Differential increase in endocannabinoid levels at the plasma and intracellular membranes

Summary

Endocannabinoids (eCBs) modulate the activity of proteins expressed at the plasma and intracellular membranes. Nothing is known about the dynamic changes in eCB levels in these subcellular compartments. We leveraged the eCB sensor, GRAB_eCB2.0, to establish the stimulus-induced increases in the eCB, 2-arachidonoyl glycerol (2-AG), at the plasma and intracellular membranes of undifferentiated Neuro2a cells in culture. Activating G protein-coupled B₂ receptors with bradykinin increased 2-AG levels at both the plasma and intracellular membranes within ≈5 and ≈15 s, respectively. By contrast, the activation of G proteins by the small peptide mastoparan and the ensuing opening of plasma membrane calcium channels increased 2-AG levels in plasma membrane within ≈1-2 s and in intracellular membranes after ≈30 s. While both these stimuli-induced increases in 2-AG production involved canonical lipases, they required distinct sources of calcium. Thus, distinct stimuli differentially increase 2-AG levels at plasma and intracellular membranes via distinct molecular mechanisms.

Graphical abstract

Highlights

• •GRAB_eCB2.0 sensor tracked 2-AG dynamics at subcellular membranes
• •Bradykinin increased 2-AG at plasma (≈5s) and intracellular (≈15s) membranes
• •Mastoparan increased 2-AG at plasma (≈1-2s) then intracellular (≈30s) membranes
• •Distinct stimuli differentially elevate 2-AG via unique mechanisms and Ca²⁺ sources

Biochemistry; Biochemistry applications; Biological sciences; Natural sciences

Introduction

Stimulus-dependent increases in the production of the eCB, 2-AG, controls multiple cellular functions, including neuronal differentiation, energy metabolism, neurotransmitter release, and neuronal plasticity.¹ It is well-known that 2-AG production is increased by stimuli that couple to G proteins and increase phospholipase C (PLC) and diacylglycerol lipase (DAGL) activities in a calcium-dependent manner.^2,3 Evidence suggests that increases in 2-AG activates receptor targets expressed at the plasma membrane and intracellularly.^1,2 Plasma membrane receptors activated by 2-AG include cannabinoid 1 receptors (CB₁R) expressed on presynaptic terminals that regulate neurotransmitter release.^3,4 CB₁Rs are also expressed on mitochondria and their activation by 2-AG controls energy metabolism in neuronal cells.^5,6 Other 2-AG intracellular targets include transcription factors such as peroxisome proliferator-activated receptors (PPAR).^7,8 Several important aspects of 2-AG signaling remain unexplored, including (1) how distinct stimuli might differentially increase 2-AG production, (2) to what extent does stimulating G proteins and opening of calcium channels at the plasma membrane control 2-AG production, and (3) whether 2-AG levels are differentially increased at the plasma and intracellular membranes. Unraveling such subcellular dynamic changes in 2-AG production is crucial to increasing understanding of its biological functions, including whether 2-AG might activate plasma membrane and intracellular targets with different time courses.

GRAB_eCB2.0 is a genetically encoded eCB sensor that was recently developed starting from CB₁R and introducing a circularly permuted-green fluorescent protein (cpGFP) in its third intracellular loop to elicit an increase in fluorescence upon eCB binding.⁹ Thus, GRAB_eCB2.0 can detect changes in the levels of both eCBs, arachidonoylethanolamide (AEA) and 2-AG, with sub-second resolution in cells in culture, mouse brain slices, and in mouse brain during behavior.^{10,11,12,13,14} Here, we expressed GRAB_eCB2.0 in undifferentiated Neuro2a cells in culture to study whether activation of G proteins by either bradykinin or the small peptide mastoparan differentially regulates 2-AG production at the plasma and intracellular membranes and determined if the sources of calcium, calcium channels, phospholipase C (PLC), and diacylglycerol lipase (DAGL) are differentially involved.

Results

BK and mastoparan differentially increase 2-AG levels in undifferentiated Neuro2a cells

Neuro2a cells are a commonly used model system to study eCB signaling since they express the main enzymes involved in 2-AG synthesis, as well as B₂Rs that couple to G_q proteins (Figure 1A).^15,16,17,18 As an alternative stimulus to increase G protein signaling in Neuro2a cells, we selected mastoparan, a positively charged small peptide that inserts into lipid bilayers and activates G proteins (Figure 1A).^19,20 We first confirmed that BK and mastoparan increase 2-AG production in Neuro2a cells by using liquid chromatography—mass spectrometry (LC-MS) (Figure 1B).¹⁶ Figures 1C and 1D show that both BK and mastoparan significantly increased 2-AG levels within minutes, as expected when using this analytical approach.^21,22,23,24 Specifically, BK increased 2-AG levels by 95% and 72% after 2- and 10-min treatment, respectively, and mastoparan increased 2-AG levels by 56% and 151% after 5- and 15-min treatment, respectively (Figures 1C and 1D). Of note, we also monitored anandamide and found that its amounts remained below detection limits (Figure S1).

Figure 1.

Bradykinin and mastoparan differentially increase 2-AG levels and GRAB_eCB2.0 signal in Neuro2a cells

Neuro2a cells were treated and changes in eCB levels measured by LC-MS and changes in GRAB_eCB2.0 signal measured using live cell confocal microscopy.

(A) Schematic of LC-MS workflow to detect 2-AG levels: Neuro2a cells were treated with either vehicle, BK, or mastoparan for specified time intervals, cells were harvested to obtain a pellet, proteins were precipitated, and fraction containing lipids analyzed by LC-MS (limit of detection was 0.17 pmol for 2-AG and 0.02 pmol for AEA) (see Figure S1).

(B and C) BK (1 μM for 2- and 10-min) and mastoparan (10 μM for 5- and 15-min) increase 2-AG levels in Neuro2a cells when measuring changes in 2-AG levels by LC-MS. Average 2-AG amounts under vehicle treatment were 0.93 pmol/mg protein in (C) and 1.07 pmol/mg protein in (D). n = 6–18 independent measurements. ∗p < 0.05, ∗∗p < 0.01, and ∗∗∗p < 0.001 significantly different from vehicles (One-way ANOVA followed by Tukey’s).

(D) Schematic of live-cell confocal microscopy workflow to detect changes in GRAB_eCB2.0 signal. Neuro2a cells were transfected with plasmid encoding GRAB_eCB2.0, cells were then treated by adding BK or mastoparan directly into the buffer, and changes in GRAB_eCB2.0 signal detected using live-cell confocal microscopy.

(E) Representative images of BK- (1 μM) and mastoparan- (10 μM) induced an increase in GRAB_eCB2.0 signal measured by live-cell microscopy. GRAB_eCB2.0 signal was first captured 30 s before drug treatment (baseline) and then 3 min following drug treatment. Scale bars, 40 μm in original image and 20 μm in inset.

(F and G) Time course of BK (1 μM)- and mastoparan- (10 μM) induced changes in GRAB_eCB2.0 signal (ΔF/F₀) over 10 min measured by live-cell confocal microscopy. Bold traces show averages of GRAB_eCB2.0 signals and thin traces show GRAB_eCB2.0 signal in individual cells.

(H and I) Frequency distribution analysis of results in g-h indicating a unimodal Gaussian distribution for the BK response and a bimodal distribution for the mastoparan response. n = 35–48 cells from 3 to 4 independent experiments.

