Endothelial atypical cannabinoid receptor: do we have enough evidence?

Abstract

Cannabinoids and their synthetic analogues affect a broad range of physiological functions, including cardiovascular variables. Although direct evidence is still missing, the relaxation of a vast range of vascular beds induced by cannabinoids is believed to involve a still unidentified non-CB₁, non-CB₂ G_i/o protein-coupled receptor located on endothelial cells, the so called endothelial cannabinoid receptor (eCB receptor). Evidence for the presence of an eCB receptor comes mainly from vascular relaxation studies, which commonly employ pertussis toxin as an indicator for GPCR-mediated signalling. In addition, a pharmacological approach is widely used to attribute the relaxation to eCB receptors. Recent findings have indicated a number of GPCR-independent targets for both agonists and antagonists of the presumed eCB receptor, warranting further investigations and cautious interpretation of the vascular relaxation studies. This review will provide a brief historical overview on the proposed novel eCB receptor, drawing attention to the discrepancies between the studies on the pharmacological profile of the eCB receptor and highlighting the G_i/o protein-independent actions of the eCB receptor inhibitors widely used as selective compounds. As the eCB receptor represents an attractive pharmacological target for a number of cardiovascular abnormalities, defining its molecular identity and the extent of its regulation of vascular function will have important implications for drug discovery. This review highlights the need to re-evaluate this subject in a thoughtful and rigorous fashion. More studies are needed to differentiate G_i/o protein-dependent endothelial cannabinoid signalling from that involving the classical CB₁ and CB₂ receptors as well as its relevance for pathophysiological conditions.

Table of Links

TARGETS	LIGANDS
5-HT receptor	Abn-CBD
α1 adrenoceptor	Acetylcholine
Akt	Anandamide (AEA)
AT1 receptor	AM251
BKCa channels	Apamin
CaV2.2	Bradykinin
CaV3.1	Cannabidiol
CaV3.2	Carbachol
CaV3.3	Charybdotoxin
CB1 receptor	Forskolin
CB2 receptor	HU-210
ERK1/2	Iberiotoxin
Glycine receptors	L-NAME
GPR18	LPI
GPR55	NaGly
GPR119	NO
Ionotropic glutamate receptor	NS1619
IP3 receptor	O-1602
KCa channels	Oleamide
M1 muscarinic receptor	Oleoylethanolamide
M2 muscarinic receptor	Rimonabant (SR141716)
MAPK	Ryanodine
Na+/Ca2+ exchanger (NCX)	THC
NaV channel	WIN55212-2
Nicotinic acetylcholine receptors
NOS
Opioid receptors
PI3K
PPARγ
ROCK
TRP channels
TRPV channels
VEGF receptor

This Table lists key protein targets and ligands in this document, which are hyperlinked to corresponding entries in http://www.guidetopharmacology.org, the common portal for data from the IUPHAR/BPS Guide to PHARMACOLOGY (Pawson et al., 2014) and are permanently archived in the Concise Guide to PHARMACOLOGY 2013/14 (Alexander et al., 2013a,b,c,d,f,g,,,,,).

Introduction

Due to the diverse physiological effects of cannabinoids, the endocannabinoid system has attracted major attention as a potential therapeutic target for a broad range of diseases. While two G_i/o protein-coupled cannabinoid receptors, CB₁ and CB₂R receptors, commonly mediate the physiological actions of cannabinoids (Alexander et al., 2013a), their vasodilator effects in a wide range of vascular beds are not thought to involve these classical cannabinoid receptors (Jarai et al., 1999; Wagner et al., 1999; Ho and Hiley, 2003a; Offertaler et al., 2003). A site of action for cannabinoids distinct from CB₁/CB₂ receptors has also been demonstrated in the endothelium of different vascular beds, both in micro- and macrovessels, including the rat (Milman et al., 2006; Herradon et al., 2007; Lopez-Miranda et al., 2010) and rabbit aorta (Mukhopadhyay et al., 2002; McCollum et al., 2007), rat (Baranowska-Kuczko et al., 2012) and human pulmonary artery (Kozlowska et al., 2007). The molecular mechanisms, by which cannabinoids produce vasodilatation, have not been fully elucidated, but are believed to involve still unidentified CB receptors located on endothelial cells, so called endothelial cannabinoid receptors (eCB receptors).

Identification of the eCB receptor and its particualr signalling cascade is not merely a pure theoretical challenge. The selection of compounds with reduced psychoactivity has emerged as a promising therapeutic strategy for a vast number of diseases. Much interest in the identification of the eCB receptor and its pharmacological characterization stems from its promising therapeutic potential in a large number of disorders (Robson, 2013), especially considering its lack of psychiatric side effects, which result from stimulation of central CB₁ receptors. Identification of the mechanisms of the CB₁/CB₂ receptor-independent actions of cannabinoids in the cardiovascular system is fuelled by discoveries of antioxidant, anti-inflammatory (Booz, 2011) and cardioprotective properties of cannabinoids that occur independently of CB₁/CB₂ receptors (Fouad et al., 2013). Accordingly, considerable effort is being assigned to identify the signalling mechanisms of eCB receptor-attributed effects, as selective eCB receptor targeting is thought to have great therapeutic potential.

Although 15 years have passed since the first publication of indications of the novel eCB receptor (Jarai et al., 1999; Wagner et al., 1999), its molecular identity is still unclear. Moreover, it is not yet conclusive that vascular effects commonly attributed to eCB receptor actually require G-protein coupled receptor (GPCR). The weak link is that our current knowledge on the eCB receptor, its pharmacology and signalling profile almost exclusively relies on the interpretation of data obtained in numerous arterial relaxation studies. These studies (i) commonly use a G_i/o protein inhibitor, pertussis toxin, to demonstrate the involvement of GPCRs (Herradon et al., 2007; Hoi et al., 2007; Kozlowska et al., 2007; O'Sullivan et al., 2009; Parmar and Ho, 2010; Alsuleimani and Hiley, 2013) and (ii) presume the eCB receptor agonists and antagonists used are selective. The sensitivity of the vasodilatation to pertussis toxin, the phytocannabinoid cannabidiol, its analogue O-1918 and the CB₁ receptor antagonist rimonabant (SR141716A) became classic tools to propose and in further studies support the presence of non-CB₁/non-CB₂ endothelial G protein-coupled CB receptors referred to in the literature as ‘abnormal cannabidiol’, ‘endothelial anandamide’ or ‘atypical endothelial cannabinoid receptor’ (Jarai et al., 1999; Wagner et al., 1999; Mukhopadhyay et al., 2002; Herradon et al., 2007; Baranowska-Kuczko et al., 2012). Due to the abundant pharmacological effects of cannabinoids and cannabinoid-like compounds outlined by Alexander and Kendall (2007), possible interactions of these compounds with unspecified CB receptor-independent targets will not always be detected in relaxation studies. Vascular responsiveness to cannabinoids varies not only between various vascular beds/species, but also even within the same vascular bed. For example, in the rat isolated aorta, the maximal relaxation to anandamide varies from 22 (O'Sullivan et al., 2005b) to 60% (Milman et al., 2006) of the imposed contraction and may (Herradon et al., 2007) or may not depend (O'Sullivan et al., 2005b) on the presence of the endothelium. Similarly, in rat small mesenteric arteries, the relaxation to anandamide was shown to be either endothelium-dependent (O'Sullivan et al., 2004) or endothelium-independent (White and Hiley, 1997). A critical look at studies published reveals a number of further inconsistencies in the mechanisms of vasodilatation commonly attributed to eCB receptors, including the sensitivity to CB₁ receptor antagonists rimonabant (Jarai et al., 1999; Harris et al., 2002; Ho and Hiley, 2003a; Milman et al., 2006) and AM251 (Ho and Hiley, 2003a; Hoi et al., 2007), NO synthase inhibitors (Jarai et al., 1999; Harris et al., 2002; Ho and Hiley, 2003a; Kozlowska et al., 2007; McCollum et al., 2007) and gap junction inhibitors (Brandes et al., 2002; Harris et al., 2002; Randall et al., 2002; Ho and Hiley, 2003a). Thus, anandamide was reported either to promote (Chaytor et al., 1999; Randall et al., 2002) or inhibit (Fleming et al., 1999; Brandes et al., 2002) gap junction communications, while rimonabant has an inhibitory effect (Chaytor et al., 1999). An ability of cannabinoids to affect gap junctions, a critical player in vascular electrical and mechanical responses, further emphasizes certain restrictions when using wire myography as the only approach to characterize the mechanisms of action of cannabinoids on endothelial cells and vascular pharmacology of CB receptor ligands. Table 1 summarizes the key findings on the pharmacology of relaxant responses to cannabinoids, highlighting substantial disparities in the pharmacological profiles of these responses.

