Genetically Encoded CB₂R-Based Fluorescent Sensor Enables Rapid Screening and Functional Assessment of Cannabinoid Modulators

Abstract

The G-protein-coupled receptor cannabinoid receptor 2 (CB₂R) initiates a key signaling pathway in mammalian physiology and pathophysiology. CB₂R signaling holds significant therapeutic potential in ameliorating many pathologies, particularly in inflammatory conditions, neurodegenerative disorders, fibroproliferative and ocular diseases. CB2 modulators have been studied for their anti-inflammatory and tissue protective effects in preclinical animal models of cardiovascular, gastrointestinal, liver, kidney, lung and neurodegenerative disorders with numerous compounds undergoing clinical evaluation. Existing ligands can be classified as endocannabinoids, cannabinoid-like natural products and synthetic CB₂R ligands. A genetically encoded G-protein-coupled receptor activation-based (GRAB) sensor for CB₁R—GRAB_eCB2.0 was developed recently. This current study extends the sensor’s development to allow for a GPCR activation-based sensor for CB₂R. The sensor, GRAB-CB2, will facilitate the evaluation of pharmacological characteristics and responses of various functionally selective and indiscriminate cannabinoid ligands acting on CB2.

1. Introduction

The endocannabinoid system (ECS) is a complex biological maze of signaling pathways and endogenous molecules that intricately modulate myriad processes [1,2,3]. Two well-characterized endocannabinoids (eCBs), 2-arachidonoylglycerol (2-AG) and arachidonoyl ethanolamine (AEA), tonically activate G-protein-coupled cannabinoid receptors, CB1 and CB2, in discrete ways and influence various physiological processes [1,4,5]. These two 7-transmembrane receptors also bind the exogenous psychoactive component of marijuana, Δ⁹-tetrahydrocannabinol (THC), uniquely. Both CB1 and CB2 receptors play a crucial role in various physiological and pathophysiological processes in the central nervous system and peripheral tissues [6]. Predominantly expressed in the central nervous system (CNS), particularly in brain regions like the basal ganglia, hippocampus, and cerebellum, CB₁R is now known to have significant expression in peripheral tissues such as the lung, liver, kidney reproductive organs, and gastrointestinal tract [7,8]. CB₂R on the other hand is primarily expressed in immune cells and tissues, including the spleen and circulating immune cells, and in the CNS, but to a lesser extent than CB1, mainly in the brainstem and hippocampal CA2/3 pyramidal neurons [9,10,11].

Brain CB₁R mediates most of the psychoactive effects of Δ⁹-tetrahydrocannabinol (THC) [12]. It is involved in regulation of neurotransmitter release and plays a role in pain modulation, memory, and appetite [13]. Because CB2 receptors are present in immune cells, they likely help regulate inflammation, and this anti-inflammatory control can contribute to neuroprotection in certain diseases [10,14,15,16].

Both receptors primarily couple to Gi/o proteins, leading to inhibition of adenylyl cyclase, decreased cellular cAMP levels and activation of the Ras-MEK-ERK pathway [5,17,18]. The stimulation of these GPCRs also leads to the phosphorylation and subsequent activation of several mitogen-activated protein kinases (MAPKs), including p42/p44 MAPK, p38 MAPK, and c-Jun N-terminal kinase. These MAPKs play a crucial role in regulating nuclear transcription factors. Additionally, once the receptor is activated and phosphorylated, it forms complexes with β-arrestin [5,17]. These complexes are instrumental in the processes of receptor desensitization and internalization, which are key mechanisms for regulating receptor activity and cellular responsiveness to cannabinoid signaling. The expression of CB₂R on innate and adaptive immune cells indicates that this receptor may be a potential target for modulation of immune responses [10,19]. In inflammatory states, CB2 mRNA is highly expressed and activation of CB2 has been reported to have anti-inflammatory effects in experimental models of neurodegenerative and fibrotic disorders [19,20]. In addition, CB₂R agonists result in increased levels of the sphingolipid messenger ceramide, particularly in tumor cell lines, which induces apoptotic cell death [21]. Unlike CB1, the absence of psychotropic effects upon activation of CB2 makes it a strong candidate for pharmaceutical targeting, which may open new avenues for the secure modulation of the ECS through the CB2 receptors [7]. Hence an unprecedented opportunity exists to develop tools and probes advancing the understanding of the CB₂R signaling pathway and to guide the rational discovery of new and selective CB₂R ligands [22,23,24,25,26]. CB2 agonists have been shown to be efficacious in animal models of cancer and, also, CB2 agonists have been suggested as potential therapeutic agents for the treatment or management of many pain-related disorders such as acute pain, chronic inflammatory pain, and neuropathic pain, neuroinflammation and/or neurodegeneration such as multiple sclerosis [27,28,29,30]. Epidemiological and preclinical data suggest that stimulation of the CB2 receptor has protective effects against osteoporosis. However, despite showing efficacy in a number of animal models, CB2 agonists have failed to live up to expectations in human clinical trials [31]. In parallel, CB2 receptor antagonism could be beneficial in certain disease states where CB2 receptor signaling maintains an undesirable immune or metabolic status. In obesity-related metabolic inflammation, CB2 receptor antagonism was found to reduce inflammation in adipose tissue macrophages, which improves insulin sensitivity [32]. In cancer immunology/immune suppression, the CB2 receptor antagonism might be helpful in overcoming cannabinoid-induced suppression of immune effector cells, suggesting that CB2 inhibition could, in some contexts, enhance anti-tumor immunity [33]. In general, the antagonism of CB2 receptors might be considered beneficial for therapeutic purposes when the goal is to modulate or disrupt CB2 receptor-related immunosuppression or inflammation in a specific disease state.

