Activation of GPR18 by cannabinoid compounds: a tale of biased agonism

Introduction

The endogenous cannabinoid system, including cannabinoid receptors and endocannabinoids, has been implicated in many physiological processes; including neurotransmission, inflammation/migration, intraocular pressure, neuropathic and inflammatory pain, obesity, osteoporosis, drug addiction, cardiovascular disorders, and liver disease (Pacher et al., 2006; Qin et al., 2011; Battista et al., 2012; McHugh, 2012; Caldwell et al., 2013; Oliere et al., 2013). Presently there are two bonafide cannabinoid receptors, CB₁ and CB₂, several endocannabinoids, of which anandamide (AEA) and 2-arachidonoyl glycerol are the most extensively studied, and their synthetic and degradative enzymes (Pertwee et al., 2010; Alexander et al., 2013; Console-Bram et al., 2012). Cannabinoid receptors were so named as they were found to bind, and elicit the effects of, the main component of the Cannabis sativa plant, Δ⁹-tetrahydrocannabinol (Δ⁹-THC). However, it is widely accepted that the actions of cannabinoid compounds are not limited to these two receptors. Consequently, it is imperative to elucidate the existence of additional cannabinoid receptors, the compounds with which they interact, and the mechanism(s) of signal transduction. The term cannabinoid compound encompasses three categories of compounds, those derived from the C. sativa plant (phytocannabinoids), synthetically derived compounds (synthetic cannbinoids), as well as endogenous compounds that were shown to bind to CB₁ and CB₂ receptors (endocannabinoids).

Current opinion describes GPR18 as a candidate cannabinoid receptor (for receptor nomenclature see Alexander et al., 2013). Transfection of GPR18 into CHO, K562 and L929 cells resulted in the mobilization of calcium in the presence of N-arachidonyl glycine (NAGly), a metabolite of the endocannabinoid AEA (Kohno et al., 2006; Bradshaw et al., 2009). Consequently, identification of NAGly as the endogenous ligand of GPR18 was postulated (Kohno et al., 2006). Further support for GPR18 as a candidate cannabinoid receptor has recently emerged as Δ⁹-THC was found to activate GPR18 in migration assays (McHugh et al., 2012) and in the PathHunter β-arrestin assay (Rempel et al., 2013). In addition, as a consequence of the overlap in the pharmacology of NAGly and abnormal cannabidiol (AbnCBD), and the inhibition of these ligand-induced effects by O-1918, McHugh et al. (2010) proposed that the AbnCBD receptor and GPR18 are one in the same.

Anatomical localization of GPR18 was initially reported to be highest in spinal cord and small intestine, with lower levels in testis and cerebellum (Gantz et al., 1997). In a more recent study, localization of GPR18 mRNA in mouse tissue revealed highest levels in spleen and bone marrow, followed by thymus, lung and cerebellar tissue (Regard et al., 2008). This receptor is also expressed in endometrial cells (McHugh et al., 2012), murine microglial cells (McHugh et al., 2010), and is one of the most highly expressed GPCRs in metastatic melanoma (Qin et al., 2011). As a consequence of its localization, it is quite possible that GPR18 may be involved in a myriad of cellular functions. However, very little is known about the physiological effects of GPR18. Thus far, cellular migration, apoptosis, immunostimulant activity in catfish, increased levels of intracellular calcium and augmented MAPK activity are ascribed to this GPCR (Kohno et al., 2006; McHugh et al., 2010; Takenouchi et al., 2012; Pridgeon and Klesius, 2013). The signal transduction mechanism involved in GPR18 activation by NAGly has been reported to be the consequence of Gα_i/o coupling, as Pertussis toxin (PTX) was found to attenuate GPR18-mediated inhibition of cAMP production (Kohno et al., 2006). In studies with BV-2 murine microglia, PTX was found to attenuate NAGly-induced migration (McHugh et al., 2010).

