoxib, Valdecoxib, and Lumiracoxib, have been developed with the idea of providing safer anti-inflammatory compounds with reduced gastrointestinal and renal toxicity. However, although these drugs have been proved to be potent anti-inflammatory compounds with less gastric damage than conventional, non-selective COX inhibitors, their use is still associated with significant gastrointestinal injury and renal side effects. (Warner et al., [81]; Rostom et al., [60]; Coruzzi et al., [15]). Particular concerns have emerged by the fact that some coxibs increase the risk of serious cardiovascular events, including myocardial infarction and stroke (Fosslien, [27]; Caldwell et al., [12]; Ray et al., [56]). This has led to the withdrawal of rofecoxib from the market, and has raised serious questions about the cardiovascular safety of NSAIDs (Kearney et al., [42]; Martinez-Gonzalez & Badimon, [46]; McGettigan & Henry, [47]). It is also important to note that, although it is generally accepted that these drugs exert their anti-inflammatory and analgesic actions through inhibition of COX enzymatic activity, increasing evidence has shown that certain actions of NSAIDs could be mediated through mechanisms independent of cyclooxygenase activity and prostaglandin production that may be relevant to their effects in vivo (Tegeder et al., [71]).

Despite the importance of COX isoenzymes in prostanoid generation, the actual prostanoid profile synthesized by a particular cell type is determined by the presence of different downstream PG and TX synthases and isomerases (Helliwell et al., [33]) (Figure 1). A particular cell type may predominantly express a particular isomerase, which will largely determine which prostanoid is generated. Moreover, some cell types can express more than one isomerase and a particular COX isoenzyme can couple to different isomerases in the same cell (Ueno et al., [75]). Thus, COX-1 preferentially couples with Thromboxane synthase (TXS) and PGF2α synthase (PGFS), while COX-2 is associated to Prostacyclin synthase (PGIS) and microsomal PGE2 synthase-1 (mPGES-1) (Murakami et al., [49]; Ueno et al., [75]). Evenmore, PGH2 and other endoperoxides generated in one cell type may be secreted and metabolized by isomerases present in surrounding cells by the so-called "transcellular metabolism" (Karim et al., [41]).

Three types of PGE synthases (PGES) participating in the synthesis of PGE2 have been described: one cytosolic (cPGES) and two membrane-associated PGE synthases (mPGES)-1 and -2. The (cPGES) belongs to the glutathione transferase family, is ubiquitously expressed and is preferentially functionally coupled to COX-1, thus, involved in housekeeping production of PGE2 (Murakami et al., [51]; Tanioka et al., [70]). In contrast, mPGES-1, which belongs to the membrane-associated proteins involved in Eicosanoid and Glutathione metabolism (MAPEG) superfamily, is inducible by similar stimuli that induce COX-2, being its induction also suppressed by glucocorticoids. Moreover, mPGES-1 appears functionally coupled with COX-2 and its induction is usually co-ordinated with COX-2 (Samuelsson et al., [62]). A second type of membrane-bound microsomal PGES is the mPGES-2, able to couple with both COX isoenzymes, although its functional significance is still unclear (Murakami et al., [50]).

PGD2 is synthesized from PGH2 by the action of PGD synthases (PGDS) of which there are two isoforms. The haematopoietic PGDS is present in mast cells, basophils, and in a subset of T helper cells (Th2) and may play a role in allergy (Kanaoka & Urade, [40]). The lipocalin-type PGD synthase is mostly expressed in brain and it is especially abundant in the cerebrospinal fluid (Urade & Hayaishi, [76]). Interestingly, PGD2 can be further metabolized by dehydration to PGJ2, delta12-PGJ2, and 15-deoxy-delta(12,14)-PGJ2, which is a ligand for the nuclear transcription factor peroxisomal proliferator activated receptor (PPAR)-gamma having anti-inflammatory activity (Scher & Pillinger, [63]).

PGF2α can be synthesized from PGH2 by PGH 9-11-endoperoxide reductase or PGF synthase (PGFS). In addition, it can be generated from PGE2 by PGE 9-ketoreductase or from PGD2 by PGD-11-ketoreductase (Watanabe, [83]).

Prostacyclin, PGI2, is produced by the action of PGI synthase (PGIS) which can be induced by mechanical stress, via activation of AP-1 (Wu & Liou, [85]).

