The Cannabinoid CB₂ Receptor Positive Allosteric Modulator EC21a Exhibits Complicated Pharmacology In Vitro

Abstract

Schizophrenia is a complex disease involving the dysregulation of numerous brain circuits and patients exhibit positive symptoms (hallucinations, delusions), negative symptoms (anhedonia), and cognitive impairments. We have shown that the antipsychotic efficacy of positive allosteric modulators (PAMs) of both the M₄ muscarinic receptor and metabotropic glutamate receptor 1 (mGlu₁) involve the retrograde activation of the presynaptic cannabinoid type-2 (CB₂) receptor, indicating that CB₂ activation or potentiation could result in a novel therapeutic strategy for schizophrenia. We used two complementary assays, receptor-mediated phosphoinositide hydrolysis and GIRK channel activation, to characterize a CB₂ PAM scaffold, represented by the compound EC21a, to explore its potential as a starting point to optimize therapeutics for schizophrenia. These studies revealed that EC21a acts as an allosteric inverse agonist at CB₂ in both assays and exhibits a mixed allosteric agonist/negative allosteric modulator profile at CB₁ depending upon the assay used for profiling. A series of compounds related to EC21a also functioned as CB₂ inverse agonists. Overall, these results suggest that EC21a exhibits complicated and potentially assay-dependent pharmacology, which may impact interpretation of in vivo studies.

Graphical Abstract

Introduction

The endocannabinoid system is composed of two main receptor subtypes, cannabinoid type-1 (CB₁) and type-2 receptors (CB₂), which belong to the Class A family of G Protein-Coupled Receptors (GPCRs; [1,2]). While CB₁ is the most widely expressed GPCR in the brain [3], CB₂ exhibits high expression in the immune system [4–6] and CB₂ receptors are important in immune responses and neuroinflammation via modulation of microglia [7–10]. CB₂ receptors are also expressed in neuronal circuits and evidence suggests therapeutic potential for CB₂ modulation in the treatment of several brain disorders including Alzheimer’s disease, pain and epilepsy (reviewed in [11]). Recently, we have found that cholinergic signaling regulates dopamine release in the striatum via a CB₂-dependent mechanism that involves the activation of M₄ muscarinic receptors and metabotropic glutamate receptor 1 (mGlu₁) [12,13]. Potentiation of M₄ or mGlu₁ with highly selective positive allosteric modulators (PAMs) induces antipsychotic-like activity; a mechanism key to these effects involves the retrograde signaling of the endocannabinoid, 2-AG, which acts on CB₂ receptors to decrease local dopamine release [11–13].

While there have been a number of agonists of CB₂ that have been reported in clinical development [14,15], modulation of receptor activity can also be achieved using ligands that bind to allosteric sites and either enhance or inhibit activity of endogenous agonists. Allosteric ligands represent a unique approach, particularly for CNS indications, as they have the ability to more subtly modulate neurotransmission, maintain spatial and temporal effects of endogenous neurotransmitters, and also provide enhanced selectivity for specific receptor isoforms [16]. In the case of the CB receptors, this point is paramount, as activation of the CB₁ receptor results in psychotomimetic effects [17,18].

To search for and profile new CB₂-selective PAMs, we are employing two orthogonal assays for compound evaluation: PI hydrolysis pathway via a chimeric G protein, and thallium flux induced by CB_1- and CB₂-mediated opening of G protein-coupled inwardly rectifying potassium channels (GIRKs). Using these assays, we evaluated the profile of the recently described CB₂-selective PAM, EC21a [19,20], anticipating moving the compound into rodent models of psychosis. Until the development of EC21a, the only compound with reported CB₂ PAM activity was a peptide termed pepcan-12, which also acts as a negative allosteric modulator (NAM) of CB₁ [21]. EC21a was originally reported as a selective CB₂ PAM based on in vitro pharmacology studies with [³⁵S]GTPγS binding and [³H]CP55940 competition binding assays (Gado et al, 2019). In these studies, EC21a enhanced [³⁵S]GTPγS binding induced by the agonist CP55940 and increased binding of [³H]CP55940 at low concentrations; this effect reversed at a 10 μM concentration. Additionally, Gado et al. used kinetic binding to show that EC21a significantly altered the dissociation rate of [³H]CP55940; all of these studies are consistent with an allosteric interaction between the agonist CP55940 and EC21a.

