Cells in culture

HEK293 wildtype (ATCC, Manasses, VA, USA) and HEK293 cells stably transfected with GPR18 [HEK293-GPR18; generated as previously described (McHugh et al., 2010)], CB₁ (HEK293-CB₁; a kind gift from Dr Ken Mackie, Indiana University) and CB₂ (HEK293-CB₂; a kind gift from Dr Ken Mackie, Indiana University) were grown in high-glucose DMEM (Gibco, Carlsbad, CA, USA) supplemented with FBS (10%), penicillin (100 units·mL⁻¹), streptomycin (100 µg·mL⁻¹), and passaged every 4–5 days for a maximum of 30 passages. Twenty-four hours prior to experimentation, the culture media for the HEK293 wildtype, HEK293-GPR18, HEK293-CB₁ and HEK293-CB₂ cells was replaced by serum-free high-glucose DMEM (Gibco) supplemented with penicillin (100 units·mL⁻¹),and streptomycin (100 µg·mL⁻¹). Similarly, the human endometrial cell line HEC-1B (ATCC) was grown in high-glucose DMEM (Gibco) supplemented with FBS (10%), penicillin (100 units·mL⁻¹), streptomycin (100 µg·mL⁻¹), and passaged every 4–5 days for a maximum of 30 passages. Twenty-four hours prior to experimentation, the culture media for the HEC-1B cells was replaced by phenol red-free and serum-free high-glucose DMEM (Gibco) supplemented with penicillin (100 units·mL⁻¹) and streptomycin (100 µg·mL⁻¹).

Migration assay

In vitro cell migration assays were performed using a modified 96-well Boyden Chamber and PVP-free polycarbonate filters with 10 µm diameter pores (Neuroprobe Inc., Gaithersburg, MD, USA), which are shown as the clear unstained circles in the photographed filters (Figure 1B). The upper wells of the Boyden chamber were filled with 50 µL of suspension of 1 × 10⁶ cells·mL⁻¹ in serum-free DMEM, before incubation with a 5% CO₂ atmosphere at 37°C for 3 h. Following incubation, non-migrated cells were then removed before fixation and staining with Diff Quik® stain set. Finally, the filter was sectioned and mounted onto microscope slides and the migrated cells counted in 10 non-overlapping fields (×40 magnification) with a light microscope by several workers, without knowledge of the treatments. For inhibition of induced migration, cells were pre-incubated with antagonist for 30 min at 37°C in a water bath before loading into the upper wells; the lower wells contained the equivalent concentration of antagonist and test compound to ensure that the only concentration gradient present is that generated by the test compound.

Mass spectrometric analysis of deuterium-labelled AEA metabolism

Recovery of deuterium-labelled AEA (d₄AEA) and NAGly (d₀NAGly and d₂NAGly) was measured after incubation with HEC-1B cells at four time points (5, 15, 90 and 180 min), using HPLC/MS/MS, as previously described byBradshaw et al., (2009).

Isolation of total RNA and PCR for GPR18

RNA was extracted from HEC-1B cells using the RNAqueous® small-scale phenol-free total RNA isolation kit (Applied Biosystems, Carlsbad, CA, USA) and RNA samples (2 µg) were reverse transcribed using the SuperScript II™ Reverse Transcription Kit (Invitrogen, Carlsbad, CA, USA). Expression of GPR18 in HEC-1B was determined by PCR using oligonucleotide primers based on the sequence of the Mus musculus GPR18 mRNA (GenBank Accession No.NM_182806.1). The primer sequences used were forward, TGAAGCCCAAGGTCAAGGAGAAGT and reverse, TTCATGAGGAAGGTGGTGAAGGCT (amplicon 163 bp) for GPR18. PCR reactions were subjected to an initial HotStart Taq DNA polymerase activation step of 95°C for 7 min, followed by 40 cycles of 94°C for 30 s, 55°C for 30 s and 72°C for 30 s. PCR products were analysed on a 2% agarose gel with ethidium bromide. Single bands corresponding to 163 bp for the GPR18 amplicon were recorded.

