The endocannabinoid system

The EC system constitutes an endogenous signalling system that plays a pivotal role in a variety of centrally and peripherally regulated physiological processes. The system comprises G-protein-coupled cannabinoid (CB) receptors, their endogenous CB (EC) ligands and the associated enzymatic apparatus that controls their synthesis and degradation (as reviewed in⁽ Reference Montecucco and Di Marzo ^{1 )}). The two recognised G-protein coupled receptors are the CB₁ and CB₂ receptors⁽ Reference Matsuda, Lolait and Brownstein ^{2 ,} Reference Munro, Thomas and Abu-Shaar ^{3 )}; CB₁ receptors are predominantly expressed in the central nervous system, but are also located in numerous peripheral tissues including cardiac⁽ Reference Batkai, Pacher and Osei-Hyiaman ^{4 ,} Reference Pacher, Batkai and Kunos ^{5 )} vascular⁽ Reference Gebremedhin, Lange and Narayanan ^{6 ,} Reference Daly, Ross and Whyte ^{7 )} and adipose⁽ Reference Nam, Lee and Kim ^{8 )} tissue. CB₂ receptors, while expressed primarily in the periphery, particularly by immune cells⁽ Reference Munro, Thomas and Abu-Shaar ^{3 )}, have also been reported to be present in both the myocardium⁽ Reference Defer, Wan and Souktani ^{9 )} and endothelial cells⁽ Reference Rajesh, Mukhopadhyay and Batkai ^{10 )}. However, evidence from functional studies also suggests the existence of non-CB receptor targets of the EC, including transient receptor potential vanilloid type-1 channels, PPAR⁽ Reference Pacher, Batkai and Kunos ^{5 )}, the vascular ‘anandamide’ receptor⁽ Reference Zygmunt, Petersson and Andersson ^{11 )} and the recently de-orphanised receptor GPR55⁽ Reference Pertwee ^{12 )}.

The very presence of these receptors in cardiovascular tissue implies that they are there to mediate responses to endogenous ligands, either as part of cardiovascular homoeostasis, or to participate in pathological processes either as a stress response or as a causative factor. Indeed, endogenous ligands do exist that activate these receptors, the most studied of which are anandamide (AEA) and 2-arachidonyl glycerol (2-AG) and will therefore be the main focus of this review. However, it is worthy of note that various additional endogenous CB receptor ligands have been identified, including docosatetraenyl-ethanolamide, N-arachidonoyl dopamine, virodhamine and noladin (reviewed in⁽ Reference Hiley ^{13 )}).

Endocannabinoid synthesis and sites of action

AEA was first shown to be synthesised from the phospholipid precursor N-arachidonyl-phosphatidyl ethanolamine (NAPE), via hydrolysis of NAPE by a calcium (Ca²⁺) sensitive, NAPE-selective phospholipase D⁽ Reference Di Marzo, Fontana and Cadas ^{14 )}. However, alternate routes (Fig. 1) of AEA biosynthesis have since been identified, including conversion of NAPE to 2-lyso-NAPE (via phospholipase A₂) and subsequently to AEA through a Ca²⁺ independent mechanism⁽ Reference Sun, Tsuboi and Okamoto ^{15 )}, and also hydrolysis of NAPE by phospholipase C to yield the phospho-AEA, which is further hydrolysed to produce AEA. The synthesis of 2-AG (Fig. 2), which is generated from arachidonic acid-containing phospholipids, can be synthesised via (i) the production of diacylglycerol from phosphatidylinositol via phospholipase Cβ and subsequent hydrolysis by sn1-diacylglycerol lipase to yield 2-AG⁽ Reference Kondo, Kondo and Nakane ^{16 )} and (ii) through generation (via phospholipase A₁) of lyso-phosphatidylinositol from phosphatidylinositol and subsequent hydrolysis by a lyso-phosphatidylinositol selective phospholipase C to yield 2-AG (reviewed in⁽ Reference Muccioli ^{17 )}). In terms of receptor activation by these endogenous ligands (reviewed in⁽ Reference Pertwee and Ross ^{18 )} ), AEA is a partial agonist at CB receptors, with a marginally higher affinity and markedly higher efficacy for the CB₁ receptor compared with the CB₂ receptor. 2-AG, on the other hand, has a similar affinity for both CB receptors but has a higher efficacy at the CB₂ receptor compared to AEA.

