Receptors for acylethanolamides—GPR55 and GPR119

Abstract

Acylethanolamides are lipid substances widely distributed in the body, generated from a membrane phospholipid precursor, N-acylphosphatidylethanolamine (NAPE). The recent identification of arachidonoyl ethanolamide (anandamide or AEA) as an endogenous cannabinoid ligand has focused attention on acylethanolamides, which has further increased with the subsequent identification of related additional acylethanolamides with signaling function, such as oleoylethanolamide (OEA) and palmitoylethanolamide (PEA). Most of the biological functions of anandamide are mediated by the two G protein-coupled cannabinoid receptors identified to date, CB₁ and CB₂, with the transient receptor potential vanilloid-1 receptor being an additional potential target. There has been increasing pharmacological evidence for the existence of additional cannabinoid receptors, with the orphan G protein-coupled receptor GPR55 being the most actively scrutinized, and is one of the subjects of this review. The other receptor reviewed here is GPR119, which can recognize OEA and PEA. These two acylethanolamides, although structurally related to anandamide, do not interact with classical cannabinoid receptors. Instead, they have high affinity for the nuclear receptor PPARα, which is believed to mediate many of their biological effects.

1. Acylethanolamides

Acylethanolamides or fatty acid ethanolamides are lipid substances widely distributed in the body, present both in brain and various peripheral tissues, most notably in the gastrointestinal tract [1]. They are thought to be generated from a membrane phospholipid precursor, N-acylphosphatidylethanolamine (NAPE) by phospholipase d-mediated cleavage [2], although additional, parallel enzymatic pathways have also been identified [3,4]. NAPE has been recently identified as a hormone-like anorexic factor present in circulating blood [5], although it is not clear whether this soluble form also serves as substrate for the biosynthesis of acylethanolamides. Interest in acylethanolamides has sharply increased following the identification of arachidonoyl ethanolamide (anandamide or AEA) as the first endogenous ligand of the brain type or CB₁ cannabinoid receptor in 1992 [6]. Since that time, a number of related acylethanolamides, such as oleoylethanolamide (OEA) and palmitoylethanolamide (PEA), have emerged as endogenous lipids with possible signaling and biological regulatory functions.

Most of the biological functions of anandamide and the subsequently identified ‘second’ endocannabinoid, 2-arachidonoylglycerol (2-AG) [7,8], as well as their plant-derived and synthetic cannabinoid counterparts are mediated by the two G protein-coupled cannabinoid receptors identified to date, CB₁ [9] and CB₂ [10], with the transient receptor potential vanilloid-1 (TRPV1) receptor being an additional target of anandamide [11], although one with a lower affinity than the affinity of anandamide for cannabinoid receptors. However, there has been a steady stream of pharmacological evidence for the existence of additional cannabinoid receptors [12], with the orphan G protein-coupled receptor (GPCR) GPR55 receiving most of the attention to date (see recent reviews [13–17], and is one of the subjects of the present review.

Despite its structural similarity with AEA, OEA is not a ligand at classical CB receptors. Indeed, its prominent effects as an anorexic factor that also promotes fat elimination have been convincingly linked to activation of the nuclear receptor PPARα [18]. However, some effects of OEA are independent of PPARα [19], and recent evidence indicates that OEA may also interact and produce certain biological effects via GPR119 [20], which is the second subject reviewed here. PEA also has some unique biological actions, such as its well documented analgesic and anti-inflammatory activity [21], the latter likely mediated via PPARα [22]. PEA has very low affinity for CB₁ or CB₂ receptors [23], but it may be a reasonably potent agonist for GPR55 [24].