In-Cell Western assay to quantify MAPK phosphorylation

An In-Cell Western assay was employed to simultaneously detect both the phosphorylated MAPK protein and normalize for total MAPK protein. The following primary antibodies were used to detect endogenous levels of the relevant total MAPK and phosphorylated MAPK: p44/42 MAPK rabbit pAb, and phospho-p44/42 MAPK mouse mAb (#4695 and #9106; Cell Signaling Technology). Cells were plated into 96-well plates coated with 1 µg·mL⁻¹ poly-D-lysine and treated with vehicle (Vh) (0.1% DMSO) or test compound (0.1 nM–100 µM) for 5 min. Ionomycin (10 µM) treatment was used as a positive control. Upon completion of the drug treatments, an In-Cell Western assay was conducted as previously described (McHugh et al., 2010).

Analysis of data

All data are expressed as means ± SEM and n = number of independent experiments. For HEC-1B endometrial cells, the mean number of cells migrated in response to test compounds was normalized against the mean number of migrated cells elicited by 1 nM oestradiol (in 0.1% DMSO), corrected for the number of cells migrating with vehicle (0.1% DMSO) was subtracted. For HEK293 wildtype, HEK293-CB₁, HEK293-CB₂, and HEK293-GPR18 transfected cell MAP kinase activation assays, the test compound responses were normalized against the mean effect elicited by 10 µM of the calcium ionophore, ionomycin. Concentration-response curves were generated using a sigmoidal dose-response (variable slope) curve-fitting process. The apparent K_B value of CBD for antagonism of Δ⁹-THC was calculated by substituting a single concentration ratio value into the equation: [x − 1]= B/K_B, where x (the ‘concentration ratio’) is the EC₅₀ value of Δ⁹-THC in the presence of CBD at a concentration, B, divided by the EC₅₀ value of Δ⁹-THC alone (Tallarida and Jacob, 1979). Statistical analyses were performed with GraphPad Prism 4; using one-way ANOVA with Dunnett’s post hoc tests; P < 0.05 was considered to show significance.

Materials

High-glucose Dulbecco’s modified Eagle’s medium (DMEM), high-glucose, phenol red-free DMEM, penicillin, streptomycin, dimethyl sulphoxide (DMSO), poly-D-lysine, ionomycin, DPX mountant Tween-20, Triton X-100, phosphate buffered saline, 17β-oestradiol from Sigma-Aldrich (St. Louis, MO, USA); fetal bovine serum (FBS) from Lonza, (Houston, TX, USA); Diff-Quik® from VWR, (Bridgeport, NJ, USA); methanol, 37% formaldehyde from Fisher Scientific, (Pittsburg, PA, USA); Odyssey Blocking buffer, goat anti-mouse 680 nm antibody (926-32220), goat anti-rabbit 800 nm antibody (926-32211) from Li-Cor (Lincoln, NE, USA); p44/42 MAPK rabbit polyclonal antibody (#4695) and phospho-p44/42 MAPK mouse monoclonal antibody (#9106) from Cell Signaling Technology (Danvers, MA, USA); NAGly, AEA, N-(cyclopropyl)-eicosatetraenamide (ACPA), WIN 55212-2, CP55,940 and AM251 from Enzo Life Sciences (Farmingdale, NY, USA); Abn-CBD, O-1602, SR141716A (rimonabant) SR144528, (R)-arachidonyl-1′-hydroxy-2′-propylamide (R1meth-AEA) and the deuterium labelled AEA and NAGly used as standards were supplied by Cayman Chemicals (Ann Arbor, MI, USA). JWH-133 and JWH-015 were kind gifts from Dr John Huffman, (Clemson University). Δ⁹-THC and CBD were from the National Institute on Drug Abuse.

HEC-1B cell migration

We compared oestradiol-induced HEC-1B migration with NAGly and AEA over a concentration range of 0.1 nM to 10 µM. All three signalling lipids produced response curves with a bell-shaped pattern of induced migration beginning at sub-nanomolar concentrations, peaking in the low nanomolar range and then showing a decreased response trend towards high nanomolar/micromolar concentrations (Figure 1A). NAGly and AEA exhibited greater efficacy than oestradiol for inducing HEC-1B migration (Figure 1A,B).