Table 1.

Summary of pharmacology of CB₁/CB₂ receptor-independent relaxant responses to cannabinoids

Ligands	SR141716A	AM-251	O-1918	L-NAME	De-endothelization	PTX	FAAH inhibitors	APA + ChTx	IbTx	TRPV1 antagonism	αGA	pEC50 (μM)	Rmax, %	Vascular bed	Reference
AEA	Inhibition			No effect	No effect	Inhibition		No effect	No effect			6.3	95	RSMA	White and Hiley, 1997
AEA	Inhibition			No effect	Slight inhibition					Slight inhibition	Inhibition	6.2	94	RMA	Harris et al., 2002
AEA	Inhibition	Inhibition	Inhibition	No effect	Inhibition			Inhibition		Inhibition	Inhibition	5.7	89	RSMA	O'Sullivan et al., 2004
AEA	Inhibition	Inhibition	No effect	No effect	No effect			No effect		Inhibition	No effect	5.0	33	RMA	O'Sullivan et al., 2004
AEA									No effect				80	RMA	Yang et al., 2007
AEA	Inhibition	No effect			No effect		No effect		Inhibition	No effect	No effect	6.8	45	RCA	White et al., 2001
AEA										Inhibition		6.7	97	RSMA	White et al., 2001
AEA	No effect			Inhibition	Inhibition	Inhibition	Inhibition	No effect		No effect		5.9	51	RA	Herradon et al., 2007
AEA	No effect	No effect			No effect	Inhibition				No effect		5.9	22	RA	O'Sullivan et al., 2005b
AEA		No effect										5.2	90	HPA	Kozlowska et al., 2007
AEA		No effect	Inhibition	Inhibition	Inhibition		Inhibition	Inhibition		No effect		5	90	RPA	Baranowska-Kuczko et al., 2012
methAEA	Inhibition				Inhibition	Inhibition				Slight inhibition	No effect	0.6	80	RabA	Mukhopadhyay et al., 2002
Abn-CBD	Inhibition			No effect	Inhibition			Inhibition	No effect	No effect				RPIMAB	Jarai et al., 1999
Abn-CBD			Inhibition									3.0	107	RMA	Johns et al., 2007
Abn-CBD			Inhibition	No effect		Inhibition		Inhibition		No effect		6.7	93	RMA	Offertaler et al., 2003
Abn-CBD	Inhibition	Inhibition		No effect		No effect		Inhibition	No effect		No effect	6.2	93	RSMA	Ho and Hiley, 2003a
Abn-CBD		No effect	Inhibition	Inhibition	Inhibition		Inhibition	Inhibition		No effect		4.6	82	RPA	Baranowska-Kuczko et al., 2012
Abn-CBD			Inhibition			no effect						2		RMA	Milman et al., 2006
abn-cnd			Inhibition	No effect	Inhibition	Inhibition		Inhibition				4.8	110	HPA	Kozlowska et al., 2007
NAGly		No effect	Inhibition	Inhibition	Inhibition	Inhibition	No effect	Inhibition	Inhibition	No effect		5.8	88	RSMA	Parmar and Ho, 2010
ARA-S				Inhibition					Inhibition			4.9	84	RSMA	Parmar and Ho, 2010
ARA-S			Inhibition			No effect						0.55	95	RMA	Milman et al., 2006
ARA-S	No effect		No effect			Inhibition						1.2	60	RA	Milman et al., 2006
Oleamide	Inhibition	No effect	Inhibition	Inhibition	Inhibition	Inhibition		Inhibition		Inhibition		1.2	99	RSMA	Hoi and Hiley, 2006

Abn-CBD, abnormal cannabidiol; AEA, anandamide; APA + ChTx, apamin plus charybdotoxin; FAAH, fatty acid amide hydrolase; HPA, human pulmonary artery; IbTx, iberiotoxin; L-NAME, NG-Nitro-L-arginine methyl ester; RA, rat aorta; RCA, rat coronary artery; RPA, rat pulmonary artery; RSMA, rat small mesenteric artery; TRPV, transient receptor potential cation channel subfamily V.

This paper represents a critical overview of the relevant publications, drawing particular attention to certain discrepancies in the field. Numerous GPCR-independent actions of the compounds widely used as selective agonists and antagonists of the novel eCB receptor highlight the need for a re-evaluation of the subject in a thoughtful and rigorous fashion.

CB receptor-independent effect of rimonabant

The laboratory of G. Kunos was the first to propose the novel CB receptor in the vascular endothelium in 1999 (Jarai et al., 1999; Wagner et al., 1999). In the rat isolated perfused mesenteric arterial bed precontracted with phenylephrine, a bolus injection of anandamide was shown to produce a sustained and pronounced concentration-dependent vasodilatation (Wagner et al., 1999), which had an endothelium-dependent component sensitive to 1 μM of the CB₁ receptor antagonist rimonabant. The receptor was suggested to be distinct from the CB₁ receptor as the CB₁ receptor agonists WIN55212-2 and HU-210 failed to reproduce the vasodilatation. These observations led the authors to conclude that the anandamide-induced mesenteric vasodilatation is mediated by a novel anandamide receptor located on endothelial cells. Consistent with this notion, abnormal cannabidiol (Abn-CBD), a synthetic analogue of the non-psychoactive phytocannabinoid cannabidiol, was shown to produce rimonabant-sensitive hypotension and endothelium-dependent mesenteric vasodilatation independently of NO (Jarai et al., 1999). The response was also preserved in mice lacking either CB₁ receptors or both CB₁ and CB₂ receptors and, hence, appeared to depend on molecular target(s) distinct from CB₁ and CB₂ receptors. Based on these findings, it was concluded that Abn-CBD is a selective agonist, while cannabidiol is a selective antagonist of novel anandamide receptor present in endothelial cells (Jarai et al., 1999).