TXA2 is synthesized from PGH2 by thromboxane synthase (TXS), which is constitutively expressed in platelets as well as other blood cells and various tissues as kidney, lung, liver and placenta. Its regulation takes place mainly at the transcriptional level (Wang & Kulmacz, [79]).

Prostanoids actions, with the possible exception of cyclopentenone PGs, are mediated through cell surface receptors. These are seven membrane spanning G-protein coupled receptors that are classified into five basic types, named according to their ligands: DP for PGD2, EP for PGE2, FP for PGF2, IP for PGI2, and TP for TXA2 (Alfranca et al., [1]; Hata & Breyer, [32]).

DP (also named DP1) is weakly expressed in brain despite the important action of PGD2 in this organ. It is moderately expressed in the ileum with very weak expression in the lung. DP2, also called CRTH2 (chemoattractant receptor-homologous molecule expressed on Th2 cells), is expressed on Th2 immune cells being a chemoattractant receptor for Th2 lymphocyte subset in allergic inflammation (Kabashima & Narumiya, [39]).

FP is most abundantly expressed in the corpus luteum according to the role played by PGF2α in menstrual cycle. It is also expressed in the heart, lung, kidney, and stomach, although its expression in these tissues does not vary during the estrous cycle as does in the corpus luteum (Tsuboi et al., [74]). There are several splice variants of FP (FPA and FPB) but its particular role is unclear.

IP is expressed in larger quantities by dorsal root ganglion neurons but also by platelet precursors (megakaryocytes) as well as in the smooth muscle cells of arteries consistent with the important action of PGI2 in the cardiovascular system (Smyth & FitzGerald, [66]).

PGE2 has four receptors, EP1 to 4. In addition, there are several splice variants of EP3. More importantly, each one is linked to a different transduction pathway that may even give rise to opposite effects (activation or inhibition) on cellular responses (Alfranca et al., [1]). Thus, EP1 induces an inhibition of adenylate cyclase leading to a decrease in cAMP whereas EP2 and EP4 receptors activate this enzyme. On the other hand, EP3 is coupled to Gαq and its activation results in intracellular calcium increase. EP3 and EP4 receptors have a wide distribution throughout the body, being expressed in almost all tissues examined. In contrast, of EP1 is restricted to the lung, kidney and stomach, and EP2 is scarcely expressed.

In the cardiovascular system, IP, DP, EP4 and EP2 receptors that mediate cAMP increases are termed "relaxant" receptors, whereas TP, FP, and EP1 receptors, which induce calcium mobilization, represent the "contractile" receptor group. EP3, which reduces cAMP levels, has been named the "inhibitory" receptor (Alfranca et al., [1]).

Nuclear actions of prostanoids have also been reported. Thus, PGJ2 derivatives can bind and activate peroxisome proliferator-activated receptors (PPARs) nuclear transcription factors (Straus & Glass, [67]). There are four types of PPARs (alpha, delta, gamma-1, and gamma-2), which may bind various PGs with different sensitivity (Berger & Moller, [6]). In this way, PGs may also act as intracellular signalling molecules and regulate gene expression (Ide et al., [35]). Generally PPARs induces transcription of anti-inflammatory genes as well as inhibits activation of pro-inflammatory ones (Straus & Glass, [67]). Besides, cyclopentenone PGs (that includes PGA1) have a reactive ring which may lead to have receptor-independent actions through redox alterations (Grau et al., [30]).

TXA2 and PGI2 are thought to be the most important prostanoids in controlling the homeostasis of the cardiovascular system, as proposed more than 20 years ago by Bunting, Moncada and Vane (Bunting et al., [9]; Davidge, [20]; Cheng et al., [17]). They are synthesized by blood platelets and vascular endothelium, respectively, and have opposed biological activities. TXA2 is potent vasoconstrictor as well as a potent inducer of platelet adhesion and regulates renal hemodynamics and sodium handling by the kidney (FitzGerald et al., [24]). On the other hand, PGI2 is the predominant prostanoid produced by cells of the vasculature, having an important vasodilator effect by promoting renal sodium excretion and regulating the growth of vascular smooth muscle cells. Moreover, it opposes TXA2 effects on platelets inhibiting their aggregation and action (Helliwell et al., [33]; Yokoyama et al., [87]).