Fig. 1. Synthetic pathways for anandamide (AEA). NAPE, N-arachidonyl-phosphatidyl ethanolamine; pAEA, phospho-anandamide; NAPE-PLD, NAPE-selective phospholipase D; PLA₂, phospholipase A2; PLC, phospholipase C.

Fig. 2. Synthetic pathways for 2-arachidonyl glycerol (2-AG). Lyso-PI, lyso-phosphatidylinositol; DAG, diacylglycerol; PLA₁, phospholipase A₁; PLCβ, phospholipase C-beta; sn-1-DAGL, sn-1 specific diacylglycerol lipase.

There are now numerous reports that both circulating and tissue EC levels are raised in a variety of cardiovascular-related pathologies, including cerebral⁽ Reference Berger, Schmid and Schabitz ^{19 )}, hepatic⁽ Reference Kurabayashi, Takeyoshi and Yoshinari ^{20 )} and myocardial⁽ Reference Wagner, Hu and Bauersachs ^{21 )} ischaemia/reperfusion (I/R) injury, heart failure⁽ Reference Weis, Beiras-Fernandez and Sodian ^{22 )}, diabetic cardiomyopathy⁽ Reference Rajesh, Batkai and Kechrid ^{23 )}, advanced atherosclerosis⁽ Reference Sugamura, Sugiyama and Nozaki ^{24 )} and obesity-related cardiovascular dysfunction⁽ Reference Quercioli, Pataky and Vincenti ^{25 )}. However, the nature of the specific EC, and its precise role (i.e. whether it is protective or detrimental) varies widely between pathological conditions. For the purposes of this review, the discussion will focus on the potential role of the EC in the setting of three cardiovascular disorders, namely hypertension, atherosclerosis and acute myocardial I/R.

Interactions between long-chain-PUFA and endocannabinoids

There is a significant literature concerning the cardio- and vasculo-protective effects of n-3 LC-PUFA and various underlying mechanisms have been proposed including: anti-inflammatory and anti-oxidant effects; modulation of cardiac ion channels; reduction of TAG; influence on membrane microdomains and downstream cell signalling pathways; improved cardiac mitochondrial function; antithrombotic and antiarrhythmic effects (reviewed in⁽ Reference Adkins and Kelley ^{26 –} Reference Stanley, Khairallah and Dabkowski ^{29 )} ). What is interesting is that the beneficial effects of n-3 LC-PUFA are observed in very similar settings to those in which changes in levels of EC are observed, which begs the question as to whether or not there are shared or integrated mechanisms through which both the n-3 LC-PUFA and the EC exert their effects. Indeed, there is building evidence of interactions between the n-3 LC-PUFA and activation of the EC system (recently reviewed in⁽ Reference Meijerink, Balvers and Witkamp ^{30 ,} Reference Kim, Li and Watkins ^{31 )}, although the findings are variable in terms of whether or not n-3 LC-PUFA raise or lower EC levels. Berger et al ⁽ Reference Berger, Crozier and Bisogno ^{32 )} were one of the first groups to demonstrate a definitive effect of n-3 LC-PUFA ingestion on raising brain levels of AEA in piglets, and this has since been extended in studies in mice where DHA supplementation significantly altered EC-related metabolites in plasma and brain⁽ Reference Wood, Williams and Pandarinathan ^{33 )}. In contrast, other studies have shown that n-3 LC-PUFA deficiency increases, while supplementation decreases, 2-AG levels in the brains of mice⁽ Reference Watanabe, Doshi and Hamazaki ^{34 )}. Thus the impact of n-3 LC-PUFA on EC levels appears to be dependent upon the EC in question. In terms of the effects of n-3 LC-PUFA on EC levels in peripheral tissues the picture appears to be largely opposite to the effects on central EC levels, as dietary n-3-PUFA supplementation, given as either fish oil⁽ Reference Batetta, Griinari and Carta ^{35 )} or krill oil⁽ Reference Piscitelli, Carta and Bisogno ^{36 )} in models of either high-fat feeding or metabolic syndrome, decreases both AEA and 2-AG levels in adipose and heart tissue, effects that are associated with an anti-obesogenic effect and an improvement in glucose tolerance⁽ Reference Rossmeisl, Jilkova and Kuda ^{37 )}. These findings raise the possibility that modulation of the EC system by n-3-PUFA may be an important part of their protective mechanisms of action. The remainder of this review will therefore attempt to synthesise the evidence for the protective mechanisms of ECs and n-3-PUFA to identify where similarities exist and where they do not.