Progress towards acceptance of NAGly as the endogenous ligand of GPR18, and activation of GPR18 by AEA, AbnCBD and O-1602 has been brought into question, as a recent report indicated that none of these ligands activate GPR18 expressed in sympathetic cervical ganglia (Lu et al., 2013). This study examined calcium mobilization via N-type voltage-gated calcium channels of rat superior ganglion cells heterologously expressing GPR18. Low voltage-gated channels of the superior ganglion are described as being inhibited by Gα_i/o proteins via an interaction between Gβγ dimers and β subunits of this type of calcium channel (Ikeda, 1996; Herlitze et al., 1999). Consequently, in the presence of Gα_i/o agonists, the expectation was that a decrease in calcium would be observed. However, Lu et al. (2013) observed an increase in calcium mobilization and thereby concluded neither NAGly, AbnCBD, anandamide nor O-1602 are agonists at GPR18.

As a consequence of the apparent discrepancies in the literature, the rationale of the present study was to examine the effects of phytocannabinoids (Δ⁹-THC), synthetic cannabinoids (AbnCBD, O-1602 and O-1918), and endocannabinoids (NAGly) on calcium mobilization in HEK293 cells heterologously expressing GPR18. HEK293 cells were chosen as they are known to be devoid of N-type voltage-gated calcium channels (Berjukow et al., 1996) and ryanodine receptors (Aoyama et al., 2004). Consequently, any change in calcium mobilization would be a consequence of GPR18 activation, not channel activity. In addition, we investigated MAPK activity following exposure to O-1918, NAGly, O-1602, AbnCBD, Δ⁹-THC and another endocannabinoid, N-arachidonyl serine (NARAS), to extend the findings of McHugh et al. (2012). In an attempt to establish mechanism(s) involved in these GPR18-mediated biological events, the following effector pathway inhibitors were employed: wortmannin (PI3K inhibitor), bisindolymalemide (BI, PKC inhibitor); PTX (Gα_i/o inhibitor), and the Gα_q inhibitor, [D-Trp^7,9,10]-substance P (Mukai et al., 1992). The GPR18-mediated effects of AbnCBD, NAGly, NARAS, O-1602, O-1918 and Δ⁹-THC were also investigated in the β-arrestin signal transduction pathway. In addition, the activity of AEA and CBD in this signalling mechanism was also explored.

Methods

Results

Calcium mobilization in HEK293/GPR18 cells

Extracellular application of Δ⁹-THC, NAGly, O-1602, AbnCBD or O-1918 resulted in significant transient calcium mobilization within a second of exposure (Figure 1). Real-time photographs of calcium mobilization in basal and treated states are depicted in Figure 1A. The specificity of the GPR18-mediated calcium response was evaluated through the measurement of intracellular calcium in HEK293 cells devoid of GPR18 expression in the presence of ligand (Figure 1B, upper right tracing). Concentration–response curves for the ligands are depicted in Figure 1C. Note that the concentration–response curve for AbnCBD is furthest to the right. At equimolar concentrations, this ligand was the least efficacious, whereas O-1918 was the most efficacious (O1918 > NAGly > Δ⁹-THC > O-1602 > AbnCBD). Levels of significance (P < 0.05) for AbnCBD-mediated increases in intracellular calcium were attained at 10 and 3 μM, whereas NAGly, O-1602, and Δ⁹-THC yielded significance at 10, 3 and 1 μM. O-1918-mediated increases in calcium were demonstrated over the largest range of concentrations; 0.3 through 10 μM. To identify the G-proteins involved in agonist-mediated calcium mobilization, HEK293/GPR18 cells were pre-incubated overnight with either the Gα_i/o inhibitor PTX (100 ng·mL⁻¹), or the Gα_q inhibitor [D-Trp^7,9,10]-substance P (10 μM). Calcium mobilization was completely inhibited by both G-protein inhibitors (Figure 1C). Furthermore, in the absence of extracellular calcium, significant agonist-induced GPR18-mediated calcium mobilization occurred (P < 0.05), albeit at lower levels (10 μM NAGly, 299 ± 14.51; 10 μM O-1918, 346 ± 10.92; 10 μM Δ⁹-THC, 273 ± 5.1 nM). An example of trace recordings with an associated graph of intracellular calcium levels in cells incubated with agonists in the absence of extracellular calcium is shown with Δ⁹-THC in Figure 2. Note that the time of peak intracellular calcium levels in the absence of extracellular calcium is similar to that observed in the presence of extracellular calcium.