2. GPR55

2.1. The structure of GPR55

The cloning of the human GPR55 was originally reported in 1999, when it was described as a classical intronless GPCR that maps to chromosome 2 and consists of 319 amino acids. Philogenetically, GPR55 belongs to the δ (purine) cluster of the rhodopsin family that comprises receptors for adenosine and uridine nucleotides and also contains some orphan receptors [25,26]. It displays limited amino acid sequence homology with GPR23 (30%), P2Y5 (29%), GPR35 (27%) and CCR4 (23%), homologies with CB₁ (13%) and CB₂ receptors (14%) being much lower [27]. Evidence suggesting that GPR55 may be a cannabinoid receptor first emerged in the patent literature in the early years of this decade (reviewed in [17]), and will be discussed in more detail below.

2.2. Tissue distribution of GPR55

The profile of GPR55 expression in tissues mostly comes from studies utilizing Northern blot analysis or quantitive polymerase chain reaction to detect the receptor at the message level. Thus, high levels of human GPR55 mRNA transcripts have been found in brain regions implicated in the control of memory, learning and motor functions, such as the dorsal striatum, caudate nucleus and putamen, and in several peripheral tissues, including the ileum, testis, spleen, tonsil, breast and omental adipose tissue [14,27], as well as some endothelial cell lines [28]. Rat and mouse GPR55 homologs have also been detected in other brain regions (frontal cortex, hippocampus, thalamic nuclei, brain stem and regions of the midbrain) and in peripheral tissues (jejunum, ileum, colon, adrenals and spleen) [14,24]. The significance of this distribution requires verification at the protein level and also by functional assays. So far, expression of GPR55 protein has only been documented in primary cultured mouse sensory neurons, using immunohistochemistry [29].

2.3. Pharmacology of GPR55

As for the cannabinoid-like pharmacology of GPR55, there has been a surprising variability across different studies, summarized in a recent excellent review [13] and also illustrated by the data in Table 1. There is general agreement that certain cannabinoid ligands interact with GPR55 with high affinity, and that the relative affinities of different cannabinoid ligands are distinct from the relative affinities of the same ligands established for either CB₁ or CB₂ receptors, as exemplified by a few key ligands for which results have been consistent across different studies and different test systems. For example, HU210, a potent synthetic agonist of CB₁ and CB₂ receptors, and JWH015, a selective CB₂ agonist, are both potent agonists at GPR55 [24,29],whereasWIN55212-2, another synthetic cannabinoid which is somewhat more potent at CB₂ than CB₁ receptors, is inactive at GPR55 [24,29–31]. Certain atypical cannabinoids that are not recognized by CB₁ or CB₂ receptors, such as abnormal cannabidiol (abnormal-CBD) and O-1602 [32], are potent agonists of GPR55 [24,28,31], whereas cannabidiol (CBD) and its analog O-1918 [33] are antagonists [24,28]. This would justify the labeling of GPR55 as ‘CB₃’ receptor, as long as other criteria for cannabinoid receptors are met. Such a designation, however, is not well justified when one considers that neither of the two main endocannabinoids, AEA or 2-AG, has consistent effects at GPR55. Although AEA was reported to be a potent agonist (EC₅₀ of 18 nM) using GTPγS labeling and RhoA activation assays in human embryonic kidney (HEK293) cells transfected with the human GPR55 cDNA [24], it was much less potent and efficacious [29] or had no agonist activity at all in increasing intracellular calcium ([Ca²⁺]_i) [34] or promoting extracellular signal-regulated kinase (ERK) phosphorylation [30,35]. 2-AG was active in increasing GTPγS labeling in one study (EC₅₀ of 3 nM)[24], but was inactive in promoting calcium release or ERK phosphorylation in several other reports [29,30,34,35]. The weak activity of the two main endocannabinoids at GPR55 was further confirmed by using the PathHunter™ β-arrestin binding assay designed to evaluate GPCR-ligand pairing [36]. By measuring a signal that is specifically generated via a tagged receptor, the assay eliminates interference by endogenous receptor present in the parental cell. Using this approach, GPR55 was found to respond strongly to the CB₁ antagonists rimonabant and AM251 and also to LPI, but weakly to endocannabinoids [36]. Such findings argue against GPR55 being a cannabinoid or, more importantly, an endocannabinoid receptor, an argument also supported by the lack of significant homology with CB₁ or CB₂ receptor functional fingerprints, as defined by the alignment of amino acid residues implicated in the binding of specific cannabinoid ligands, and suggesting the absence of a cannabinoid ligand binding pocket [16]. This raises the question of what is the cognate endogenous ligand of GPR55.