Drug discovery approaches require pharmacological assays with high fidelity that are easy to set up and run. Genetically encoded fluorescent biosensors allow visualization of biological processes directly within their natural environment while preserving the native context and revealing precise details about their location and behavior in cells. Using these biosensors in drug screening offers several clear advantages: they enable real-time observation of a drug’s effects within specific cellular compartments, organs, or tissues; allow high-resolution analysis at the single-cell level; and help prevent false positives that may arise in traditional in vitro assays [34,35]. To probe real-time in eCB lipid mediator dynamics in cultured cells and rodent models acting through CB₁R, a genetically encoded fluorescent sensor named GRAB_eCB2.0 was developed [36]. Its activation by eCB agents and limited phyto-cannabinoids was also characterized [37]. In an earlier study, our group analyzed the actions of various classes of synthetic cannabinoid ligands on the GRAB_eCB2.0 sensor [38]. The evaluation allowed for CB₁R-based GRAB_eCB2.0 rapid assessment of apparent binding affinities and preliminary functional characterization of ligands.

In a molecular pharmacology study involving GPCRs, the functional modulation of ligands is studied using well-known ‘gold-standard’ assays. The assays often include radioligand binding, GTPγS, cAMP, β-arrestin, and other sophisticated FRET/BRET-based assays [39]. At times these assays can pose issues when handling radioisotopes, sophisticated plate-readers, and complex assay protocols, along with expensive and hard-to-procure reagents. Traditional radioligand assays are inherently labor-intensive, which constrains their implementation during the initial phases of drug discovery. In contrast, fluorescence-based assays present several advantages, including the feasibility of developing homogeneous assay formats, enabling continuous data acquisition, and achieving superior throughput.

In an approach reported by Li and co-workers for CB₁R, we have developed a novel GRAB-CB2 sensor, engineered from CB₂R, by incorporating a circularly permutated-green fluorescent protein (cpGFP) in intracellular loop 3 (ICL3) where the wild-type receptor interacts with heterotrimeric G-proteins [36]. Through trial-based approaches, the GRAB-CB2, a robust sensor, was shown to be a fast, reliable cell-based tool that can quantitatively indicate binding of numerous modulators of the CB₂R. As known, this sensor primarily works via the confirmational change caused in the transmembrane domain when a ligand binds and causes the eGFP fluorescence intensity to change. This sensor in principle lacks the functional signaling transducers of the native receptor and allows for visual measurement through the increased fluorescence intensity upon ligand binding and activation. Our sensor can detect not only ligands that inherently activate the sensor and act as agonists but also CB2 functional antagonists/inverse agonists. We also assessed the sensor’s performance to detect allosteric interactions of a ligand. The present study was undertaken primarily to establish a fluorescence-based binding assay for the investigation of ligand–receptor interactions with CB2.

In the current study we validate the CB2 sensor’s ability to detect known CB2 agonists and show that the sensor enables a rapid assay that can uncover antagonist/inverse agonist functional fingerprints of CB2 modulators based on the ability to displace agonists. We also utilize this assay to confirm the mode switching seen in two previously reported ligands, MRI-2594 and MRI-2687, which enabled the study of molecular determinants involved in the inactive state structure of CB₂R [40].

2. Results and Discussion

To understand the signal stability after exposure to an agonist we employed kinetics measurement of the GRAB-CB2 sensor with the known orthosteric agonist CP55940. Addition of 10 µM CP55940 at 20 s resulted in a fast rise in the fluorescence of the GRAB-CB2 sensor, lasting about 60 s before leveling off. The fluorescence signal of the activated sensor was retained throughout the remaining measurement time and indicated reproducible activation kinetics under the specified conditions. Long-lasting activation would be consistent with strong binding affinity of CP55940 to the CB2 receptor, which may be significant to the interpretation of its pharmacological effects.

Further, we were interested in determining whether the agonist-induced increase in the fluorescence intensity of the GRAB-CB2 sensor would be visualized through live-cell epifluorescence microscopy. The imaging revealed that the administration of the agonist (1 µM CP55940) elicited a time-dependent increase in fluorescence intensity relative to baseline values, suggesting effective receptor activation as well as sensor engagement. Elevation of the signal was maintained over the period of imaging, possessing kinetics that were consistent with GPCR-mediated processes of activation. Notably, these results validate the utility of the GRAB-CB2 sensor as an effective tool for agonist-mediated measurements of receptor activity in live cells (Figure 1).

Direct engagement of CB2 antagonist MRI-2687 with GRAB-CB2 sensor was monitored in real-time during live-cell confocal microscopy [40]. Administration of the compound, 1 µM MRI-2687, produced a gradual decline in fluorescence output compared to baseline values, thus indicating successful receptor inhibition. The downward trajectory of the signal demonstrates the mechanism by which the antagonist reduces CB2 activity and also supports the conclusion that the GRAB-CB2 sensor is effective at reporting inhibitory actions on receptor activation within the cells (Figure 2).

2.1. GRAB-CB2 Assays for CB₂R Agonists

Agonists of the CB2 receptor have been shown to have therapeutic effects in preclinical models of inflammation and pain with numerous compounds having undergone clinical trials [23,31,41,42,43,44,45,46]. The lack of psychotropic properties was anticipated to work in favor of CB2 agonists as opposed to CB1 agents [46,47]. The GRAB-CB2 sensor was initially designed to look at ligand dynamics of endocannabinoids that are part of the ECS and show a concentration-dependent increase in fluorescence upon ligand binding. We show that synthetic, exogenous agonists reported in the literature as well components of the ECS like 2-AG respond to the sensor by activating the sensor. To better comprehend the effectiveness and performance of the sensor, we segregated the tested agonists into different classes as noted below.

