The cannabinoid agonist WIN55,212-2 increases intracellular calcium via CB₁ receptor coupling to G_q/11 G proteins

Abstract

Central nervous system responses to cannabis are primarily mediated by CB₁ receptors, which couple preferentially to G_i/o G proteins. Here, we used calcium photometry to monitor the effect of CB₁ activation on intracellular calcium concentration. Perfusion with 5 μM CB₁ aminoalkylindole agonist, WIN55,212-2 (WIN), increased intracellular calcium by several hundred nanomolar in human embryonic kidney 293 cells stably expressing CB₁ and in cultured hippocampal neurons. The increase was blocked by coincubation with the CB₁ antagonist, SR141716A, and was absent in nontransfected human embryonic kidney 293 cells. The calcium rise was WIN-specific, being essentially absent in cells treated with other classes of cannabinoid agonists, including Δ⁹-tetrahydrocannabinol, HU-210, CP55,940, 2-arachidonoylglycerol, methanandamide, and cannabidiol. The increase in calcium elicited by WIN was independent of G_i/o, because it was present in pertussis toxin-treated cells. Indeed, pertussis toxin pretreatment enhanced the potency and efficacy of WIN to increase intracellular calcium. The calcium increases appeared to be mediated by G_q G proteins and phospholipase C, because they were markedly attenuated in cells expressing dominant-negative G_q or treated with the phospholipase C inhibitors U73122 and ET-18-OCH₃ and were accompanied by an increase in inositol phosphates. The calcium increase was blocked by the sarco/endoplasmic reticulum Ca²⁺ pump inhibitor thapsigargin, the inositol trisphosphate receptor inhibitor xestospongin D, and the ryanodine receptor inhibitors dantrolene and 1,1′-diheptyl-4,4′-bipyridinium dibromide, but not by removal of extracellular calcium, showing that WIN releases calcium from intracellular stores. In summary, these results suggest that WIN stabilizes CB₁ receptors in a conformation that enables G_q signaling, thus shifting the G protein specificity of the receptor.

The CB₁ cannabinoid receptor (CB₁) mediates the majority of the psychotropic and behavioral effects of cannabis (1, 2). CB₁ is a member of the heptahelical G protein-coupled receptor (GPCR) superfamily (1). It couples via pertussis toxin (PTX)-sensitive G_i/o G proteins to inhibit adenylyl cyclase and L-, N-, and P/Q-type calcium channels and to activate potassium channels and mitogen-activated protein kinase (2). Rarely, CB₁ has also been found to couple to phospholipase C (PLC) in a PTX-sensitive manner involving the βγ subunits from G_i/o (3, 4).

Several distinct agonists activate CB₁. The classical cannabinoids are tricyclic dibenzopyran compounds. The prototype of this group is Δ⁹-tetrahydrocannabinol (THC), the main psychoactive ingredient of cannabis. THC is a low-affinity partial agonist for CB₁ (2), whereas other synthetic classical compounds such as HU-210 are both more potent and efficacious (2). Nonclassical cannabinoids lack the dihydropyran ring yet maintain some similarity to THC's stereochemistry. The best known, CP55,940 (CP), is a potent and efficacious CB₁ agonist. The third group, the eicosanoids, includes endogenous ligands for CB₁, such as anandamide and 2-arachidonoylglycerol (2-AG), and their analogs (5, 6). The aminoalkylindoles (AAIs) were synthesized as antiinflammatory drugs, with the goal to reduce gastrointestinal side effects. When tested, they were first found to have potent antinociceptive activity and weak antiinflammatory properties (7) and later discovered to bind to cannabinoid receptors (8). AAIs, and in particular WIN55,212-2 (WIN), produce the full spectrum of in vivo effects seen with THC and other cannabimimetic agonists (9).

Here, we are concerned with the G protein specificity of CB₁. Initial dogma was that each GPCR interacts with a certain class of G protein. Subsequently, more promiscuous GPCR-G protein coupling was recognized. Multiplicity in G protein coupling activates multiple second-messenger cascades, each with distinct effects. Such promiscuity should allow a cell greater range in response to activation of a single GPCR. Some examples include D₁ dopamine and α₂-adrenergic receptors, which couple to both G_s and G_i, and the PACAP type I and H₂ histamine receptors, which activate both G_s and G_q/11 (10-13). CB₁ has been shown to activate both G_i/o and G_s (14). This multiplicity in G protein coupling can be agonist-regulated. The prevailing idea here is that receptors exist in different activated states capable of coupling to different G proteins. Binding of a specific agonist stabilizes a distinct active state, favoring coupling to a certain G protein (15, 16). Examples of this have been reported for A₁ adenosine receptors and the Ca²⁺-sensing receptor (17, 18). In this study, we report the previously undescribed observation that CB₁ couples to G_q/11 G proteins in an agonist-specific manner to increase the concentration of intracellular calcium.