The role of cannabinoid receptors was examined using SR141716A (rimonabant; CB₁ receptor antagonist/inverse agonist), SR144528 (CB₂receptor antagonist/inverse agonist) and CBD (GPR18 receptor weak partial agonist/antagonist). HEC-1B cell migration to SR141716A (0.1 nM–10 µM) alone, SR144528 (0.1 nM–10 µM) alone, and CBD (0.1 nM–10 µM) alone were not different from that to vehicle (0.1% DMSO) alone. SR141716A (100 nM) had no effect on oestradiol (100 nM), AEA (100 nM) or NAGly (100 nM). SR144528 (100 nM) significantly attenuated the response to AEA (100 nM) but not oestradiol (100 nM) or NAGly (100 nM). Similar to SR144528, CBD (1 µM) was able to significantly attenuate AEA (100 nM), but in addition, it completely blocked NAGly (100 nM) -induced migration of HEC-1B cells, without effect on oestradiol-induced migration (Figure 1C). PTX (1 µg·mL⁻¹) pre-treatment abolished the migration to oestradiol, AEA and NAGly, without affecting cell viability (Trypan blue exclusion; data not shown).

AEA is metabolized into NAGly in HEC-1B cells through two biochemical pathways

In C6 glioma cells, NAGly is a product of AEA metabolism through a FAAH-dependent pathway and an alcohol dehydrogenase-dependent pathway (Bradshaw et al., 2009). As FAAH activity had not been investigated in HEC-1B cells, we first measured the rates of deuterium-labelled AEA (d₄AEA) metabolism using HPLC/MS/MS (Figure 2A). After 1.5 h of incubation with HEC-1B cells, only ~10% of d₄AEA remained (Figure 2B) and HEC-1B cells produced both d₀NAGly and d₂NAGly, indicating that both of the ‘AEA-to-NAGly’ conversion pathways demonstrated in C6 glioma cells were present and functional in HEC-1B cells (Figure 2A,B). These data suggest that the effect of AEA on migration of HEC-1B cells may be, in part, due to its conversion to NAGly, which then activates GPR18.

PCR measurement of GPR18 mRNA

To determine if GPR18 receptors are produced by HEC-1B cells, we performed standard PCR techniques with the previously optimized GPR18 primers, verified in BV-2 microglia and GPR18-transfected HEK293 cells (McHugh et al., 2010). Figure 3 shows the definitive band at 163bp that coincides with the primer product for GPR18 mRNA, demonstrating the presence of GPR18 mRNA in HEC-1B cells.

Evaluation of cannabinoid ligands at GPR18 through MAPK activation assays

We have proposed the working hypothesis that GPR18 is the molecular identity of the Abn-CBD receptor, (McHugh et al., 2010) and showed that NAGly drove MAPK (p44/42; ERK1/2) activation in GPR18-transfected HEK293 cells. Here, we have used the same expression system to screen the specificity of cannabinoid ligands for CB₁ and CB₂ receptors, at GPR18 in HEK293 cells.

In order of potency, NAGly, O-1602, Abn-CBD, Δ⁹-THC, AEA and ACPA activated GPR18 to induce p44/42 MAPK phosphorylation, as full agonists (Figure 4A) and their EC₅₀ values are shown in Table 1. CBD was a weak GPR18 partial agonist, inducing p44/42 MAPK only at concentrations >10 µM (Figure 4B; EC₅₀ value in Table 1).

WIN55212-2, CP55940, R1-methAEA, JWH-133 and JWH-015 had no effect on GPR18 receptors (Figure 4C). In HEK293-CB₁ cells, WIN55212-2, CP55940, R1-methAEA and ACPA-did activate p44/42 MAPK (Figure 4D; EC₅₀ values in Table 1). Similarly, JWH-133 and JWH-033 were both full agonists to activate CB₂ receptor -induced p44/42 MAPK (EC₅₀ values in Table 1; concentration–response curves not shown). Δ⁹-THC had no effect on HEK293 wildtype cells (Figure 5B) nor did any of the other compounds tested (Table S1).