It is interesting to note that the authors also observed an antagonistic effect of 1 μM rimonabant on endothelium-dependent relaxation elicited by administration of a low concentrations of the Ca²⁺ ionophore ionomycin (Wagner et al., 1999). In isolated vessels, ionomycin evokes endothelial cell hyperpolarization, similar to that produced by acetylcholine (Bondarenko and Sagach, 2006). The antagonistic effect of 1 μM rimonabant on endothelium-dependent relaxation induced by the Ca²⁺ ionophore A23187, carbachol and bradykinin was also demonstrated by others (Randall et al., 1996; Randall and Kendall, 1997; White and Hiley, 1997), indicating that rimonabant may act through targets other than the CB₁ receptor and putative eCB receptor. However, this possibility was discarded by the authors (Jarai et al., 1999) because of a reported failure of 1 μM rimonabant to inhibit the relaxation to NS1619, an opener of high-conductance Ca²⁺-dependent K⁺ channels (BK_Ca channels) (White and Hiley, 1998). At the same time, the contribution of BK_Ca channels to the Abn-CBD-evoked mesenteric vasodilatation was not confirmed because of its resistance to a combination of apamin plus iberiotoxin (Jarai et al., 1999). The inhibitory effect of 1 μM rimonabant on the relaxation induced by ionomycin was attributed to the stimulated release of anandamide from the endothelium in the presence of ionomycin (Wagner et al., 1999).

Of interest, earlier studies performed on the rat mesentery showed that the anandamide-induced vasodilatation is unaffected by endothelial cell denudation (Randall et al., 1996; White and Hiley, 1997) and NOS blockade (White and Hiley, 1997). Nevertheless, even in endothelium-denuded vessels, the response was antagonized by 0.1–1 μM rimonabant (Randall et al., 1996; Harris et al., 2002; Ho and Hiley, 2003b; O'Sullivan et al., 2004), strongly indicating a non-endothelial site of action of the blocker. Rimonabant was also shown to inhibit the relaxation of mesenteric arteries induced by the re-addition of Ca²⁺ (Bukoski et al., 2002). A similar blocking effect was observed for O-1918. The effect was detected in arteries from both control and CB₁ receptor-deficient mice, excluding the effect of rimonabant on CB₁ receptors. Additionally, in patch-clamp experiments, rimonabant inhibits an unidentified macroscopic K⁺ current other than that attributable to BK_Ca channels (Bukoski et al., 2002). In another early study (Chaytor et al., 1999), anandamide was shown to relax isolated rings of rabbit superior mesenteric artery through an endothelium-dependent and endothelium-independent mechanism. In this study, rimonabant at 10 μM attenuated endothelium-dependent, but not endothelium-independent, relaxation through inhibition of myoendothelial communications. Altogether, these data clearly indicate that rimonabant exhibits a number of effects independently of CB receptors and, hence, further work is needed to identify whether rimonabant-mediated suppression of eCB receptor-attributable vasodilatation is mediated via CB receptor-independent mechanisms. Pharmacological targets of putative antagonists of eCB receptors and their CB receptor-independent effect on vasodilaation are summarized in Table 2.

Table 2.

Cannabinoid receptor-independent molecular targets for the putative eCB receptor antagonists

Putative antagonists of eCB receptor	Targets or CB receptor-independent effects	Reference
SR141716A	Antagonizes endothelium-dependent relaxation to carbachol, bradykinin, ionomycin and A23187	Wagner et al., 1999 Randall et al., 1996 White and Hiley, 1997
SR141716A	Antagonizes endothelium-independent relaxation to anandamide	Randall et al., 1996 Harris et al., 2002 Ho and Hiley, 2003a
SR141716A	Inhibition of myoendothelial communications	Chaytor et al., 1999
SR141716A	Inhibits unidentified K+ channels	Bukoski et al., 2002
SR141716A	Inhibits relaxation induced by Ca2+-re-addition	Bukoski et al., 2002
O-1918	Inhibits relaxation of denuded arteries to sodium nitroprusside Inhibits BKCa channels Inhibits Na+-Ca2+ exchanger	Parmar and Ho, 2010 Godlewski et al., 2009 Bondarenko et al., 2013
Cannabidiol	Acts as PPAR γ ligand Antagonist of α1 adrenoceptors Inhibits voltage-gated Na+ channels Inhibits 5-HT3 receptors Activates TRPV channels	O'Sullivan et al., 2009 Pertwee et al., 2002 Hill et al., 2014 Yang et al., 2010 Hassan et al., 2014

BK_Ca, high-conductance Ca²⁺-dependent K⁺ channels; SR141716, rimonabant.

CB receptor-independent effect of O-1918

In 2003, the group of G. Kunos further substantiated the concept of the involvement of novel eCB receptor in the vasorelaxation induced by endocannabinoids by showing that in endothelium-intact, but not endothelium-denuded, rat mesenteric arterial segments, the cannabidiol analogue O-1918 concentration-dependently inhibits the relaxation to Abn-CBD, a ligand for the putative eCB receptor (Offertaler et al., 2003). These observations allowed the authors to propose O-1918 as a selective antagonist of the Abn-CBD receptor located on endothelial cells. Since then, numerous vascular relaxation studies replicated on different vascular beds employed O-1918 as a selective antagonist of eCB receptors (Hoi et al., 2007; Kozlowska et al., 2007; Zakrzeska et al., 2010; Baranowska-Kuczko et al., 2012; Caldwell et al., 2013), offering further support for existence of novel endothelially located CB receptors. However, in de-endothelized and permeabilized rabbit pulmonary artery strips, Abn-CBD still produced vasodilatation, the response was prevented by O-1918, and only partially reduced by AM251, SR141716A and pertussis toxin (Su and Vo, 2007), indicating a GPCR-independent action of Abn-CBD and O-1918 on molecular targets located on smooth muscle cells. Recently, GPCR-independent effects of O-1918 on distinct molecular targets located in the vasculature have been demonstrated (Godlewski et al., 2009; Bondarenko et al., 2013), indicating the possibility that the inhibitory action of O-1918 on vasodilatation is mediated by a non-specific effect. Consistent with this, at concentrations used to verify the role of eCB receptors (3–10 μM), O-1918 was also shown to significantly attenuate the relaxant responses of endothelium-denuded mesenteric arteries to sodium nitroprusside, a NO donor (Parmar and Ho, 2010). This observation suggests that O-1918 has CB receptor-independent target(s) located on smooth muscle cells. Indeed, in a concentration range used for inhibition of eCB receptor-attributable relaxation, O-1918 inhibits the activity of BK_Ca channels (Godlewski et al., 2009) and the Na⁺-Ca²⁺ exchanger (Bondarenko et al., 2013), key players regulating vascular contractility. These effects do not require GPCR, excluding O-1918 as a selective antagonist of eCB receptors and indicating that the interpretation of results where O-1918 has been assumed to be a selective eCB receptor antagonist should be treated with caution. The conclusions of authors on involvement of eCB receptors in relaxation are frequently based on the supposed selectivity of eCB receptor agonists and antagonists, while additional off-target effects of the compounds are not critically considered. For example, in a recent study performed on rat pulmonary artery, a vascular bed of endothelial cells expressing functional BK_Ca channels (Vang et al., 2010), the putative eCB receptor antagonist O-1918, which acts as a BK_Ca channel inhibitor (Godlewski et al., 2009), was shown to attenuate the anandamide-evoked relaxation (Baranowska-Kuczko et al., 2012). Conclusively, it is a general recommendation that the site of action of O-1918, a commonly used eCB receptor inhibitor claimed to be selective, needs to be unambiguously demonstrated.