We confirmed here that EC21a only weakly and incompletely displaces the binding of the radioligand [³H]CP55940, suggesting it interacts at a distinct receptor binding site. Functional profiling of EC21a, however, showed that the compound functions as an allosteric agonist of CB₂ and a mixed allosteric agonist/NAM of CB₁ depending upon the assay used for profiling. Synthesis of a range of analogs based upon the EC21a scaffold showed that all of them blocked baseline and agonist-mediated responses at CB₂. Overall, these results suggest that the EC21a series of compounds exhibits complicated pharmacology at the cannabinoid receptors.

Materials and Methods

Drugs

JWH133, 2-Arachidonylglycerol (2-AG), N-(2-Chloroethyl)-5Z,8Z,11Z,14Z-eicosatetraenamide (arachidonyl 2’-chloroethylamide, ACEA), AM251, AM630, and CP55940 were purchased from Tocris Bioscience (Minneapolis, MN). [³H]-CP55940 was obtained from Perkin Elmer (Boston, MA). EC21a (VU6043295) and structural analogs were synthesized by the Warren Center for Neuroscience Drug Discovery, Vanderbilt University, Nashville, USA (Supplemental Methods).

Cell Lines and Cell Culture

Human Embryonic Kidney (HEK) 293 cells stably expressing G_qi9 (Gladstone Institute) were maintained in Dulbecco’s modified Eagle media (DMEM, Invitrogen. Waltham MA) containing 10% FBS, 100 units/mL penicillin/streptomycin, 20 mM HEPES, 1 mM sodium pyruvate, 2 mM L-glutamine, 1x non-essential amino acids, 700 μg/mL G418 sulfate. HEK cells stably expressing G protein inwardly rectifying potassium channels (GIRK1 and GIRK 2 subunits, HEK/GIRK [22]) were maintained in DMEM/F12 containing 10% FBS, 100 units/mL penicillin/streptomycin, 20 mM HEPES, 1 mM sodium pyruvate, 2 mM L-glutamine, 1x non-essential amino acids, and 700 μg/mL G418 sulfate. Cells were monitored by mycoplasma testing every two to three months.

Construction of human and rat CB₁ and CB₂ in pIRESneo3

The coding sequences of the rat CB₁ (ENSRNOG00000008223), rat CB₂ (ENSRNOG00000009260), human CB₁ (ENSG00000118432), human CB₂ (ENSG00000188822) receptors were cloned into the pIRESpuro3 expression vector (Takara Bio USA, San Jose, CA) using AgeI and EcoRI restriction sites. All cDNA sequences were confirmed by sequencing at Azenta Life Sciences (Chelmsford, MA).

Expression of human and rat CB₁ and CB₂ in HEK293 cells

Transient transfections were performed with ViaFect^™ Transfection Reagent (Promega, Madison, WI). Briefly, 5 X 10⁵ HEK293 cells stably expressing G_qi9 (HEK/G_qi9) or GIRK (HEK/GIRK) per well in 6-well dish grown overnight at 37 °C. Cells were incubated with 4.5 μL of ViaFect^™ Transfection Reagent and 1.5 μg pIRESpuro3-based plasmid construct in Opti-MEM I Reduced Serum Medium. Cells were selected with DMEM growth medium containing 1 μg/mL of puromycin.

Assay of Inositol Phosphate Accumulation by chromatography

Data in Figures 2B and 3A were generated using this assay. Selected HEK/G_qi9 cells stably expressing CB₁ or CB₂ were seeded in 24-well plates at 1.5 X 10⁵ cells/well and assayed after 48 hours. Inositol lipids were radiolabeled by overnight incubation of cells with 200 μL inositol-free DMEM containing 0.4 μCi myo-[³H]inositol (specific activity = 20 Ci/mmol, PerkinElmer) and 5% dialyzed FBS. Agonists were added at 5 x concentration in 50 μL of 50 mM LiCl and 250 mM HEPES, pH 7.25. Following a 30-minute incubation at 37 °C, the medium was aspirated, and the reaction was terminated by adding 0.75 mL boiling EDTA, pH 8.0. [³H]Inositol phosphates were resolved on Dowex AG1-X8 (Sigma-Aldrich) columns and counted by scintillation counting as described previously [23].