The most surprising of the outcomes centred on the novel pharmacological activity of Δ⁹-THC at GPR18 receptors (Figure 4A). Given that both NAGly and Δ⁹-THC were found to be full agonists at GPR18, and that our HEC-1B cell migration experiments demonstrated NAGly’s effects were blocked by CBD, we next tested the ability of CBD to antagonize Δ⁹-THC activity at GPR18. A concentration of 20 µM was chosen, which was below the observed threshold of GPR18 activation for CBD (Figure 4B, Table 1). CBD (20 µM) caused a significant rightward shift in the concentration–response curve for Δ⁹-THC-induced phosphorylation of p44/42 MAPK (Figure 5A). An apparent K_B value of 57.1 nM was calculated for CBD. It should be noted that CBD might not be acting competitively here. Concentrations of Δ⁹-THC greater than 1 mM would be required to establish the E_max and ascertain this.

The data in Figure 5C demonstrate that AM251 was a very weak GPR18 partial agonist, able to induce p44/42 MAPK only at concentrations above 50 µM (EC₅₀ value in Table 1). AM251 also antagonized Δ⁹-THC (100 nM) and NAGly (100 nM) -induced p42/44 MAPK phosphorylation in HEK293-GPR18 cells with IC₅₀ values of 47.4 nM and 143.8 nM respectively (Figure 5D).

Δ⁹-THC drives HEC-1B migration and is blocked by CBD and AM251

We next sought to replicate these interesting findings with the HEC-1B cell migration assay. CBD and AM251 alone had no effect on HEC-1B cell migration, but Δ⁹-THC exhibited a concentration-dependent bell-shaped pattern of induced migration, peaking at 100 nM, and was less efficacious than NAGly (Figure 6A). CBD and AM251 both antagonized (100 nM) NAGly-induced cell migration with IC₅₀ values of 28 nM and 112 nM respectively (Figure 6B). With regard to Δ⁹-THC-induced migration, the IC₅₀ values for CBD and AM251 were 18 nM and 79 nM respectively. Collectively, these values and the difference in starting efficacy reflect the greater potency of NAGly over Δ⁹-THC as a GPR18 agonist, and CBD over AM251 as a GPR18 antagonist.

Cannabinoids and endometrial migration

Recent work by Gentilini et al. (2010) reported that primary human endometrial stromal cells migrated towards R1-methAEA (10 µM) and that this was blocked by AM251 (10 µM), suggesting a CB₁ receptor-mediated signalling mechanism. This is in contrast to our findings with HEC-1B endometrial epithelial cells, where the data point towards migration occurring through CB₂and GPR18 receptors, and independently of CB₁ receptors. Our HEK293-GPR18 MAPK studies show that while AM251 was a mixed CB₁/GPR18 antagonist, R1-methAEA remained a selective agonist for CB₁ receptors, supporting the conclusions of Gentilini et al. (2010). Stromal cells are part of the support structure of the endometrium, forming thick connective tissue, as distinct from endometrial epithelium, which is parenchymal. Therefore, differences in phenotype are likely to underlie this apparent discrepancy in regulation of migration by the endocannabinoid system. Like the endometrial epithelium, stroma also undergoes many changes to mediate the dynamic proliferation, secretion and regression events associated with the various phases of the menstrual cycle (Chard and Grudzinskas, 1994). However, Tayloret al. (2010) have observed notable dissimilarities in both CB₁ and CB₂receptor immunoreactivity between stroma and epithelium in human endometrial biopsies. Maccarrone et al. (2000) found FAAH localized in endometrial epithelium and this is compatible with our HPLC/MS/MS data, which indicated the functional presence of a potential FAAH-dependent ‘AEA-to-NAGly’ conversion in HEC-1B endometrial epithelial cells.