Table 1.

Pharmacological profile of compounds tested as GPR55 ligands.

Compound	Cell type	Measured response [references]
		[35S]GTPγS	[Ca2+]i mobilization	pERK 1/2	RhoA	NAFT	β-Arrestin [36]
Cannabinoids
AEA	hGPR55-HEK293	EC50 18 nM [24]	5 µM [29]; no effect [32,35]	No effect [30,35]	1 µM [24]	-	Very weak
	EA.hy926	-	EC50 7.3 µM [28]	10 µM [28]	-	10 µM [28]	-
Meth-AEA	hGPR55-HEK293	-	5 µM [29]	-	-	-	-
	EA.hy926	-	EC50 10.6 µM [28] 5 µM [29]	-	-	-	-
OEA	hGPR55-HEK293	EC50 440 nM [24]	-	No effect [30]	-	-	-
PEA	hGPR55-HEK293	EC50 3–4 nM [24]	No effect [29]	No effect [30]	-	-	-
Virodhamine	hGPR55-HEK293	EC50 10–12 nM [24]	No effect [29]	No effect [30]	-	-	-
2-AG	hGPR55-HEK293	EC50 3 nM [24]	No effect [29,32,35]	No effect [30,35]	-	-	No effect
Noladin ether	hGPR55-HEK293	EC50 10–11 nM [24]	-		-	-
abnormal-CBD	hGPR55-HEK293	EC50 2.5 nM to 2.8 µM [20,24,31]	No effect [29]	No effect [30]	-	-	No effect
O1602	hGPR55-HEK293	EC50 2.5–13 nM [24,31]	No effect [35]	No effect [30]	1 µM [24]	-	No effect
CBD	hGPR55-HEK293	IC50 350 nM [24]	No effect [29]	-	Antagonist-10 µM [6]	-	-
	EA.hy926	-	10 µM [28]	-	1 µM [24]	-	No effect
O1918	EA.hy926	-	Antagonist [28]	-	-	-	-
Δ9THC	hGPR55-HEK293	EC50 8 nM [24]	5 µM [29]	5 µM [29]; No effect [30]	5 µM [29]	-	Weak
	Mouse DRG	-	5 µM [29]	-	-	-	-
HU210	hGPR55-HEK293	EC50 26–33 nM [10,24]	-	No effect [30]	-	-	No effect
CP55940	hGPR55-HEK293	EC50 5–7 nM [24]	No effect [29,35]; antagonist [32]	No effect [30]	-	-	No effect
WIN55212-2	hGPR55-HEK293	No effect [24,31]	No effect [29]	No effect [30]	No effect [24]	-	-
AM251	hGPR55-HEK293	EC50 39 nM [24]	-	-	-	-	EC50 2.7 µM
Rimonabant	hGPR55-HEK293	EC50 600 nM [24]	-	No effect [30]	-	-	EC50 9.3 µM
	Mouse DRG	-	Antagonist-2 µM [29]	-	-	-	-
JWH015	hGPR55-HEK293	EC50 4 nM [24]	3 µM [29]	3 µM [29]	3 µM [29]	-	-
	Mouse DRG	-	3 µM [29]	-	-	-	-
Lysophospholipids
LPI	hGPR55-HEK293	EC50 1 µM [30]	EC50 49 nM to 1 µM [30,32]	EC50 200 nM [30]	1 µM [32]	1 µM [32]	EC50 3.6 µM
	EA.hy926	-	10 µM [28]	-	-	-	-
	Mouse DRG	-	3 µM [29]	-	-	-	-
2-AG-LPI	hGPR55-HEK293	-	EC50 30 nM [35]	EC50 1 µM [35]	-	-	-