CB receptor-independent effect of cannabidiol

Cannabidiol has a relatively low affinity for classical cannabinoid receptors, and is commonly used as an antagonist of the novel CB receptor. However, the relevant experimental data is also controversial and indicates the presence of multiple target sites for cannabidiol. Thus, unlike in rat mesentery (Jarai et al., 1999) and rat (Baranowska-Kuczko et al., 2012) and human (Kozlowska et al., 2007) pulmonary artery, where cannabidiol antagonizes the eCB receptor-attributed relaxation, in rat aorta, cannabidiol actually produces vasodilatation (O'Sullivan et al., 2009) by acting as a PPAR ligand. In contrast to findings of the earlier study (Jarai et al., 1999), cannabidiol was reported to produce relaxation in rat isolated mesenteric arteries (Offertaler et al., 2003), suggesting that this compound may act both as a silent antagonist and a partial agonist of eCB receptors. At 5–10 μM, concentrations commonly used to antagonize the eCB receptor-attributed vasodilatation, cannabidiol attenuates contractile responses to agonists of α₁ adrenoceptors, as demonstrated in mouse isolated vas deferens (Pertwee et al., 2002). In addition, cannabidiol was shown to inhibit Na_v channels (Hill et al., 2014), 5-HT_3A receptor-mediated currents (Yang et al., 2010), and via activation of one or more transient receptor potential cation channel subfamily V (TRPV), enhance phagocytosis of mouse microglial BV-2 cells (Hassan et al., 2014).

K_Cachannels in the eCBR-attributable relaxation

The identity of the K⁺ channel subtypes involved in the relaxation to cannabinoids is still controversial. In rat isolated small mesenteric artery, the relaxation to N-arachidonylglycine (NAGly) is sensitive to iberiotoxin, a blocker of Ca²⁺-activated K⁺ (K_Ca) channels of large (BK_Ca) conductance, and has endothelium-dependent component (Parmar and Ho, 2010). In contrast, the relaxation to anandamide was reported to be endothelium-independent and insensitive to iberiotoxin either alone or in combination with apamin (White and Hiley, 1997). In rat isolated mesenteric arteries, charybdotoxin, the BK_Ca and IK_Ca channel blocker, inhibited the relaxation to Abn-CBD (Offertaler et al., 2003). The effect was slightly potentiated by the additional presence of apamin, while apamin alone had no effect. The combination of charybdotoxin plus apamin was effective at inhibiting the eCB receptor-attributed vasodilator responses in a number of other vascular beds, including human (Kozlowska et al., 2007) and rat (Baranowska-Kuczko et al., 2012) pulmonary artery, implicating K_Ca channels of small (SK_Ca), intermediate (IK_Ca) and large conductance. In rat coronary arteries, the relaxation to anandamide was shown to be endothelium-independent and sensitive to iberiotoxin (White et al., 2001). Interestingly, the endothelial cells of rat aorta express functional IK_Ca and SK_Ca channels (Marchenko and Sage, 1996; Bondarenko et al., 2012), and anandamide-induced aortic relaxation is insensitive to a combination of apamin and charybdotoxin (Herradon et al., 2007).

The role of NO in relaxation to cannabinoids

In rat isolated small mesenteric artery, the relaxation attributable to novel CB receptors was reported to be either sensitive (Hiley and Hoi, 2007; Parmar and Ho, 2010) or insensitive to NOS inhibition (Ho and Hiley, 2003a; 2004,; O'Sullivan et al., 2004). In macrovessels, such as rat (Herradon et al., 2007) and rabbit (Mukhopadhyay et al., 2002) aorta and rat pulmonary artery (Baranowska-Kuczko et al., 2012), the CB₁/CB₂ receptor-independent endothelium-dependent relaxation is diminished or fully inhibited by inhibition of NOS and was shown to be accompanied by an increased NO release (McCollum et al., 2007; Baranowska-Kuczko et al., 2012). Conversely, in human pulmonary artery, the relaxation was reported to be insensitive to NOS inhibition (Kozlowska et al., 2007). The reasons for these discrepancies are not clear but probably represent regional/species differences.

Is GPCR 55 (GPR55) a novel CB receptor?

In addition to CB₁ and CB₂ receptors, GPR55 has been described as a target for cannabinoid ligands (Ryberg et al., 2007; Pertwee, 2010; Liu et al., 2014). Initially, GPR55, which, in addition to classical cannabinoids and cannabinoid-like compounds [Abn-CBD, O-1602, N-arachidonyl serine (ARA-S)], is activated by lysophosphatidylinositol (LPI), an endogenous lysophospholipid not related to CB receptor ligands, has been proposed to represent the eCB receptor (Baker et al., 2006). Activation of GPR55 by several ligands was shown to increase intracellular Ca²⁺ in several cell types (Kohno et al., 2006; Lauckner et al., 2008), including endothelial cell line EA.hy926 (Waldeck-Weiermair et al., 2008), where GPR55 stimulation by LPI produces transient hyperpolarization (Bondarenko et al., 2010), suggesting the functional importance of GPR55 in vascular physiology. In human dermal microvascular endothelial cells, GPR55 knockdown partially inhibited ARA-S-induced migration and activation of Akt and ERK1/2 (Zhang et al., 2010). Additionally, in GPR55-knockdown cells, the ARA-S-induced VEGF production was significantly decreased as compared with the control GPR55-expressing cells (Zhang et al., 2010). However, the BP lowering effect and the vasodilator responses of isolated mesenteric arteries to O-1602 and Abn-CBD appeared to be unaltered in GPR55-deficient mice (Johns et al., 2007). These results indicate that although current knowledge implicates GPR55 in cell migration and growth, GPR55 is not critically involved in the vasodilatation to O-1602 and abn-cbd, and accordingly, GPR55, perhaps, is unlikely to represent the molecular identity of the eCB receptor. An additional set of observations at first glance arguing against GPR55 as the eCB receptor is a discrepancy between the reported G_i/o protein requirement for the vasodilator responses (Mukhopadhyay et al., 2002; Offertaler et al., 2003) and non-G_i/o protein-mediated GPR55 signalling cascade reported in a number of studies (Lauckner et al., 2008; Henstridge et al., 2009; Brown et al., 2011). However, given the uncertainties over the requirement of G_i/o proteins in the eCB receptor-attributed vasodilatation discussed further below and the ability of shared GPR55/GPR18 signalling through G_q protein to increase intracellular Ca²⁺ (Lauckner et al., 2008; Console-Bram et al., 2014), the association between GPR55 and the eCB receptor should still be considered.