Differences in the activity curves of AEA and NAGly-induced HEC-1B migration suggest that the increased potency of AEA is due in part to AEA potentially activating CB₂ receptors (as demonstrated by the attenuated response with pre-incubation of SR144528 and no effect of SR141716A) as well as GPR18 through the conversion of AEA to NAGly. In addition, AEA and NAGly each demonstrate the same bell-shape response with a maximization or ‘ceiling effect’ of migration, which does not appear to be additive as neither compound caused migration levels above ~275% of the maximum effect of oestradiol at amounts above 1 µM. Therefore, additional activation by NAGly (probably acting on GPR18) in the presence of AEA (acting on CB₂ receptors) does not increase this ceiling effect. Taken together, these results demonstrate a sophisticated role for CB₁, CB₂ receptors and GPR18 in endometrial signalling.

Novel cannabinoid pharmacology

The most current understanding of the endocannabinoid system holds that the pharmacology of endogenous and phytocannabinoids is complex. For several years, well-documented evidence has supported the existence of additional cannabinoid receptors, other than CB₁ and CB₂ receptors, contributing to the system (Begg et al., 2005; Mackie and Stella, 2006; Brown, 2007; McHugh et al., 2008). Our recent publication was the first to demonstrate that the G_i/o-coupled GPCR, GPR18, is one of these unknown receptors in that AEA and its metabolite NAGly exert potent control of CNS microglia (McHugh et al., 2010). Moreover, in that same publication we proposed the working hypothesis that GPR18 is the molecular identity of the Abn-CBD receptor, the most prominent of the non-CB₁, non-CB₂ receptors. A portion of the data reported showed that NAGly drove MAPK (p44/42; ERK1/2) activation in GPR18-transfected HEK293 cells. Using this same expression system we screened the specificity of cannabinoid ligands including traditional CB₁ and CB₂ receptor agonists and antagonists at GPR18. In order of potency, NAGly, O-1602, Abn-CBD, Δ⁹-THC, AEA and ACPA are full agonists at GPR18; CBD was a weak GPR18 partial agonist. WIN55212-2, CP55940, R1-methAEA, JWH-133 and JWH-015 had no effect. The ACPA data in Figure 4A suggested that there may be a two-site curve in the GPR18 cells, possibly reflecting two populations of GPR18 receptors (coupled and non-coupled) that have different affinity for the agonist. In order to test this, using GraphPad Prism, the fit of the ACPA data was compared between one-site curve versus a two-site curve. The F-test resulted in a P-value of <0.05, indicating that the one-site model is preferred.

It is becomingly increasingly clear that a proper understanding of GPR18 pharmacology and its physiological role is essential firstly, for an adequate understanding of the complexity of the endocannabinoid system. Secondly, to avoid misinterpretation of effects observed with ligands previously regarded as CB₁ or CB₂ receptor selective, which are also pharmacologically active at GPR18. For instance, R1-methAEA is the most potent agonist in the methanandamide series. It is approximately fourfold more potent at CB₁receptors and more resistant to hydrolytic inactivation by FAAH, than AEA (Abadji et al., 1994). These characteristics have made this ligand a convenient substitute for AEA in various tissues. However, now on the basis of our results, caution should be exercised when interpreting the outcome. Our data indicate activity for AEA but not R1-methAEA at GPR18 and that FAAH-dependent hydrolysis of AEA is likely to be an integral biosynthetic step towards the production of NAGly, which is a more potent agonist at GPR18. An appreciation of this will avoid misconstruing effects as solely CB₁ receptor- versus GPR18-mediated, especially given the antagonism of GPR18 by AM251 we also report here. It will also allow for a more nuanced use of R1-methAEA to selectively target CB₁ receptor signalling over that of GPR18 and CB₂receptors. Thirdly, a more complete picture of GPR18 pharmacology will help to generate future hypotheses in cannabinoid research.

There are many reports of AEA and Abn-CBD activity associated with Abn-CBD receptors (also known as the ‘endothelial anandamide receptor’), that are either insensitive to SR141617A antagonism (Mukhopadhyay et al., 2002;Walter et al., 2003; Milman et al., 2005; O’Sullivan et al., 2005; McHugh et al., 2010) or require substantially greater concentrations than those required to block CB₁ receptors (White and Hiley, 1998; Járai et al., 1999; Wagner et al., 1999; McHugh et al., 2008). Our data showed that SR141716A had no effect on migration of HEC-1B cells (Figure 1C).