Initially we intended to check the GRAB-CB2 sensor efficacy with CP55940, which is known to be a full agonist for the CB2 receptor. In CB2-expressing cells, it generates a sharp, concentration-responsive effect with nanomolar potency (single-digit nM EC₅₀ in most assays) and Emax close to 100% for both G-protein-coupled (e.g., cAMP/GTPγS) and β-arrestin downstream pathways. CP55940 has also been shown to have therapeutic effects as both an antinociceptive and an antiemetic. However, reported psychotropic effects of CP55940 preclude its use as a therapeutic. Since its intrinsic efficacy is strong, CP55940 is commonly used as the positive control to normalize Emax when comparing novel ligands. As a potent CB2 receptor agonist, the reported EC₅₀ of CP55940 ranges from the sub-nanomolar value of 0.04 nM to 31 nM. As anticipated, we obtained a robust concentration-dependent fluorescence enhancement at varied concentrations, which in turn generated a sigmoidal curve showing CB2 agonist potency when tested through the GRAB-CB2. With the GRAB-CB2 sensor, under our reported conditions, CP55940 showed an apparent EC₅₀ of 10–16 nM (Figure 3).

2.1.1. Natural (Phytocannabinoids/Plant Terpenes)

Cannabinoids are naturally occurring substances found in several cannabis species, including Cannabis sativa, Cannabis indica, and Cannabis ruderalis. These plants generate a broad range of chemical components, including over 100 chemicals now classed as cannabinoids. However, many of these compounds have yet to be completely characterized. The first scientifically discovered compounds were Δ⁹-tetrahydrocannabinol (THC) and cannabidiol (CBD), both of which were isolated from Cannabis sativa flowers. THC is also considered a partial agonist at CB2, but has a lower efficacy than for CB1 [48]. Relative to CP55940 (a high-affinity CB1/CB2 full agonist), Δ⁹-THC exhibits ~20–70-fold lower binding affinity at human CB₁R and ~20–60-fold lower affinity at human CB₂R.

Like THC, the natural bicyclic sesquiterpene phytocannabinoid β-Caryophyllene (BCP) is a common constituent of essential oils in numerous spice and food plants, a major component in the cannabis sativa plant and reported to act as a selective agonist of CB2 receptors [49]. The anti-inflammatory effects of BCP were reported to be mediated through CB2 receptors [50]. In a departure from reported data, β-caryophyllene failed to elicit any detectable agonism on our CB2 sensor. Nonetheless, we cannot rule out weak interaction on the CB₂R for BCP as was reported by Glass et al. [50].

To detect partial agonism of THC, we used different concentrations of the full agonist CP55940 and obtained the anticipated sigmoidal response curve. We then added 0.1 µM, 0.5 µM, and 10 µM partial agonist THC to the variable CP55940 concentration. We observed a clear rightward shift in the sigmoidal curve with increasing concentrations of THC in the variable concentration of CP55940. Our interpretation was that at the lowest concentration of THC, only a small fraction of the receptor is occupied by THC, whereas at higher concentrations, there would be more receptor occupancy causing a rightward shift. The shift scales with the concentration of THC, indicating that it is competing with the full agonist for the same site. This rightward shift indicates a competitive interaction at the orthosteric binding site (Figure 4).

2.1.2. CB2 Compounds as Research Tools

We next tried additional well-characterized compounds and compounds reported as agonists on CB2 in the literature. Compounds JWH-133, HU-308, WIN-55,212-2 and A836339 are well-known for their CB2 agonist behavior in different functional assays. Among them, WIN 55,212-2, an aminoalkylindole derivative, is a potent cannabinoid receptor 1 and 2 (CB1 and CB2) agonist at concentrations in the nanomolar range and with 20-fold higher affinity for CB₂R and has been found to be a potent analgesic in a rat model of neuropathic pain. HU 308 is a very effective and selective agonist for the CB2 receptor, with Ki values of 22.7 nM for CB2 and more than 10 μM for CB1 receptors, and an EC₅₀ of 5.57 nM (cAMP) for CB2. JWH 133 is a potent CB2 selective agonist (Ki = 3.4 nM) with approximately 200-fold selectivity over CB1 receptors and is active in vivo, reducing spasticity in a murine model of multiple sclerosis. A-836,339 is a compound developed by Abbott Laboratories that acts as a potent cannabinoid receptor full agonist. It is selective for CB2, with Ki values of 0.64 nM for CB2 vs. 270 nM for the psychoactive CB1 receptor, exhibiting selective analgesic, anti-inflammatory and anti-hyperalgesic effects at low doses (Figure 5) [51,52].

MRI-2594 and PB-133 were developed by our research group and are known CB2 agonists [16,53]. Unsurprisingly, our GRAB-CB2 sensor showed a robust and potent CB2 agonism for MRI-2594 and PB-133 (structural analog of MRI-2594) with an EC₅₀ value of 18.9 nM and 11.5 nM respectively, validating our previously reported values for MRI-2594 as a potent CB2 agonist (Figure 5) [40,54].

Other tool compounds also showed concentration-based enhancement of fluorescent intensity from lowest to highest concentration, attesting to agonist identification and validation through GRAB-CB2 (

Table 1. Selected compounds used for the GRAB-CB2 assay.

