Pharmacological Characterization of the Endocannabinoid Sensor GRAB_eCB2.0

Abstract

Introduction:

The endocannabinoids (eCBs), 2-arachidonoylglycerol (2-AG) and arachidonoyl ethanolamine (AEA), are produced by separate enzymatic pathways, activate cannabinoid (CB) receptors with distinct pharmacological profiles, and differentially regulate pathophysiological processes. The genetically encoded sensor, GRAB_eCB2.0, detects real-time changes in eCB levels in cells in culture and preclinical model systems; however, its activation by eCB analogues produced by cells and by phyto-CBs remains uncharacterized, a current limitation when interpreting changes in its response. This information could provide additional utility for the tool in in vivo pharmacology studies of phyto-CB action.

Materials and Methods:

GRAB_eCB2.0 was expressed in cultured HEK293 cells. Live cell confocal microscopy and high-throughput fluorescent signal measurements.

Results:

2-AG increased GRAB_eCB2.0 fluorescent signal (EC₅₀=85 nM), and the cannabinoid 1 receptor (CB₁R) antagonist, SR141716 (SR1), decreased GRAB_eCB2.0 signal (IC₅₀=3.3 nM), responses that mirror their known potencies at the CB₁R. GRAB_eCB2.0 fluorescent signal also increased in response to AEA (EC₅₀=815 nM), the eCB analogues 2-linoleoylglycerol and 2-oleoylglycerol (EC₅₀=632 and 868 nM, respectively), Δ⁹-tetrahydrocannabinol (Δ⁹-THC), and Δ⁸-THC (EC₅₀=1.6 and 2.0 μM, respectively), and the artificial CB₁R agonist, CP55,940 (CP; EC₅₀=82 nM); however their potencies were less than what has been described at CB₁R. Cannabidiol (CBD) did not affect basal GRAB_eCB2.0 fluorescent signal and yet reduced the 2-AG stimulated GRAB_eCB2.0 responses (IC₅₀=9.7 nM).

Conclusions:

2-AG and SR1 modulate the GRAB_eCB2.0 fluorescent signal with EC₅₀ values that mirror their potencies at CB₁R, whereas AEA, eCB analogues, THC, and CP increase GRAB_eCB2.0 fluorescent signal with EC₅₀ values significantly lower than their potencies at CB₁R. CBD reduces the 2-AG response without affecting basal signal, suggesting that GRAB_eCB2.0 retains the negative allosteric modulator (NAM) property of CBD at CB₁R. This study describes the pharmacological profile of GRAB_eCB2.0 to improve interpretation of changes in fluorescent signal in response to a series of known eCBs and CB₁R ligands.

Introduction

Many physiological functions and behaviors are differentially controlled by endogenously produced arachidonoyl ethanolamine (AEA) and 2-arachidonoylglycerol (2-AG) that activate cannabinoid 1 receptor (CB₁R).¹ Specifically, AEA or 2-AG production by select cells will partially or fully activate CB₁Rs (EC₅₀=10–100 nM and 30–300 nM, respectively) in an autocrine and paracrine manner.² The effects of CB₁R activation are dependent on both cell type and coupling to intracellular signaling systems: CB₁R are expressed at remarkably different levels by distinct excitatory and inhibitory neurons and by glial cells where they couple to specific signaling pathways. Thus, cell-specific, differential activation of CB₁Rs by AEA and 2-AG in the brain fine tunes excitatory and inhibitory neurotransmission and neuromodulation, and regulates neuronal metabolism and phenotype.^3,4

This fundamental signaling mechanism is modulated by Δ⁹-tetrahydrocannabinol (Δ⁹-THC) via its binding to the orthosteric binding site of CB₁R where it acts as a partial agonist. For example, free CB₁Rs expressed throughout the brain are partially activated by THC, whereas CB₁R signaling stimulated by localized activity-dependent increases in 2-AG may be inhibited by THC.^5–8 CB₁R signaling is also modulated by cannabidiol (CBD) that interacts with a putative allosteric binding site on CB₁R.⁹ Multiple lines of evidence suggest that CBD acts as a negative allosteric modulator (NAM) of CB₁R, although its direct binding to CB₁R has still not been demonstrated.^10,11 Thus, CBD reduces 2-AG-stimulated CB₁R activity without influencing basal/tonic CB₁R signaling. This premise emphasizes a need to better understand the role of endogenously produced AEA and 2-AG, their dynamics, and how the presence of THC and CBD affects their activation of the CB₁R.

Mass spectrometry has demonstrated that 2-AG is 10–1000 times more abundant than AEA in select cell types and tissues; however, little is known about the activity-dependent and spatiotemporal changes of 2-AG and AEA that occur within seconds.^1,12 Of importance, increased cellular activity not only enhances the production of AEA and 2-AG but also enhances the production of lipid analogues synthesized by the same enzymatic pathways. For example, 2-linoleoylglycerol (2-LG) and 2-oleoylglycerol (2-OG) activate CB₁R yet with lower potency and efficacy than 2-AG.^13–15 Thus, localized activity-dependent increases in endocannabinoids (eCBs) and their analogues will differentially activate CB₁R within seconds. The recent development of genetically encoded fluorescent sensors has enabled the real-time detection of changes in the levels of neurotransmitters and neuromodulators in live tissues.¹⁶

This technology leverages the selective binding of endogenous agonists to specific receptors that stabilize their conformation. For example, the GRAB_eCB2.0 sensor was recently engineered starting from hCB₁R by introducing a circularly permutated-green fluorescent protein (cpGFP) in its third intracellular loop.^17,18 Thus, GRAB_eCB2.0 was developed by screening for constructs with functional insertion sites of cpGFP, followed by individual randomized mutations of amino acids that increased the fluorescent signal in response to 2-AG specifically.¹⁹ Several laboratories reported that GRAB_eCB2.0 signal increases within seconds when exogenously applying 2-AG or AEA to cells in culture, as well as when endogenously stimulating eCB production in cells in culture, mouse brain slices, and behaving animals.^17,20–22

In this study, we measured GRAB_eCB2.0 signal in HEK293 cells in culture using live cell fluorescence microscopy and a high-throughput fluorescence plate reader assay. We found that 2-AG and SR141716 (SR1) formulated in buffer containing bovine serum albumin (BSA), a lipid binding protein known to assist eCB's activation of CB₁R, modulate GRAB_eCB2.0 fluorescent signal in HEK293 cells with potencies that closely mirror their reported activities at CB₁R.^17,23 Thus, we leveraged this experimental approach to characterize the pharmacological profile of eCB analogues and phyto-cannabinoids (phyto-CBs) at GRAB_eCB2.0.

Materials and Methods

Chemicals and reagents

2-AG (Cayman), 1-arachidonoylglycerol (1-AG; Cayman), Δ⁹-THC (NIDA Drug Supply Program), CP55,940 (CP; Cayman), SR1 (NIDA Drug Supply Program), AEA (Cayman), goat anti-CB₁R antibody (1:1000 for immunocytochemistry and 1:2500 for immunoblotting; gift from Dr. Ken Mackie); AlexaFluor 647 conjugated donkey anti-goat (1:1000; Invitrogen, CA); IRDye 800 CW conjugates donkey anti-goat (1:10,000; LI-COR).

Cloning

GRAB_eCB2.0 and mut-GRAB_eCB2.0 DNA were subcloned into an AM/CBA-WPRE-bGH plasmid using the BamHI and EcoRI restriction sites. The plasmid was purified (Purelink HiPure Plasmid Maxiprep Kit; Invitrogen) from transformed Stellar Competent Cells (Takara Bio, Inc., Japan). The DNA was sequenced and verified (CLC Sequence Viewer 8) before use in transfection.