The lack of consistency for SR141716A antagonism at Abn-CBD receptors may be explained as a product of its affinity for Abn-CBD receptors and the number that need to be blocked in order to observe a reduction in signalling, which will in turn vary according to the degree of receptor reserve in any given tissue. AM251 is structurally very similar to SR141716A. In light of this, we hypothesized that AM251 may be able to antagonize GPR18 receptors, where SR141716A failed to do so. SR141617A and AM251 are both biarylpyrazole antagonists. In AM251, the p-chloro group attached to the phenyl substituent at C-5 of the pyrazole ring of SR141716A is replaced with a p-iodo group – essentially a halogen substitution (Lan et al., 1999). Structure–activity studies where comparable halogen substitutions were made on the vanillyl moieties of the TRPV1 agonists resiniferatoxin and nordihydrocapsaicin, have yielded potent TRPV1 antagonists (Wahl et al., 2001; Appendino et al., 2003). In addition, Liu and Simon (1997) found that the inhibitory potency of TRPV1 antagonists correlated directly with the size of the halogen substituent (I > Br > Cl) and inversely with its electronegativity. Based on these data, and in order to further test our hypothesis that GPR18 is the molecular identity of the Abn-CBD receptor, we predicted that AM251 would be a more potent antagonist at GPR18 than SR141716A. Our results are in support of this. AM251 antagonized NAGly and Δ⁹-THC activation of p44/42 MAPK in HEK293-GPR18 cells with IC₅₀ values of ~144 nM and ~47 nM respectively (Figure 5D). AM251 also antagonized NAGly and Δ⁹-THC-induced migration of HEC-1B cells, where SR141716A had no effect (Figures 1C and 6B,C); AM251’s IC₅₀ values here were ~ 112 nM and ~ 80 nM respectively. Furthermore, the IC₅₀ values for AM251 antagonism of GPR18 (Abn-CBD receptors) are lower than the reported IC₅₀for SR141716A, of ~ 600 nM at Abn-CBD receptors (Jung et al., 1997; Bukoskiet al., 2002).

In conclusion, the discovery of the endocannabinoid system began with the identification of a single GPCR, the CB₁ receptor and the first endogenous ligand, anandamide. However, a very few additional components have accrued over the last two decades to produce a small group of endocannabinoid receptors and ligands. Recent discussions have centred on what it means to be a member of the ‘cannabinoid’ family and whether or not the levels of inclusion and exclusion are beneficial in terms of getting to a better understanding of the science related to how the phytocannabinoids either mimic or interfere with endogenous mammalian signalling systems (see Pertwee et al., 2010). Here, we provide data that the G_i/o-coupled GPCR, GPR18, was activated by Δ⁹-THC and additional data showing that CBD acted as an antagonist at this same receptor. These results demonstrate that greater concentrations of Δ⁹-THC are required to activate GPR18 receptors, than of CBD to produce GPR18 antagonism, a difference that is likely to affect the therapeutic outcome of existing pharmacotherapies that combine both Δ⁹-THC and CBD.

We also provide evidence that an endogenous metabolite of anandamide, NAGly, mimicked the effects of Δ⁹-THC in driving migration of the human endometrial cell line, HEC-1B, which was also blocked by <100 nM of CBD. These data potentially explain the ‘idiosyncratic’ outcomes from some previous studies in the field in which the effects of CB₁ and CB₂ receptors were unable to account for the observed phenomena.

The endocannabinoid system branch of neuroscience research is still evolving; however, it is already clear that it powerfully regulates neuronal, vascular, immune and reproductive functions. An understanding of the expression, function and regulation of the hitherto unidentified cannabinoid receptors such as GPR18, their molecular interactions with endogenous ligands, and how phytocannabinoids play a role with their signalling is vital if we are to comprehensively assess the function of the cannabinoid signalling system in human health and disease.

This work was supported by NIH DA018224.