Compound	Class	CB2 Affinity (Ki/pKi)	Literature-Reported Functional Potency for CB2R (EC50/IC50; Assay)	GRAB-CB2 (EC50 nM)
Virodhamine	Endocannabinoid	NA	EC50~1.4 µM (CB2) CB2 agonist; CB1 partial agonist/antagonist	4548 ± 652
2-AG	Endocannabinoid	Ki = 1.3–1.4 µM (hCB2; binding)	Partial agonist; pEC50 often < 5.5 (assay-dependent) Partial agonist; pathway bias reported	1389 ± 134
Δ9-THC	Phytocannabinoid	Ki = 30–40 nM/pKi: 6.4  ±  0.2 (HEK-293 T cells)	Partial agonist (assay-dependent) [35S]GTPγS binding (CHO-K1; “partial agonist”): EC50 ≈ 12–12.3 nM (pEC50 7.91–7.92)	397 ± 99
β-Caryophyllene	Phytocannabinoid	Ki = 155 nM	EC50~38 nM (cAMP, CHO-CB2) CB2-selective agonist	>1000
JWH-133	Synthetic	Ki = 3.1–3.4 nM	Potent; often full agonist (assay-dependent), highly CB2-selective full agonist, pEC50 value = 7.54 ± 0.16	84.5 ± 1.5
HU-308	Synthetic	Ki = 22.7 nM; CB1 Ki > 10 µM	EC50~5.6 nM (cAMP inhibition) CB2-selective full agonist	45.1 ± 5.3
A-836339	Synthetic	High CB2 affinity Ki 0.64 nM	EC50~1.6 nM (reported) CB2 full agonist	8.9 ± 0.6
GW842166X	Synthetic	NA	EC50~63–91 nM (human/rat CB2; cyclase/FLIPR) full agonist	224 ± 35
Olorinab (APD-371)	Synthetic (clinical)	Ki = 6 nM	EC50~6.2 nM (human CB2), highly selective full CB2 agonist, EC50: 6–8 nM for the rat, and dog CB2 receptors	3.5 ± 0.2
Tedalinab (GRC-10693)	Synthetic	NA	EC50~50.7 nM (CB2) > 4700-fold functional selectivity for CB2 over CB1	8.9 ± 1
LY-2828360	Synthetic (clinical)	Ki = 40.3 nM	EC50~20.1 nM (GTPγS, CB2) G-protein-biased CB2 agonist	35.3 ± 12.6
MRI-2594	Synthetic	Ki = 0.031 nM	EC50~310 ± 54 nM (β-arrestin2, CB2)	18.9 ± 1.6
CP55940	Synthetic (reference)	Ki = 0.7–2.6 nM	EC50~3.2–3.4 nM (β-arrestin2, CB2), potent, efficacious CB2 agonist	14 ± 2.4
WIN-55,212-2	Synthetic	Ki = 3–4.5 nM	Potent agonist (assay-dependent), full agonist; CB2-preferred vs. CB1 (often), 3.3 nM for human recombinant CB2, EC50: 0.52 ± 0.11 nM (cAMP)	16.1 ± 6.3
AZD1940	Synthetic (clinical)	pKi~9.06 → Ki~0.87 nM	Peripherally acting mixed CB1/CB2 agonist	1.95 ± 0.07

). Additionally, the consistent agonist behavior observed reinforces the reliability of the GRAB sensor as a tool for evaluating CB2 ligand–receptor interactions.

2.1.3. Compounds in Clinical Trials

We proceeded to test many reported compounds that were in clinical trials for CB2 agonist activity as their mechanism of action (Table 1). GW842166X is a potent and selective CB2 agonist with literature EC₅₀ values of 63 and 91 nM for human and rat CB2, respectively [55,56]. The GRAB-CB2 sensor showed that GW842166X compound is an agonist with an EC₅₀ value of 224 nM. Olorinab (APD 371) is a highly potent, selective and fully efficacious CB2 agonist, with an EC₅₀ of 6.2 nM for hCB2 [57]. We found it is also a strong agonist with a 3 nM potency to CB₂R. Tedalinab (GRC-10693) developed by Glenmark was in clinical trials for the treatment of osteoarthritis and neuropathic pain, which acts as a potent and selective cannabinoid CB2 receptor agonist. It has a very high selectivity (>4700-fold) for CB2 over the related CB1 receptor. The compound LY2828360 (hCB₂: Ki 40 nM, EC₅₀, [³⁵S]-GTPγS assay 20 nM), developed by Eli Lily as an oral analgesic for osteoarthritic knee pain, is based upon a substituted purine core, and behaved as a CB2 agonist with our GRAB-CB2 sensor validating previous reports [58,59] (Table 1, Figure 5). AZD1940 is an orally bioavailable, high-affinity agonist for CB1 and CB2 receptors, with pKi values of 7.93 and 9.06 for human CB₁R and CB₂R, respectively. AZD1940 is novel peripherally acting drug candidate with significant analgesic efficacy (Figure 5) [60,61].

2.1.4. Endogenous Cannabinoid Ligands (Endocannabinoids)

We also tested our GRAB-CB2 sensor with endogenous endocannabinoids derived from arachidonic acid that act as lipid signaling molecules. 2-AG generally drives CB1/CB2 activation, whereas virodhamine often restrains CB1 while favoring CB2. Virodhamine was reported as a weak CB1 antagonist and CB2 agonist [62,63]. Applying them complementarily can evoke opposing responses in the endocannabinoid system. Indeed, we could detect CB2 agonism for virodhamine on our CB2 sensor. 2-AG is a naturally occurring endocannabinoid with a relatively high concentration in the central nervous system. Using the GRAB-CB2 sensor in HEK293 cells, an EC₅₀ of 1.4 μM was obtained for 2-AG, showing an increase in fluorescence upon ligand binding. Similarly, by employing the GRAB-CB2 sensor with virodhamine, we observed an EC₅₀ value of 4.5 μM (Figure 5).