Cell culture

HEK293 cells were grown in Dulbecco's modified Eagle medium (DMEM; supplemented with 10% fetal bovine serum and 1% penicillin/streptomycin) at 37°C and 5% CO₂. To passage cells for experiments, a confluent 10 cm plate of cells was detached by incubating with 0.25% Trypsin- ethylenediaminetetraacetic acid (EDTA) for 2–3 min at 37°C, adding 4–5 mL of supplemented DMEM and using gentle pipetting to remove any cells still attached, and then added to a new plate with fresh supplemented DMEM. Cells were passaged every 3–4 days, and for no more than 25 passages.

Transfection

All transfections were performed by incubating DNA with the transfection reagent polyethylenimine (PEI; 25K linear, Polysciences 23966) in a 1:3 ratio in serum-free DMEM, incubating for 20–30 min. The DNA/PEI mixture was then added to cells in a dropwise manner without changing the growth media. Cells were transfected when reaching >50% confluent and were incubated for 24 h after transfection before harvesting or using for GRAB_eCB2.0 assays.

Immunocytochemistry

Glass coverslips (Fisher Scientific; 12-545-82) in a six-well plate were coated with poly-d-lysine (50 ng/mL; Sigma-Aldrich; P6407) for 1–2 h at 37°C, after which the coverslips were washed three times with sterile water and one time with DMEM. HEK293 cells were detached and resuspended in supplemented DMEM as described previously, plated at a density of 100,000 cells/well, and were transfected after 24 h with 0.75 μg DNA.

Twenty hours after transfection, media was removed, and cells were fixed with 4% paraformaldehyde in phosphate-buffered saline (PBS) (Alfa Aeser) for 20 min at room temperature. After fixation, cells were washed five times with PBS and permeabilized and blocked with 0.1% saponin (Sigma-Aldrich) made and 1% BSA in PBS for 30 min at room temperature. Cells were incubated in goat anti-CB₁R antibody (1:1000) overnight at 4°C. Cells were then washed with PBS six times and incubated in AlexaFluor647-conjugated donkey anti-goat secondary antibody (Invitrogen; 1:1000) for 1 h at room temperature. Cells were washed with PBS six times, air dried overnight, and mounted using ProLong Diamond Antifade Mountant with 4′,6-diamidino-2-phenylindole (ThermoFisher; P36966). Images were captured using a LeicaSP8X confocal microscope and a 40×oil objective.

Western blotting

HEK293 cells were plated as described previously at a density of 500,000 cells/well in a six-well plate 24 h after plating and were transfected the following day with 0.75 μg DNA. Twenty hours after transfection, cells were washed three times with ice-cold PBS, and in the last wash cells were harvested with a cell scraper and pelleted by centrifuging at 500 g for 10 min. The supernatant was discarded, and cell pellets kept at −80°C until further use. Thus, pellets were thawed on ice, resuspended in lysis buffer (25 mM HEPES [pH 7.4], 1 mM EDTA, 6 mM magnesium chloride, and 0.5% CHAPS), Dounce homogenized on ice (20–30 strokes), and incubated on a rotator at 4°C for 1 h. Homogenates were then centrifuged at 700 g for 10 min at 4°C, supernatant was collected, and protein concentration of supernatant was determined using a DC protein assay.

Samples were then mixed with 4×Laemmli sample buffer containing 10% β-mercaptoethanol and incubated at 65°C for 5 min. Twenty-five micrograms of protein were loaded onto a 10% polyacrylamide gel and transferred to polyvinylidene difluoride membrane. After transfer, membranes were washed once with Tris-buffered saline (TBS) and incubated in blocking buffer (5% BSA in TBS) for 1 h at room temperature, followed by incubation in goat anti-CB₁R antibody (1:1000) overnight at 4°C. After incubation in primary antibody, membranes were washed with TBS with 0.05% Tween-20 (TBST) three times, 10 min each. Membranes were then incubated in IRDye 800 CW conjugates donkey anti-goat (1:10,000) for 1 h at room temperature. Blots were then washed with TBST three times, 10 min each followed by three washes with TBS, 10 min each. Fluorescent signal was detected using a Chemidoc MP (Biorad). Primary antibodies were diluted in blocking buffer. Secondary antibodies were diluted in 1:1 TBS:Odyssey blocking buffer (LI-COR; 927-50000).

Live-cell imaging

Glass bottom cell culture plates (MatTek; P35G-1.5-14-C) were coated with poly-d-lysine (50 ng/mL; Sigma-Aldrich; P6407) for 1–2 h at 37°C, after which the poly-d-lysine was removed, and coverslips were washed three times with sterile water and one time with DMEM. HEK293 cells were detached and resuspended in supplemented DMEM as described previously, counted using a hemocytometer, plated (250,000 cells/well), and were transfected after 24 h with 0.75 μg DNA.

Twenty-four hours after transfection, cells were imaged at room temperature after replacing media with PBS. The plates were transferred to the microscope (Leica SP8X) and cells were imaged using a 40×oil objective with the following settings: 485 excitation and 525 emission wavelength, 5% laser power, HyD hybrid detector, and a scan speed of 200 lines Hz (0.388 frames per second) with bidirectional scanning. All agents were prepared in PBS supplemented with 1 mg/mL BSA and added directly into buffer of cell on the microscope stage (final concentration BSA=0.1 mg/mL).

Ninety-six-well plate reader assay

Clear-bottom, black 96-well plates (USA Scientific; 5665-5087) were coated with poly-d-lysine (50 ng/mL; Sigma-Aldrich; P6407) for 1–2 h at 37°C, and coverslips were washed three times with sterile water and one time with DMEM. HEK293 cells were detached using trypsin, resuspended in supplemented DMEM as described previously, plated (20,000 cells/well), and transfected after 24 h with 0.1 μg DNA and 0.3 μg of PEI in 10 μL of serum-free DMEM. Twenty-four hours after transfection, media were replaced with 90 μL PBS. Cells were incubated at room temperature for 20 min, and then a 1 min baseline fluorescent signal reading was obtained using a fluorescent plate reader (i.e., 485 excitation and 525 emission filter settings with a 515 nm cutoff, and a speed of 1 reading every 20 sec). Immediately after baseline reading, 10 μL of agents were formulated in 1 mg/mL BSA and PBS were immediately added into buffer in wells.

Approximately 2 min after addition of treatment, the fluorescent signal of the plate was read for up to 30 min. For pretreatment with SR1, this antagonist was prepped in PBS (room temperature) and added into the media. All conditions were tested in triplicate in each experiment.

Statistical methods

All GRAB_eCB2.0 fluorescent signals are expressed as ΔF/F₀ as calculated by MATLAB for the high-throughput fluorescence assay or FIJI ImageJ for live cell confocal imaging. For live cell imaging, changes in GRAB_eCB2.0 signal were quantified by averaging baseline fluorescence for each cell (F₀: average of fluorescent signal averaged over 30 sec, ∼30 sec before agonist treatment) and calculating relative fold-change in fluorescent signal (ΔF/F₀) by subtracting the signal at every time point measured (to construct a time course) or at a specific time point (i.e., relative change 1 min post-treatment) by F₀ and then dividing by F₀.