To study the dynamics of BK- and mastoparan-induced increase in 2-AG production, we monitored changes in GRAB_eCB2.0 fluorescent signal in Neuro2a cells in culture using live-cell confocal microscopy (line scanning frequency: 200 Hz, 0.78 frames per sec) (Figure 1E). Figures 1F and 1G show that the basal fluorescence signals in Neuro2a cells prior to treatment were low, and that BK (1 μM) rapidly increased the average GRAB_eCB2.0 signal up to 0.87 ΔF/F₀ after 3 min, followed by a gradual decline. Mastoparan (10 μM) also induced a rapid increase in the average GRAB_eCB2.0 signal that reached a comparable maximum fluorescence signal of 0.84 ΔF/F₀ after 3 min, which remained elevated for >12 min (Figures 1F and 1H). To test for the biological relevance of these responses, we compared the BK- and mastoparan-induced increases in ΔF/F₀ GRAB_eCB2.0 signal to the increase in ΔF/F₀ GRAB_eCB2.0 signal induced by exogenous 2-AG. Thus, 2-AG (1 μM) increased GRAB_eCB2.0 signal to 1.99 ΔF/F₀ after 5 min, and BK and mastoparan increased GRAB_eCB2.0 signal reached 79% and 71% of this response, emphasizing the strong amplitude of the stimuli-induced production of 2-AG detect with the sensor. As expected, the level of GRAB_eCB2.0 signal varied in each Neuro2a cell due to heterologous transfection efficiency; nevertheless, we accounted for this by calculating ΔF/F₀ values for each cell prior to averaging and performing further analysis (Figure 1F).

As previously reported, mastoparan triggered changes in cell morphology within minutes of treatment, which we accounted for when analyzing stimulus-induced changes in the GRAB_eCB2.0 signal (Figure S2).²⁵ Using this analysis, we found that all GRAB_eCB2.0-Neuro2a cells responded to BK, and that these responses fit a single mode Gaussian distribution; whereas GRAB_eCB2.0-Neuro2a cells responded to mastoparan in a more heterogeneous fashion that followed a bimodal distribution (Figures 1I and 1J). Together, these results establish that both BK and mastoparan induce rapid increases in 2-AG production that peak within min, that these responses differ in their late kinetics and in the proportion of responding cells, and that these stimuli do not couple to increasing anandamide production.

Characterization of the BK- and mastoparan-induced increases in 2-AG production

To characterize the pharmacological profile of changes in GRAB_eCB2.0 signal in Neuro2a cells, we switched to a 96-well fluorescent plate reader assay (3 Hz scanning frequency) (Figure 2A).¹⁶ Figures 2B and 2C show that both BK and mastoparan induced rapid, concentration-dependent increases in the GRAB_eCB2.0 signal that reached a maximum ΔF/F₀ signal of 0.17 and 0.34, respectively, approximately 5 min after treatment. Based on the concentration-dependent responses within the first 5 min and during the later phase (i.e., between 5 and 30 min), which confirmed differences in these dynamic responses, the following mechanistic studies used 1 μM BK and 10 μM mastoparan (Figures 2D–2G and S3). Next, we determined whether the BK effect was mediated by B₂R using the selective B₂R antagonist, HOE 140 (300 nM), and found that it blocked the BK-induced increase in GRAB_eCB2.0 signal by 90% (Figure 2H). As selectivity controls, we confirmed that both the BK and mastoparan responses were absent in mutant-GRAB_eCB2.0-Neuro2a cells and reduced by the CB₁R antagonist SR141716 (300 nM) (Figure 2I).^10,14,16

Figure 2.

BK and mastoparan increase GRAB_eCB2.0 signal in Neuro2a cells with distinct pharmacological profiles

Neuro2a cells were treated and changes in GRAB_eCB2.0 signal measured using fluorescence plate reader.

(A) Workflow for measuring changes in GRAB_eCB2.0 signal using a 96-well fluorescence plate reader.

(B and C) Time course of the concentration-dependent increase in GRAB_eCB2.0 signal following treatment with BK (B) and mastoparan (C). Dotted line indicates 5 min timepoint. n = Mean ± SEM of 3–13 independent experiments performed in duplicate.

(D–G) Analysis of the concentration-dependent increase in GRAB_eCB2.0 signal induced by BK and mastoparan measured in B and C at the 5 min time point (D and F) and during the late response (area under the curve between 5 and 30 min) (E and G).

(H and I) BK (1 μM) response in Neuro2a cells either expressing mutant-GRAB_eCB2., pretreated with the B₂R antagonist HOE 140 (HOE, 300 nM) or pretreated with SR1 (300 nM) as determined by averaging ΔF/F₀ at the 5 min time point (H). Mastoparan (10 μM) response in Neuro2a cells either expressing the mutant-GRAB_eCB2.0 or pretreated with SR1 (300 nM) as determined by averaging ΔF/F₀ between at the 5 min time point (I). ∗∗∗p < 0.001 significantly different from vehicle (ANOVA followed by Dunnett). n = 3–5 independent experiments performed in duplicate.

(J–M) Effect of pertussis toxin (PTX, 1 μg/mL for 5 h) on the time-course of BK- (1 μM) and mastoparan- (10 μM) induced increase in GRAB_eCB2.0 signal (J and L). (K and M) show the quantification of the early phase (area under curve between 0 and 5 min) and late phase (area under curve between 5 and 30 min) presented in J and L. All results are from n = 3–4 independent experiments performed in triplicate; Statistics: two-way ANOVA followed by Bonferroni. Non-significant = ns > 0.05 and ∗∗∗p < 0.001 significantly different from stimuli in corresponding phase.

While it is known that while B₂R often couple to G_q-proteins, mastoparan is thought to preferentially activate G_i/o proteins.²⁶ Therefore, we tested the involvement of G_i/o in these responses by pretreating GRAB_eCB2.0-Neuro2a cells with pertussis toxin (PTX) to ADP-ribosylate G_i/o and block its function.²⁷ Figures 2J–2M show that PTX (1 μg/mL for 5 h) reduced both the early phase (0–5 min) and the late phase (5–30 min) of the BK response by −30% and −40%, respectively; however, PTX only reduced the late phase (5–30 min) of the mastoparan response by −30%. Similar results were obtained when treating Neuro2a cells with PTX at 100 ng/mL for 24 h (Figure S4). Thus, the B₂R-mediated increase in 2-AG production is partially sensitive to PTX, supporting studies showing the B₂R coupling to multiple G proteins, including PTX-sensitive G_o.^28,29,30 By contrast, only the late phase of the mastoparan-induced increase in 2-AG involves G_i/o proteins, here suggesting that mastoparan activates different G protein complexes as a function of time, namely PTX-insensitive G proteins within seconds of treatment and PTX-sensitive proteins over the course of several minutes.