2.2. GRAB-CB2 Assays for CB₂R Antagonist

Cannabinoid receptor type 2 antagonists constitute an intriguing area of pharmaceutical research, which may offer yet untapped benefits in multiple diseases [64]. CB2 antagonists were discovered in the late 1990s to have been used to study the pharmacology of CB2 receptor systems. The recent literature precedence suggests that antagonists of the CB₂R may have promising beneficial effects in modulating the tumor microenvironment [65]. CB2 antagonists work by inhibiting the receptors, which is useful in scenarios where inflammation plays a pivotal role in disease development. Schering-Plough research showed that a CB2 inverse agonist SCH336 blocked macrophage recruitment in an in vivo model of inflammation [66]. This is consistent with the general idea that any disruption of CB2 function will result in a suppression of immune cell-based inflammation. CB2 antagonist functions were shown to slow down cognitive decline [65]. Hence CB2 antagonists could be used as vital tools for elucidating the diverse roles of the CB2 receptor, and a range of CB2 antagonists/inverse agonists spanning different structural scaffolds have been reported [40,67]. Accordingly, CB2 inverse agonists like TT-816 and SMM-189 have undergone extensive preclinical investigations as immune checkpoint inhibitor and for the treatment of central nervous system disorders, and for conditions like colitis respectively [33,68,69,70].

Previous work with our GRAB_eCB2.0 CB1-based sensor has shown the responses of various CB1 antagonists belonging to differing chemotypes [38]. Prototypical CB1 antagonists bearing three-arm properties like rimonabant, taranabant, and otenabant as well as our four-arm class of CB1 antagonists gave robust dose-dependent antagonist responses with the sensor. Here we sought to investigate the behavior of reported CB2 antagonists on our engineered GRAB-CB2 sensor.

Initially we started our investigation with our ‘in-house’ developed molecule MRI-2687, which is a synthetic, super-high-affinity ligand for the human CB2 receptor derived from the benzothiazoles series [40,54]. The ligand exhibits a sub-nanomolar affinity with K_i-values that range between ~0.053 and ~0.1 nM. Functionally, MRI-2687 is a CB2 receptor inverse agonist, since it significantly inhibited basal CB2 receptor signaling in both recruitment of β-arrestin2 and [³⁵S]GTPγS binding displacement assays, (EC₅₀~0.2–0.3 nM) [40]. In the GRAB-CB2 sensor assay, the CP55940 quenching by MRI-2687 could be seen in a clear concentration-dependent manner with decreasing fluorescence, which indicates antagonism. A potent CB2 antagonist curve for MRI-2687 was observed in the same experimental system with an IC₅₀ of 2.6 nM. When tested in the absence of an agonist, we also saw a clear concentration-dependent decrease in basal GRAB-CB2 fluorescence intensity, which could be attributed to inverse agonist behavior with an IC₅₀ of 6.7 nM (Figure 6).

We were intrigued to assess whether such inverse agonism could be detected by our newly developed GRAB-CB2 sensor for some of the well-known, literature-reported ligands of CB2. One of the earliest CB2 antagonists, SR-144528, was identified and developed by Sanofi [71]. It was reported as a potent and selective CB2 antagonist, with a K_i of 0.6 nM. It is an orally bioavailable compound with high affinity, and selectivity to CB2 receptors (700-fold selectivity for CB2 receptor binding over CB1 receptor binding). SR144528 inhibits CP55940-mediated forskolin-sensitive adenylyl cyclase activity as well as CP55940-mediated MAPK in Chinese hamster ovary cells (CHO) expressing CB2 receptors. Furthermore, it inhibits B-cell activation mediated by CP55940. SR 144528 has been determined to inhibit constitutive mitogen-activated protein kinase activity, which is associated with the spontaneously autoactivated peripheral cannabinoid receptor (CB2). A well-known CB2 tool compound SCH336 was developed by Schering and is an extremely potent (K_i = 1.8 nM, EC₅₀ = 2 nM) inverse agonist of CB2 with 100-fold selectivity for CB2 over CB1. This molecule works as a putative selective anti-inflammatory drug by reducing the migration of leukocytes (white blood cells) into inflamed areas by targeting CB2 receptors. SCH336 is under consideration as an anti-inflammatory drug for those diseases that relate to CB2 because of its anti-inflammatory property [66]. Another tool compound, AM630, has a K_i of 31.2 nM for CB2, with 165-fold selectivity for CB2 versus CB1 binding. AM630 behaves as an inverse agonist for CB2, reducing the antinociceptive actions of a range of different cannabinoids, while acting as a weak partial agonist at CB1 [72].

Using our GRAB-CB2 sensor, we have found evidence of inverse agonism with the above-mentioned tool compounds (SCH336, SR144528, and AM630). When treated with various concentrations, strong inverse agonism was found with values in the range of 17.48 ± 6.2, 15.16 ± 16, and 6.2 ± 1.8 for SCH336, SR144528, and AM630, respectively. Again, mimicking the antagonist functional assay, the sensor could be activated by agonist CP55940 at the EC₈₀ concentration, and was dose-dependently reduced by variable concentrations of the above molecules. This is a clear indication of the antagonistic behavior of the tool compounds, which aligns with previous literature reports (Figure 7).

Previous studies on benzothiazole-based compounds developed in our lab have shown that they are extremely potent CB2 ligands, with the affinity of the MRI series for the CB2 receptor ranging from 0.053 to 0.1 nM in hCB2-expressing CHO cell membranes, as measured by displacement binding experiments (Figure 8). The report suggested that MRI-2646 behaves as a partial agonist or neutral antagonist, while MRI-2653, MRI-2654, and MRI-2659 act as inverse agonists [54]. With our GRAB-CB2 sensor, these three arm-type compounds reduced the fluorescence signal in a concentration-dependent manner below baseline, which is consistent with an inverse agonist profile. We also examined their ability to block the agonist-induced fluorescence intensity increase in the presence of CP55940. The compounds other than MRI-2646 showed clear inverse agonism in the GRAB-CB2 sensor assay. In addition, the previously reported MRI-1995, MRI-2007, MRI-2213, and MRI-2006 compounds all behaved as inverse agonists when tested, further supporting that our GRAB sensor system is quite robust for identifying CB2 antagonists and agonists (Figure 8).