For the 96-well plate assay, changes in GRAB_eCB2.0 signal were quantified by averaging baseline fluorescence for each well (F₀: fluorescent signal averaged over 1 min, ∼2 min before start of agonist treatment) and calculating relative fold-change in fluorescent signal (ΔF/F₀) by subtracting the signal at every time point measured (to construct a time course) or at a specific time point (i.e., relative change 1 min post-treatment) by F₀ and then dividing by F₀.

Since every condition was tested in triplicate, the ΔF/F₀ values for all three technical replicates for each condition were averaged to obtain the final ΔF/F₀ used in the analysis. GRAB_eCB2.0 activation was determined by calculating ΔΔF/F₀ between specific time points (i.e., time=0 and peak signal). This calculation involved first calculating the ΔF/F₀ at each time point, then subtracting the two values. To facilitate a data analysis pipeline, we developed a MATLAB R2021a algorithm that averages the fluorescent signal value of each well over time, for multiple experiments, and at select timepoints (see the following link for the code: https://github.com/StellaLab/StellaLab.git). Data are given as mean + s.e.m. and significance was determined by running a two-way analysis of variance with Dunnett's multiple comparison test using GraphPad Prism.

Results

Real-time activation and antagonism of GRAB_eCB2.0 fluorescent signals in HEK293 cells in culture: live cell microscopy

HEK293 cells in culture were transfected with eCB2.0 DNA plasmid constructs containing the chimeric cytomegalovirus-chicken β-actin promoter to drive expression.^24–28 GRAB_eCB2.0 expression was confirmed by western blot using a goat polyclonal antibody developed against C-terminal amino acids of CB₁R that remained unchanged in GRAB_eCB2.0 (Fig. 1A and Supplementary Fig. S1 for amino acid sequence alignment).¹⁷ Fluorescence confocal microscopy analysis of GRAB_eCB2.0 expression in fixed HEK293 cells using the CB₁R antibody showed abundant expression in many cells, consistent with transient transfection approaches (Fig. 1B).

FIG. 1.

Agonist triggered changes in GRAB_eCB2.0 fluorescent signal in HEK293 cells detected by live-cell confocal microscopy. HEK293 cells were transfected with eCB2.0 DNA plasmid and GRAB_eCB2.0 expression was measured by western blot and ICC, and changes in fluorescent signal measured by live-cell confocal microscopy when treated with the CB₁R agonists 2-AG, AEA, or CP. (A) Detection of GRAB_eCB2.0 expression by western blot using lysates from HEK293 cells transfected with pcDNA3, myc-CB₁R (positive control), CBA plasmid, or GRAB_eCB2.0. Both CB₁R and GRAB_eCB2.0 were detected using an antibody against the CB₁R. Loading control: Ponceau stain. (B) Detection of GRAB_eCB2.0 (i) or mut-GRAB_eCB2.0 (ii) expression by ICC of fixed and permeabilized HEK293 cells transfected with eCB2.0 DNA plasmid (i), mut-eCB2.0 DNA plasmid (ii), or CBA plasmid (iii). Scale bar=20 μm (inset scale bar=20 μm). (C) Schematic of live cell confocal imaging of HEK293 cells: cells were transfected with eCB2.0 DNA plasmid (1); 24 h later, cell growth media was exchanged for imaging PBS buffer and cells were placed on a confocal microscope (2); changes in GRAB_eCB2.0 fluorescent signal was determined by measuring fluorescent signal during baseline, spiking in treatments formulated with BSA (0.1%, 10×), and lastly spiking in SR1 (3 and 4). (D) Live cell images of GRAB_eCB2.0-expressing HEK293 cells during baseline recording (i, iii, and v) and after 60 sec of treatment with 2-AG (1 μM; ii), AEA (10 μM; iv), or CP (1 μM; vi). Arrows indicate cells with different levels of fluorescent signal to show heterogeneous expression of GRAB_eCB2.0. Scale bars=40 μm (inset scale bar=20 μm). (E, F) Change in plasma membrane GRAB_eCB2.0 fluorescent signal following 1- and 5-min treatment with 2-AG (1 μM; E) or CP (1 μM; F), as compared to baseline (basal) fluorescent signal. N=23 cells from 3 independent experiments. (G) Time courses of GRAB_eCB2.0 activation (ΔF/F₀) following treatment with 2-AG (1 μM), AEA (10 μM), and CP (1 μM) as measured by live-cell confocal microscopy. (H) Effect of SR1 (2 μM) on agonist-stimulated increase in GRAB_eCB2.0 fluorescent signal. Shaded area represents s.e.m. n=42–66 cells from 3 to 5 independent experiments. 2-AG, 2-arachidonoylglycerol; AEA, arachidonoyl ethanolamine; BSA, bovine serum albumin; CB₁R, cannabinoid 1 receptor; CP, CP55,940; ICC, immunocytochemistry; PBS, phosphate-buffered saline; SR1, SR141716.

We used live-cell confocal microscopy (line scanning frequency=200 Hz) to establish the time-course of changes in the dynamics of GRAB_eCB2.0 signal in response to 2-AG (the ligand was used to develop this sensor) by adding a 10×concentration formulated in BSA directly into PBS buffer inside the imaging chamber (Fig. 1C). Figure 1D_i,ii shows that HEK293 cells exhibited low, yet clearly detectable basal fluorescent signal at the plasma membrane (treated with vehicle control, dimethylsulfoxide 0.1%); and that 2-AG (1 μM, 60 sec) increased this signal. Similarly, AEA (10 μM, 60 sec) and the potent, artificial CB₁R agonist CP (1 μM, 60 sec) increased GRAB_eCB2.0 signal at the plasma membrane (Fig. 1D_iii,iv). Of note, both basal fluorescent signal and all agonist-triggered increases in GRAB_eCB2.0 signal reached different levels in HEK293 cells as expected by heterologous expression (e.g., see Fig. 1D_iv arrowheads).

It is important to note that significant levels of the GRAB_eCB2.0 protein were also present in the intracellular compartment of ∼70% of the HEK293 cells (see arrow in Fig. 1B, insert, and Supplementary Fig. S2A, B). In these cells, 2-AG (1 μM) significantly increased the intracellular fluorescent signal; however, unlike the increase in the plasma membrane fluorescent signal, the intracellular signal was increased to a lesser extent (e.g., ∼124% increase in peak fluorescent signal after 5 min of treatment compared with 255% increase of the peak signal at the plasma membrane in the same cells; see Fig. 1E, F and Supplementary Fig. S2E).

Furthermore, maximal increases in intracellular fluorescent signal occurred after ∼5 min of treatment, compared with the plasma membrane signal that peaked within 1–2 min (see Fig. 1E, F). This delay may be a result of the time it takes 2-AG to travel to these intracellular compartments. Similarly, CP induced increases in both the plasma membrane and intracellular fluorescent signal (∼180% and 110% increase in peak fluorescent signal, respectively, after 5 min treatment) (Supplementary Fig. S2C, D, and F), indicating that activation of intracellular GRAB_eCB2.0 signal is likely not ligand selective. This indicates that activation of intracellular GRAB_eCB2.0 signal is likely not ligand selective specific and despite having a smaller activation and differing kinetics than the plasma membrane signal, it is detectable and has the potential be relevant for studying intracellular eCB signaling.²⁹

Analysis of the time course of the increase in GRAB_eCB2.0 signal (ΔF/F₀) at the plasma membrane triggered by these agonists indicated that each agonist increased GRAB_eCB2.0 signal within seconds and that these responses plateaued after ∼60 sec (Fig. 1G). Figure 1H shows that the CB₁R antagonist, SR1 (2 μM), reduced these responses within 100 sec, and this reduction reached levels below basal. Calculation of the onset of activation (i.e., slope of the fluorescent signal increase within the first 20 sec after start of treatment), the magnitude of the response (i.e., peak fluorescent signal and area under the curve), and the decay after SR1 treatment (i.e., τ value following start of SR1 treatment) showed that each agonist had significantly different pharmacological profiles at GRAB_eCB2.0 (Table 1).