Due to this, many aspects of cardiovascular disease have traditionally been explained by alterations in the balance between PGI2 and TXA2 during the interactions between platelets and vessel wall (Vane, [77]) although recent studies argue about this theory (Flavahan, [26]). Since platelets express COX-1 but not COX-2, it is inferred that TXA2 production by TXAS uses PGH2 derived from COX-1. In this regard, platelet TXA2 is blunted in COX-1 deficient mice, which have decreased platelet aggregation and thrombosis (Yu et al., [88]). Mice deficient in the TP (Cheng et al., [17]; Thomas et al., [72]) or human patients with a genetic disorder in the TP receptor (Fuse et al., [29]; Hirata et al., [34]) exhibit an increase in bleeding tendency and resistance to platelet aggregation, confirming the role of TXA2 in those activities and consistent with the effects of TP antagonists in humans. Conversely, transgenic mice specifically over-expressing TP in the vasculature results in placental ischemia during pregnancy and suppression of TXA2 formation rescue the phenotype (Rocca et al., [58]). Since TXA2 is a mitogen of vascular smooth muscle cells, TP KO mice have decreased vascular proliferation and platelet response after artery injury to the carotid although they are normotensive (Cheng et al., [17]).

Whether PGI2 is synthesized through COX-1 or COX-2 has been of paramount importance in classical cardiovascular research. Initial reports indicated that vascular endothelial cells and smooth muscle cells have COX-1, and that PGI2 can be formed through COX-1 which was supported by the fact that in endothelial cells, PGIS co-localizes with COX-1 but not with COX-2 (Flavahan, [26]; Warner & Mitchell, [82]). However, PGIS expression can be induced together with COX-2 in endothelial cells by haemodynamic shear stress or by oxidized low density lipoproteins (oxLDLs) leading to PGI2 production (Norata et al., [54]; Topper et al., [73]). Moreover, in healthy individuals COX-2 specific inhibitors reduce PGI2 metabolites in urine without affecting TXA2 metabolites (FitzGerald & Patrono, [25]). This suggests that COX-2 is a major source of PGI2 in the human cardiovascular system and plays a role under physiological conditions. The predominant coupling of PGI2 to COX-2 suggests that COX-2 specific inhibitor could be prothrombotic, since anti-aggregatory PGI2 circulating levels within would be reduced. This has been proposed to explain the putative associations between COX-2 inhibitors and the development of cardiovascular complications (Belton & Fitzgerald, [4]; FitzGerald, [23]). In contrast to TP deficient mice, IP knockout mice have normal blood pressure and develop normally without suffering from spontaneous thrombosis. However, after endothelial damage they show an enhanced thrombotic response compared to control littermates (Murata et al., [52]), indicating that PGI2 does play a major role only in response to stress and not in the basal systemic circulation. Mice lacking IP have enhanced vascular smooth cell proliferation and platelet activation in response to vascular injury likely caused by enhanced TXA2/PGI2 ratio (Cheng et al., [17]). Simultaneous deletion of the TP and the IP abrogated both increased responses (Cheng et al., [17]). Selective inhibition, knockout, mutation or deletion of COX-2 or IP has been shown to accelerate thrombogenesis. These responses were attenuated by COX-1 knock down (Cheng et al., [18]). IP KO mice also show vascular hyperplasia and increased vascular remodeling (Rudic et al., [61]) as well as cardiac hypertrophy and fibrosis (Francois et al., [28]). Taken together those results indicate that PGI2 regulates the cardiovascular activity of TXA2 and further support the hypothesis that cardiovascular homeostasis is resulting from the balance between these two eicosanoids.

Atherosclerosis is one of the most important diseases of the developed countries and can be considered as a multifactorial inflammatory disease triggered by high levels of cholesterol in serum leading to an inflammation in the intima of large arteries and involving several cell types including immune cells as T lymphocytes and monocytes/macrophages as well as endothelial and smooth muscle cells and platelets (Ross, [59]; Hansson & Libby, [31]). Among those, monocytes/macrophages play an important key role in many phases of atherogenic process. Thus, after an atherogenic stimulus, monocyte-macrophages reversibly adhere to the endothelium and migrate across it, leading to a prolonged retention of those cells in the intima which is central in atherogenesis. A variety of substances including prostanoids have been implicated in the pathogenesis of atherosclerosis.