MAPK activity in HEK293/GPR18 cells

Initial time course studies revealed that levels of ERK1/2 phosphorylation remain elevated for 2 h post exposure to NAGly whereas levels returned to basal levels within 40 min following exposure to AbnCBD and O-1602 (Figure 3). Peak increases were attained 5 min post ligand exposure; consequently, levels of MAPK activity were quantitated at this time point. Significant concentration-dependent increases in ERK1/2 phosphorylation were observed in HEK293/GPR18 cells following exposure to NAGly, O-1602, AbnCBD and Δ⁹-THC (Figure 4A, C, D and E), similar to previous reports (McHugh et al., 2012). In addition, we demonstrate significant concentration-dependent increases in ERK phosphorylation in the presence of O-1918 (Figure 4B). As shown in Figure 4, pre-incubation with 100 μg·mL⁻¹ of PTX or 1 μM NARAS inhibited the ligand-induced increases in ERK1/2 phosphorylation. No effect was observed with incubation of either NARAS or PTX in the absence of cannabinoid ligands as illustrated in Figure 4C.

Further characterization of the MAPK response was performed using the most efficacious ligand, O-1918, with the PI3K inhibitor wortmannin (100 nM), or the PKC inhibitor BI (10 nM). As illustrated in Figure 5, application of either inhibitor attenuated O-1918-dependent ERK1/2 phosphorylation. Because activation of Gα_q leads to production of diacylglycerol and inositol-3-phosphate via PLC cleavage and concomitant PKC activation, the results of these experiments supports the involvement of Gα_q in GPR18 activation.

β-Arrestin activation

Of all the agonists investigated in the MAPK assay only Δ⁹-THC, had a concentration-dependent effect on β-arrestin translocation (Figure 6). Increases in β-arrestin activity were significantly different at 10 and 30 μM Δ⁹-THC in comparison with the other cannabinoid ligands (AbnCBD, AEA, NAGly, NARAS, O-1602 and O-1918). It is interesting to note that slight significance in β-arrestin activity was observed at 30 μM CBD, yet the CBD analogues O-1602 and O-1918 had no effect on β-arrestin translocation. These findings further demonstrate multiple signalling avenues for GPR18 and suggest biased agonism at this GPCR.

Cellular localization of GPR18

Distribution of GPR18 in HEK293 and HEK293/GPR18 cells using immunocytochemistry is illustrated in Figure 7. Figure 7A (live, nonpermeabilized) and 7B (permeabilized) depict staining of cells with an anti-HA antibody illustrating both plasma membrane and intracellular localization of GPR18. A similar pattern was observed using an anti-GPR18 antibody in permeabilized cells (Figure 7C). Figure 7D illustrates the absence of detectable GPR18 in untransfected HEK293 cells using both antibodies thereby demonstrating specificity of the antibodies.

Discussion

Findings from the study presented here indicate that GPR18 signalling is rather complex. Results from our study suggest the occurrence of biased agonism at GPR18 as activation of this receptor by a particular ligand was dependent upon the biological endpoint. Increases in both intracellular calcium and MAPK, two cellular signalling pathways, were observed in HEK/GPR18 cells in the presence of five cannabinoid ligands; Δ⁹-THC, AbnCBD, NAGly, O-1602 and O-1918. However, the only agonist that exhibited concentration-dependent GPR18-mediated β-arrestin activity was Δ⁹-THC.

Our research is the first to suggest multiple G-protein involvement in GPR18 activation. The Gα_i/o pathway is implicated in view of the findings that ligand-induced increases in MAPK and calcium mobilization were attenuated by PTX treatment. Given that Δ⁹-THC-induced increases in MAPK were not completely blocked by PTX, and that the GPR18-mediated rises of intracellular calcium induced by NAGly, Δ⁹-THC and O-1918 were also inhibited by the Gα_q inhibitor, [D-Trp^7,9,10]-substance P (Mukai et al., 1992), the Gα_q pathway is probably involved as well. Additional support for involvement of both G-proteins comes from the observation that the Gα_q inhibitor completely blocked the initial and secondary rises in intracellular calcium in the presence of NAGly, Δ⁹-THC and O-1918; whereas, only the initial rise in intracellular calcium was inhibited by PTX (the secondary rise, observed 10 min following agonist exposure, was not; data not shown). Furthermore, the demonstrated inhibition of ERK1/2 phosphorylation by inhibition of PKC activity, known to be activated via Gα_q, reinforces the involvement of Gα_q in GPR18 activation. Participation of both Gα_i and Gα_q transduction pathways suggests effector redundancy; perhaps for indispensable cellular function(s) and/or maintenance of homeostasis. Several GPCRs have been shown to exhibit promiscuity in coupling to G-proteins; for example, rhodopsin and β₂-adrenoceptor (Cerione et al., 1985), GPR55 (Lauckner et al., 2008; Obara et al., 2011), melanin-concentrating hormone receptor (Hawes et al., 2000) and CB₁ receptor (Ahn et al., 2013). Furthermore, crosstalk among G-protein subunits has been observed for Gα_s- and Gα_q-coupled adenosine receptors (Benovic et al., 1985). Hence, the involvement of multiple G-proteins in signal transduction via GPR18 is quite plausible.