Table 1.

Parameters of GRAB_eCB2.0 Activation and Antagonism

PD parameter	Units	CB1R agonists
		2-AG	CP	AEA
Initial response: slope	(×10–2 ΔF/F0/sec)	6.6	4.5	10.9
Maximal response: peak	(ΔF/F0)	2.5	1.7	2.7
Overall response	(Area under the curve)	649	453	760
Antagonism response: τ	(sec)	42.7	24.3	36.8

Calculation of the activation onset (i.e., the slope of the fluorescent signal increase within the first 20 sec after start of agonist treatment), the magnitude of the response (i.e., peak response and area under the curve), and the decay following SR1 treatment (i.e., τ value following treatment). Analysis of data presented in Figure 1, n=42–66 cells from 3 to 5 independent experiments.

2-AG, 2-arachidonoylglycerol; AEA, arachidonoyl ethanolamine; CB₁R, cannabinoid 1 receptor; CP, CP55,940; PD, pharmacodynamic; SR1, SR141716.

Specifically, (1) AEA triggered a 1.7- and 2.2-fold faster initial response compared with 2-AG and CP, respectively; (2) 2-AG and AEA reached similar peak responses, and these were 47–58% greater than the CP maximal response; (3) 2-AG and AEA triggered a 1.4–1.6-fold overall greater response (area under the curve) compared with CP; and (4) SR1 antagonized the CP and AEA responses with faster decay than the 2-AG response (24 sec <37 sec <43 sec, respectively). Together, these results indicate that GRAB_eCB2.0 expressed by HEK293 cells are activated by the CB₁R agonists 2-AG=AEA > CP, and these responses are differentially antagonized by SR1.

High-throughput measures of GRAB_eCB2.0 fluorescent signal in HEK293 cells in culture

To further define the pharmacological profile and dynamics of GRAB_eCB2.0 signal, we developed and validated a fluorescent plate reader assay (96 wells, 3 Hz scanning frequency) (Fig. 2A). Figure 2B and C shows that 2-AG and AEA induced concentration-dependent increases in GRAB_eCB2.0 signals that were detected at the 0 sec timepoint (i.e., when the first stimulation fluorescent signal was measured). For example, both the 10 and 100 nM 2-AG responses at 0 sec reached a GRAB_eCB2.0 signals value of 0.19 and 0.96 ΔF/F₀ over basal, respectively; and both these responses reached an initial inflection point at 0.66 and 0.33 min, respectively (Fig. 2B, see color coded arrows). Of note, 2-AG at 3 μM induced the strongest response that reached an inflection point at ∼0.66 min, whereas 2-AG at 10 μM induced a response that had already reached a maximum plateau response at the 0 min timepoint (Fig. 2B).

FIG. 2.

2-AG, AEA, and CP differentially activate GRAB_eCB2.0: high-throughput fluorescence assay. (A) Schematic of high-throughput fluorescent plate reader assay: HEK293 cells were plated in a 96-well plate and transfected with eCB2.0 DNA plasmid; 24 h post-transfection, growth media was replaced with imaging PBS buffer and incubated for 20 min. For the last 2 min of this incubation, the plate was placed in the fluorescent plate reader and a basal fluorescent signal was measured. Treatments were added immediately after the basal reading and the pate was reinserted in the plate reader and fluorescent signal was recorded for 5 min. (B, C, F, G and I) Kinetics of GRAB_eCB2.0 activation (ΔF/F₀) following treatment with increasing concentrations of 2-AG (B), AEA (C), CP (F), SR1 (G), and AA and Gly (I). Arrows represent inflection points and dotted lines represent initial change in response (slope between t=0 and time at inflection). (D) Initial change in ΔF/F₀ response elicited by increasing concentrations of 2-AG, AEA, or CP shown in (B, C, and F). (E) Concentration-dependent responses and EC₅₀ values of 2-AG, AEA, and CP at inducing GRAB_eCB2.0 fluorescent signal and of SR1 at reducing GRAB_eCB2.0 fluorescent signal as determined by averaging ΔF/F₀ between 4 and 5 min. (H) 2-AG (1 μM), CP (1 μM), and AEA (10 μM)-induced increases in GRAB_eCB2.0 fluorescent signal was reduced by SR1 (300 nM) and absent in HEK293 cells expressing mut-GRAB_eCB2.0. Data represent mean ΔF/F₀ between 4 and 5 min. Shaded areas on time-course plots and error bars on histograms represent s.e.m. Statistics: (D) **p<0.01 significantly different from basal (two-way ANOVA followed by Dunnett's). (H) ***p<0.001 significantly different from corresponding CTR treatment (two-way ANOVA followed by Dunnett's). n=9–50 independent experiments performed in triplicate. AA, arachidonic acid; ANOVA, analysis of variance; CTR, control; EC₅₀, 50% effective concentration; Gly, glycerol.

Thus, 2-AG rapidly activated GRAB_eCB2.0 in a concentration-dependent manner as calculated by its initial response (i.e., slope: ΔΔF/F₀ between 0 min and the initial inflection time point) (Fig. 2D). To calculate EC₅₀ values, we averaged the GRAB_eCB2.0 signals between 4 and 5 min, and found 85 nM for 2-AG, a value that is consistent with its reported potency at CB₁R (e.g., EC₅₀=12–100 nM for inhibiting cyclic adenosine monophosphate [cAMP] production) (Fig. 2E).^30,31 AEA also induced a concentration-dependent and rapid increase in GRAB_eCB2.0 signal that was detected at the 0 sec timepoint, but only reached a significant initial response starting at 30 nM, and a maximum plateau response at 10 μM (Fig. 2C, D). Thus, AEA induced a concentration-dependent increase in GRAB_eCB2.0 with an EC₅₀=815 nM, an activity that is ∼10-fold less potent than AEA's potency at CB₁R (e.g., EC₅₀=69 nM for inhibiting cAMP production) (Fig. 2E).^31–33

To further characterize these high-throughput measures, we tested the effect of CP and SR1 on GRAB_eCB2.0 signal. Figure 2F shows that CP induced a concentration-dependent increase in GRAB_eCB2.0 signal that was detected at 0 sec and reached an initial inflection time point within ∼1.33 min and an EC₅₀=82 nM, which is less potent than CP's potency at CB₁R (e.g., EC₅₀=1–3 nM as determined by inhibition of cAMP production) (Fig. 2D, E).^34–36 As expected, SR1 induced a rapid and concentration-dependent decrease in GRAB_eCB2.0 signal that was first detected at 0 sec and reached a plateau below basal levels starting at 30 nM for at least 5 min (Fig. 2G). The IC₅₀ for SR1 measured between 4 and 5 min was 3.3 nM, a value that mirrors SR1's potency at CB₁R (IC₅₀=5.6–7.8 nM as determined by SR1's inhibition of CP-induced decrease of cAMP levels) (Fig. 2E).^34,36–38