Could the endocannabinoid system be a mechanism for the effects of n-3 long-chain-PUFA in hypertension?

Both AEA⁽ Reference Varga, Lake and Huangfu ^{38 –} Reference Pacher, Batkai and Osei-Hyiaman ^{40 )} and 2-AG⁽ Reference Jarai, Wagner and Varga ^{41 )} elicit complex vasodilatory⁽ Reference Malinowska, Baranowska-Kuczko and Schlicker ^{42 )} and cardio-depressive responses in vivo that are sensitive to inhibition by CB₁ receptor antagonists⁽ Reference Varga, Lake and Martin ^{43 )} and are absent in CB₁-knockout mice⁽ Reference Jarai, Wagner and Varga ^{41 )}. Together this implies that the CB₁ receptor is responsible for mediating the cardiovascular responses to the EC. The simple fact, however, that CB receptor antagonists do not elicit significant blood pressure responses⁽ Reference Batkai, Pacher and Osei-Hyiaman ^{4 ,} Reference Hiley ¹³ ,Reference Lake, Martin and Kunos ^{39 ,} Reference Varga, Lake and Martin ^{43 ,} Reference Wagner, Varga and Ellis ^{44 )}, and that blood pressure in CB₁ knockout mice is comparable with wild-type controls⁽ Reference Jarai, Wagner and Varga ^{41 )}, suggests that EC do not play a tonic role in blood pressure maintenance. However, in hypertensive rats CB₁ receptor antagonists increase blood pressure, whereas inhibitors of AEA metabolism normalise blood pressure⁽ Reference Batkai, Pacher and Osei-Hyiaman ^{4 )}, suggesting that under situations of pathophysiological (i.e. hypertensive) stress the EC are produced as a compensatory mechanism. While, as far as the authors are aware, no determination of EC levels has been made in either hypertensive animal models or in clinical samples from hypertensive patients, the increased expression of CB₁ receptors in both the heart and endothelium of hypertensive rats⁽ Reference Batkai, Pacher and Osei-Hyiaman ^{4 )} is suggestive of an up-regulation of the EC system aimed at redressing the balance. The blood pressure lowering effects of EC in hypertension have been attributed to a combination of a tonic suppression of cardiac contractility (and thus cardiac output) along with a direct vasodilator effect.

The effect of n-3 LC-PUFA on blood pressure has long been an issue of controversy. Recent analysis of clinical and epidemiological studies⁽ Reference Cabo, Alonso and Mata ^{45 )} shows that, while very high (≥3 g/d) doses of n-3 LC-PUFA do produce a small but significant decrease in blood pressure (especially systolic blood pressure), this is evident only in certain groups such as older hypertensive subjects and individuals with hypertriglyceridaemia⁽ Reference Cicero, Derosa and Di Gregori ^{46 )}. Moreover, lower doses of n-3 LC-PUFA as a single treatment are ineffective in lowering blood pressure in mild essential hypertensive patients⁽ Reference Russo, Olivieri and Girelli ^{47 )} or individuals with metabolic syndrome⁽ Reference Cicero, Derosa and Di Gregori ^{46 )}. Thus, like the EC system, n-3 LC-PUFA only appear to modify blood pressure in the setting of hypertension.