Prostanoids may be involved in atherosclerosis through its ability to regulate a variety of mechanisms potentially involved in the pathogenesis such as inflammation, vasodilatation, vasoconstriction, platelet aggregation and leukocyte–endothelial cell adhesion, and leukocyte migration among others (Vila, [78]; Kobayashi et al., [43]). In this regard, COX-2 expression has been observed in symptomatic atherosclerotic lesions (Cipollone et al., [14]). This enzyme may play a dual role in the pathogenesis of the atherosclerosis. Initially, COX-2 expression is induced in monocytes by pro-inflammatory cytokines and several growth factors. Then, COX-2-mediated PG production by activated macrophages may promote atherosclerosis in the artery wall through several mechanisms, as induction of other pro-inflammatory mediators or by favouring migration of macrophages and other immune cells or by induction of adhesion molecules. Later on, COX-2-derived PGI2, likely from the endothelial cells, may have a protective role in atherogenesis by favouring vasodilatation. Moreover, in atherosclerotic patients, PGI2 can be formed through the action of both COX-1 and COX-2 (Flavahan, [26]). COX-1 but not COX-2 is expressed in normal arteries, whereas both isoforms are expressed in atherosclerotic lesions. In those lesions COX-2 is expressed not only by monocyte/macrophages, but also by endothelial and proliferating smooth muscle cells (Belton & Fitzgerald, [5]). Atherogenic lipoproteins, such as oxidized low density lipoproteins (oxLDL) may promote atherosclerosis first by contributing to inflammation by activating monocytes/macrophages and later by stimulating lipid uptake by macrophages, leading to foam cell formation. Interestingly, oxLDL has apparently contradictory effect on macrophages, since they can activate the expression of some pro-inflammatory genes whereas reduce COX-2 expression (Banfi et al., [3]). In agreement with this is the fact that macrophage-derived foam cells from the atherosclerotic lesions in mice, did not express COX-2. Thus induction COX-2 expression in macrophages takes place before their transformation into foam cells in the plaques.

In spite of the clear involvement of COX-2 derived prostanoids in vascular atherosclerosis, results on the effect of COX-2 selective inhibitors on the formation and progression of atherosclerotic plaques are controversial. Results from different studies have shown increased, reduced as well as unaltered atherogenesis (Belton & Fitzgerald, [5]; Burleigh et al., [10]; FitzGerald & Patrono, [25]; Fosslien, [27]; Ray et al., [56]; Martinez-Gonzalez & Badimon, [46]). Those discrepant effects of COX-2 may reflect the dual role of COX-2 in promoting early, but protecting late atherosclerotic lesions, mentioned above. Results about the influence of COXs in atherosclerosis in animal models, apart from those using pharmacological inhibitors, have been obtained by means of cell transplantation from fetal liver or bone marrow from COX-1 or COX- 2 deficient mice into ApoE or LDLR KO mice (Babaev et al., [2]; Burleigh et al., [10]; Burleigh et al., [11]). The size of the atherosclerotic lesions was significantly reduced when cells deficient in COX-2 were transplanted compared to the mice transplanted with wild type fetal liver cells (Burleigh et al., [10]; Burleigh et al., [11]). Those results implicate COX-2 expression in the macrophage and not in other cells as endothelial cells, smooth muscle cells, or T-cells in promoting atherosclerotic lesion formation. Efforts to obtain information from mice deficient in both COX-2 and ApoE or LDLR genes have been unsuccessful due to the severe renal defects of COX-2 mice. Regarding to COX-1, both proatherosclerotic and anti-atherosclerotic roles also have been reported (Babaev et al., [2]; Pratico et al., [55]).

Various prostanoids may play a significant role in the atherogenic process (Reiss & Edelman, [57]; Vila, [78]). Patients with extensive disease as well as murine models of atherosclerosis have enhanced formation of TXA2. Even more, TP antagonists decrease atherogenesis in mice. TP deficiency in atherosclerotic mice models induced a significant delay in atherogenesis, compared with mice deficient in apoE alone (Kobayashi et al., [43]).