2.4. Endogenous ligand(s)

Recent studies have provided strong evidence for lysophos-phatidylinositol (LPI) being such a ligand. Earlier observations have indicated that various species of LPI are present in tissues and biological fluids and can modulate a variety of cellular functions, including intracellular calcium levels [37]. LPI promoted ERK phosphorylation, a rise in [Ca²⁺]_i and GTPγS labeling in HEK293 cells transfected with hGPR55, but not in mock-transfected cells [30]. The LPI-induced rise in [Ca²⁺]_i depended on G^α13 and required activation of the small GTP-binding protein RhoA, whereas neither AEA nor 2-AG were active in this paradigm [34]. LPI also increased [Ca²⁺]_i in mouse large diameter dorsal root ganglion neurons, which express high levels of native GPR55 [29]. In this system, the effect depended on phospholipase Cβ activation via G_αq/G_α12, and the source of the calcium released could be traced to IP₃ receptor-gated intracellular stores [29]. Additional findings indicate that among the individual molecular species of LPI, the 2-arachidonoyl species is the most potent in phosphorylating ERK and releasing intracellular calcium (EC₅₀ is in the 10–100 nM range), and is thus the most likely endogenous ligand [35].While 2-arachidonoyl LPI was found to be present in the brain, representing 8.3% of all LPI species [35], its physiological functions and the mechanism of its biosynthesis remain to be clarified.

2.5. GPR55 signaling

As evidence for G protein coupling of GPR55, its agonist ligands have been consistently reported to increase [³⁵S]GTPγS binding with nanomolar potencies in membrane fractions prepared from HEK293 cells transfected with the human GPR55 gene, but not in vector-transfected control cells (hGPR55-HEK293; Table 1) [24,30]. Using peptides directed against the C termini of G protein α subunits, Ryberg and coworkers have demonstrated that GPR55 couples to G_α13 and activates small GTPases (RhoA, cdc42 and rac1) [24]. The putative endogenous ligand of GPR55, LPI, was also found to activate G_α13 and RhoA, resulting in oscillatory release of [Ca²⁺]_i as well as further downstream activation the nuclear factor of activated T cells (NFAT) family of transcription factors, that regulate gene expression [34]. Others reported agonist-induced association ofGPR55withG_α12 and G_αq, and the resulting increase in intracellular calcium from IP₃ receptor-gated stores requiring PLC activation in the case of G_αq, or RhoA activation and actin polymerization in the case of G_α12, although these responses required higher, micromolar concentrations of agonists [29]. Although activation of G_aq and rise in [Ca²⁺]_i are generally associated with vascoconstriction, the opposite response is triggered when this happens in the endothelium, where a rise in [Ca²⁺]_i stimulates the activity of intermediate and small calcium-activated potassium channels, leading to hyperpolarization and vasorelaxation [38]. An intriguing, further complexity of GPR55 signaling has been revealed in a recent study, which provided evidence for the role of integrins in regulating anandamide signaling via CB₁ receptors and GPR55 in endothelial cells [28]. The human endothelial cell line EA.hy926 expresses both CB₁ and GPR55 receptors, each of which can be activated by anandamide. In this study, preferential activation of one or the other was found to depend on the activity status of integrins, cell surface receptors for adhesion molecules with a key role in communications between the intra- and extracellular milieu [28]. In the presence of extracellular calcium, which inhibits integrin clustering, anandamide preferentially activated G_i/o-coupled CB₁ signaling with downstream activation of spleen tyrosine kinase (Syk), which then activated NFκB and inhibited phosphoinositide 3-kinase (PI3K). Since PI3K activation plays a key role in GPR55 signaling, under these conditions GPR55 activation was suppressed. On the other hand, in the nominal absence of extracellular calcium, which results in the clustering of integrins, anandamide as well as the GPR55 ligand O-1602 evoked a rise in intracellular calcium from IP₃R-gated stores via PI3K and PLCγ activation [28]. The specific involvement of αvβ3 and α5β1 integrins was indicated by the ability of crosslinkers or neutralizing antibodies to these integrins to abrogate the GPR55-mediated calcium mobilization [28]. The presence of such alternative pathways may explain the ability of ligands that recognize both receptors to elicit opposite functional responses. For example, in human umbilical vein endothelial cells (HUVEC), anandamide acting via CB₁ receptors was reported to inhibit endothelial cell proliferation and angiogenesis [39], whereas selective activation of a non-CB₁/CB₂ receptor, possibly GPR55, by abnormal-CBD leads to increased endothelial cell migration and angiogenesis [40].