The results presented here corroborate the agonist ligand profiles of AbnCBD, NAGly, O-1602 and Δ⁹-THC as reported by McHugh et al. (2012). Our study is the first report of NARAS inhibition of MAPK activity induced by AbnCBD, NAGly, O-1602 and O-1918. This finding complements the reported inhibition of NAGly–GPR18-mediated cellular migration by NARAS (McHugh et al., 2010). However, our finding that O-1918 mediates increases in MAPK activity via GPR18 activation is in contrast with previously reported GPR18 antagonist activity of O-1918 in migration studies using BV-2 and HEK/GPR18 cells (McHugh et al., 2010). Additionally, attenuation of AbnCBD-induced MAPK activity by high concentrations of O-1918 has been reported in HUVEC cells (Offertaler et al., 2003).

Further characterization of O-1918-mediated increases in MAPK activity was performed as a consequence of enhanced calcium mobilization with this ligand. Increases in intracellular calcium, such as triggered by the activation of GPR18, are known to activate several isoforms of PKC. Once activated, either via calcium and/or directly by Gα_q activation, PKC may lead to activation of the MAPK and PI3K pathways (Breitkreutz et al., 2007; Rozengurt, 2007). In addition, βγ dimer formation is known to activate PI3K (Naor et al., 2000). We therefore examined whether inhibition of PKC and PI3K pathways would affect O-1918-mediated ERK1/2 phosphorylation. Both the PKC inhibitor (BI), and the PI3K inhibitor (wortmannin), attenuated increases in ERK1/2 phosphorylation. Thus, our findings demonstrate the multiplicity of cellular pathways involved in GPR18-mediated increases in MAPK activity.

As indicated earlier, increases in MAPK induced by AbnCBD, NAGly, O-1602, O-1918 and Δ⁹-THC were sensitive to PTX. In both BV-2 and human endometrial cells, PTX also attenuates migration induced by NAGly (McHugh et al., 2010; 2012); however, sensitivity to PTX by the other ligands was not demonstrated. In addition, there are discrepancies in the literature with respect to PTX sensitivity of endothelial-dependent AbnCBD-induced vasorelaxation of mesenteric arteries. For example, AbnCBD-induced vasorelaxation of mesenteric arteries has been reported to be both sensitive (Offertaler et al., 2003) and insensitive to PTX treatment (Ho and Hiley, 2003). Whereas aortic relaxation induced by AbnCBD has been demonstrated to be sensitive to PTX inhibition (Milman et al., 2006). NARAS also induced vasorelaxation of aortic and mesenteric vessels; however, these relaxant effects were less endothelium-dependent in mesenteric as compared with aortic arteries, and the effect of NARAS in aortic vessels, but not mesenteric vessels, was antagonized by PTX (Milman et al., 2006). Conversely, O-1918 antagonized the effect of NARAS in mesenteric vessels, but not in aortic vessels. Milman et al. (2006) also reported that AbnCBD- and NARAS-induced increases in MAPK activity were attenuated by PTX in HUVEC cells. As a consequence of the similarity between aortic vessels and HUVEC cells, with respect to PTX sensitivity of these ligand-induced effects, the authors suggested that the vascular receptor mediating these effects were one and the same. However, because of the differences between mesenteric and aortic arteries, they suggested the existence of more than one vascular receptor for these ligands. Alternatively, a single receptor coupling to different G-protein subunits and/or signal transduction pathways, depending upon the local milieu, may satisfy these apparent discrepancies in PTX and O-1918 sensitivity of the ‘AbnCBD’ receptor. Additionally, this explanation may also serve to support the identification of GPR18 as the ‘AbnCBD’ receptor.