Pretreatment of HEK293 cells with SR1 (300 nM, 20 min) followed by treatment with 2-AG, AEA, and CP significantly reduced GRAB_eCB2.0 activation (Fig. 2H). Figure 2H also shows that 2-AG, AEA, and CP failed to elicit increases in fluorescent signals in HEK293 cells expressing the mut-GRAB_eCB2.0, which has a similar expression profile as GRAB_eCB2.0 as determined by fluorescence confocal microscopy (Fig. 1B). Thus, mut-GRAB_eCB2.0, which has a phenylalanine 177 to alanine mutation in the region within the orthosteric binding pocket to impair ligand binding, represents a valid negative control.^39,40 As previously reported, the products of 2-AG hydrolysis, arachidonic acid, and glycerol, did not influence the GRAB_eCB2.0 signal (Fig. 2I).¹⁷

Together, these results provide strong support for use of this high-throughput experimental approach to study the pharmacology and dynamics of changes in GRAB_eCB2.0 fluorescence when expressed by cells in culture, and show that (1) pronounced increases and decreases in GRAB_eCB2.0 signal are reproducibly detected using a 96-well plate-reader format and promptly spiking agents in media, (2) 2-AG and SR1 modulate GRAB_eCB2.0 signal with EC₅₀ values comparable with their potencies at the CB₁R, and (3) AEA and CP also increase GRAB_eCB2.0 signal, although with 2–10-fold lower potency than at CB₁R.

The EC₅₀ values of 2-AG and AEA for GRAB_eCB2.0 activation measured here are lower than their previously reported values in cell culture model systems (i.e., 2-AG=3.1–9.0 μM and AEA=0.3–0.8 μM), which is likely owing to our inclusion of BSA in the buffer to facilitate solubility and interaction with GRAB_eCB2.0.^17,41 To confirm the effect of BSA on GRAB_eCB2.0 sensitivity with 2-AG, we tested how BSA affected the 2-AG–induced increase in GRAB_eCB2.0 signal. Figure 3A–D shows that increasing concentrations of BSA alone did not influence baseline GRAB_eCB2.0 signal, yet BSA (0.1 mg/mL) increased the 2-AG (10 and 100 nM) responses by 7.3- and 4-fold, respectively, when measured 10 min following treatment.

FIG. 3.

BSA enhances 2-AG–induced activation of GRAB_eCB2.0. (A) Effect of increasing concentrations of BSA on GRAB_eCB2.0 activation (ΔF/F₀). (B, D) Kinetics of GRAB_eCB2.0 activation (ΔF/F₀) following treatment with increasing concentrations of 2-AG diluted in PBS alone (−BSA) or in 0.1 mg/mL BSA (+BSA). (C) Effect of BSA (0.1 mg/mL) on 2-AG–induced increases in GRAB_eCB2.0 fluorescent signal (ΔF/F₀ after 10 min of treatment). Statistics: ***p<0.001 significantly different (two-way ANOVA followed by Sidak's). n=3 independent experiments performed in triplicate. Error bars represent s.e.m.

BSA (0.1 mg/mL) had a lesser effect on the 1 and 10 μM 2-AG–induced increase in GRAB_eCB2.0 signal, enhancing the ΔF/F₀ by 5% and 7% after 10-min treatment, respectively, likely because these are close to saturating concentrations at GRAB_eCB2.0 (Fig. 3C, D). This confirms that BSA increases 2-AG's availability at the GRAB_eCB2.0 sensor and its ability to detect mid-to-low nanomolar concentrations of 2-AG.

Pharmacological activity of 2-AG analogues at GRAB_eCB2.0

We leveraged the high-throughput approach to determine whether 2-LG and 2-OG, which are produced by cells concomitantly to 2-AG, change GRAB_eCB2.0 signals when expressed in HEK293 cells.^13,14 Specifically, although 2-LG (18:2) and 2-OG (18:1) are lipid analogues of 2-AG (20:4) and are produced by the same biosynthetic and metabolic pathways as 2-AG, they are likely to activate CB₁R signaling with much lower potency and efficacy than 2-AG. For example, 2-LG partially activates CB₁R (EC₅₀=16.6 μM for arrestin recruitment) or antagonizes CB₁R depending on the model system, whereas there is still no evidence that 2-OG might also activate CB₁R.^13,14 Figure 4A and B shows that both 2-LG and 2-OG induced a concentration-dependent increase in GRAB_eCB2.0 signal that was detected at 0 sec and that these responses rapidly reached their initial inflection time points within 1 min.

FIG. 4.

2-LG, 2-OG, and 1-AG activate GRAB_eCB2.0. HEK293 cells were transfected with eCB2.0 DNA plasmid and changes in fluorescent signal measured by high-throughput fluorescence assay. (A, B, E) Kinetics of GRAB_eCB2.0 activation (ΔF/F₀) following treatment with increasing concentrations of 2-LG (A), 2-OG (B), and 1-AG (E). Shaded areas in time courses and error bars represent s.e.m. Arrows represent inflection points and dotted lines represent initial change in response (slope between t=0 and time at inflection). (C) Initial response (slope between time=0 and inflection point) of ΔF/F₀ response. (D) Concentration-dependent responses and EC₅₀ values of 2-LG, 2-OG, 1-AG at inducing GRAB_eCB2.0 fluorescent signal as determined by averaging ΔF/F₀ between 4 and 5 min. (F) 2-LG (1 μM), 2-OG (1 μM), and 1-AG (1 μM)-stimulated increases in GRAB_eCB2.0 fluorescent signal are reduced by SR1 (300 nM) and absent in HEK293 cells expressing mut-GRAB_eCB2.0. Statistics: (C) **p<0.01 significantly different from basal (two-way ANOVA followed by Dunnett's). (F) ***p<0.001 significantly different from corresponding CTR treatment (two-way ANOVA followed by Dunnett's). n=3–11 independent experiments performed in triplicate. 1-AG, 1-arachidonoylglycerol; 2-LG, 2-linoleoylglycerol; 2-OG, 2-oleoylglycerol.

Thus, 2-LG and 2-OG activated GRAB_eCB2.0 with increasingly rapid onset and EC₅₀=0.63 and 0.87 μM, respectively, indicating that 2-LG and 2-OG have higher potency when activating the GRAB_eCB2.0 sensor compared with their activation of CB₁R (Fig. 4C, D). We next tested 1-AG (20:4), which is a product of a nonenzymatic isomerization of 2-AG that is not endogenously produced by mammalian cells, but is commonly used to study the structure activity relationship of monoacylglycerols (MAGs) at CB₁R as its acyl chain length and saturation match that of 2-AG.^30,42,43 We found that 1-AG induced a concentration-dependent increase in GRAB_eCB2.0 signal that was detected at 0 sec, reached its initial inflection timepoint within 1 min, and had an EC₅₀=1.8 μM, which mirrors its potency at CB₁R (EC₅₀=1.45 μM for inhibiting cAMP production) (Fig. 4D, E).

The responses to these three MAGs were blocked by pretreatment with SR1 (300 nM) and absent in cells expressing the mut-GRAB_eCB2.0 (Fig. 4F). Thus, 1-AG, 2-LG, and 2-OG increase GRAB_eCB2.0 signal (rank order of potency: 2-LG >2-OG >1-AG; maximum efficacy: 1-AG >2-LG >2-OG). Of note, all three lipid analogues had lower potencies than 2-AG and AEA, indicating that the GRAB_eCB2.0 has greater sensitivity for these two eCBs compared with similar MAG lipids.