2.6. Is GPR55 the ‘endothelial’ cannabinoid receptor?

A question of particular interest is whether or not GPR55 fits the bill of the putative endothelial anandamide receptor, also dubbed the abnormal-CBD receptor for the ability of this synthetic analog to cause CB₁/CB₂-independent, endothelium-dependent vasodilation in certain vascular beds [12,15,41–43], as first documented 10 years ago in our laboratory [32,33,44]. Additional studies suggest that a putative receptor with the same or similar pharmacology is also involved in regulating microglia migration [45] and microglia-mediated neuroprotection [46], endothelial cell [40] and neutrophil migration [47], endothelial cell transformation induced by Kaposi sarcoma-associated herpesvirus infection [48], and decreasing cardiac contractility [49]. There are several shared key features of GPR55 and the abnormal-CBD receptor: with regard to the pharmacology of these sites, anandamide, but not 2-AG, appears to be an agonist ligand for both receptors, albeit with low affinity in both cases, whereas the synthetic cannabinoid WIN55212-2 is inactive at both; furthermore, abnormal-CBD and its analog, O-1602 [32], act as selective agonists, whereas CBD, which is similarly devoid of activity at CB₁ or CB₂ receptors, acts as antagonist at both receptors [24,32]. The potent, CB₁-selective antagonist rimonabant, but not its close structural analog AM251, can antagonize responses triggered via both GPR55 [29] and the abnormal-CBD receptor [32,50], but with an inhibitory potency one to two orders of magnitude lower than its inhibitory potency at CB₁ receptors.

Despite these similarities, there are some important differences that have been often cited as evidence for the distinct molecular nature of these two sites. First, the endothelium-dependent vasodilator effect of anandamide is pertussis toxin-sensitive [32,51], suggesting the involvement of G_i/o proteins, whereas effects mediated by GPR55 are not, and are instead linked to G_α12, G_α13 or G_αq activation and rises in intracellular calcium, as discussed above. Second, there are some notable differences in the pharmacology of signaling via GPR55 versus the abnormal-CBD receptor, such as LPI being a potent agonist at the former [35] but inactive at the latter [47]. Third, the hypotensive/vasodilator actions of abnormal-CBD and its antagonism by the synthetic cannabinoid O-1918 persist in GPR55 knockout mice [31]. However, these differences are not immutable. For example, the vasodilator response to abnormal-CBD is much less sensitive to pertussis toxin than the effect of anandamide [33,52,53], or not sensitive at all [43]. A possible explanation of such discrepant findings is that the pertussis toxin sensitivity of the effect of anandamide may reflect the involvement of CB₁/GPR55 heterodimers in these effects. Cannabinoid receptors can form heterodimers with other GPCR [54], but the possibility of their dimerization with GPR55 remains to be tested. One might speculate that to the extent such heterodimers occur, they may also account for some of the differences in the pharmacology of GPR55 and the abnormal-CBD receptor, as detected in different studies. Finally, although abnormal-CBD retained its hypotensive and mesenteric vasodilator actions in GPR knockout mice [31], this may not exclude the identity of GPR55 with the abnormal-CBD receptor. In vivo hypotension is a complex response with little or no relation to localized vasodilation. For example, anandamide-induced hypotension, but not mesenteric vasodilation, is absent in CB₁ receptor knockout mice [32,55]. As for the mesenteric vasodilator effect of abnormal-CBD and related ligands, only the endothelium-dependent component has been linked with a putative G_i/o-coupled GPCR [32,33], the endothelium-independent component is likely mediated by a direct activation of BK_Ca channels in vascular smooth muscle [56]. Unpublished observations in our laboratory indicate that mesenteric artery segments of mice, as opposed to rats, lack the endothelium-dependent component in the vasodilator response to atypical canabinoids. Thus, the possible identity of GPR55 with the putative endothelial abnormal-CBD receptor, although elusive, remains to be further tested. Another unanswered question is whether N-arachidonoyl-l-serine, a recently identified lipid brain constituent with vasodilator properties similar to those of abnormal-CBD [57] is also a ligand for GPR55.