Is GPCR 18 (GPR18) a novel CB receptor?

Another recently emerged candidate for the eCB receptor is GPR18 activated by Δ⁹-tetrahydrocannabinol (THC), anandamide, Abn-CBD and the anandamide metabolite NAGly (Kohno et al., 2006; McHugh et al., 2012; Takenouchi et al., 2012). Stimulation of GPR18 by several ligands increases intracellular Ca²⁺ in several cell types (Kohno et al., 2006; Console-Bram et al., 2014), suggesting that vascular GPR18-dependent signalling may have functional implications. In microglial cells, at picomolar and low nanomolar concentrations range, NAGly and Abn-CBD drive cell migration in a GPR18-dependent manner (McHugh et al., 2010). Pertussis toxin was found to attenuate the Abn-CBD- and NAGly-induced cell migration (Mo et al., 2004; McHugh et al., 2010). Additionally, the toxin antagonized the NAGly-induced inhibition of forskolin-stimulated cAMP production (Kohno et al., 2006) and activation of MAPK by Abn-CBD (Offertaler et al., 2003), anandamide and NAGly (McHugh et al., 2010; 2012,), suggesting that the effect is mediated by G_i/o protein.

Based on observations made on microglial cell migration, GPR18 was proposed to be a molecular candidate for the Abn-CBD receptor (McHugh et al., 2010). Notably, both GPR18 mRNA and protein have recently been identified in isolated arteries and endothelial cells (Ho and Yeung, 2009; MacIntyre et al., 2014; Wilhelmsen et al., 2014). However, a correlation between the GPR18 level and the dilator responses is missing and, hence, the role of GPR18 in vasoactivity and other vascular functions is still unclear. There is also inconsistency in the effects of GPR18 ligands on endothelium-dependency of the dilator responses even within the same vascular bed. Thus, in rat isolated small mesenteric arteries, the relaxation to GPR18 agonists anandamide (O'Sullivan et al., 2004), NAGly (Parmar and Ho, 2010) and Abn-CBD (Ho and Hiley, 2003a) has a strong endothelium-dependent component, while the dilator responses of the same arterial bed to THC (O'Sullivan et al., 2005a) and, curiously, anandamide (White and Hiley, 1997; Ho and Hiley, 2003b) were shown to be endothelium-independent.

The stimulating effects of NAGly and Abn-CBD on cell migration were insensitive to rimonabant, but antagonized by O-1918 and ARA-S, a structurally related to anandamide non-psychoactive endocannabinoid with weak affinity for GPR18 (McHugh et al., 2010). However, in rat isolated aorta and mesenteric artery, ARA-S produces a significant endothelium-dependent vasodilatation with potency similar to that evoked by Abn-CBD (Milman et al., 2006), an observation which is not consistent with the ARA-S being a GPR18/Abn-CBD receptor antagonist. Within the same concentration range, ARA-S directly stimulates the BK_Ca channel (Godlewski et al., 2009). Unlike in the mesentery, the aortic relaxation to ARA-S was blocked by pertussis toxin, suggesting the involvement of a GPCR other than GPR18 in aortic, but not mesenteric, vasodilatation. Oleamide, an endogenous lipid that is structurally related to anandamide, produces vasorelaxation with an endothelium-dependent component in rat isolated small mesenteric arteries (Hoi and Hiley, 2006). Consistent with the pharmacological profile of eCB receptors, the relaxation was sensitive to rimonabant, O-1918 and pertussis toxin pretreatment. There are, unfortunately, no reports that oleamide binds GPR18 and, hence, it is difficult to assign the mechanism of relaxation to GPR18 or another putative GPCR. The identity of eCB receptor as GPR18 was further questioned by the demonstration that oleoylethanolamide, the anandamide-like monounsaturated fatty acid, which was shown to be a GPR119, but not a GPR18 agonist (Overton et al., 2008), produces O-1918-sensitive endothelium- and NO-dependent relaxation in third branches of rat superior mesenteric artery (Alsuleimani and Hiley, 2013). The recent demonstration of an inability of the GPR55/GPR18 agonist O-1602 to mimic the effects of Abn-CBD and NAGly in ocular signalling system (Caldwell et al., 2013; MacIntyre et al., 2014) casts further doubts on the selectivity of the compounds involved, further challenging the concept that the effects of NAGly and Abn-CBD are solely mediated by GPR18. Given the discrepancy between the levels of GPR18 expression and NAGly in different tissues (Alexander, 2012), a link between NAGly and GPR18 and the identification of GPR18 as the Abn-CBD receptor, the stimulation of which causes vasodilation, seems not so straightforward. Activation of GPR18 by NAGly was not established in a study employing high-throughput β-arrestin-based screening (Yin et al., 2009). The application of NAGly to GPR18-expressing rat sympathetic neurons did not inhibit N-type (Ca_v2.2) Ca²⁺ currents, a primary downstream effector of G_i/o proteins, but instead potentiated the currents (Lu et al., 2012). Recent findings indicate that GPR18 is coupled to several effector pathways through both G_i/o and Gα_q proteins (Console-Bram et al., 2014). Signalling via a specific pathway is likely to be context-specific. Signalling via a single receptor coupled to different G-protein subunits may partially explain the controversies in eCB receptor pharmacology. Nevertheless, the association between GPR18 and the eCB receptor-attributed signalling is not yet proven. A schematic figure explaining the current status of the possible existence and signalling through eCB receptors is shown in Figure 1.

Figure 1.

Schematic illustration of current status of research of CB₁/CB₂ receptor-independent endothelial cannabinoid signalling. Red arrows indicate stimulating effects, blue arrows indicate inhibitory effects.

Limits of cell culture models

An additional caveat in vascular cannabinoid research is the common use of cell culture models, a condition that may profoundly affect the signalling profile, not reflecting the situation in situ but rather a diseased state. Many aspects of endothelial cell signalling, such as functional expression of ion channels/receptors, change dramatically in culture conditions (Sandow and Grayson, 2009). It is well known that endothelial cells grown in culture cease to produce mRNA for M₁ and M₃ muscarinic receptors and, consequently, cultured endothelial cells are unresponsive to acetylcholine (Marchenko and Sage, 1993). Cell culture-associated abnormalities in endothelial cell signalling are also associated with TRPV3 channels and ryanodine-sensitive stores (Bondarenko and Sagach, 1997; Kohler et al., 2001). In addition, the alterations are undoubtedly relevant for endocannabinoid signalling mechanisms. Notably, while some reports utilizing the human endothelial cell line EA.hy926 indicate the expression of the CB₁ receptor and its involvement in the mobilization of intracellular Ca²⁺ in response to anandamide (Waldeck-Weiermair et al., 2008), other studies failed to detect CB₁ receptor expression in the same cell line (Liu et al., 2000). A further potential limitation is associated with the studying of GPCR signalling pathways in heterologous expression systems. It is worth noting that in transfected cells, recombinant opioid and CB receptors operate through the same pool of G proteins, while in cells endogenously co-expressing CB₁ with opioid receptors, the signalling occurs through distinct pools of G proteins (Shapira et al., 2000).