Finally, we evaluate whether our newly developed GRAB-CB2 sensor performs reliably with pyridazine series of molecules previously reported by our group. The compounds PB-228E2 and PB-95E2 are tetrahydro-pyridazine-based four-arm compounds that were previously characterized as CB₁R antagonists and showed potency for obesity and metabolic disease treatment (Figure 8) [73,74]. In this new GRAB-CB2 assay we specifically wanted to examine their functional activity toward CB2. Based on their behavior on the sensor, these compounds behaved as weak antagonists, with IC₅₀ values in the micromolar range (

Table 2. Selected CB2 antagonists tested for GRAB-CB2.

Compound	Class	CB2 Affinity (Ki/pKi)	Literature-Reported Functional Potency for CB2R (EC50/IC50; Assay)	IC50 for Inhibition of Basal GRAB-CB2 Fluorescence Signal (nM)	IC50 for Antagonism (in Presence of CP55940 at EC80) (nM)
SCH336	Synthetic	Ki = ~1.8 nM (human CB2)	Highly CB2-selective (≈100- to >2000-fold over CB1, forskolin-stimulated cAMP in hCB2-expressing CHO cells, EC50 ≈ 2 nM	17.48 ± 6.2	14.48 ±4.1
SR144548	Synthetic	Ki = 0.6 nM (rat spleen CB2 and human cloned CB2	~700-fold lower affinity for CB1 (Ki ≈ 400 nM), cAMP (CHO-hCB2): EC50 ≈ 26 nM	15.16 ± 16	21.5 ± 10.9
AM630	Synthetic	Ki = 31.2 nM (human CB2)	~150–165-fold CB2-selective vs. CB1 (CB1 Ki ≈ 5–5.2 µM), [35S]GTPγS binding (CB2-CHO membranes): EC50 ≈ 76.6 nM	6.2 ± 1.8	86 ± 15
MRI-2687	Synthetic	Ki = 0.1 ± 0.01 nM	GTP-γ-S binding EC50= 1.23 ± 0.09 with Emax (%) = −35 ± 1	6.7 ± 1.9	2.6 ± 0.5
MRI-2646	Synthetic	Ki = 0.079 ± 0.011 nM	N.D.	14 ± 2.9	3.99 ± 0.6
MRI-2653	Synthetic	Ki = 0.094 ± 0.011 nM	GTP-γ-S binding EC50= 0.5 ± 0.09 with Emax (%) = −46 ± 2	9 ± 0.65	8.46 ± 1.4
MRI-2654	Synthetic	Ki = 0.079 ± 0.009 nM	GTP-γ-S binding EC50= 0.76 ± 0.04 with Emax (%) = −23 ± 3	6.5 ± 0.9	17.7 ± 2.7
MRI-2659	Synthetic	Ki = 0.053 ± 0.011 nM	GTP-γ-S binding EC50= 0.25 ± 0.01 with Emax (%) = −55 ± 1	31 ± 5.1	30.6 ± 1.3
MRI-1995	Synthetic	Ki = 28 ± 1.7 nM	ND	77.5 ± 9.6	1200 ± 279
MRI-2007	Synthetic	Ki = 36± 5 nM	ND	154 ± 44	2408 ± 1504
MRI-2213	Synthetic	Ki = 100 nM	ND	42 ± 3.8	1097 ± 117
MRI-2006	Synthetic	Ki = 2.4 ± 0.3 nM	ND	18.7 ± 5.3	157 ± 81
PB-228E2	Synthetic	Ki = 21 ± 1.9 nM	ND	837 ± 390	>5000
PB-95E2	Synthetic	Ki = 43 ± 3.4 nM	ND	118 ± 24	>5000
CBD	Phytocannabinoid	Ki = ≫1 µM	Antagonizes CB2 agonists; inverse agonism reported. Low-affinity CB2 antagonist/negative allosteric effects reported	3100 ± 392	>5000

, Figure 9).

Further we wanted to check the behavior of cannabidiol (CBD) on the GRAB-CB2 sensor. Cannabidiol behaves as a high-potency antagonist of cannabinoid receptor agonists in mouse brain tissue and in membranes from CHO cells transfected with human CB2 receptors. Cannabidiol is known to partially inhibit the migration of immune cells via CB2 receptor activation and is associated with various anti-inflammatory and neuroprotective benefits in central nervous system inflammation models [75]. Furthermore, it was shown that cannabidiol in hCB2-CHO cell membranes appears to behave as an inverse agonist in hCB2 receptors [75,76]. Intrigued by these reports, the next objective was to utilize the GRAB-CB2 model system to evaluate the inverse agonist and antagonist properties of CBD. Here, concentration–response curves for CBD were obtained both alone and in the presence of fixed amounts of CP55940 (EC₈₀). As observed from the data in Figure 10, CBD alone caused a concentration-dependent response curve with a reduction in the GRAB-CB2 basal fluorescence, indicating inverse agonism by CBD (Figure 10a). Also, concentration–response curves for CBD obtained in the presence of fixed amounts of CP55940 (EC₈₀) showed an antagonistic effect with higher levels of CBD (Figure 10b). This agrees with previous reports that CBD functions as a CB2 receptor antagonist/inverse agonist [75,77], and further confirms that the GRAB-CB2 sensor can reliably discern the activity of CB2 receptor ligands.