Figure 2. EC21a only weakly displaces [³H]CP55940 and blocks basal signaling in both the rat CB₂ PI hydrolysis and GIRK assays.

A. Increasing concentrations of EC21a or AM630 were added to membranes and competition with [³H]CP99540 was performed. Data are representative of three experiments performed in triplicate or duplicate (pIC₅₀ values = 6.08+/− 0.86 for EC21a, and 7.14 +/−0.12 for AM630, respectively). B. Increasing concentrations of JWH133 alone (gray) or 10 μM EC21a plus JWH133 (red) were applied to CB₂/HEK/G_qi9 cells and PI hydrolysis responses were measured ([³H]inositol radiolabeled chromatography assays). EC21a decreased the basal level of inositol phosphates and shifted the potency of JWH133 to the right (pEC₅₀ value = 7.54 ± 0.16 for JWH133 alone, > 5μM for JWH133 + EC21a,). When applied alone (black), EC21a induced a concentration-dependent decrease in PI hydrolysis (pEC₅₀ value = 6.66 ± 0.23). C. When assessed in the rat CB₂/HEK/GIRK assay, EC21a blocked the response to JWH133 (red versus gray). As a control, the known inverse agonist AM630 was also applied with JWH133 and similarly blocked the response. D. When assessed alone in the rat CB₂/HEK/GIRK thallium flux assay, EC21a decreased basal thallium flux levels concentration-dependently (pEC₅₀ of 7.00 ± 0.11). Data are the Mean ± SEM of three independent experiments performed in triplicate.

Figure 3. EC21a exhibits agonist activity at rat CB₁ in PI hydrolysis and negative allosteric modulator activity at rat CB₁ in thallium flux.

A. Increasing concentrations of ACEA were applied in either the absence (gray, pEC50 = 5.28 ± 0.64) or presence (black, pEC50 = > 5 μM; note that maximal responses were not achieved in this assay) of 10 μM EC21a; additionally, increasing concentrations of EC21a were applied to rat CB₁/HEK/G_qi9 cells alone (white, pEC₅₀ = 5.77 ± 0.36, [³H]inositol radiolabeled chromatography assays). B. ACEA-induced thallium flux responses in rat CB₁/HEK/GIRK cells were blocked by increasing concentrations of EC21a (Emax = 100 ± 2 and 22 ± 2 in the absence and presence 10 μM EC21a, p< 0.0001). Data are the Mean ± SEM of three independent experiments performed in triplicate. Increasing concentrations of EC21a did not significantly shift the ACEA potencies in the thallium flux assay (B, pEC₅₀ = 7.67 ± 0.05, DMSO versus 6.93 ± 0.28, ACEA + 10 μM EC21a, p = 0.06).

Assay of Inositol Phosphate Accumulation by IP-ONE kit

Due to a lack of routine availability of some reagents for radiolabeling chromatography assays, IP-ONE kits were used in some PI hydrolysis experiments; data in Table 1 were generated using this assay. This method produced comparable data to chromatography (data not shown). One day prior to experimentation, selected HEK/G_qi9 monoclonal cells stably expressing rat CB₁ or rat CB₂ were plated onto poly-D-lysine coated clear-bottom 384-well plates (20,000 cells per well) in DMEM supplemented with 10% FBS and 20 mM HEPES. Ten minutes prior to the assay, cell culture medium was replaced with 20 μL 37 °C Hank’s balanced salt solution (HBSS, with Ca²⁺, Mg²⁺). IP₁ stimulation was then initiated by adding 5 μL of 5x agonists dissolved in HBSS plus 50 mM Li⁺, and cells were incubated for an additional hour before aspiration and addition of Lysis Buffer. IP₁ levels were determined using the Cisbio HTRF IP-ONE assay kit per the manufacturer’s instructions, and fluorescence was measured using an Envision plate reader (PerkinElmer). Data were acquired as HTRF ratio (665/620) and expressed as nanomolar levels of IP₁.