In contrast, PGI2 may theoretically have a beneficial effect in the atherogenic process by limiting platelet adhesion to the endothelium and activation in the plaques. In accordance with this, local delivery of PIGS gene through an adenoviral vector reduces the platelet deposition seen following vascular injury (Yamada et al., [86]). Genetic deletion of the IP receptor, both in the LDRL and in the ApoE KO mice models, aggravated atherogenesis (Egan et al., [22]; Kobayashi et al., [43]). As platelets are though to contribute to the development and progression of atherosclerosis in the late phase, PGI2 may likely suppress lesion formation by limiting platelet deposition. Those mice exhibited a significant acceleration in atherogenesis with enhanced platelet activation and increased rolling of leukocytes on the vessel walls. Those results indicate that TXA2 promotes whereas PGI2 prevents the initiation and progression of atherogenesis by modulating platelet activation and leukocyte-endothelial cell interaction.

In addition to PGI2 and TXA2, other PGs, such as PGE2 and PGD2, could also play an important role in the pathophysiology of atherosclerosis. In this regard, vasoconstrictor responses to PGE2 are greatly increased in atherosclerosis (Lopez et al., [45]) and PGE2 may also promote angiogenesis (Iniguez et al., [38]). Moreover, inducible mPGES-1 has been detected in activated macrophages in atherosclerotic lesions where it colocalizes with COX-2 both in mice and humans (Cipollone et al., [14]). Disruption of this enzyme in mice reduced foam cell formation and atherosclerosis in fat-fed LDLR−/− mice. mPGES-1 deletion augmented both PGIS and TXS expression in endothelial cells (Wang, M. et al., [80]). On the other hand, COX-2 derived PGs as PGD2 may possess anti-inflammatory and anti-atherosclerotic properties in such a way that the balance between PGDS and PGES has been shown to be a major determinant of atherosclerotic plaque instability (Cipollone et al., [13]).

Based on the hypothesis that atherosclerosis is an inflammatory disease, it was proposed that COX inhibition by NSAIDs, and in particular selective inhibition of COX-2 by the Coxibs family of NSAIDs, might have protective and even anti-atherogenic effects. However, clinical studies have indicated that there is an increased risk of atherothrombosis in individuals taking these drugs in such a way that some of them as Rofecoxib have been recently withdrawn from the market (Bombardier et al., [8]; McGettigan & Henry, [47]; Caldwell et al., [12]; Kearney et al., [42]). These undesirable effects of COX-2 selective inhibitors have been explained by the TXA2/PGI2 balance theory by which selective inhibition of COX-2 lead to a reduction in the production of the vasodilator PGI2 whereas production of the vasoconstrictor and pro-aggregatory TXA2, mostly COX-1–dependent, remains unaffected. Thus, disruption of the physiological balance between TXA2 and PGI2 accelerates atherosclerosis and increases the risk of thrombosis and other cardiovascular complications (Figure 2).

Graph: Figure 2. The balance hypothesis for the cardiovascular effects of NSAIDs. Cardiovascular homeostasis is controlled by the balance between COX-1 – dependent production of TXA2 by platelets, and COX-2 – mediated PGI2 (prostacyclin) production by endothelial cells. Whereas TXA2 function as a potent platelet activator and vasoconstrictor, PGI2 inhibits platelet aggregation and thrombosis. Classical non-selective NSAIDs reduce the production of TXA2 by platelets through their ability to inhibit COX-1 activity thus displaying anti-thrombotic properties. In contrast, COX-2 selective NSAIDs reduce PGI2 formation by endothelial cells consequently disturbing the equilibrium between TXA2 and PGI2 and potentially favouring pro-thrombotic cardiovascular events.

Nevertheless, the mechanisms underlying the pathogenesis of cardiovascular complications upon NSAIDs administration remain to be clarified as the TXA2/PGI2 balance theory is somewhat simple and it does not consider some important clinical and experimental evidences as the increased risk of cardiovascular events by some non-selective NSAIDs, the contribution of other prostanoids to the overall effect in cardiovascular physiopathology, and the COX-2 independent effects of NSAIDs. In this sense, a more profound knowledge of the complex relations between prostanoids-mediated actions and function of cells in the cardiovascular system is required to clearly understand the benefits and risks of NSAIDs on cardiovascular diseases.

The authors thank the financial support of Comunidad Autónoma de Madrid (S-SAL-0159-2006), Ministerio de Educación y Ciencia-FEDER (SAF 2004-05109, BFU2004-04157 and BFU2007-62659), European Commission (EICOSANOX integrated project LSH-CT-2004-005033; and MAIN network of excellence) and Laboratorios del Dr. ESTEVE S.A. We apologize that many valuable studies, especially original contributions, have not been cited due to space limitation.

Declaration of interest: The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.