2.3. GRAB-CB2 Assays for CB₂R PAM

Allosteric sites are specific sites on a receptor that are located outside the orthosteric (main) binding pocket. Such sites are capable of binding small molecules or interacting proteins, and their binding alters the responsiveness of the receptor to the signal. In the major families of G protein-coupled receptors (GPCRs), i.e., A, B, and C, such sites have been identified, reflecting the wide-ranging importance. Molecules that act on such sites are grouped as follows: molecules that directly activate the receptor are called allosteric agonists, and molecules that change the effect of the orthosteric (main) ligand without inducing the molecule to exhibit activity autonomously are labeled positive (PAM) or negative (NAM) modulators. From a drug point of view, pure allosteric modulators are often viewed positively because their effect depends on the availability of natural/endogenous or administered orthosteric ligands. This requirement offers an intrinsic check of their activity as well as dampens the likelihood of overstimulation [78]. Allosteric modulation has recently been shown to be a promising approach for the fine-tuning of CB₁R receptor signaling with fewer deleterious side effects related to the engagement of the receptor’s orthosteric site [79]. Rather than competing with the endogenous ligand for the receptor’s binding site, allosteric modulators interact with alternate sites that are topographically different and modulate the magnitude and/or time course of receptor signaling. Despite their therapeutic promise for achieving more selective, tissue- and context-dependent CB2 modulation with potentially fewer side effects, the number of well-characterized CB2 allosteric modulators remains very limited. Few PAMs and NAMs on CB2 have been reported, many with incomplete pharmacological profiles, modest potency, or poor drug-like properties. As a result, the CB2 allosteric ligand toolbox is still sparse compared with orthosteric CB2 agonists/antagonists, which significantly constrains both mechanistic and structure–activity relationship (SAR) studies.

With the aim of further characterizing GRAB-CB2 within the framework of the 96-well plate reader assay, the next direction of this study focused on the examination of literature-reported CB2 allosteric modulators. By doing so, we aimed to better understand how GRAB-CB2 reports allosteric effects at CB₂R and to assess whether this platform can be used to distinguish between different classes of allosteric modulators, such as PAMs and NAMs. Therefore, we aimed to detect allosterism using the GRAB-CB2 sensor. Utilizing the cannabinoid full agonist CP55940, we generated a typical sigmoidal concentration–response curve. We then introduced a known positive allosteric modulator Ec2la (GRAB-CB2, EC₅₀: 101 ± 7 nM) at two concentrations (0.3 µM and 10 µM) while varying CP55940 [80]. In both cases, the agonist curve of CP55940 shifted leftward, a clear indicator that the sensor can effectively capture allosteric modulation at the CB2 receptor and also a modest increase in Emax (greater maximal response). The concurrent leftward shift in the CP55940 concentration–response relationship is consistent with the augmentation of agonist potency (reduced EC₅₀) produced by the PAM, while the small augmentation of E_max suggests either heightened efficacy or improved receptor–sensor coupling during allosteric stabilization (Figure 11).

One significant advantage of the GRAB-CB2 sensor is that it enables us to understand CB2 pharmacology in real-time in intact cells, allowing rapid differentiation of orthosteric agonism from allosteric modulation and providing immediate readouts of potency, and modulatory ‘fingerprints.’ This is especially helpful for CB2 allosteric ligands because their effects are often subtle and can vary depending on the agonist level and experimental conditions. Furthermore, this could be useful for comparisons between different compounds under varied conditions (e.g., different concentrations of PAM) without changing the type of assay. Additionally, the mode switching or ‘functional flip’ observed with CB2 ligands can be rapidly discerned by such fluoresence-based assay [40,81].

In order to determine the selectivity of the GRAB-CB2 sensor, the known CB1 receptor-selective agonist arachidonyl-2′-chloroethylamide (ACEA) was tested in the GRAB-CB2 sensor assay. However, ACEA did not produce any measurable enhancement in fluorescence at all, even at a concentration of 10 µM, thus supporting the specificity of the GRAB-CB2 sensor assay for discerning CB1 receptor-selective compounds (Figure 12). ACEA, based on the literature and from our earlier work with GRAB_eCB2.0, showed clear CB1 agonsits.

3. Materials and Methods

Chemicals and reagents: 2-AG (Cayman, Ann Arbor, MI, USA), AEA (Cayman), CP55940 (Cayman), β-Caryophyllene (Cayman), Virodhamine (Cayman), JWH-133 (Cayman), HU-308 (Tocris, Bristol, UK), WIN-55,212-2 (Tocris), A836339 (Cayman), GW842166X (Cayman), Olorinab (MedChem Express, Monmouth Junction, NJ, USA), Tedalinab (MedChem Express), LY-2828360 (MedChem Express), AZD1940 (MedChem Express), Ec2la (Tocris), SCH336 (Tocris), SR144528 (MedChem Express), AM630 (Tocris), ACEA (Tocris), SR141716 (MedChem Express), MRI-2594, MRI-2687, MRI-2653, MRI-2654, MRI-2646, MRI-2659, MRI-1995, MRI-2006, MRI-2007, MRI-2213, PB-133, PB-95E2 and PB-228E2 were synthesized in-house according to previously published procedures [53,82,83].

Design of GRAB-CB2 Sensor

The GRAB-CB2 sensor was constructed in the previously described eCB2.0 (hCB1-based GRAB) pEGFP-C1 backbone [36]. Briefly the open reading frame of hCB2 (NM_001841) was amplified from human whole brain cDNA (Takara) and inserted in place of the hCB1 sensor portion of eCB2.0 using the In Fusion cloning system (Takara, San Jose, CA, USA). The IgK signal sequence was retained on the amino terminus of the new hCB2 construct. The entire sensor containing ICL3 of eCB2.0 was amplified and inserted to replace the entire ICL3 of the hCB2 construct. The hCB2 ICL3 was identified by transmembrane prediction using the protein analysis toolbox in MacVector and compared to previous predictions [84]. This construct was used to amplify the entire GRAB-CB2 open reading frame for insertion into the PiggyBAC transposon-based shuttle vector PB514b (SBI) as described for the eCB2.0 stable cell line generation³¹. The co-transfection of transposase and GRAB-CB2 shuttle for generation of stable tsA201-HEK293 cells was as described for eCB2.0 [38].