The blood pressure lowering effects of EC are thought to be mediated through an endothelial mechanism involving the release of nitric oxide (NO)⁽ Reference Rossmeisl, Jilkova and Kuda ^{37 )}, via both CB₁ ⁽ Reference Malinowska, Kwolek and Gothert ^{48 )} and non-CB₁/CB₂ (possibly vanilloid) receptors⁽ Reference Malinowska, Kwolek and Gothert ^{48 ,} Reference McCollum, Howlett and Mukhopadhyay ^{49 )}. Conversely, the proposed mechanism(s) by which n-3 LC-PUFA reduce blood pressure are numerous (reviewed in⁽ Reference Cicero, Ertek and Borghi ^{50 )} ) and include effects on sodium excretion, interference with the renin-angiotensin system and enhancement of endothelial NO production, although with respect to the latter there is a degree of ambiguity between experimental findings. In vitro studies on isolated bovine and ovine blood vessel preparations show that EPA relaxes blood vessels through endothelial NO release⁽ Reference Omura, Kobayashi and Mizukami ^{51 ,} Reference Singh, Kathirvel and Choudhury ^{52 )}. However, studies in human subjects are inconsistent with this and have reported no effect of long-term n-3 LC-PUFA supplementation (0·45–3·4 g/d EPA+DHA) on flow-mediated dilatation and/or arterial stiffness (surrogate markers of endothelial function) in either normal subjects⁽ Reference Sanders, Hall and Maniou ^{53 ,} Reference Skulas-Ray, Kris-Etherton and Harris ^{54 )}, hypertensives⁽ Reference Woodman, Mori and Burke ^{55 )} or patients with peripheral vascular disease⁽ Reference Mackay, Ford and Thies ^{56 )}, although microvascular endothelial function is improved in individuals with type II diabetes⁽ Reference Stirban, Nandrean and Gotting ^{57 )}.

At present there is no data around whether or not n-3 LC-PUFA influence EC levels in the vasculature in hypertension. However, the profiles of EC and n-3 LC-PUFA in the setting of hypertension differ substantially: n-3 LC-PUFA are not particularly effective as either anti-hypertensive or NO-releasing agents, while EC are potent endothelium (and presumably NO) dependent vasodilators. On this basis, current evidence suggests that there is no correlation between n-3 LC-PUFA and EC in hypertension.

Could the endocannabinoid system contribute to the anti-atherosclerotic effects of n-3 long-chain-PUFA?

There is a growing evidence base for a role of the EC system in atherosclerotic lesion progression. CB₂ activation has been associated with an anti-atherogenic effect on the basis of the following: (i) the phytocannabinoid ∆⁹-tetrahydrocannabinol reduces atherosclerotic lesion progression and suppresses leucocyte adhesion to the vascular wall in high-fat fed ApoE^−/− mice, an effect that is abolished by co-administration with a CB₂ selective antagonist (SR144528)⁽ Reference Steffens, Veillard and Arnaud ^{58 )} ; (ii) the CB₂ agonist JWH-133 attenuates smooth muscle cell proliferation and migration⁽ Reference Rajesh, Mukhopadhyay and Hasko ^{59 )} and (iii) the suppressant effect of the CB₁/CB₂ receptor agonist WIN 55,212-2 on TNFα and superoxide production in human peripheral blood mononuclear cells is sensitive to CB₂ receptor blockade⁽ Reference Germain, Boichot and Advenier ^{60 )}. However, more recent data have shown that neither CB₂ receptor activation with a selective agonist (JWH-133) nor CB₂ receptor gene deletion modulate atherogenesis in high-fat fed LDLR^−/− mice⁽ Reference Willecke, Zeschky and Ortiz Rodriguez ^{61 )}. In contrast to the picture with CB₂ receptors, evidence that CB₁ activation results in atherogenesis is derived from the findings that patients with advanced coronary artery disease exhibit elevated circulating levels of both AEA and 2-AG along with increased CB₁ receptor expression in coronary plaques⁽ Reference Sugamura, Sugiyama and Nozaki ^{24 )}. Taken together with the findings that the CB₁ selective antagonist rimonabant (i) attenuates lesion development in a murine model of atherosclerosis via mechanisms involving suppression of proinflammatory gene expression and macrophage recruitment⁽ Reference Dol-Gleizes, Paumelle and Visentin ^{62 )} and (ii) inhibits vascular smooth muscle cell proliferation and migration⁽ Reference Rajesh, Mukhopadhyay and Hasko ^{63 )}, this has led to the proposal that EC exert a pro-atherogenic effect, signalling through CB₁ receptors. However, it should be borne in mind that all of the afore-mentioned studies employed phyto- or synthetic CB ligands, which may behave differently from the endogenous EC, AEA and 2-AG. In fact, AEA attenuates TNFα-induced expression of inter-cellular adhesion molecule 1 and vascular cell adhesion molecule 1 in human coronary artery endothelial cells and attenuates TNFα-stimulated human acute monocytic leukaemia cells monocyte adhesion, both actions being sensitive to CB₁ and CB₂ receptor blockade⁽ Reference Rajesh, Mukhopadhyay and Batkai ^{64 )}, suggesting that AEA at least exhibits effects that are more in line with an anti-, rather than pro-, atherogenic effect. With regard to 2-AG, there is in fact a dearth of literature about its effects on either atherogenesis or the cellular events involved in the atherogenic process.