Determining THC and CBD's pharmacological activity at GRAB_eCB2.0

To extend the pharmacological characterization of GRAB_eCB2.0 using the high-throughput assay, we tested Δ⁹-THC, the principal psychoactive ingredient in Cannabis that activates CB₁R as a high-affinity partial agonist,³⁴ and Δ⁸-THC, which activates CB₁R with a comparable pharmacology as Δ⁹-THC but represents a minor product of most Cannabis strains and remains poorly characterized.⁴⁴ Δ⁹-THC increased GRAB_eCB2.0 signal in a concentration-dependent manner that was detected at 0 sec starting at 1 and 3 μM, and these responses reached their initial inflection points within 1 min (Fig. 5A). Of note, higher concentrations of Δ⁹-THC, that is, 10 and 30 μM, triggered a response that continuously increased for at least 5 min without apparent inflection (Fig. 5A).

FIG. 5.

Δ⁹-THC and Δ⁸-THC activate GRAB_eCB2.0, and CBD act as a negative allosteric modulator. HEK293 cells were transfected with eCB2.0 DNA plasmid and changes in fluorescent signal measured by high-throughput fluorescence assay. (A, B) Kinetics of GRAB_eCB2.0 activation (ΔF/F₀) induced by increasing concentrations of Δ⁹-THC (A) or Δ⁸-THC (B). Shaded areas in time courses and error bars represent s.e.m. Arrows represent inflection points and dotted lines represent initial change in response (slope between t=0 and time at inflection). (C) Concentration-dependent responses and EC₅₀ values of Δ⁹-THC and Δ⁸-THC at inducing GRAB_eCB2.0 fluorescent signal as determined by averaging ΔF/F₀ between 4 and 5 min. (D) Δ⁹-THC (10 μM) and Δ⁸-THC (10 μM)-stimulated increases in GRAB_eCB2.0 fluorescent signal are reduced by SR1 (300 nM) and absent in HEK293 cells expressing mut-GRAB_eCB2.0. (E, F) CBD (100 pM–10 μM) does not modulate GRAB_eCB2.0 fluorescent signal (F) and decreased the 2-AG (1 μM)-stimulated increase in GRAB_eCB2.0 fluorescent signal (IC₅₀ of 9.7 nM) (G). (H) Effect of CBD (1, 10, and 100 nM) on 2-AG concentration dependent activation of GRAB_eCB2.0 (ΔF/F₀). Statistics: (D) ***p<0.001 significantly different from corresponding CTR treatment (two-way ANOVA followed by Dunnett's). n=3–7 independent experiments performed in triplicate. CBD, cannabidiol; THC, tetrahydrocannabinol.

Figure 5B shows that Δ⁸-THC induced a concentration-dependent increase in GRAB_eCB2.0 signal that was detected at 0 sec starting at 1 μM, and that all higher concentrations continuously increased for at least 5 min without inflection. Δ⁹-THC and Δ⁸-THC activate GRAB_eCB2.0 with comparable EC₅₀ values as calculated between 4 and 5 min (1.6 and 2 μM, respectively; Fig. 5C). These two responses were blocked by pretreatment with SR1 (300 nM) and absent in cells expressing the mut-GRAB_eCB2.0 (Fig. 5D). Thus, Δ⁹-THC and Δ⁸-THC similarly increase GRAB_eCB2.0 signal, although with a 10–100-fold lower potency than their activation of CB₁R, and most of these responses at micromolar concentrations continuously increase for at least 5 min.

CBD acts as an NAM of CB₁R.^10,45 Accordingly, we found that CBD, at up to 3 μM, did not significantly modulate basal GRAB_eCB2.0 signal but did significantly reduce the 2-AG–induced increase in GRAB_eCB2.0 signal by 26% at 1 μM and with an IC₅₀=9.7 nM, values that are similar to CBD's activity at CB₁R (Fig. 5E, G). Thus, we calculated the Schild plot of this response and found a slope of 0.33, as expected for an NAM (Fig. 5H). These results indicate that GRAB_eCB2.0 retains the molecular mechanism that mediates the proposed NAM activity of CBD at CB₁R.

Comparing changes in GRAB_eCB2.0 fluorescent signals

We sought to compare key pharmacological and dynamic parameters that characterize changes in GRAB_eCB2.0 signal elicited by the agents that we tested in this study (chemical structures in Supplementary S6A). Figure 6A shows GRAB_eCB2.0 activation by each agonist applied at their EC₅₀ values results in different GRAB_eCB2.0 activation dynamics. Specifically, when analyzing the ΔΔF/F₀ of these responses between 3 and 5 min, we found that the responses induced by MAGs were decaying between 3 and 5 min and thus, had reached an earlier maximum response, whereas the GRAB_eCB2.0 signals induced by AEA, CP, and Δ⁹-THC continuously increased between 3 and 5 min and thus had not reached a maximum response within this time period (Fig. 6B).

FIG. 6.

Monoacylglycerols have distinct dynamics and pharmacological properties at GRAB_eCB2.0 compared with THC, CP, and AEA. HEK293 cells were transfected with eCB2.0 DNA plasmid and changes in fluorescent signal measured by high-throughput fluorescence assay. (A) Comparison of GRAB_eCB2.0 activation kinetics elicited by each ligand at a concentration that approaches their respective EC₅₀ values. Shaded areas in time courses and error bars represent s.e.m. (B) Comparison of the EC₅₀ of each ligand and their corresponding end response at a concentration that approached their respective EC₅₀ and measured between 3 and 5 min. Y axis positive value represent increase in fluorescent signal (i.e., AEA, CP and D⁹-THC) and negative values represent decrease in fluorescent signal (i.e., 2-AG, 1-AG, 2-LG and 2-OG). (C) Comparison of the maximum increase in GRAB_eCB2.0 fluorescent signal as expressed by area under the curve and as a percent of 3 μM 2-AG produced by each ligand. (D–F) Antagonism of each response by SR1 (100 and 300 nM).

Analysis of concentration responses indicated that 2-AG at 3 μM resulted in the most pronounced increase in GRAB_eCB2.0 signal as measured by area under the curve between 0 and 5 min, and thus we calculated the relative efficacy of each agent at the concentration that induced their maximal response as compared with the 2-AG response set at 100%. Figure 6C shows that the maximal GRAB_eCB2.0 responses to AEA (30 μM), 1-AG (10 μM), and CP (3 μM) reached 91%, 81%, and 71% of the 2-AG response, respectively. Figure 6C also shows that the maximal GRAB_eCB2.0 responses to 2-LG (10 μM) and 2-OG (10 μM) only reached 37% and 17% of the 2-AG response, respectively, and that Δ⁹-THC (30 μM) and Δ⁸-THC (30 μM) only reached 22% and 16% of the 2-AG response, respectively. Finally, we compared the antagonism of SR1 (100 and 300 nM) for each ligand.

Figure 6D and E shows that the 2-AG (1 μM), CP (1 μM), 2-LG (10 μM), as well as Δ⁹-THC (10 μM) and Δ⁸-THC (10 μM) were partially antagonized by SR1 100 nM and antagonized by >80% with SR1 300 nM. By contrast, 1-AG (10 μM) and 2-OG (10 μM) were strongly antagonized by SR1 at both 100 and 300 nM, and AEA (10 μM) was antagonized by only 65% by both 100 and 300 nM SR1. These results reveal differences in key pharmacological and dynamic parameters that describe changes in GRAB_eCB2.0 signal elicited by eCBs and CB₁R ligands.