To add further complexity to the activation of GPR18, a recent study of rat superior cervical ganglion neurons heterologously expressing GPR18 reported that neither NAGly nor AbnCBD resulted in activation of this receptor (Lu et al., 2013). As activation of Gα_i-coupled GPCRs is generally associated with a decrease in calcium current via N-channel modulation, yet an increase in calcium current following exposure to NAGly and AbnCBD was observed, these investigators reported that these ligands did not activate GPR18. We chose to examine the properties of GPR18 in HEK293 cells, as they are known to be devoid of N-type voltage-gated calcium channels (Berjukow et al., 1996), and ryanodine receptors (Aoyama et al., 2004). Consequently, any change in calcium mobilization would be a consequence of GPR18 activation, not channel activity. Our findings clearly demonstrate that GPR18 is activated by NAGly, AbnCBD, O-1602, O-1918 and Δ⁹-THC, resulting in increased calcium mobilization and MAPK activity. Nonetheless, it is feasible that the properties of GPR18 may be different in neuronal cells.

Our findings that PTX totally blocks MAPK activity induced by AbnCBD, NAGly, O-1602 and O-1918, but not Δ⁹-THC, supports the involvement of multiple signal transduction pathways in GPR18 activation. The attenuation (as opposed to blockade) of Δ⁹-THC-induced increases in ERK1/2 phosphorylation by PTX suggests involvement of more than Gα_i/o coupling. Furthermore, the calcium mobilization studies presented here suggest that GPR18 activation can occur via differential coupling of G-proteins. The complete inhibition of calcium mobilization in the presence of either PTX, or [D-Trp^7,9,10]-substance P, suggests a dual pathway of Gα_i and Gα_q transduction. Attenuation of MAPK by PI3K and PKC inhibitors substantiate the involvement of PKC in GPR18 activation, a known effector of Gα_q-coupled receptors. Findings that AbnCBD, NAGly, O-1602, O-1918 and Δ⁹-THC enhance intracellular calcium and MAPK activity, yet only Δ⁹-THC activates β-arrestin signalling implies that these agonists are biased. Curiously, at a single concentration the phytocannabinoid CBD, but not synthetic analogues O-1602 or O-1918, engage β-arrestin signalling. As a consequence of an intracellular and plasma membrane localization of GPR18, it is possible that GPR18 mediates cellular events at both of these locations via different signalling pathways. Both CB₁ receptors and GPR55 have previously been described as having functional intracellular and plasma membrane locations with unique signalling pathways (Brailoiu et al., 2011; Yu et al., 2013).

Collectively, the results presented herein complement the discoveries that: activation of GPR18 increases MAPK activity, and migration of BV-2, HEK293/GPR18 and endometrial cells (McHugh et al., 2010; 2012), GPR18 mobilizes calcium in the presence of NAGly (Kohno et al., 2006), and NAGly is devoid of β-arrestin activity at GPR18 (Yin et al., 2009). Our findings that O-1918 enhances both GPR18-mediated MAPK activity and calcium mobilization, but not β-arrestin translocation, along with previous reports that O-1918 antagonizes other GPR18-mediated biological events, is suggestive that this ligand is biased. Garnering further support for biased agonism at GPR18 are the findings that NAGly, O-1602 and AbnCBD lead to augmented levels of GPR18-mediated MAPK activity and calcium mobilization, but not β-arrestin signalling. This revelation sheds light on the multiplicity of mechanisms involved in GPR18 signalling, and will ultimately enable identification of additional ligands capable of targeting transduction pathways specific for inhibition or activation of a desired GPR18-mediated biological event. Regardless of whether NAGly is the principal endogenous ligand for GPR18, investigations by several independent laboratories, including ours, strengthen the evidence for activation of GPR18 by NAGly and other cannabinoid ligands. Importantly, activation of GPR18 by Δ⁹-THC reinforces the classification of GPR18 as a cannabinoid receptor.

Author contributions

L.C.B. designed and performed experiments, analysed data and wrote the paper; E.B. designed and performed experiments, and analysed data; G.C.B. implemented experiments; H.S. designed experiments and contributed to the writing of the paper; and M.E.A. designed experiments and wrote the paper.

Conflict of interest

None.