Table 1.

Agonist potencies (pEC₅₀) in HEK/G_qi9 cells expressing cannabinoid receptors.

	Rat CB1 pEC50 ± SEM	Rat CB2 pEC50 ± SEM
2-AG	6.07 ± 0.09	5.69 ± 0.01
ACEA	6.53 ± 0.05	NA
JWH133	NA	7.64 ± 0.06

Agonist-promoted PI hydrolysis was measured in HEK/G_qi9 cells expressing rat CB₁ or CB₂ receptors. Data are Mean ± SEM of three to five independent experiments performed in triplicate using IP-One assays. NA=not assessed.

Thallium Flux Assay

Thallium flux assays were performed as previously described (Niswender et al., 2008). HEK/GIRK cells stably expressing cannabinoid receptors (15,000 cells/20 μL/well) were seeded in 384-well, amine-coated assay plates (Greiner BioOne, Monroe, NC) and incubated overnight at 5% CO₂ cell and 37 °C. Media was replaced with 20 μL/well of a dye loading solution containing assay buffer (Hanks Balanced Salt Solution plus 20 mM HEPES, pH 7.3), and 1.2 μM solution of the thallium-sensitive dye Thallos-AM (ION biosciences, San Marcos, TX), prepared as a DMSO stock and mixed first in a 1:1 ratio with (w/v) Pluronic F-127 (Sigma-Aldrich, St. Louis, MO). Following 1 hour at RT, dye solution was replaced with 20 μL/well assay buffer and the plates were loaded into a Hamamatsu FDSS 7000 (Bridgewater, NJ). Data were acquired at 1 Hz (excitation 470 ± 20 nm, emission 540 ± 30 nm) for ten seconds, followed by the addition of 20 μL/well of compound or DMSO (to test for PAM activity), followed by an additional 4 minutes of data collection. 140 seconds after compound addition, 10 μL/well of a thallium stimulus buffer (125 mM NaHCO₃, 1.8 mM CaSO₄, 1 mM MgSO₄, 5 mM glucose, 12 mM Tl₂SO₄, 10 mM HEPES pH 7.4), along with serial dilution of agonist (5X), was added and data collection continued for an additional 2 minutes.

Raw kinetic data were analyzed in a multi-step process (see Supplemental Figure 1). 1) Fluorescence readings for each time point in a well were divided by the fluorescence reading at the initial time point to account for differences in cell number, non-uniform illumination, and dye-loading. 2) The slope value for each kinetic trace was calculated for the time window of 145–155 seconds, a window occurring directly after addition of agonist. 3) The average slope was calculated for wells containing vehicle and this value was subtracted from all wells. 4) Vehicle-subtracted slope was normalized to the relevant maximal agonist signal for each assay. Normalized data were fit to a four-parameter logistic equation using GraphPad Prism. Data shown represent the mean ± standard error of the pEC₅₀ or maximal response. Experiments were performed in triplicate and repeated three to four separate times. To assess effects on inverse agonism, slope values were calculated after compound addition alone.

Radioligand Binding Assays

Membranes were made from HEK/GIRK cells stably expressing rat CB₂. Radioligand competition binding assays were performed as previously described [24] with minor modifications. In brief, compounds were serially diluted into assay buffer with 0.1% bovine serum albumin (BSA) and added to each well of a 96-well plate, along with 20 μg/well cell membrane and approximately 500 pM [³H]-CP55940 (specific activity = 104 Ci/mmol, Perkin Elmer, Boston, MA). Following a 3-hour incubation period on shaker at room temperature, the membrane-bound ligand was separated from free ligand by filtration through glass fiber 96-well filter plates (Unifilter-96, GF/B; PerkinElmer, Boston, MA). Forty microliters of scintillation fluid were added to each well, and the membrane-bound radioactivity was determined by scintillation counting (Microbeta2; Revvity, Hopkinton, MA). Nonspecific binding was determined using 10 μM of cold CP55940.