The BK- and mastoparan-induced increases in 2-AG production require distinct calcium sources

Increased 2-AG production is tightly coupled with increases in intracellular calcium (_i[Ca²⁺]) that stimulates the activity of select lipases involved in its biosynthesis.^3,31 Both BK and mastoparan treatment of neuronal cells have been shown to increase _i[Ca²⁺]^32,33,34,35; yet the calcium sources involved in these responses has not been characterized. We first determined to what extent omitting calcium in the treatment media affected the BK- and mastoparan-induced increase in 2-AG production. Figure 3A shows that omitting calcium in the treatment media did not affect the BK response during the early phase and had a nominal effect on the late phase (−18%) (Figures 3I and 3J). By contrast, both phases of the mastoparan response were profoundly reduced by omitting calcium in the treatment media (−117% and −86%, respectively) (Figures 3B, 3K, and 3L). Omitting calcium in the treatment media combined with adding EGTA (1 mM) to further chelate any residual calcium did reduce the BK response during the early and late phases (−59% and −63%, respectively); as well as the early and late phases of the mastoparan response (−87% and −61%, respectively) (Figures 3I–3L and S3A). These results indicate that the BK response requires _e[Ca²⁺] to a lesser extent than the mastoparan response. Accordingly, buffering _i[Ca²⁺] with BAPTA-AM reduced the BK responses during the early and late phase by −65% and −66%, respectively (Figures 3C, 3I, and 3J); whereas it reduced the early and late phase of the mastoparan response to a lesser extent (−49% and −20%, respectively) (Figures 3D, 3K, and 3L). Depleting calcium from intracellular stores with thapsigargin (10 μM, 20 min pretreatment) reduced the early and late phase of the BK response by −73% and −70%, respectively, without affecting the mastoparan response (Figures 3E–3F and 3I–3L). Furthermore, neither the BK nor the mastoparan responses were affected when cotreated with thapsigargin (i.e., no prior depletion of intracellular calcium stores (Figures 3I–3L and S3B). Finally, 2-ABP (100 μM), an inhibitor of IP₃ receptors on intracellular calcium storers, strongly reduced the BK response and hardly affected the mastoparan response (Figures 3I–3L and S3C).

Figure 3.

Distinct calcium sources are required for the BK- and mastoparan-induced increases in 2-AG

Neuro2a cells were treated and changes in GRAB_eCB2.0 signal measured using fluorescence plate reader.

(A) Diagram depicting the pharmacological treatments used to determine the source of calcium involved in the BK and mastoparan induced changes in GRAB_eCB2.0.

(B–E) Effect of omitting calcium in buffer (Without _ext[Ca²⁺]) on BK- and mastoparan-stimulated increases in GRAB_eCB2.0 signal.

(J–M) Effect of BAPTA-AM (30 μM) on BK- and mastoparan-stimulated increases in GRAB_eCB2.0 signal.

(F–I) Effect of thapsigargin (Thapsi; 10 μM) on BK- and mastoparan-stimulated increases in GRAB_eCB2.0 signal.

Treatments in F–M were added 20 min prior to stimulation with BK or mastoparan.

(C, E, G, I, K, and M) Quantification of the area under the curve of GRAB_eCB2.0 signal in corresponding time-courses. None of these treatments affect basal and CP55940 (1 μM)-induced increase in GRAB_eCB2.0 signal.¹⁶ All results are Mean ± SEM from n = 3–6 independent experiments performed in triplicate; Statistics: two-way ANOVA followed by Bonferroni. Non-significant = ns > 0.05, ∗p < 0.05, ∗∗p < 0.01 and ∗∗∗p < 0.001 significantly different from stimuli in corresponding phase.

Together, these results show that the molecular mechanism involved in the BK- and mastoparan-induced increases in 2-AG production rely on different sources of calcium: the BK response relying mainly on increases in _i[Ca²⁺] while the mastoparan response relying mainly on the influx of _e[Ca²⁺]. Considering that BAPTA-AM can buffer _i[Ca²⁺] close to the plasma membrane,³⁶ this result also suggest that the mastoparan response depends more on the early influx of calcium close to the plasma membrane.

The BK- and mastoparan-induced increases in 2-AG production differentially involve channels permeable to calcium

Since little is known about the molecular mechanism involved in the mastoparan-induced increase in _i[Ca²⁺], we sought to determine if calcium channels are involved. We first tested the broad calcium channel blocker, cadmium (100 μM),^37,38 and found that it reduced both the early and late phases of the BK response (−38% and 58%, respectively) (Figures 3G, 3I, and 3J); as well as the mastoparan responses (−55% and 58%, respectively) (Figures 3H, 3K, and 3L), providing initial evidence for both responses opening calcium channels at the plasma membrane. Nifedipine blocks Ca_V1.1, Ca_V1.2, Ca_V1.3, and Ca_V1.4 channels.³⁹ Figures 3I and 3J show that nifedipine (10 μM) did not affect the early phase and inhibited the late phase of the BK response (−33%) (Figure S3D). By contrast, nifedipine did not affect the mastoparan (Figures 3K, 3L, and S3). Blocking Ca_V2.2 channels with ω-conotoxin (10 μM)³⁹ did not affect the BK and mastoparan responses (Figures 3I–3L and S3E). Together, these results suggest that the late phase of the BK response involves calcium influx via Ca_V1 channels, and that the early phase of its response and the entire mastoparan response involved opening of other subtypes of ion channels at the plasma membrane permeable to calcium and sensitive to cadmium.

As our pharmacological approach did not identify the molecule responsible for the mastoparan-induced calcium increase, we took an alternative strategy. We performed live cell calcium imaging of Neuro2a cells loaded with the fluorescence sensor Cal-520 (Figure 4A) and tested the effect of cadmium. Mastoparan induced an increase in _i[Ca²⁺] in Neuro2a cells that lasted approximately 100 s, which contrasted with the BK response that lasted for approximately 20 s (Figures 4B–4E). The specific parameters of these distinct calcium responses are in Table 1, including that the BK-induced increase in _i[Ca²⁺] had a 54% greater (peak ΔF/F₀) and was 2.6-fold faster (τ_1/2,rise) than the mastoparan-induced increase in _i[Ca²⁺]. In contrast to the BK response that occurred in 100% of the cells, the mastoparan response occurred only in 70% of cells. Strikingly, the application of cadmium blocked the calcium signals induced by mastoparan but not by BK (Figures 4B–4E). Together, these results suggest that the mastoparan-induced increase in 2-AG production requires a pronounced influx of calcium that is detected by the fluorescence sensor Cal-520. In contrast, the calcium elevation induced by BK and subsequent increase in 2-AG, comes mainly from intracellular stores.

Figure 4.

Distinct calcium-permeable channels are involved in the BK- and mastoparan-induced increases in 2-AG production

Neuro2a cells were treated and changes in _i[Ca²⁺] measured by live cell confocal fluorescence microscopy using Cal-520. Changes in 2-AG levels measured using GRAB_eCB2.0 signal measured using fluorescence plate reader.

(A) Schematic depicting calcium imaging experiment workflow: Neuro2a cells were loaded with Cal-520 and changes in _i[Ca²⁺] in response to mastoparan and BK were measured using live-cell confocal fluorescence microscopy.

(B–E) Representative images of the mastoparan- (B) and BK- (D) induced an increase in intracellular calcium levels in Cal-520-loaded Neuro2a cells, and the effect of cadmium (100 μM) on these responses. Traces of calcium transients of the mastoparan- (C) and BK- (E) induced an increase in _i[Ca²⁺] and the effect of cadmium (100 μM) on these responses. The treatment window indicated by the orange shaded area. Numbering of images corresponds to (1) before agonist, (2) at the peak of the response, and (3) after the agonist and once recovery was observed. Same numbers are depicted in the traces. Values represent the mean ± s.e.m. of n = 91 cells from 2 to 3 independent experiments.

(F–I) Effect of cadmium (100 μM) on BK- and mastoparan-stimulated increases in GRAB_eCB2.0 signal. (G and I) Quantification of the area under the curve of GRAB_eCB2.0 signal in corresponding time-courses. This treatments did not affect basal and CP55940 (1 μM)-induced increase in GRAB_eCB2.0 signal. All results are mean ± SEM from n = 4 independent experiments performed in triplicate; Statistics: two-way ANOVA followed by Bonferroni. ∗∗p < 0.01 and ∗∗∗p < 0.001 are significantly different from stimuli in corresponding phase.

Table 1.