2.7. GPR55 and pain sensitivity

Apart from the possible, but as yet unproven, involvement of GPR55 in cardiovascular regulation (see above), recent studies indicate its possible role in the control of pain sensitivity. GPR55−/− mice were found to lack mechanical hyperalgesia in rodent models of inflammatory and neuropathic pain [58], extending earlier observations by Lauckner and coworkers who documented abundant expression of GPR55 in large dorsal root ganglion neurons in mice and its ability to modulate the activity of these neurons [29]. Further studies are necessary to explore the potential therapeutic usefulness of selective GPR55 ligands as analgesic agents.

3. GPR119

3.1. Structure of GPR119

The human orphan receptor GPR119 was identified through a bioinformatics approach [26]. Its sequence was found to align with the earlier reported hGPCR2 receptor [59] and it is present in other mammalian species [26,60]. The 335 amino acid hGPR119 protein is encoded by an intronless gene located on chromosome X [26,59]. Philogenetically, GPR119 has been assigned to the MECA (melanocortin; endothelial differentiation gene; cannabinoid; adenosine) receptor cluster, which designates cannabinoid receptors among its closest relatives [25,26].

3.2. Tissue distribution of GPR119

GPR119 displays a relatively narrow expression pattern. The mRNA transcripts of human and rodent GPR119 have been found predominantly in pancreatic and intestinal tissues [61–63]. Rodents also express GPR119 in some brain regions [60–64]. Autoradiographic and immunohistochemical data indicate that localization of GPR119 is largely confined to a subset of cells in the pancreatic islets of Langerhans where it co-localizes with insulin [62], although another study indicates predominant GPR119 localization in pancreatic polypeptide-releasing cells [65]. In the gastrointestinal tract, GPR119 immunoreactivity was observed within villi of the small intestine where it co-localizes with glucagon-like peptide-1 (GLP-1) in proglucagon positive cells [62]. Consistent with these reports, GPR119 was also documented in the pancreatic β cell lines NIT-1, MIN6, RIN5, HIT-T15, the insulinoma cell line MIN6c4, in enterocrine L-cell models such as FRIC, mGLUTag, hNCI-H716, and in mouse L-cell primary cultures [61,62,64,66,67].

3.3. GPR119 signaling

Cells expressing GPR119 at high levels display a constitutive increase in intracellular cAMP [62], which implies coupling of GPR119 to G_s. Putative GPR119 agonists have also been reported to increase cAMP, stimulate adenylyl cyclase and enhance protein kinase A activity in cells expressing native or recombinant GPR119 [20,61–63,67,68] see also Table 2). There is also evidence for the involvement of ATP-sensitive K⁺ and voltage-dependent Ca²⁺ channels in GPR119-mediated responses [66].

Table 2.