Chemistry

Compounds were prepared using previously published procedures in the case of compound EC21a, compound 1 [19], and compound 9 [25]. Several closely related analogs, compounds 5–8 and 10, were prepared by similar procedures reported in [19] (see supplemental information for chemical synthesis).

Results

The compound EC21a acts as an inverse agonist of CB₂ receptors in both the PI hydrolysis and thallium flux assays.

To explore the pharmacology of the CB receptors, we employed two assays, phosphoinositide hydrolysis in the presence of the chimeric G protein, G_qi9, and thallium flux through G protein inwardly rectifying potassium (GIRK) channels [22]. These assays revealed the expected pharmacology for a range of receptor ligands for both the rat and human receptors (Tables 1 and 2), and agonist activities were blocked by the antagonists AM251 (for CB₁) and AM630 (for CB₂).

Table 2.

Agonist potencies (pEC₅₀) in HEK/GIRK cells expressing cannabinoid receptors.

	Rat CB1		Rat CB2		Human CB1		Human CB2
	pEC50	Emax (%)	pEC50	Emax (%)	pEC50	Emax (%)	pEC50	Emax (%)
2-AG	7.77±0.03	91±9	7.31±0.14	130±12	7.18±0.09	110±2	6.84±0.08	117±9
ACEA	7.70±0.07	97±2	6.60±0.12	98±21	7.22±0.04	98±1	5.77±0.21	76±12
JWH133	5.70±0.17	47±10	6.27±0.09	103±2	5.91±0.18	23±4	6.56±0.08	104±3

Agonist-promoted thallium influx signals were assayed in HEK/GIRK cells expressing the rat CB₁, rat CB₂, human CB₁, and human CB₂ receptors. Data are Mean ± SEM of three to five independent experiments performed in triplicate. The maximum efficacies (E_max) were normalized to ACEA (for CB₁) and JWH 133 (for CB₂) respectively. Data are Mean ± SEM of three independent experiments performed in triplicate.

We next evaluated the activity of the newly reported CB₂ PAM, EC21a (Figure 1, (Gado et al, 2019; 2021)). We performed binding with [³H]CP55940 and observed that AM630 almost completely displaced the binding of [³H]CP55940 (84.9 ± 6.9%); in contrast, EC21a only exhibited partial displacement (37.6 ± 4.9%), consistent with compound binding to an alternate site compared to AM630 and CP55940 (pIC₅₀ = 6.08 ± 0.86 for EC21a, and 7.14 ± 0.12 for AM630, respectively) (Figure 2A). We next evaluated EC21a functional activity in the PI hydrolysis and thallium flux assays (Figures 2B–D). Surprisingly, EC21a did not potentiate agonist responses in either assay. As shown in Figure 2B, addition of a 10 μM concentration of EC21a to rCB₂/HEK/G_qi9 cells in the presence of the agonist JWH133 decreased basal levels of inositol phosphates and shifted the concentration-response of JWH133 to the right compared to the response of JWH133 alone (gray, pEC₅₀ JWH133 alone = 7.54 ± 0.16; red, pEC₅₀ JWH133 + EC21a = EC50 > 5μM). Application of increasing concentrations of EC21a alone resulted in a concentration-dependent decrease in basal PI hydrolysis with a pEC₅₀ of 6.66 ± 0.23 (black, n=3, Figure 2B), consistent with blockade of constitutive activity/allosteric inverse agonist activity. In the rCB₂/HEK/GIRK assay (Figure 2C), EC21a again significantly decreased basal responses (30.0 ± 5.9%, n=3 experiments) and shifted the concentration-response of JWH133 to the right with a nearly full blockade of the maximal signal (red versus gray). This response was similar to the response observed when the known inverse agonist AM630 was applied with JWH133 (black, Figure 2C). Application of increasing concentrations of EC21a alone to rCB₂/HEK/GIRK cells resulted in a concentration-dependent decrease in basal thallium flux with a pEC₅₀ of 7.00 ± 0.11 (Figure 2D). To eliminate the possibility that EC21a directly interacts with GIRK channels, increasing concentrations of EC21a were applied prior to the addition of the group III metabotropic glutamate receptor agonist L-AP4 to cells expressing the human mGlu₇ receptor in the HEK/GIRK background (Niswender et al, 2008). As shown in Supplemental Figure 2, EC21a had no effect on thallium flux responses promoted by the group III metabotropic glutamate receptor agonist L-AP4 in hmGlu₇/HEK/GIRK cells, indicating that effects of E21a are not due to non-specific effects.