The impact of n-3 LC-PUFA intervention in atherosclerosis progression has long been a subject of contention, but there is now sufficient evidence from several large-scale randomised trials (DART Trial⁽ Reference Burr, Fehily and Gilbert ^{65 )}, GISSI-Prevenzione Study^{( 66 )}, JELIS Study⁽ Reference Yokoyama, Origasa and Matsuzaki ^{67 )} ) to demonstrate the effectiveness of n-3 LC-PUFA supplementation in the primary and secondary prevention of CHD. The main effects of n-3 LC-PUFA appear to involve modulation of processes key to atherosclerosis progression and plaque stablisation⁽ Reference von Schacky, Angerer and Kothny ^{68 ,} Reference Sekikawa, Curb and Ueshima ^{69 )} via both direct (i.e. at the level of the plaque) and indirect (i.e. through alterations in lipid metabolism) mechanisms. Since the aim of this article is to identify similarities between the actions of n-3 LC-PUFA and EC, and since EC influences in atherogenesis are related to direct effects on the developing plaque, this discussion will be confined to the direct effects of n-3 LC-PUFA on plaque stability, which are predominantly of an anti-inflammatory and antioxidant nature.

Dietary fish-oil intake has been documented to lower chemoattractant (platelet-derived growth factor and monocyte chemoattractant protein 1) production in mononuclear cell fractions⁽ Reference Baumann, Hessel and Larass ^{70 )} and decrease surface expression of the adhesion molecules inter-cellular adhesion molecule 1 and vascular cell adhesion molecule 1 in cultured human aortic endothelial cells⁽ Reference Wang, Chen and Lee ^{71 )}, both of which actions would serve to reduce the inflammatory response and subsequent leucocyte infiltration that fuels plaque progression. n-3 LC-PUFA also directly alter plaque morphology by inducing structural changes consistent with increased stability, characterised by an increased fibrous cap thickness and reduced macrophage infiltration⁽ Reference Thies, Garry and Yaqoob ^{72 )}, reduced foam cell infiltration and mRNA expression of matrix metalloproteinase-7, -9 and -12⁽ Reference Cawood, Ding and Napper ^{73 )}, and increased collagen content of the plaques⁽ Reference Matsumoto, Sata and Fukuda ^{74 )}.

These positive effects of n-3 LC-PUFA on atherosclerotic plaque resemble very closely those seen with agents that modulate CB₂-mediated effects in the EC system. Although there is evidence of increased CB receptor expression in atherosclerotic plaques, there are no definitive data to show that EC levels (as opposed to the receptors that mediate their effects) are similarly elevated. However, n-3 LC-PUFA do reduce cardiac EC levels in high-fat fed mice⁽ Reference Piscitelli, Carta and Bisogno ^{36 )} and unpublished results from our own laboratory have shown that both heart and vascular tissue levels of both AEA and 2-AG are markedly elevated in high-fat fed ApoE-/- and that these are normalised by dietary intervention with EPA. At face value it could therefore be argued that the anti-atherogenic effect of n-3 LC-PUFA could be explained by removal of EC, and thus their pro-atherogenic influence mediated by CB₁ receptors; however, this would not take account of the possible protective role of the EC exerted via CB₂.