Bradykinin and mastoparan induce distinct increases in i[Ca²⁺] in Neuro2a cells

Treatment	Area Under Curve	Peak ΔF/F0	τ1/2,rise	# of responders/total cells
Bradykinin	42.1 ± 1.7	5.4 ± 0.2	0.7 ± 0.02	91/91
Mastoparan	113.3 ± 20.6	3.7 ± 0.5	2.5 ± 0.2	64/91

Neuro2a cells were loaded with Cal-520 and changes in fluorescent signal (ΔF/F0) in response to bradykinin (1 μM) or mastoparan (10 μM) were measured using live cell confocal microscopy. The fluorescent traces were used to calculate 4 parameters of the calcium signal: (1) overall response (area under curve), (2) maximal response (Peak ΔF/F₀), (3) Rise kinetics (τ1/2,rise: time to reach half peak ΔF/F0), and (4) number of cells responding to treatment (cell classified as responder if cell experienced ≥10% increase in ΔF/F0). All results represent the Mean ± SEM of n = 17–19 cells from 3 independent experiments.

The BK-induced increase in 2-AG production involves PLC and DAGL whereas the mastoparan-induced increase in 2-AG production predominantly involves DAGL

Increasing PLC and DAGL activities results in increased 2-AG levels (Figure 5A).^{3,31,40,41,42} To study the involvement of PLC in the BK-induced increase of 2-AG, we pre-treated GRAB_eCB2.0-Neuro2a cells with the PLC inhibitor, U73122.^3,16,43 Figures 5B and 5C show that U73122 (3 μM) profoundly inhibited the BK response (early and late phases: −76% and −69%, respectively), did not affect the mastoparan-response during the early phase, and decreased the mastoparan-response during the late phase by only −38% (Figures 5D and 5E). We then tested the involvement of phosphoinositide-PLCs using their broad-spectrum inhibitor, ET-18-OCH3 (ET-18).^3,16,44,45 Remarkably, ET-18 (3 μM) slightly reduced both the BK (1 μM)- and mastoparan (10 μM)-induced increases in GRAB_eCB2.0 signal, and this to the same extent during both early and late phases (−23% and −24%) (Figures 5F–5I). These results suggest that different PLC subtypes are sequentially involved in the BK- and mastoparan-induced increases in 2-AG production: the BK response involves PLC subtypes that are sensitive to both U73122 and ET-18; whereas the mastoparan response involves PLC subtypes that are sensitive only to ET-18 during the early phase and sensitive to both U73122 and ET-18 during the late phase.

Figure 5.

PLC and DAGL are differentially involved in the BK- and mastoparan-stimulated 2-AG production

Neuro2a cells were treated and changes in GRAB_eCB2.0 signal measured using fluorescence plate reader.

(A) Diagram of BK (1 μM) and mastoparan (10 μM) stimulated 2-AG production and pharmacological interventions (U73122, ET-18, and DO34) to test the role of PLCs and DAGL.

(B–E) Effect of U73122 (3 μM) on BK- and mastoparan-stimulated increase in GRAB_eCB2.0 signal: kinetics (B and D) and early and late phases (area under the curves; C and E).

(F–I) Effect of ET-18 (3 μM) on BK- and mastoparan-stimulated increase in GRAB_eCB2.0 signal: kinetics (F and H) and early and late phases (area under the curves; G and I).

(J–M) Effect of DO34 (10 nM) on BK- and mastoparan-stimulated increase in GRAB_eCB2.0 signal: kinetics (J and L) and early and late phases (area under the curves; K and M). All results are mean ± SEM from n = 4–6 independent experiments performed in triplicate; Statistics: two-way ANOVA followed by Bonferroni. Non-significant = ns > 0.05, ∗p < 0.05, ∗∗p < 0.01 and ∗∗∗p < 0.001 significantly different from stimuli in corresponding phase.

To test the involvement of DAGL, we pre-treated GRAB_eCB2.0-Neuro2a cells with DO34.^16,46,47 Figures 5J–5M show that DO34 (10 nM) greatly decreased both the BK- and mastoparan-induced increase in GRAB_eCB2.0 signal. Specifically, the early phase of the BK response was reduced by 79% and its late phase was blocked by 99% (Figures 5J and 5K) Of note, DO34 treatment reduced the GRAB_eCB2.0 signal below basal. These results suggest that DAGL represents a rate limiting step in the BK-induced increase in 2-AG production. Regarding the mastoparan-induced increase in GRAB_eCB2.0 signal, both the early and late phases were also inhibited by DO34 (i.e., −70% and −82%, respectively); yet here DO34 treatment did not reduce GRAB_eCB2.0 signal below basal (Figures 5L and 5M).

Together, these results indicate that the acute activation of B₂R, a receptor known to couple to several G proteins, rapidly increases 2-AG production by enhancing the activity of PLCs that are sensitive to both U73122 and ET-18, and that DAGL activity represents a key rate-limiting step of this response. The mastoparan-induced increase in 2-AG production involves PLC subtypes that are sensitive to ET-18 and predominantly involves DAGLs. Thus, distinct stimuli that activate different G proteins result in increasing 2-AG production by recruiting different PLC subtypes that then converge at DAGL.

BK and mastoparan differentially increase 2-AG levels at the plasma and intracellular membranes

Here, we leveraged the fact that Neuro2a cells express GRAB_eCB2.0 at both plasma and intracellular membranes to determine whether BK and mastoparan differentially increase 2-AG production within these subcellular compartments. Specifically, we co-stained GRAB_eCB2.0-Neuro2a cells with wheat germ agglutinin (WGA), which stains both the plasma and intracellular membranes and allows for the isolation of changes in GRAB_eCB2.0 signal at the plasma membrane versus the intracellular space (Figure 6A). A line scan analysis showed that the WGA and the GRAB_eCB2.0 fluorescence signal colocalize strongly at the plasma membrane as indicated by sharp peak fluorescence signals and that this signal was distinct from intracellular fluorescence signals (Figures 6B and S6). Thus, we developed a subcellular segmentation method to isolate and measure changes in the GRAB_eCB2.0 signal at the plasma versus intracellular membranes as we switched from basal to stimulated conditions (Figure 6C). Line scan analysis of the GRAB_eCB2.0 signals showed that a 10-min treatment with both mastoparan (10 μM) and BK (1 μM) increased the GRAB_eCB2.0 signal at both the plasma membrane and intracellularly (Figures 6D and 6E). When comparing the kinetics of the mastoparan and the BK responses, we discovered changes in GRAB_eCB2.0 signals at the plasma- and intracellular membranes that had comparable dynamics but key differences depending on time points. (1) Both stimuli induced a rapid increase of the GRAB_eCB2.0 signal at the plasma membrane: mastoparan within ≈1–2 s and BK within ≈5 s of treatment (Figure 6E). (2) With both stimuli, the initial GRAB_eCB2.0 responses at intracellular membranes were delayed compared to that at the plasma membrane. Specifically, BK increased the intracellular GRAB_eCB2.0 signal after ≈15 s and mastoparan increased the intracellular GRAB_eCB2.0 signal after ≈40 s (Figure 6F). (3) Both stimuli induced an increase of the GRAB_eCB2.0 signal at the plasma membrane that was greater than at intracellular membranes over the course of 10 min (Table 2, as measured by the area under the curve of the ΔF/F₀). (4) The proportion of the response that occurred intracellularly was comparable for both stimuli (maximal plasma membrane ΔF/F₀ was 2.4-fold and 2.8-fold greater than the maximal intracellular ΔF/F₀ for BK and mastoparan, respectively) (Table 2). 5) However, BK induced a 2.5-fold greater increase at the plasma membrane relative to intracellular membranes whereas mastoparan induced a 3.7-fold greater increase at the plasma membrane relative to intracellular membranes (Table 2 and Figure S6). Thus, an increase in 2-AG production can occur with different dynamics and proportions at the plasma and intracellular membranes depending on the type of stimuli and signaling mechanism (Figure 6G).