Figure 1.

Structure of EC21a.

EC21a was originally reported as a selective CB₂ receptor PAM [19]. As shown in Figure 3A, EC21a induced IP accumulation in rCB₁/HEK/G_qi9 cells when applied alone with a pEC₅₀ of 5.77 ± 0.36 (white). When 10 μM EC21a was applied with a concentration-response of the agonist ACEA, EC21a raised the baseline PI hydrolysis response without an apparent shift in the pEC₅₀ of ACEA (Figure 3A, black), which is consistent with distinct binding sites for EC21a and ACEA. We would note, however, that we did not achieve maximal PI hydrolysis responses in this assay. In the rCB₁/HEK/GIRK assay, EC21a significantly and concentration-dependently decreased the maximal response to ACEA (E_max = 100 ± 2, DMSO versus 22 ± 2, ACEA + 10μM EC21a; p<0.0001) in a concentration-dependent manner without significantly shifting ACEA potency (pEC₅₀ = 7.67 ± 0.05, DMSO versus 6.93 ± 0.28, ACEA + 10 μM EC21a, p = 0.06, Figure 3B), consistent with the profile of a negative allosteric modulator (NAM). These findings suggest that EC21a is a mixed agonist/NAM of CB₁ in these assays and its activity is consistent with activity via an allosteric site.

EC21a analogs all inhibit, rather than potentiate, thallium flux in rCB₂/HEK/GIRK cells.

Within scaffolds of allosteric modulators, subtle chemical modifications can induce dramatic shifts in pharmacology, from positive to negative to silent allosteric modulators (e.g., [26–28]). We synthesized a variety of compounds based on the EC21a scaffold (Figure 4) and tested them in single point format in either the absence or the presence of a 10 μM concentration of the CB₂ agonist JWH133. As shown in Figure 5, all of these compounds decreased both basal and JWH133-promoted thallium flux responses, suggesting that, similar to EC21a, these analogs block constitutive receptor activity; based on structural similarity with EC21a, we would anticipate that these effects are mediated by allosteric binding sites.

Figure 4.

Structures of EC21a analogs

Figure 5. EC21a analogs block agonist-induced and constitutive thallium flux in the rat CB₂/HEK/GIRK assay.

10 μM of each EC21a analog was applied prior to vehicle (white) or 10 μM JWH133 (gray). Similar to EC21a itself, none of these compounds induced potentiation; rather, they all decreased basal levels of thallium flux and blocked JWH133 responses. Data are the Mean ± SEM of three independent experiments performed in triplicate.

Discussion

The CB₁ and CB₂ receptors have been studied for years as the targets of cannabis and both receptors have been investigated for their therapeutic potential in neurological and psychiatric diseases. While CB₁ is thought to mediate psychoactive effects, CB₂ is an intriguing target that is highly expressed in cells of the immune system and microglia [4–6], as well as in neuronal populations in hippocampal, cortical, and dopaminergic circuits [7–10,29–36].