Could the endocannabinoid system contribute to the beneficial effects of long-chain-PUFA in IHD?

I/R injury represents the principal cause of tissue damage following acute myocardial infarction (AMI), while the electrical disturbances that occur during an AMI pose a threat to life through sudden arrhythmic death.

Can the beneficial effects of n-3 long-chain-PUFA in CVD be linked to those of the endocannabinoid system?

From the afore-mentioned discussion, while it is evident that a link between n-3 LC-PUFA and EC is unlikely to exist in the setting of hypertension, there are sufficient similarities regarding the effects of both in atherosclerosis and IHD to consider that a connection exists. Considering the cardinal role that inflammatory cells play in many CVD states, and the similar influence of both n-3 LC-PUFA and EC on inflammatory processes in atherogenesis, this raises the intriguing possibility that EC may be the ‘missing link’ in understanding the mechanisms underlying the vasculo-protective effects of n-3 LC-PUFA. Similarly, common mechanisms exist for both the tissue-sparing and antiarrhythmic effects of both n-3 LC-PUFA and EC, again raising the notion that modulation of the EC system by n-3 LC-PUFA may play a part in the underlying mechanisms. However, counterintuitive to this hypothesis is that while EC levels are consistently seen to be up-regulated in these pathological conditions, n-3 LC-PUFA tend to suppress EC levels. While this may be interpreted as an overall reduction of activity of the EC system by n-3 LC-PUFA, it is probably more probably that these changes are a consequence of a shift in the n-3/n-6 balance of membrane lipids, resulting in compensatory increases in the n-3 LC-PUFA-derived acyl conjugates docosahexaenoyl ethanolamide and eicosapentaenoyl ethanolamine⁽ Reference Wood, Williams and Pandarinathan ^{33 )}. Indeed, it is plausible that these n-3 LC-PUFA–EC conjugates may in part be responsible for some of the beneficial effects of n-3 LC-PUFA, since docosahexaenoyl ethanolamide and eicosapentaenoyl ethanolamine both bind to CB₁ and CB₂ receptors in human and mouse leucocytes⁽ Reference Yang, Fredman and Krishnamoorthy ^{108 ,} Reference Anagnostopoulos, Rakiec and Wood ^{109 )} and docosahexaenoyl ethanolamide has been shown to exert anti-inflammatory effects in mouse peritoneal and RAW264·7 macrophages⁽ Reference Meijerink, Plastina and Vincken ^{110 )}. Moreover, in contrast to AEA, docosahexaenoyl ethanolamide appears to have a greater affinity for CB₂ than for CB₁, in human inflammatory cells⁽ Reference Yang, Fredman and Krishnamoorthy ^{108 )} ; what this likely means at a cellular level is that, in the presence of high n-3 LC-PUFA concentrations, EC are converted to n-3 LC-PUFA–EC conjugates, which then act as ‘surrogate’ CB₂ agonists and thus alter the balance between activation of CB₁ v. CB₂ receptors in favour of CB₂ (Fig. 3). However, it must not be overlooked that the EC are known to act at sites other than the classical CB₁/CB₂ receptors and therefore the n-3 LC-PUFA–EC conjugates may similarly exert actions at sites distinct from CB receptors, although as far as we are aware this has not yet been tested.

Fig. 3. (Colour online) Proposed mechanism of n-3 LC-PUFA–endocannabinoid interaction in atherogenesis and myocardial ischaemia. AEA, anandamide; 2-AG, 2-arachidonyl glycerol; DHEA, docosahexaenoyl ethanolamide; EPEA, eicosapentaenoyl ethanolamine; CB ₂ , cannabinoid receptor type 2.