Figure 6.

BK and mastoparan increase intracellular GRAB_eCB2.0 signal

Live cell confocal fluorescence microscopy of GRAB_eCB2.0-expressing Neuro2a cells.

(A) Representative images of GRAB_eCB2.0 signal (green) and WGA signal (red) under basal conditions and after treatment and mastoparan (10 μM) for 10 min as detected by live-cell confocal microscopy. Arrows in the merged image indicate the position of line-scan analysis. Scale bar, 10 μm.

(B) Line-scan plots of a representative cell depicting the co-localization of GRAB_eCB2.0 signal (green) and WGA signal (red) at the plasma membrane (plasma membrane peaks indicated by arrows of corresponding colors).

(C and D) Line-scan analysis of the representative cell depicting changes in GRAB_eCB2.0 signal from basal conditions (green) and following treatment with mastoparan (10 μM; 10 min; purple; C) or following treatment with bradykinin (1 μM; 10 min blue, D).

(E) Images of masking protocol to differentiate between and calculate changes in plasma membrane and intracellular GRAB_eCB2.0 signal. (1) The fluorescence intensity was averaged over 12 min imaging period (2 min basal signal and 10 min treatment with BK or mastoparan). (2) Average intensity image was processed by using a Gaussian blur convolution function. (3) Processed images were used to obtain a mask for the plasma membrane (red) and intracellular signal (blue) using thresholding function.

(F and G) Time course of BK (1 μM)- and mastoparan (10 μM)-induced changes in GRAB_eCB2.0 signal at the plasma membrane (F) and intracellular membrane (G) during the first 60 s of treatment. Bold lines indicate mean response from n = 17–19 cells from 3 individual experiments, and thin traces indicate individual cell responses. Bold orange and blue lines indicate the corresponding responses.

(H) Model of 2-AG activation of GRAB_eCB2.0 localized to intracellular membranes following 2-AG production at the plasma membrane stimulated by BK or mastoparan.

Table 2.

Parameters of intracellular and plasma membrane (PM) GRAB_eCB2.0 activation induced by bradykinin and mastoparan

PD parameter	Units	Bradykinin		Mastoparan
		PM	Intracellular	PM	Intracellular
Initial response: slope	(x 10−1 ΔF/F0/min)	10.5	2.2	10.6	0.87
	(95% CI)	9.1–12.0	1.7–2.6	8.6–12.6	0.38–1.36
Response at 2 min	(ΔF/F0)	1.24	0.32	1.58	0.28
	(std. error)	0.10	0.03	0.15	0.06
Maximal response	(ΔF/F0)	1.26	0.37	1.84	0.49
	(std. error)	0.10	0.04	0.18	0.08
Overall response	(area under the curve)	9.95	2.86	15.52	3.30
	(std. error)	0.25	0.10	0.43	0.17

Neuro2a cells expressing GRAB_eCB2.0 and co-stained with WGA were treated with bradykinin (1 μM) or mastoparan (10 μM) and changes in fluorescent signal (ΔF/F₀) at the plasma membrane (PM) and intracellular were detected using live cell confocal microscopy. Results are Mean ± SEM of n = 39–70 cells from 4 independent experiments.

Bold entries are specific values and non-bold are variance and standard error.

Discussion

Our study shows that the stimuli-dependent increases in 2-AG levels in undifferentiated Neuro2a cells can be spatially segregated between the plasma and intracellular membranes. Both undifferentiated and differentiated Neuro2a cells are a commonly used model system to study eCB signaling, including the stimuli-dependent of 2-AG, for example 2-AG production increased by activating P2X₇ ionotropic receptors,^48,49 the molecular interactions that occur between PLC, DAGL, and intracellular calcium signaling.^50,51,52 Thus, undifferentiated Neuro2a cells express the molecular machinery involved in regulating 2-AG production in cells while lacking neurites and synaptic connections present in differentiated neurons in culture and in brain slices. However, the results presented here are likely also relevant to 2-AG production in differentiated neurons when considering that the overall molecular mechanisms involved in 2-AG production at the plasma membrane and its ability to reach intracellular membrane is likely conserved in most cell types, though the precise dynamics and time course may vary depending on the type of stimuli and subtype of cell. Acute activation of presynaptic CB₁R at the plasma membrane by 2-AG inhibits neurotransmitter release.^53,54 More prolonged activation of presynaptic CB₁R at the plasma membrane by 2-AG induces several forms of synaptic plasticity, including DSI\DSE and LTD.^54,55 Relevant to this study, 2-AG induced short-term synaptic plasticity, e.g., DSI and DSE, may involve the opening of calcium channels and the release of calcium from intracellular stores depending on stimuli and their duration.^56,57,58 To our knowledge, our study is the first to provide direct evidence that opening of calcium channels and influx of extracellular calcium results in increasing 2-AG production. 2-AG also regulates synaptic transmission and plasticity by activating CB₁R expressed by intracellular mitochondria.⁵ The extent to which increased 2-AG production at the plasma membrane might also reach CB₁R expressed by mitochondria remains to be established, for example by targeting GRAB_eCB2.0 to this organelle. Together, our results suggest that distinct stimuli that differentially increase 2-AG levels and ensuing activation of receptor targets at the plasma and intracellular membranes represent a mechanism underlying activation of distinct receptor targets in a time- and location dependent manner.

The relative involvement of lipases involved in 2-AG production varies when activating either G_q or G_i/o proteins

Stimulation of GPCRs that couple to G_q results in increasing PLC activity, which generates IP₃ and DAG, and the latter is then hydrolyzed by DAGL to produce 2-AG.^2,59 For example, group 1 glutamate metabotropic receptors that couple to G_q increase 2-AG production from post-synaptic neurons and this response requires functional PLC and DAGL.³¹ Here, we show that stimulation of B₂R known to couple to G_q also rapidly increases 2-AG production, and that this response is also profoundly reduced by PLC and DAGL inhibitors (Figure 5). A limitation to the use of U73122 and ET-18 as PLC inhibitors is their potential PLC-independent off-target.^60,61,62,63 A better understanding of the dynamics and molecular mechanism involved in the BK-induced increases in 2-AG production by neurons is relevant as, for example, activation of B₂R and CB₁R in peripheral sensory neurons have been shown to control their sensitization to nociceptive pain.^42,64,65 Remarkably, we found that PLC is not involved in the activation phase of the mastoparan-induced increase in 2-AG production and only partially involved in its prolonged increase in 2-AG production. Since the mastoparan response is blocked by the DAGL inhibitor, these results suggest the direct activation of DAGL by increasing _i[Ca²⁺] (52–54). A limitation to the use of mastoparan as G protein activator is that this small peptide is known to activate additional targets.^20,26,66,67 Thus, distinct biosynthetic pathways resulting in 2-AG production are recruited by either B₂R or the opening of calcium channels. In line with this interpretation, different PLC subtypes with distinct sensitivities to U73122 and ET-18 are involved during the BK- and mastoparan-induced dynamic increases in 2-AG production (Figure 5). Finally, both the BK- and mastoparan-induced increase in 2-AG production were strongly inhibited by a DAGL inhibitor, indication that DAGL likely represents a point of convergence and a rate limiting step for disparate biosynthetic pathways.