An additional caveat associated with utilizing heterologous expression systems is an essential dependency of CB₁ receptor- and GPR55-mediated cannabinoid signalling on the lipid composition of the plasma membrane (Gasperi et al., 2012). Keeping in mind that cannabinoids target distinct GPCR-independent ion transport systems, the functional activity of which is regulated by membrane cholesterol, cell culture models may not be able to represent the ‘native’ environment and, most likely, the evoked response is context-dependent. Conclusively, studying cannabinoid signalling in a non-native environment may not reflect a true picture and complicates data interpretation, indicating a need for alternative, more physiological, methodological approaches.

GPCR-independent targets of CB receptor ligands

The mechanism of action of endocannabinoids is not limited to GPCRs. Recently, a growing number of studies have convincingly shown that, at physiologically relevant concentrations, in addition to GPCR-dependent effects, bioactive lipids affect the properties of various ion transport mechanisms independent of GPCRs. The targets affected include voltage- and ligand-gated ion channels, namely the nicotinic acetylcholine, 5-HT and glycine receptors and ionotropic glutamate receptors (Oz, 2006; Alexander and Kendall, 2007; Barana et al., 2010). Anandamide in the concentration range of 1–10 μM causes a significant inhibition of voltage-gated Na⁺ channels in mouse cortical (Nicholson et al., 2003) and rat dorsal root ganglion neurons (Nicholson et al., 2003; Kim et al., 2005). Various lipoamino acids, including NAGly, were recently shown to inhibit Ca_v3.1, Ca_v3.2 and Ca_v3.3 channels (Ross et al., 2009) and currents mediated by recombinant GABA_A receptors (Baur et al., 2013). NAGly enhances the inhibitory glycinergic synaptic transmission by blocking glycine uptake (Jeong et al., 2010).

While much is known about the GPCR-independent effects of endocannabinoids and bioactive lipids on ion transport mechanisms in the nervous system, the vascular molecular targets are much less explored. In recent years, evidence has accumulated suggesting that endocannabinoids and lipid compounds exert their effect on endothelial cells by targeting both GPCR-dependent and GPCR-independent targets, including non-selective transient receptor potential (TRP) channels (Bondarenko et al., 2010), Na⁺-K⁺-ATPase (Bondarenko et al., 2010) and different K_Ca channels (Bondarenko et al., 2011a,b,). In a GPCR-independent manner, anandamide and NAGly inhibit both forward and reverse modes of Na⁺-Ca²⁺ exchanger in endothelial cells (Bondarenko et al., 2013). In addition, NAGly was shown to directly potentiate the BK_Ca channels (Bondarenko et al., 2013) and inhibit store-operated Ca²⁺ entry in different cell types (Deak et al., 2013). Because vascular signalling and function are chiefly regulated by Na⁺-Ca²⁺ exchanger (Bondarenko, 2004; Andrikopoulos et al., 2011), IK_Ca, BK_Ca (Vang et al., 2010; Bondarenko et al., 2012; Wulff and Kohler, 2013) and different members of TRP channels, identification of their contribution to GPCR-dependent and -independent cannabinoid signalling still needs to be clarified.

There is also a certain disparity in EC₅₀ values of some endocannabinoids as GPR18 ligands and the EC₅₀ values of the evoked relaxation. The original study of Kohno et al. (2006), who first identified NAGly as a ligand of GPR18, showed that this lipoamino acid concentration-dependently inhibits cAMP formation in a pertussis toxin-sensitive manner, with a calculated IC₅₀ value of 20 nM. However, in rat small mesenteric arteries, NAGly evoked a concentration-dependent relaxation with an EC₅₀ value of 5.8 μM (Parmar and Ho, 2010), which is several orders higher than that required to activate GPR18. At these concentrations, NAGly exhibits a direct GPCR-independent effect on a number of ion transport systems expressed in both endothelial and smooth muscle cells (Bondarenko et al., 2013; Deak et al., 2013). In endometrial cell line, HEC-1B, THC was shown to be a full agonist for GPR18, with a calculated EC₅₀ value of 0.96 μM (McHugh et al., 2012). However, the THC-induced relaxation occurred at higher concentrations, with an EC₅₀ value of 5.8 μM (O'Sullivan et al., 2005b). Similar observations were made with another GPR18 agonist, Abn-CBD. Abn-cbd binds GPR18 with an EC₅₀ value of 835.8 nM (McHugh et al., 2012), while it produces relaxation with EC₅₀ value of 6.2 μM (Ho and Hiley, 2003a). As the concentrations needed for the vasodilator effects are far beyond the values sufficient to activate GPR18, the conclusion that the vascular effects of GPR18 agonists are indeed GPR18-dependent is rather speculative. Altogether, the data available do not yet allow the eCB receptor, the stimulation of which is required for endothelium-dependent relaxation to cannabinoids, to be classified as GPR18.

Sensitivity of the relaxant responses to pertussis toxin

In addition to sensitivity of the relaxant responses to the ‘selective’ antagonists of eCB receptors, the concept of the involvement of novel eCB receptors in the responses is largely supported by inhibition of cannabinoid-induced vasodilatation following pretreatment with pertussis toxin. However, a number of studies did not confirm the pertussis toxin sensitivity of mesenteric vasodilatation to Abn-CBD (Ho and Hiley, 2003a; Milman et al., 2006), ARA-S (Milman et al., 2006) and a synthetic cannabinoid-like compound VSN16 (3-(5-dimethylcarbamoyl-pent-1-enyl)-N-(2-hydroxy-1-methyl-ethyl)benzamide), which produces endothelium-dependent vasodilatation sensitive to rimonabant, O-1918 and AM-251 (Hoi et al., 2007), thus questioning the involvement of a GPCR. Interestingly, unlike in the mesentery, the aortic relaxant effects of ARA-S and Abn-CBD in the study of Milman et al. were antagonized by pertussis toxin pretreatment. However, the vasorelaxation was insensitive to O-1918 and rimonabant, commonly used selective antagonists of eCB receptors, suggesting distinct mechanisms for the vasodilatation in rat aorta and mesenteric artery (Milman et al., 2006). In rat small mesenteric artery, pertussis toxin was shown to antagonize the relaxation to the Ca²⁺ ionophore A23187 (White and Hiley, 1997). The recently demonstrated multiplicity of GPR18- and GPR55-activated cellular pathways, which include either G_i/o- and Gα_q- (Console-Bram et al., 2014) or Gα_12/13- and Gα_q- (Lauckner et al., 2008) mediated signalling, respectively, may also partially explain the discrepancies in the reported sensitivity of eCB receptor-attributed relaxant responses to pertussis toxin.