Pharmacological profile of compounds tested as GPR119 ligands.

Compound	Cell type	Measured response	Effect, EC50 value	[Reference]
Cannabinoids
Anandamide	hGPR119-yeast	Fluorimetric assay	very weak	[57]
OEA	hGPR55-HEK293	↑cAMP	EC502.9 µM	[57]
	hGPR55-yeast	Fluorimetric assay	EC503.2 µM	[57]
	mGPR55-yeast	Fluorimetric assay	EC502.9 µM	[57]
PEA	hGPR55-yeast	Fluorimetric assay	EC503.2 µM, low efficacy	[57]
2-AG	hGPR55-yeast	Fluorimetric assay	no effect	[57]
CP5594	hGPR55-yeast	Fluorimetric assay	no effect	[57]
WIN55212-2	hGPR55-yeast	Fluorimetric assay	no effect	[57]
JWH133	hGPR55-yeast	Fluorimetric assay	no effect	[57]
Lysophospholipids
Oleoyl LPC	hGPR119-RH7777	↑cAMP	EC501.5 µM	[58]
	hGPR55-yeast	Fluorimetric assay	EC50 >30 µM	[57]
	mGPR55-yeast	Fluorimetric assay	EC50 >30 µM	[57]
Palmitoyl LPC	hGPR119-RH7777	↑cAMP	EC501.6 µM	[58]
Stearoyl LPC	hGPR119-RH7777	↑cAMP	EC503.3 µM	[58]
LP-ethanolamine	hGPR119-RH7777	↑cAMP	EC505.7 µM	[58]
LPI	hGPR119-RH7777	↑cAMP	EC505.7 µM	[58]
Synthetic ligands
PSN632408	hGPR55-HEK293	↑cAMP	EC501.9 µM	[57]
	hGPR55-yeast	Fluorimetric assay	EC505.6 µM	[57]
	mGPR55-yeast	Fluorimetric assay	EC507.9 µM	[57]
AR231453	hGPR55-HEK293	↑cAMP	EC505.7 nM	[59]
	HIT-T15	↑cAMP	EC504.7 nM	[59]
	HIT-T15	insulin release	EC503.5 nM	[59]
	GLUtag	↑cAMP	EC504.3 nM	[59]
	GLUtag	GLP-1 release	EC5056 nM	[59]

3.4. Pharmacology of GPR119

Because of the close philogenetic proximity of GPR119 to cannabinoid receptors, substances related to endocannabinoids were first considered as potential GPR119 ligands. Thus, using a yeast reporter assay, Overton and coworkers found that among cannabinoids only fatty acid amides induced fluorescence in cells transfected with human or mouse GPR119, with OEA being the most active, followed by PEA, and AEA displaying very weak effect [20] (Table 2). OEA was also found to stimulate cAMP production in HEK293 cells stably transfected with hGPR55, as well as in cell lines expressing native GPR119, while cells lacking GPR119 failed to respond to OEA [20,63]. Additionally, low micromolar concentrations of lysophospholipids, particularly LPC species, activated hGPR119 transgenically expressed in rat hepatoma cells (Table 2) [61].

Using high throughput screening, two groups independently identified oxadizone analogues, typified by PSN632408 [20] and AR231453 [68], as small synthetic GPR119 agonists (Table 2). These compounds were found to increase intracellular cAMP [20,62,68,69] and the secretion of insulin [62] and GLP-1 [69]. The compound AR231453 is particularly noteworthy for its nanomolar affinity for GPR119 (Table 2), while PSN632408 can also elicit some GPR119-independent effects at pharmacologically relevant concentrations [66].