Isolating different sources of calcium involved in the BK- and mastoparan-induced increase in 2-AG production

The BK-induced increase in 2-AG production depends on increased _i[Ca²⁺] from mobilization of intracellular calcium stores, as suggested by its sensitivity to BAPTA-AM, thapsigargin, and 2-ABP; however, the mastoparan-induced increase in 2-AG production depend on _e[Ca²⁺] and opening of calcium channels, as indicated by its dependence on _e[Ca²⁺] and sensitivity to cadmium (Figures 3 and 4). What is the potential molecular mechanism that mediates the mastoparan-induced activation of calcium channels? A possibility may involve a G protein-mediated depolarization relying on activation of Gαq proteins, activation of phospholipase C, hydrolysis of phosphatidylinositol 4,5 bisphosphate (PIP₂), and inhibition of KCNQ channels. Specifically, KCNQ channels depend on PIP₂ for their opening, and these channels close when PIP₂ levels are reduced, which leads to membrane depolarization.⁶⁸ An alternative mechanism that does not involve the activation of G proteins is the formation of calcium-permeable pores by mastoparan, a mechanism previously described for a specific isoform.⁶⁷ Thus, the calcium influx mediated by opening of calcium channels results in localized increase of i[Ca²⁺] within intracellular microdomains adjacent to the plasma membrane that directly activate DAGL. In line with this interpretation, opening of calcium channels has been shown to increase i[Ca²⁺] within intracellular microdomains adjacent to the plasma membrane⁶⁹ and the calcium channels-mediated increase in 2-AG production is merely reduced when chelating i[Ca²⁺] by BAPTA-AM (Figure 3). Furthermore, DAGL is associated with the plasma membrane via its four transmembrane domains and its 2-AG producing activity is directly increased by calcium.^70,71,72 Together, our results show that stimuli using different mechanisms, differentiated by the use of discrete calcium sources restricting calcium signaling into cellular spaces, converge in the production of 2-AG.

Several lines of evidence indicate that the BK-induced increase in 2-AG production also involves increases in _i[Ca²⁺] microdomains. First, the rapid increase in 2-AG production occurring within 5 min is sensitive to BAPTA-AM, and BK induces a rapid and transient increase in _i[Ca²⁺] as measured by Cal-520 imaging (Figures 3 and 4). Considering that the early BK response is also sensitive to DAGL inhibition, our results raise the possibility that DAGL expressed on intracellular membranes might be directly activated by calcium released from intracellular calcium stores to increase 2-AG levels on intracellular space. Whether such subcellular lipid biosynthetic pathway is directly regulated by increases in _i[Ca²⁺] and results in producing 2-AG from its precursors at the intracellular membrane or organelles remains to be determined.

Using fluorescence sensors to detect 2-AG production at the plasma membrane that reaches intracellular membranes

Our study is the first to demonstrate that distinct stimuli acting at the plasma membrane targets (here B₂R and calcium channels) will differentially increase 2-AG levels at the plasma and intracellular membranes. The BK-induced increase in 2-AG production at the plasma membrane results in concomitant activation of its receptor targets present at both the plasma and intracellular membranes (Figure 6). By contrast, the opening of calcium channels induced a fast increase in 2-AG levels at the plasma membrane and a slight delayed increase in 2-AG levels at the intracellular membranes (Figure 6). In line with these results, studies indicate only prolonged opening of calcium channels results in activating intracellular signaling mechanisms, such as activation of transcription factors that regulate cell phenotype.^73,74

The adoption of newly developed sensors of lipid signaling molecules and their ability to precisely determine the dynamics of the stimulus-dependent production of such hydrophobic signals raise new questions and hypotheses. For example, does a fully functional, stimulus-dependent, biosynthetic pathway on intracellular membranes exist, and if so, is it responsible for increasing 2-AG levels to directly activate its intracellular targets? Do changes in cellular and energy metabolism affect 2-AG levels on intracellular membranes? In summary, our study shows that fluorescence sensors developed to detect changes in the production of lipid mediators, here GRAB_eCB2.0, represent powerful tools to unravel the precise subcellular dynamic of changes in their levels and considering the proximity to their receptor targets.

Limitations of the study

Critical broader questions remain to be addressed. (1) What functional purpose would spatially compartmentalized production of 2-AG serve in neurons? Most likely, spatially compartmentalized production of 2-AG represents a mechanism to differentially activate receptor targets at either the plasma membrane or intracellularly in a time- and location dependent manner. (2) How do the findings reconcile with the established roles of synaptic versus extrasynaptic CB₁R signaling? Most likely, intracellular increase of 2-AG levels will impact subcellular organelles, such as mitochondria, present in synapses and neuronal somas. (3) Are there implications for disease states where 2-AG signaling is dysregulated, such as epilepsy or chronic pain? Our approach should be able to answer such questions in cell culture models of such diseases, for example human iPS cells.

Resource availability

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Nephi Stella (nstella@uw.edu).

Materials availability

This study did not generate new unique reagents.

Data and code availability

• •Data: No data were deposited.
• •Code: See key resources table.
• •All other items: None.

Acknowledgments

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). We would also like to acknowledge support from the University of Washington Center of Excellence in Opioid Addiction Research/Molecular Genetics Resource Core (P30DA048736), Dale Whittington for help with LC-MS analysis and Dr. Bertil Hille for critical commenting on our study.

Author contributions

S.S., D.S., and J.Y. performed 96-well fluorescent imaging. S.S. performed cloning and live-cell confocal microscopy. S.S., D.S., and D.W. performed LC-MS/MS. S.S. and O.V. performed calcium imaging experiments and data analysis. A.E. performed MATLAB analysis. S.S. and N.S. designed experiments, analyzed data, and wrote the manuscript. B.B.L., M.R.B., and L.Z. contributed to data interpretation and provided feedback on the manuscript. N.S. supervised and coordinated research as well as secured funding for the study. All authors edited the MS.

Declaration of interests

N.S. is employed 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.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Chemicals, peptides, and recombinant proteins
1,2-bis(2-aminophenoxy)ethane-N,N,N′,N′-tetraacetate-acetoxymethyl ester (BAPTA-AM)	Sigma-Aldrich	A4926
2-Arachidonoyl Glycerol-d5	Cayman Chemical	362162
2-aminoethoxydiphenyl borate (2-APB)	Tocris	1224
acetonitrile	ThermoFisher Scientific	A9551
bovine serum albumin	Sigma-Aldrich	A3803
bradykinin	Sigma-Aldrich	05-23-0500
cadmium	Sigma-Aldrich	C3141
Cal-520	ATT Bioquest	21230
Dulbecco’s Modified Eagles Medium (DMEM)	ThermoFisher Scientific	SH30243FS
Dimethyl sulfoxide	Sigma-Aldrich	D2650
DO34	AOBIOUS	AOB8060
Ethylene glycol-bis(2-aminoethylether)-N,N,N′,N′-tetraacetic acid (EGTA)	Sigma-Aldrich	E3889
ET-18-OCH3	R&D Systems Inc	7462
HOE 140	R&D Systems Inc	A740003
mastoparan	Sigma-Aldrich	444898
DPBS, no calcium, no magnesium	ThermoFisher Scientific	14-190-250
Penicillin-Streptomycin (10,000 U/mL)	ThermoFisher Scientific	15-140-122
pertussis	Sigma-Aldrich	P2980
pluronic acid F127	Biotium	59004
Poly-D-lysine	Sigma-Aldrich	P6407
Polyethylenimine, Linear, MW 25000, Transfection Grade	Polysciences	23966
SR141716	NIDA Drug Supply
thapsigargin	R&D Systems Inc	1138
Trifluoroacetic acid	ThermoFisher Scientific	A11650
Trypsin-EDTA	Sigma-Aldrich	T4049
U73122 hydrate	Sigma-Aldrich	U6756
Wheat germ agglutinin Alexa Fluor 594	ThermoFisher Scientific	W11262
Experimental models: Cell lines
Neuro2a cells	This lab	N/A
GRABeCB2.0 in AM/CBA-WPRE-bGH plasmid	This lab	N/A
Mut-GRABeCB2.0 in AM/CBA-WPRE-bGH plasmid	This lab	N/A
Software and algorithms
Fiji	N/A	https://imagej.net/
Prism	GraphPad	N/A
MATLAB	MathWorks	N/A
Deposited data
Code: available at	https://github.com/StellaLab/StellaLab.git	See ref.16
Other
35 mm glass coverslip imaging cell culture dish	Mattek	P35G-1.5-10-C
Confocal Microscope	Leica	SP8X
Triple quadrupole mass spectrometer	Waters	Xevo TQ-S
Zorbax C18 2.1 × 50 mm, 3.5 μm reverse-phase column	Agilent	N/A
Fluorescent plate reader	Molecular devices	SpectraMax M2e
96-Well Optical-Bottom Microplate, black, TC surface	ThermoFisher Scientific	165305