Notably, the use of pertussis toxin in myographic studies as an indicator for GPCR involvement is associated with several flaws. Firstly, pertussis toxin is not a universal GPCR inhibitor but is only an inhibitor of the majority of the members of G_i/o protein family. The ADP-ribosylating toxin, pertussis toxin, catalyzes the ADP-ribosylation of the α subunits of the heterotrimeric G_i/o protein family (Gα_i, Gα_o and Gα_t, but not Gα_z), thereby preventing the G proteins from interacting with the related GPCRs (Mangmool and Kurose, 2011). It should be noted that cannabinoid signalling is not limited to G_i/o proteins and involves Gα_s and Gα_q proteins. Both CB₁ and CB₂ receptors activate multiple signal transduction pathways such as both inhibition and stimulation of adenylyl cyclase and protein kinase pathways via G_i/o and Gα_s proteins respectively (Bosier et al., 2010). Coupling of the CB receptor to Gα_q proteins results in stimulation of the activity of PLC (Bosier et al., 2010), which hydrolyzes phosphatidylinositol 4,5-bisphosphate to diacyl glycerol and inositol 1,4,5-trisphosphate (IP₃) with subsequent activation of IP₃-operated Ca²⁺ permeable channels. Hence, pertussis toxin as a tool for detection of the involvement of Gα_s and Gα_q in the response has limited benefits, unless suppression of the signalling via G_i/o proteins unmasks coupling of CB receptors to non-G_i/o proteins. Secondly, pertussis toxin is not specific for G_i/o. proteins and exhibits a number of GPCR-independent effects on key signalling mechanisms, including modulation of Ca²⁺ entry, cAMP and diacyl glycerol synthesis (Garcia et al., 2001; Mangmool and Kurose, 2011), activation MAPK (Garcia et al., 2001) and up-regulation of angiotensin type 1 (AT₁) receptor signalling (Nishida et al., 2010). Consistent with these non-specific effects, pre-incubation of mesenteric arteries with pertussis toxin was found to inhibit endothelium-dependent relaxation to the Ca²⁺ ionophore A23187 (White and Hiley, 1997). Collectively, the sensitivity of the relaxant responses to pertussis toxin is likely to serve as an unreliable indicator for GPCR involvement.

In view of the conflicting pharmacology on the presence of eCB receptors, the data either supporting or non-supporting the concept of an eCB receptor was summarized in Tables 2013a and 2013b respectively.

Table 3.

Data supporting the presence of a novel endothelial cannabinoid receptor

Read-out assay	Species and vascular bed	Agonist	pEC50	Reference
PTX-sensitive CB1/CB2 receptor-independent endothelium-dependent relaxation	RSMA	AEA	6.3 μM	White and Hiley, 1997
	RSMA	NAGly	5.8 μM	Parmarand Ho, 2010
	RSMA	ARA-S	4.9 μM	Parmar and Ho, 2010
	RA	AEA	5.9 μM	Herradon et al., 2007
	RA	THC	5.81 μM	O'Sullivan et al., 2005b
	RA	Abn-CBD	not presented	Milman et al., 2006
	RA	ARA-S	1–50 μM	Milman et al., 2006
	RMA	Abn-CBD	5.6 μM	Offertaler et al., 2003
	RabA	AEA	0.6 μM	Mukhopadhyay et al., 2002
PTX-sensitive activation of p42/44 MAPK	Cell type	Agonist	Concentrations	Reference
	HUVEC	Abn-CBD	30 μM	Mo et al., 2004
	HUVEC	Abn-CBD	10 μM	Offertaler et al., 2003;
	HEK293-GPR18, BV-2	NAGly	10 nM–10 μM	McHugh et al., 2010
	HEC-1B	NAGly, AEA	10 nM	McHugh et al., 2012
PTX-sensitive cell migration	Cell type	Agonist	Concentrations	Reference
	HUVEC	Abn-CBD	10–50 μM	Mo et al., 2004
	BV-2	NAGly	0.17 nM	McHugh et al., 2010
	BV-2	Abn-CBD	13.1 nM	McHugh et al., 2010
	HEK293-GPR18	NAGly	1 μM	McHugh et al., 2010
	HEK293-GPR18	Abn-CBD	1 μM	McHugh et al., 2010
Cell migration sensitive to PI3K/Akt inhibitors	Cell type	Agonist	Concentrations	Reference
	HUVEC	Abn-CBD	30 μM	Mo et al., 2004

Abn-CBD, abnormal cannabidiol; AEA, anandamide; ARA-S, N-arachidonyl serine; BV-2, mouse microglial cells; CB₁R-cannabinoid receptor type 1; CB₂R-cannabinoid receptor type 2; PTX, pertussis toxin; RA, rat aorta; RabA, rabbit aorta; RMA, rat mesenteric artery; RSMA, rat small mesenteric artery; THC, Δ⁹-tetrahydrocannabinol.

Table 4.

Data non-supporting the presence of novel endothelial cannabinoid receptor

Read-out assay	Species/vascular bed	Agonist	pEC50	Reference
PTX-insensitive relaxation to CB	RSMA	Abn-CBD	6.2 μM	Ho and Hiley, 2003a;
	RMA	Abn-CBD	2 μM	Milman et al., 2006
	RMA	ARA-S	0.55 μM	Milman et al., 2006
	RSMA	VSN16	88 nM	Hoi et al., 2007
	RSMA	THC	5.37 μM	O'Sullivan et al., 2005a
PTX-sensitive EDR to A23187	RSMA	A23187	6.46 μM	White and Hiley, 1997
			pEC50
Lack or minor EDR to CB	RabA	Abn-CBD	0.01–0.3 μM tested	Su and Vo, 2007
	RCA	AEA	6.8 μM	White et al., 2001
	RSMA	AEA	6.3 μM	White and Hiley, 1997
	RMA	AEA	5.0 μM	O'Sullivan et al., 2004
	RMA	AEA	0.3 μM tested	Randall et al., 1996
	RA	AEA	5.9 μM	O'Sullivan et al., 2005b
	RMA	AEA	6.2 μM	Harris et al., 2002
	RSMA	AEA	6.7 μM	Ho and Hiley, 2003b
	RSMA	THC	5.37 μM	O'Sullivan et al., 2005a
	RMA	THC	1–10 μM tested	O'Sullivan et al., 2005a
			Rimonabant, μM
Rimonabant inhibits vasodilatation to non-CB agonists/Ca2+ ionophores	RMA	Ionomycine	5 μM	Wagner et al., 1999
	RCA	Bradykinin	1 μM	Randall et al., 1997
	RMA	Bradykinin	1 μM	Randall et al., 1996
	RMA	A23187	1 μM	Randall et al., 1996
	RMA	Carbachol	1 μM	Randall et al., 1996
	RSMA	Carbachol	1 μM	White and Hiley, 1997
			O-1918, μM
No effect of O-1918 on EDR O-1918 inhibits endothelium-independent relaxation	RA	ARA-S	10 μM	Milman et al., 2006
	MA	ARA-S	10 μM	Milman et al., 2006
	RabPA	Abn-CBD	0.1–1 μM	Su and Vo, 2007
	RSMA	NAGly	3 μM	Parmar and Ho, 2010
	RSMA	SNP	3 μM	Parmar and Ho, 2010
GPCR-independent targets for O-1918	Targets	Effect	Concentration tested
	NCX	Inhibition	10 μM	Bondarenko et al., 2013
	BKCa	Inhibition	pIC50 = 6.6 μM	Godlewski et al., 2009