Assay Procedure: The DMEM buffer ((Thermo Fisher, Waltham, MA, USA: Gibco #10569-010)) was removed from the T25 flat-sided cell culture flask by decantation. Then, 3 mL of TrypLE (Gibco #12604-021) was added and allowed to incubate for 2–3 min at 25 °C in a laminar flow hood; 7 mL of fresh DMEM was added, and cells were gently pipetted to remove any remaining cells from the flask wall. The total volume (10 mL) was transferred to a 15 mL Falcon Tube and centrifuged at 800× g for 4 min at 23 °C. Upon completion of centrifugation, the supernatant was removed and discarded, leaving behind a small pellet of cells that were resuspended in 4 mL of fresh DMEM. The cell suspension was mixed in a 1:1 ratio with 0.4% TrypanBlue (10 μL each) and 10 μL of the combined solution was added to each side of a Countess Cell Counting Chamber and inserted into the Countess II Automated Cell Counter (Thermo Fisher, Waltham, MA, USA). The concentration of cells was adjusted to 0.8–1.0 million cells/mL with DMEM complete media and 100 μL was added to the desired wells of a clear bottom black plastic tissue culture-treated 96-well plate. The plate was incubated at 37 °C and 8% CO₂ for a 24–48 h period. After incubation, the 100 μL DMEM buffer was removed and replaced with 100 μL of DPBS buffer (with 0.5 mg/mL BSA) containing calcium and magnesium (Gibco). After the addition of the DPBS buffer (Gibco #14040-133), a baseline reading of the cells was taken using the PHERAstar FSX microplate reader (BMG/Labtech, Cary, NC, USA to establish the F₀ or initial fluorescence (Figure 13).

When running the assay for an agonist, 25 μL of the compound in DPBS buffer was added to the wells. A concentration curve was made for the modulator with the final concentrations in the wells holding the cells ranging from 1 pM to 10 mM with 1% DMSO in a total volume of 125 μL. The procedure was modified for compounds which did not show an increase in fluorescence upon testing (presumed antagonist/allosteric modulators); 25 μL of the compound and 25 μL of additional standardized agonist CP55940 was used to assess fluorescence alterations in the absence and presence of standardized agonist. A concentration curve was made for the modulator with the final concentration in the wells holding the cells ranging from 1 pM to 10 mM, and the final concentration of the CP55940 in the well was 50 nM with 1% DMSO and 0.5 mg/mL BSA in a total volume of 150 μL. In either assay, the cannabinoid ligands were added, followed by incubation of the compound on the cells at 37 °C for 5 min before taking a fluorescence reading using the PHERAstar FSX microplate reader equipped with a GFP filter cube module. The recorded values were the F or final fluorescence.

For kinetic measurements, stable GRAB-CB2 cells were plated as described above. The agonist CP55940 was prepared as a 50 μM stock solution in DPBS buffer with calcium and magnesium and loaded onto the injector system of the PHERAstar FSX microplate reader. Plating media was removed from wells and replaced with 100 μL of DPBS buffer with calcium and magnesium. Running in well mode, fluorescence was measured in each well for 20 s to determine baseline (F₀) and then for 240 s post injection of 25 μL of CP55940. This injection provided mixing and yielded a final agonist concentration of 10 μM CP55940 (Figure 14).

Statistical Methods: All GRAB_eCB2.0 fluorescent signals (unless otherwise noted) are expressed as normalized ΔF/F₀ as calculated for the fluorescence assay in the PHERAstar FSX Plate reader. EC₅₀/IC₅₀ values were calculated using GraphPad Prism 10 by fitting data to a nonlinear regression model with normalized variable slope (3-parameter, normalized where indicated) or using three-parameter response. Data represent mean ± SEM from at least three independent experiments performed in technical duplicate or triplicates.

4. Conclusions

In the current study, we extend the toolbox of genetically encoded sensors to a GPCR activation-based sensor for CB₂R. Our goal was to develop an expedient assay for the discovery stage assessment of cannabinoid modulators, and our findings with newly engineered GRAB-CB2 sensors validate the usefulness of fluorescent sensor platform and reveals its future potential for pharmacological studies. Moreover, the findings reported here demonstrate not only the utility of the GRAB-CB2 to detect ligand potencies but also the feasibility of future studies to gain a deeper understanding of the role of the CB2 receptor within differing physiological functions. Using the GRAB-CB2 sensor, we were able to differentiate between agonists and antagonists in a rapid fashion, and also validate that the sensor can robustly identify CB2 PAMs as well as weak or partial agonists. While the sensor lacks signaling transducers (e.g., G proteins, β-arrestins, etc.), thereby limiting the secondary signaling events impacted by ligand binding, overall, this platform appears quite reliable for rapidly identifying CB2 agonists and antagonists with assured confidence. In addition, the GRAB-CB2 assay offers a practical advantage for medium-throughput screening, where quick discrimination between different classes of ligands is essential. The ability to capture subtle differences in ligand activity (full vs. partial agonist, PAM, mode switch, etc.) may also support structure–activity relationship (SAR) optimization in medicinal chemistry programs. Along with careful calibration and in parallel to gold-standard functional assays, we believe that this sensor system could become a useful complementary tool for early-stage CB2 drug discovery and pharmacological profiling.