Experimental model and study participant details

Neuro2a cells can be maintained either as undifferentiated, proliferating cells or undergo differentiation following the select activation of both plasma membrane and intracellular receptor targets, such as CB₁R and PPARγ, respectively.^75,76,77 Here, Neuro2a cells were maintained as undifferentiated and transfected as previously described.¹⁶ Briefly, Neuro2a cells were grown in DMEM (Gibco, supplemented with 10% fetal bovine serum and 1% penicillin/streptomycin) at 37°C and 5% CO₂. Cells were passaged every 3–4 days by detaching with 0.25% Trypsin-EDTA and gentle pipetting before adding to a new plate with fresh supplemented DMEM. Cells were transfected using PEI 24 h before an experiment with plasmids encoding the GRAB_eCB2.0 and mut-GRAB_eCB2.0 constructs, which have been previously described.^14,16

Method details

Live-cell confocal microscopy measurement of GRAB_eCB2.0 fluorescent signal

Live cell confocal microscopy was performed as previously described.¹⁶ 250,000 Neuro2a cells were plated on each glass bottom cell culture plate (Mattek) coated with poly-D-lysine (50 ng/mL) and after 24 h cells were transfected with 0.75 μg DNA and 2.25 of μg PEI. The next day, cells were serum starved by exchanging the growth media for serum free DMEM and incubated at 37°C and 5% CO₂ for 1–2 h. In the last 5 min of serum starvation, Wheat germ agglutinin conjugated to Alexa Fluor 594 was added directly to the media for a final concentration of 10 μg/mL. To image, the serum free DMEM was exchanged for room temperature phosphate-buffered saline containing 1 mM CaCl₂ and 0.55 mM MgCl₂. Imaging was performed at room temperature with a line-scanning, confocal microscope (Leica SP8X) using a 40× oil objective with the following settings: 485 excitation, 525 emission wavelength, 5% laser power, HyD hybrid detector and a line scan speed of 200 Hz (acquisition rate of 0.388 frames/second) with bidirectional scanning. All treatments were made in BSA in PBS to achieve a final concentration of 0.1 mg/mL BSA; treatment onset time represented in Figure 1 represents time at which treatments were pipetted directly into the imaging chamber.

Fluorescence plate reader assay

Measurement of changes in GRAB_eCB2.0 fluorescent signal using a 96-well fluorescence plate reader was performed as previously described.¹⁶ Neuro2a cells were plated on clear-bottom, black 96-well plates coated with poly-D-lysine (50 ng/mL) at a density of 20,000 cells/well. Twenty-four hours after plating, cells were transfected with 0.1 μg DNA and 0.3 μg of PEI in 10 μL. The next day, cells were serum starved for 1 h by exchanging growth media with serum free DMEM (DAGL inhibitor treatment occurred during serum starvation), before replacing media with PBS supplemented with 1 mM CaCl₂ and 0.55 mM MgCl₂ (unless specified). Cells were incubated at room temperature for 20 min then a 1 min baseline fluorescence reading was obtained using a fluorescence plate reader with the following settings: 485 excitation, 525 emission, 515 nm cutoff, and a speed of 1 reading every 20 s. Immediately after baseline reading, treatments (made in 1 mg/mL BSA and PBS) were added to buffer in wells and the plate was reinserted into the plate reader and read with same filter settings for 30 min. The following inhibitors and pharmacological agents were added at the start of the 20 min incubation in PBS prior to the baseline reading: SR141617, HOE 140, Pertussis toxin, BAPTA-AM, thapsigargin (pre-treatment), 2-APB, U73122, ET-18, and cadmium. The following agents were added as co-treatments along with BK and mastoparan after baseline reading: EGTA and thapsigargin. Calculations of A.U.C. and statistical analysis of kinetics responses and S.E.M. were analyzed using PRISM.

LC-MS/MS detection of 2-AG

Measurement of 2-AG in Neuro2a cells using LC-MS/MS was performed as previously reported.¹⁶ Neuro2a cells were plated at a density of 1,000,000 cells per well in a 6-well plate. Next day, the cells were serum starved by replacing growth media with serum free DMEM and incubating for 1 h at 37°C, the media was then replaced with PBS supplemented with 1 mM CaCl₂ and 0.55 mM MgCl₂ and incubated at room temperature for about 10 min, and finally cells were treated with vehicle, bradykinin, or mastoparan for specified time intervals. To stop treatment, buffer was removed, and cells were washed three times with ice-cold PBS. Cells were harvested with a cell scraper and centrifuged at 500 x g for 10 min at 4°C to obtain a cell pellet (PBS was discarded). The pellet was resuspended in 20 mL of 0.02% trifluoroacetic acid and 100 mL acetonitrile with 1 pmol of 2-AG-d⁵ internal standard (Cayman Chemical) on ice before transferring to 2.5 mL of acetonitrile in a glass vial. Samples were vortexed and incubated overnight at −20°C. The next day, the homogenate was centrifuged at 2,000 x g for 5 min to remove debris, the supernatant was evaporated under nitrogen stream at 35°C, resuspended in 50 μL of acetonitrile, and samples were capped under nitrogen stream and stored at −80°C until ready for the LC-MS/MS. LC-MS/MS detection of 2-AG, AEA, and arachidonic acid was performed by first using a Zorbax C18, 2.1 × 50 mm, 3.5 μm reverse-phase column (Agilent) for chromatographic separation and then directing the output into the electrospray ionization source of a Waters Xevo TQ-S mass spectrometer in multiple reaction monitoring mode. 2-AG and AEA were detected using positive ionization mode and arachidonic acid was detected using negative mode.

Calcium imaging

We chose Cal-520 for its high signal-to-noise ratio due to its low background signal and because it is retained in the cell.⁷⁸ Neuro2a cells were seeded on 25 mm, 1.5 coverslips coated with poly-D-lysine (50 ng/mL) at a density of 200,000 cells/well in a 6-well plate (one coverslip per well) to achieve a confluency of 30–50% for the calcium imaging experiment. The next day, the calcium indicator was loaded into cells by washing cells with Ringer’s buffer, incubating in 5 μM Cal-520 indicator and pluronic acid (30 min at 37°C), and finally washing dye with Ringer’s buffer for 15 min. Following wash, coverslips were transferred to a gravity perfusion system (rate of 2 mL/min) on a confocal Zeiss LSM 880 inverted confocal microscope with AiryScan detector and imaged with a 25× objective lens using 488 nm excitation and 525 nm emission settings, an acquisition rate of 0.5 frames/second. During image acquisition, cells were perfused in PBS supplemented with 1 mM CaCl₂ and 0.55 mM MgCl₂ for at least 2 min, followed by manually starting and stopping treatment. Treatment times in Figure 4 and the resulting temporal alignment used in data analysis represent time at which treatment was manually initiated; thus, a delay of several sec is expected between time of treatment and effect onset.