Abn-CBD, abnormal cannabidiol; AEA, anandamide; ARA-S, N-arachidonyl serine; BK_Ca, high-conductance Ca²⁺-dependent K⁺ channels; CB, cannabinoid; EDR, endothelium-dependent relaxation; NCX, Na⁺-Ca²⁺ exchanger; PTX, pertussis toxin; RA, rat aorta; RabA, rabbit aorta; RabPA, rabbit pulmonary artery; RCA, rat coronary artery; RMA, rat mesenteric artery; RSMA, rat small mesenteric artery; SNP, sodium nitroprussid; THC, Δ⁹-tetrahydrocannabinol; VSN16, 3-(5-dimethylcarbamoyl-pent-1-enyl)-N-(2-hydroxy-1-methyl-ethyl)benzamide.

Pertinent unsolved questions

Despite great efforts, our understanding of the vascular pharmacology of cannabinoid signalling and the mechanisms through which endocannabinoids affect vascular function is limited by assessment of vascular contractility as the technique of choice and utilization of non-vascular cell culture models. The variable effects of putative eCB receptor inhibitors and pertussis toxin indicate the complexity of the mechanisms of vascular cannabinoid signalling and possible simultaneous activation of several molecular targets, which may be expressed both in endothelial and smooth muscle cells. The largely accepted view on the presence of a novel CB receptor in vascular endothelium depends on key assumptions on the selectivity of eCB receptor agonists/antagonists, which appear to be rather premature. Recently identified multiple GPCR-independent targets for compounds, which have been used as selective agonists and antagonists for the eCB receptor, challenge the concept of the existence of an eCB receptor and its role in vascular responses to cannabinoids needs further support. Direct evidence for the existence of an eCB receptor is still missing and the mechanisms responsible for eCB receptor-attributed relaxation still need to be elucidated. Meanwhile, several key questions remain unanswered.

1. Do the vasodilator effects of cannabinoids commonly attributed to eCB receptors require a novel GPCR or do the effects, at least partially, depend on the stimulation of GPCR-independent targets? Clear dissection of GPCR-dependent and -independent endothelial signalling to cannabinoids at the level of ion channels and the molecular identity of the respective targets are ultimately required.

2. If a novel GPCR is involved in the vasodilatation induced by cannabinoids, is it only one or several eCB receptors? What is the molecular identity of the GPCRs and the respective signal transduction pathways? What is their role in vascular reactivity, angiogenesis and inflammation?

3. What are the multiple intracellular targets for endocannabinoids, the identity of these targets and their functional role?

4. Considering that in blood vessels, endothelial cell signalling and function is largely influenced by the smooth muscle cells, what are the molecular targets located on smooth muscle cells and their role in endothelial cell responses to cannabinoids?

Concluding remarks and future directions

In conclusion, substantial data have been accumulated suggesting that a site distinct from CB₁ and CB₂ receptors mediates the vasodilator effect of cannabinoids. There is evidence that cannabinoids bind to non-CB₁/CB₂ GPCRs, including GPR18 and GPR55, and their expression has been shown in vascular cells. Ligands and blockers of the putative eCB receptor widely used as selective compounds, often have GPCR-independent effects, influence functional properties of a number of ion channels/transporters expressed in the vasculature with potencies similar to that required to influence vascular reactivity, but higher than that required to activate respective GPCR. The observations that demonstrate Abn-CBD and NAGly are capable of activating GPR18 and inducing vasodilatation do not unequivocally indicate that the vasodilator effect of these compounds is mediated by GPR18. In this regard, it is unhelpful that due to a lack of selective GPR18 antagonists, O-1918 and cannabidiol are increasingly used to support the presence of eCB receptors in in vitro and in vivo experiments. Clearly, there is a need to attribute cannabinoid-induced vascular responses to stimulation of a specific orphan GPCR. To tackle this problem, it would be very helpful to develop and test selective agonists and antagonists of GPR18 and GPR55. Future studies on arteries isolated from GPR18 knockout mice would help to understand the role of GPR18 in the vasodilator effects of cannabinoids commonly attributed to eCB receptors. In addition, as endothelial electrical signalling has a key role in cell function and as there are limitations in cell culture models, systematic characterization of electrical events at the level of in situ vascular endothelium initiated by CB₁/CB₂ receptor-inactive cannabinoids with the identification of GPCR-dependent signalling is ultimately required. This would pinpoint the responses either to specific GPCR stimulation or GPCR-independent targets. It would be important to determine whether manipulations in GPR18 expression in excised vessels affect endothelial cell electrical responses and endothelium-dependent relaxation and other vascular functions. It is also important to extrapolate any signalling crosstalk between heterologously expressed GPR55 and CB₁/CB₂ receptors (Kargl et al., 2012; Martinez-Pinilla et al., 2014) and determine the biased agonism of GPR18 signalling reported for GPR18-expressing HEK293 cells (Console-Bram et al., 2014) to native GPCRs expressed in the vascular endothelium.

In vascular preparations, molecular targets for cannabinoids may be located both on endothelial and smooth muscle cells. In the latter case, the signal initiated can be transmitted to the endothelial cell layer either through gap junctions or a diffusible messenger to produce an endothelium-dependent response. It is essential, therefore, to differentiate endothelium-derived signalling from that originating within the smooth muscle cells. Further work on understanding the mechanisms of the regulation of CB₁/CB₂ receptor-independent vascular effects of cannabinoids would provide further insights into the elucidation of the role of novel GPCRs sensitive to cannabinoids in health and disease.

Conflict of interest

None.

Glossary

Abn-CBD

abnormal cannabidiol

ARA-S

N-arachidonyl serine

BK_Ca

high-conductance Ca²⁺-dependent K⁺ channels

CB₁ receptor

cannabinoid receptor type 1

CB₂ receptor

cannabinoid receptor type 2

eCB receptor

endothelial cannabinoid receptor

GPR18

GPCR 18

NAGly

N-arachidonyl glycine

O-1918

1,3-dimethoxy-5-methyl-2-[(1R,6R)-3-methyl-6-(1-methylethenyl)-2-cyclohexen-1-yl]-benzene

SR141716

rimonabant

THC

Δ⁹-tetrahydrocannabinol

TRP

transient receptor potential