Recently, CB₂ has been genetically linked to schizophrenia; for example, reduced CB₂ receptor expression and/or activity is correlated with an increased risk for the manifestation of schizophrenia [37–40]. We have recently shown that striatal cholinergic signaling regulates dopamine release via a CB₂-dependent mechanism [12]. Potentiation of the M₄ muscarinic receptor or mGlu₁ with highly selective PAMs induces antipsychotic-like activity in numerous rodent behavioral assays [12,41–43]. One important process by which M₄ and mGlu₁ PAMs mediate antipsychotic-like efficacy is by reducing striatal extracellular dopamine [13,44], and M₄ and mGlu₁-mediated effects on striatal dopamine release and antipsychotic efficacy are blocked by the CB₂-selective antagonist AM630 and lost in CB₂ knockout animals, indicating that CB₂ receptors are required for the efficacy of these compounds [12,13]. CB₂ activation or potentiation may represent a potential alternate, downstream strategy to M₄ or mGlu₁ potentiation that engages similar circuity but may have additional advantages. For example, CB₂ activation has been shown to be cardioprotective [45] and anti-inflammatory by suppressing microglia activation and the release of inflammatory cytokines (reviewed in [46–48]). Here, we have focused on profiling a PAM of CB₂ as PAMs are predicted to avoid receptor desensitization that may occur with direct agonists and maintain spatial and temporal signaling of endogenous neurotransmitters’ plasticity [16,49–51].

Our work provides further evidence demonstrating that CB receptor signaling pathways are complex and diverse. For example, the CB₁ receptor has been shown to couple to G_i/o, G_s, β-arrestin, phospholipase C and inositol phosphate pathways, p38 and p42 Mitogen Activated Proteins Kinases (MAPKs), Phosphoinositide 3 Kinase (PI3K), potassium channels (including GIRKs), calcium channels (including N-, L-, and P/Q-type), and β-arrestins [52–56]. In addition, CB signaling pathways are often biased depending on the cellular context and ligands themselves [53]. Here, we used two assays to interrogate the activity of ligands at CB₁ and CB₂ receptors: PI hydrolysis and thallium flux mediated by the βγ subunit stimulation of the GIRK channels. The latter assay complements studies in neurons which endogenously express the GIRKs or oocytes in which the channel is co-expressed with cannabinoid receptors showing that activation of the CB receptors induces GIRK channel opening [53,57–60].

In this study, we examined the activity of EC21a, a recently reported PAM of CB₂ [19,61], as we anticipated advancement of the compound to in vivo studies to evaluate the therapeutic potential of CB₂ PAMs in rodent schizophrenia models. EC21a was originally described in a GTPγS assay as a PAM in the presence of the agonists CP55940 or 2-AG, but not anandamide, an endogenous cannabinoid [19]. It was also proposed to bind to the receptor at an allosteric site based on increased binding of the ligand [³H]CP55940 at low concentrations and displacement at only high concentrations (~10 μM [19]). We confirmed partial displacement of [³H]CP55940 at high concentrations of EC21a (Figure 2A). Functionally, however, EC21a behaved as an inverse agonist, presumably via an allosteric site based on its binding profile [19], of the CB₂ receptor in both PI hydrolysis and thallium flux assays. In addition, despite its reported selectivity for the CB₂ receptor, EC21a displayed agonist activity in PI hydrolysis and negative allosteric modulation in thallium flux at the CB₁ receptor; all analogs of EC21a also decreased GIRK channel activity when applied alone to CB₂/GIRK/HEK cells. Overall, our findings suggest that EC21a exhibits complicated pharmacology in vitro at both CB receptor isoforms.

EC21a has been shown to exhibit efficacy in a neuropathic pain model, and effects were lost when the animals were pretreated with a CB₂ antagonist but not a CB₁ antagonist [19]. EC21a was also shown to potentiate the activity of a CB₂ agonist, B2, in an in vitro model of lipopolysaccharide (LPS)-induced cytokine release and the effect was also blocked by a CB₂ antagonist [62]. EC21a had no efficacy alone in this model. EC21a was also shown to be effective in reducing seizures and these effects were blocked by the CB₂ antagonist AM630 [63]; additionally, in a genetic model of seizures (R1648H SCN1A mice), EC21a was also effective. This efficacy was not observed with AM630 [63]. These findings suggest that the activity of EC21a in vivo is distinct from AM630 and suggests that additional systematic studies are warranted to understand the signaling pathways that underlie the in vivo efficacy of EC21a, AM630, and CB₂ modulators.

Data Availability Statement

The authors declare that all processed data supporting the findings of this study are available within the paper and its Supplemental Data. Raw data are available on request from the corresponding author.