Because GRAB_eCB2.0 activation by 2-AG and its analogues reached a peak maximal response followed by a decay and HEK293 cells endogenously express one or more enzymes capable of hydrolyzing eCBs,⁴⁶ we tested the involvement of enzymatic hydrolysis in this decay. Thus, we treated HEK293 cells with methoxy arachidonoyl fluorophosphonate (MAFP), a broad-spectrum inhibitor that targets multiple enzymes that hydrolyze 2-AG, including monoacylglycerol lipase (MAGL), α/β domain containing 6 (ABHD6), and ABHD12.⁴⁷ MAFP (10 nM) potentiated the overall 2-AG (100 nM) response (calculated using area under the curve) by 12.5% without significantly affecting the maximal response (Supplementary Fig. S6B, C). Furthermore, MAFP (10 nM) increased the decay time constant (τ) by 1.7-fold (Supplementary Fig. S6C). Together, these results indicate that serine-hydrolases endogenously expressed by HEK293 cells may control the activity of 2-AG at its targets, here GRAB_eCB2.0.

Discussion

Similarities and differences in GRAB_eCB2.0 and CB₁R pharmacology

GRAB_eCB2.0 was developed by screening for genetic constructs and individual randomized mutations starting with the CB₁R backbone to improve the change in fluorescent signal in response to 2-AG.¹⁹ The initial pharmacological characterization of GRAB_eCB2.0 was performed in the absence of BSA in the buffer and resulted in EC₅₀ values that are higher than their EC₅₀ values at CB₁R: 2-AG=3–9 μM, AEA=0.3–0.8 μM, CP=20 nM, and THC=2 μM.¹⁷ BSA is known to facilitate the solubility and interaction of CB agents with CB₁R.⁴¹ We found that inclusion of BSA in the buffer improves 2-AG's potency at GRAB_eCB2.0 to 85 nM, which is comparable with published EC₅₀ values at the CB₁R, but did not affect the activities of AEA, THC, and CP at GRAB_eCB2.0 compared with published values. This result suggests that BSA might preferentially facilitate the solubility and interaction of select CB agents with CB₁R and GRAB_eCB2.0, in this case 2-AG.

2-AG interacts with specific amino acids within the orthosteric binding site of CB₁R that are likely conserved in GRAB_eCB2.0. This is reflected in the similarity of 2-AG's potency at GRAB_eCB2.0 and its potency at the CB₁R, because the development process likely selected for a sensor where the binding and activation properties of 2-AG were retained, such as the conservation of amino acids 2-AG may interact with. Although there is an overlap between which CB₁R residues are predicted to interact with 2-AG, AEA, CP, and THC (such as F170, F177, W279, and F379 as determined by molecular docking simulations using the crystal structure of agonist-bound hCB₁R), other amino acid interactions may be ligand specific.^9,39,48 For instance, CP and AEA, but not 2-AG and THC, are predicted to interact with F268. The disparity in potency of AEA and CP at the sensor compared with their potency at the CB₁R indicates that mutating the CB₁R and incorporating the cpGFP to create GRAB_ecb2.0 may have impacted the structure of the sensor and subsequent conformation of the ligand binding pocket, resulting in ligand-specific effects on binding and potency that reflect the unique binding properties these agonists have at the native CB₁R.

Furthermore, the efficacy of these ligands at GRAB_eCB2.0 (rank order of maximal efficacy: 2-AG > AEA >1-AG > CP >> 2-LG >2-OG > Δ⁹-THC > Δ⁸-THC) indicates that GRAB_eCB2.0 expressed by cells in culture and mouse tissues reliably senses nanomolar changes in 2-AG levels, and micromolar amounts of AEA, eCB analogues, THC, and CP. However, owing to the differences in potency and maximal efficacy of these ligands, in model systems with a mixture of agonists (i.e., rodents treated with Δ⁹-THC), the GRAB_eCB2.0 may preferentially detect some ligands (2-AG and AEA) over others (Δ⁹-THC). Further interrogation of GRAB_eCB2.0's response to simultaneous treatment by multiple ligands is needed to decipher the meaning of changes in GRAB_eCB2.0 signal in these in vivo model systems.

We detected the presence of intracellular GRAB_eCB2.0 and showed that agonist treatment increases its signal, although to a lesser extent than GRAB_eCB2.0 at the plasma membrane. The possibility that GRAB_eCB2.0 undergoes internalization was indirectly tested by Dong et al. in their original publication by quantifying β-arrestin2 binding.¹⁷ Significantly, they found no increase in β-arrestin2 binding when activating GRAB_eCB2.0 with CB agonists, in contrast to agonists-induced CB₁R internalization in HEK293 cells that depend on β-arrestin2. This suggests that extracellular agonists slowly cross the plasma membrane and may increase intracellular GRAB_eCB2.0 signal, although to a lesser extent and with slower kinetics.⁴⁹ Therefore, it is unlikely that the sensor will undergo significant internalization, particularly during short incubation times (i.e., 5 min).

GRAB_eCB2.0 pharmacology relative to abundance of eCB analogues and CB agents

Diacylglycerol lipase (DAGL) produces several MAGs in addition to 2-AG that exhibit significant agonist-like activity at CB₁R, such as 2-LG.^13,14 Furthermore, MAGL and ABHD6 hydrolyze several MAGs in addition to 2-AG, including 2-LG and 2-OG.⁴⁷ However, in the mouse brain, 2-LG is 10 times less abundant than 2-AG, whereas 2-OG is ∼2–3 times more abundant compared with 2-AG, and in both cases, the mechanism of how 2-LG and 2-OG production is stimulated is not well understood.^50,51 Although there is potential that under physiological conditions, GRAB_eCB2.0 may detect 2-LG and 2-OG, considering that 2-LG and 2-OG activate the GRAB_eCB2.0 with potencies ∼10 times lower than that of 2-AG, we suggest that the activity-dependent increases in GRAB_eCB2.0 signal measured in cells in culture and mouse tissues that are blocked by DAGL inhibitors and increased by MAGL/ABHD6 inhibitors are more likely to involve change in 2-AG levels than changes in 2-LG and 2-OG levels.

Δ⁹-THC increases GRAB_eCB2.0 signal with an EC₅₀ that is ≈100-fold higher than its EC₅₀ at CB₁R when measuring inhibition of cAMP production. Since intraperitoneal injections of Δ⁹-THC (5 mg/kg) in adult mice results in ≈10 ng/mL of THC in mouse brain (31.8 nM),^52,53 we conclude that GRAB_eCB2.0 expressed in mouse brain will therefore be able to sense intraperitoneal injections of Δ⁹-THC and Δ⁸-THC and enable studying their pharmacokinetic profile and how this correlates with changes in behavior, or when either ligand arrives at a particular circuit or brain region in real time.

GRAB_eCB2.0 plateau versus decay dynamics

When applied at a concentration that approaches their EC₅₀ values, CB agents increased GRAB_eCB2.0 signal with different kinetics: 2-AG, 1-AG, 2-LG, and 2-OG triggered rapid increases in fluorescent signal that were followed by slow decays, whereas AEA, CP, Δ⁹-THC, and Δ⁸-THC triggered progressive increases in fluorescent signal that did not reach peak response within 5 min. The broad-spectrum serine hydrolase inhibitor MAFP reduced this decay, indicating that HEK293 cells express one or more hydrolases that metabolize MAGs,^54,55 and underlie the decay response of 2-AG, 1-AG, 2-LG, and 2-OG. HEK293 cells also lack CYP2C9 and CYP3A4 enzymes that metabolize Δ⁹-THC and Δ⁸-THC. Thus, Δ⁹-THC and Δ⁸-THC are likely not enzymatically degraded by HEK293 cells in culture and induced a steadily increasing GRAB_eCB2.0 signal.