Quantification and statistical analysis

Analysis of mass spectrometry data

Data was processed using (software). Calibration curves ranging from X to Y 2-AG were used to calculate 2-AG content in samples. Data was analyzed using Prisma and statistical significance is represented by ∗p < 0.05, ∗∗p < 0.01, and ∗∗∗p < 0.001 significantly different from vehicles as determined using one-way ANOVA followed by Tukey’s post hoc test.

Cell segmentation and quantification of intracellular and plasma membrane GRAB_eCB2.0 fluorescent signal

To quantify changes in GRAB_eCB2.0 fluorescent signal (ΔF/F₀) at the plasma membrane and intracellular membranes from live cell confocal microscopy images, an intensity-based plasma membrane segmentation protocol using FIJI ImageJ differentiated the plasma membrane GRAB_eCB2.0 fluorescent signal from the intracellular GRAB_eCB2.0 fluorescent signal. Due to irregularities in cell shape, size, and fluorescence intensity over time, we performed a manual segmentation method using FIJI ImageJ in a blinded manner by using de-identified image files and employing personnel not involved in image acquisition. First, a single image representing the average fluorescent signal over a 12 min imaging period (2 min baseline and 10 min treatment with either BK or mastoparan) was obtained. An average image was used to account for changes in signal intensity and cell morphology of the GRAB_eCB2.0-Neuro2a cells over time (Figure S2). This average fluorescent signal image was processed by applying a smoothing step using a Gaussian blur convolution function (σ = 2) to ensure that the plasma membrane region and intracellular region appear as continuous, uninterrupted regions despite morphological changes or fluctuations in signal intensity and allowed for the analysis to include regions that had lower signal intensities. Next, plasma membrane segmentation was performed on the processed image by setting a minimum intensity-based threshold such that the pixels included formed a continuous ring as would be expected from the plasma membrane, without the presence of intracellular structures. This visual marker for thresholding was used since the cells exhibited differences in fluorescence intensities and a single threshold could not accurately define the plasma membrane in the heterogeneous population. Finally, a particle analysis function was used to create a binary mask for the plasma membrane as it has a greater intensity than the signal in the intracellular region. A mask for the whole cell was created in the same way, except with the change that a “include holes” setting was used during the particle analysis step. The intracellular mask was obtained by inverting the threshold settings used to obtain the plasma membrane mask; a different threshold setting was not applied to ensure that there was no overlap between the plasma membrane mask and intracellular region mask. To analyze changes in GRAB_eCB2.0 fluorescent signal (ΔF/F₀) at the plasma membrane and intracellular regions, the live cell microscopy timelapse image set was combined with the corresponding mask (plasma membrane or intracellular), the fluorescent signal was measured for each image and the baseline and change in fluorescence was performed using the ΔF/F₀ calculation described in the Data Analysis section. The cells analyzed using this protocol were filtered such that cells that overlapped or were clustered and could not be differentiated from one another were excluded from the analysis. The membrane marker wheat germ agglutinin (WGA) was used to confirm plasma membrane colocalization of GRAB_eCB2.0 fluorescent signal and this colocalization was visualized using a line scan analysis of a representative cell.

Quantification of coefficient of variation

To quantify coefficient of variation of the WGA signal for each GRAB_eCB2.0-expressing Neuro2a cell, a timelapse series that included a 2-min baseline, and 10-min treatment was first grouped into bins of 10 images. Next, for each the bin the mean and standard deviation of the WGA fluorescent signal was calculated for a manually defined region encompassing the entire cell, and finally the average coefficient of variation of all the pixels in this defined area was calculated for each bin using the following equation: $\text{coefficient}\text{of}\text{variation}=\frac{standarddeviation}{mean}$.

Quantification of GRAB_eCB2.0 fluorescent signal

The fold change in GRAB_eCB2.0 fluorescent signal (ΔF/F₀) was calculated using the following equation: $\frac{\Delta F}{F_0}=\frac{F_{x-}{F}_0}{F_0}$. F₀ is baseline fluorescent signal and F_x is the fluorescent signal at a specific time point. The GRAB_eCB2.0 fluorescent signals measured using live cell confocal microscopy were calculated using FIJI ImageJ using the following method: cells were manually identified and a region of interest encompassing each cell was selected, fluorescence level of each cell was then calculated over an entire time course, and finally baseline fluorescence (F₀) was determined by averaging the fluorescent signal over a 30 s interval occurring approximately 30 s prior to start of agonist treatment. The GRAB_eCB2.0 fluorescent signals measured using the 96-well fluorescence plate reader was calculated using MATLAB as previously described¹⁴ and the baseline fluorescence (F₀) was determined by averaging the fluorescent signal over a 1 min basal reading and the plate reader results represent the average ΔF/F₀ values for each set of technical replicates, performed in triplicate. Identification of maximum GRAB_eCB2.0 responses induced by BK and mastoparan at 5 min was done using PRISM and their GRAB_eCB2.0 decay responses calculated as A.U.C. between 5- and 30-min. Biological and technical replicates were combined. Error bars are smaller than each dot. No concentration response curves were fitted. To facilitate data analysis, we developed a MATLAB algorithm that averages the fluorescence value measured in each well over time, for multiple experiments and at select timepoint as recently published.⁴⁸ Data are shown as mean +S.E.M and significance was determined by running a two-Way ANOVA followed by Bonferroni test using GraphPad Prism. Non-significant = ns > 0.05, ∗p < 0.05, ∗∗p < 0.01 and ∗∗∗p < 0.001.

Data and statistical analyses

The fold change in GRAB_eCB2.0 fluorescent signal (ΔF/F₀) was calculated using the following equation: $\frac{\Delta F}{F_0}=\frac{F_{x-}{F}_0}{F_0}$. F₀ is baseline fluorescent signal and F_x is the fluorescent signal at a specific time point. The GRAB_eCB2.0 fluorescent signals measured using live cell confocal microscopy were calculated using FIJI ImageJ using the following method: cells were manually identified and a region of interest encompassing each cell was selected, fluorescence level of each cell was then calculated over an entire time course, and finally baseline fluorescence (F₀) was determined by averaging the fluorescent signal over a 30 s interval occurring approximately 30 s prior to start of agonist treatment. The GRAB_eCB2.0 fluorescent signals measured using the 96-well fluorescence plate reader was calculated using MATLAB as previously described¹⁴ and the baseline fluorescence (F₀) was determined by averaging the fluorescent signal over a 1 min basal reading and the plate reader results represent the average ΔF/F₀ values for each set of technical replicates, performed in triplicate. Identification of maximum GRAB_eCB2.0 responses induced by BK and mastoparan at 5 min was done using PRISM and their GRAB_eCB2.0 decay responses calculated as A.U.C. between 5- and 30-min. Biological and technical replicates were combined. Error bars are smaller than each dot and thus do not appear on the graphs. Considering the multiple molecular steps that are involved between receptor activation and increase in GRAB_eCB2.0 signal, no concentration response curves were fitted. To facilitate data analysis, we developed a MATLAB algorithm that averages the fluorescence value measured in each well over time, for multiple experiments and at select timepoint as recently published.⁴⁸ Data are shown as mean + S.E.M and significance was determined by running a Two-Way ANOVA with Dunnett’s Multiple Comparison test using GraphPad Prism.

Published: October 28, 2025