3.5. Physiological function(s) of GPR119

The expression of GPR119 in pancreatic islet β-cells and intestinal enterocrine L-cells focused attention on the possible role of GPR119 in the control of glucose homeostasis and obesity. The first evidence for this came from the observation that theGPR119 agonist LPC improved glycemic control by enhancing glucose-stimulated insulin secretion from the perfused rat pancreas and that a similar effect of LPC in cultured mouse pancreatic β-cells was blunted by GPR119-specific siRNA [61]. The direct insulinotropic effect of GPR119 activation was subsequently documented in vitro using OEA [66] and AR231453 [62,68]. AR231453 was also effective in an in vivo model where it improved glucose tolerance by acting directly on pancreatic β-cells to enhance glucose-dependent insulin release [69]. This effect occurred in lean, diabetic, Lepr^db/db and KKA_y mice, but was entirely lost in GPR119-deficient mice [69]. The observation that AR231453 was less effective when glucose had been delivered intraperitoneally focused attention on the possible involvement of intestinal incretins in its effect. Indeed, AR231453 was found to increase GLP-1 release in vitro, and when it was administered to mice in vivo, it enhanced GLP-1 secretion after an oral glucose load and improved glucose tolerance [69]. This effect was greatly amplified when AR231453 was administered in combination with sitagliptin, an inhibitor of the GLP-1 inactivating enzyme [69]. Consistent with this observation, OEA also increases GLP-1 secretion in vitro and in hyperglycemic rats only after intraluminal administration, and this effect could be augmented by the FAAH inhibitor URB597 [63]. The above data suggest that orally effective GPR119 agonists and/or modulators of OEA metabolism may be used to improve glucose homeostasis.

In contrast to these observations, genetic deletion of GPR119 in mice resulted in no clearcut change in glucose homeostasis. These animals retain normal islet morphology and normal responsiveness to glucose and GLP-1, as well as normal fed/fasted glucose levels [62]. When maintained on a low fat diet, GPR119-deficient mice displayed reduced body weight and lower post-prandial GLP-1 levels relative to wild type littermates. However, these differences were no longer noticeable when the animals were on a high fat diet [64], suggesting that GPR119 is required for physiological regulation of GLP-1, but not for metabolic homeostasis. Nevertheless, the involvement of GPR119 in the physiological modulation of glucose homeostasis cannot be completely excluded, as such regulatory function may manifest under different experimental and/or pathological conditions, in animals with different genetic background, or after combined administration with sitagliptin to accentuate circulating GLP-1 levels, as reported earlier [69]. So far, basal OEA level in GPR119−/− mice has not been measured or whether it correlates with AEA, 2-AG or other lipid molecules. The elucidation of OEA–GPR119 interaction in intestinal tissue is of particular importance, because the duodenal infusion of fat has been shown to mobilize OEA in the small intestine [70] and OEA increases GLP-1 secretion from intestinal L-cells through a GPR119-dependent mechanism [63].

3.6. Does GPR119 play a role in obesity?

The idea that GPR119 may represent a potential target for the treatment of obesity comes from the de-orphanization of GPR119 as the receptor for OEA [20]. OEA had been formerly identified as a hypophagic agent that regulates feeding and body weight through activation of the nuclear receptor peroxisome proliferator-activated receptor α (PPARα), a proposal strongly supported by the loss of OEA effects in PPARα deficient mice [18]. Like OEA, the synthetic GPR119 agonist PSN632408 was found to suppress food intake and body weight gain but, unlike OEA, it is devoid of PPARα activity [20]. On the other hand, OEA was equipotent in suppressing food intake in wild type and GPR119-deficient mice [64], which suggests that the two molecules may act through different mechanisms. Further studies will have to test whether PSN632408 and/or AR231453 retain their effects in GPR119-deficient animals. Moreover, in addition to well recognized PPARα-mediated effects such as anorexia [18,22,71], lipolysis [72] and neuroprotection [73,74], OEA has also been shown to cause a number of effects through PPARα-independent mechanisms, such as inhibition of intestinal motility [19] or reduction of visceral and inflammatory pain [75]. Whether or not such effects may be mediated via GPR119, remains to be determined.