NAM activity of CBD at GRAB_eCB2.0

Allosteric binding sites are distinct protein domains from orthosteric sites that bind small molecules and either increase or decrease orthosteric site-mediated changes in protein conformations and activities. Early in vitro and in vivo studies showed that CBD reduces CB₁R signaling at concentrations well below its reported affinity (Ki) to the orthosteric agonist site of CB₁R, providing the initial evidence that CBD acts as a NAM at this receptor.^56–58 Accordingly, in neurons in cell culture, CBD reduces the efficacy and potency of 2-AG and THC at increasing CB₁R signaling and inhibits eCB-mediated synaptic plasticity without influencing basal neurotransmission.^10,59,60 Mutagenesis of CB₁R indicated that several N-terminal residues of the CB₁R, namely Cys98, Cys107, and Met1, interact with CBD when it occupies the putative allosteric site of the CB₁R that CBD may target, and that this allosteric site overlaps with the orthosteric site that is near the second extracellular loop.^10,11 In silico modeling of CB₁R with an intact N-terminus revealed a potential binding pocket for NAMs of CB₁R in close proximity to its N-terminus, one of the longest among class A G protein–coupled receptors, and molecular docking studies suggest that binding to this site may result in a change in the three-dimensional structure of the orthosteric binding site and thus in THC's and 2-AG's potencies.^11,61

Although these results suggest that CBD inhibits CB₁R signaling by directly interacting with an allosteric binding site on this target, direct demonstration of CBD binding to CB₁R is still needed. We found that CBD does not affect baseline GRAB_eCB2.0 signal but reduces 2-AG's activity at GRAB_eCB2.0, a result consistent with the premise that the molecular mechanism involved in mediating CBD's allosteric modulation of CB₁R remains functional in GRAB_eCB2.0. Accordingly, mutagenesis optimization of CB₁R constructs to generate GRAB_eCB2.0 did not significantly affect its N-terminus (Supplementary Fig. S1). Our results suggest that structural and mechanistic comparisons of CBD activity at CB₁R and GRAB_eCB2.0 might help us better understand that molecular mechanism is involved in the allosteric modulation of CB₁R, that is, how allosteric ligands produce a distinctive receptor conformation with unique signaling and therapeutic value.

Conclusion

In this report, we show that 2-AG increases GRAB_eCB2.0 fluorescent signal as a full agonist and with an EC₅₀ similar to its activity at CB₁R, and that GRAB_eCB2.0 responds to AEA, 2-LG, and 2-OG with EC₅₀ values that are higher than their EC₅₀ values at CB₁R. Considering the lower amount of AEA, 2-LG, and 2-OG produced by cells, our results suggest that activity-dependent increases in GRAB_eCB2.0 fluorescent signal measured in cells in culture and mouse tissues will mainly reflect change in 2-AG levels, especially when this response is blocked by inhibitors of 2-AG production and inactivation. We also show that SR1 blocks GRAB_eCB2.0 fluorescent signal with an IC₅₀ similar to its reported potency at CB₁R, and that this leads to levels below baseline fluorescent signal. Thus, GRAB_eCB2.0 exhibits basal fluorescence and SR1 may act as an inverse-like agonist and represents a useful pharmacological tool to validate GRAB_eCB2.0 functionality when expressed by various model systems.

THC and CP increase GRAB_eCB2.0 fluorescent signal with EC₅₀ values lower than their activity at CB₁R indicating that only high brain concentration of these agents will be detected in cell culture and mouse tissue model systems. CBD reduces the 2-AG–induced increase in GRAB_eCB2.0 fluorescent signal but not its basal fluorescent signal, suggesting that the molecular mechanism of CBD allosterism present in CB₁R is maintained in GRAB_eCB2.0. Thus, GRAB_eCB2.0 provides an opportunity to study how changes in THC and CBD concentration and coactivity at CB₁R might occur in cell culture and mouse tissue model systems. Our results presented here outline the pharmacological profile and activation dynamics of GRAB_eCB2.0 to improve the interpretation of changes in its fluorescent signal when expressed in various model systems.

Data Availability Statement

Article available on bioRxiv; doi: 10.1101/2023.03.03.531053

Abbreviations Used

1-AG

1-arachidonoylglycerol

2-AG

2-arachidonoylglycerol

ABHD6

α/β domain containing 6

AEA

arachidonoyl ethanolamine

BSA

bovine serum albumin

cAMP

cyclic adenosine monophosphate

CB

cannabinoid

CBD

cannabidiol

CB₁R

cannabinoid 1 receptor

CP

CP55,940

cpGFP

circularly permutated-green fluorescent protein

DMEM

Dulbecco's modified Eagle medium

DAGL

diacylglycerol lipase

eCBs

endocannabinoids

EC₅₀

50% effective concentration

EDTA

ethylenediaminetetraacetic acid

IC₅₀

50% inhibitory concentration

2-LG

2-linoleoylglycerol

MAFP

methoxy arachidonoyl fluorophosphonate

MAGs

monoacylglycerols

MAGL

monoacylglycerol lipase

NAM

negative allosteric modulator

2-OG

2-oleoylglycerol

PBS

phosphate-buffered saline

PEI

polyethylenimine

SR1

SR141716

Δ⁹-THC

Δ⁹-tetrahydrocannabinol

TBS

Tris-buffered saline

TBST

TBS with 0.05% Tween-20

Authors' Contributions

S.S.: Conceptualization, methodology, validation, investigation, and writing—original draft, and visualization. D.S.: Investigation. M.M.: Formal analysis and software. A.E.: Formal analysis and software. A.D. and Y.L.: Resources and writing—review and editing. L.Z.: Resources and writing—review and editing. M.R.B.: Conceptualization and writing—review and editing. B.B.L.: Conceptualization and writing—review and editing. N.S.: Conceptualization, writing—original draft, writing—review and editing, visualization, supervision, and funding acquisition.

Author Disclosure Statement

N.S. is employed by the University of Washington, Seattle, and by Stella Consulting LLC. The terms of this arrangement have been reviewed and approved by the University of Washington in accordance with its policies governing outside work and financial conflicts of interest in research. M.R.B. is a co-founder and SAB member of Neurolux, Inc. None of the technology or work described here is related to those efforts. All other authors declare that they do not have any known competing financial interests or relationships that could have influenced the work in this article.

Funding Information

This work was supported by the National Institutes of Health (NS118130 and DA047626 to N.S., DA055448 to A.E., DA033396 to M.R.B., and T32GM007750, AT011524 to B.B.L.). The authors also acknowledge support from the University of Washington Center of Excellence in Opioid Addiction Research/Molecular Genetics Resource Core (P30DA048736). National Natural Science Foundation of China (31925017 and 31871087), the Beijing Municipal Science & Technology Commission (Z181100001318002 and Z181100001518004), the NIH BRAIN Initiative (1U01NS113358), the Shenzhen-Hong Kong Institute of Brain Science (NYKFKT2019013), the Science Fund for Creative Research Groups of the National Natural Science Foundation of China (81821092) and grants from the Peking-Tsinghua Center for Life Sciences and the State Key Laboratory of Membrane Biology at Peking University School of Life Sciences (to Y.L.).

Cite this article as: Singh S, Sarroza D, English A, McGrory M, Dong A, Zweifel L, Land BB, Li Y, Bruchas MR, Stella N (2023) Pharmacological characterization of the endocannabinoid sensor GRAB_eCB2.0, Cannabis and Cannabinoid Research 9:5, 1250–1266, DOI: 10.1089/can.2023.0036.