Materials and Methods

Cells. Human embryonic kidney (HEK) 293 cells stably expressing rat CB₁ cannabinoid receptor (CB₁-HEK293) were generated by using standard techniques (19). Most experiments were done with cells expressing CB₁ at a density of 2.2 pmol/mg protein. Before an experiment, cells were plated onto poly d-lysine-coated coverslips and, when required, transfected the next day. cDNA for additional genes (M₁R and dominant-negative G_αq) were transiently transfected with Lipofectamine 2000. DsRed (0.5 μg) was cotransfected as a transfection marker. Cells were used for photometry experiments 24-36 h after transfection. Hippocampal neurons were cultured according to the method of Brewer (20).

Dye Loading and Photometry. Cytosolic Ca²⁺ was monitored with the ratiometric calcium indicator fura-2. Cells were loaded at room temperature for 15-20 min with fura-2 acetoxymethyl ester (AM) dissolved in DMSO, dispersed in 20% pluronic 127, and diluted to 8 μM in 0.5 ml of Ringer's solution. Test solutions were applied to the incubation chamber with a solenoid-controlled gravity-fed multibarreled local perfusion device.

During [Ca²⁺]_i measurements, the dye was excited by 340 and 380 nm light and a photodiode collected emission above 520 nm (Polychrome IV TILL Photonics, Planegg, Germany). The standard calibration parameters (21), R_min (0.35), R_max (2.5), and K^* (613 nM), were determined from HEK293 cells equilibrated in KCl-based internal solutions containing ionomycin (10 μM) and either 20 mM EGTA, 15 mM CaCl₂, or 20 mM EGTA + 15 mM CaCl₂ (149 nM free Ca²⁺). For experiments with hippocampal neurons, the relative [Ca²⁺]_i is given as the fluorescence ratio F₃₄₀/F₃₈₀. Photometric measurements were analyzed in Igor (WaveMetrics, Lake Oswego, OR) and statistical analyses performed in excel (Microsoft), except ANOVAs, which were performed on the statistical methods web site, developed by Tom Kirkman at the College of St. Benedict/St. John's University (St. Joseph and Collegeville, MN). All results are presented as mean ± SEM. Each drug was tested on at least 2 different days, with concurrent interleaved controls. Each calcium measurement represents a single cell from an individual coverslip. The measurement “rise in [Ca²⁺]_i” was calculated as: maximum [Ca²⁺]_i during drug application - basal [Ca²⁺]_i, where basal [Ca²⁺]_i was the mean [Ca²⁺]_i for 47 s before drug application.

Receptor Expression Studies and Measurement of [³H]Inositol Phosphate (IP) Formation. To determine the level of CB₁ expression, CB₁-HEK293 cells were grown in 10-cm plates and membrane binding performed with [³H]CP, as described (22). [³H]Inositol phosphate production was measured in cells loaded with myo-[2-³H]inositol by using established techniques (23).

Solutions and Materials. Fura-2-AM and pluronic 127 were from Molecular Probes, and thapsigargin (TG), xestospongin D (XeD), and ET-18-OCH₃ were from Calbiochem. PTX was from List Biological Laboratories (Campbell, CA); and HU210, U73122, and 1,1′-diheptyl-4,4′-bipyridinium dibromide were from Tocris Cookson (Ellisville, MO). SR141716A (SR), THC, and CP were supplied by the National Institute on Drug Abuse Research Resources Drug Supply System. Zeocin, DMEM, FBS, Lipofectamine 2000, and penicillin/streptomycin were from Invitrogen. The dominant-negative G_αq (Q209L/D277N) construct was obtained from the Guthrie Research Institute (Sayre, PA). AM356 was a gift of A. Makriyannis (Northeastern University, Boston). Cannabidiol, [³H]CP, and myo-[2-³H]inositol were gifts from N. Stella (University of Washington, Seattle). All other chemicals were from Sigma.

The control mammalian Ringer's bath solution contained 160 mM NaCl, 2.5 mM KCl, 2 mM CaCl₂, 1 mM MgCl₂, 10 mM Hepes, and 8 mM glucose adjusted to pH 7.4 with NaOH. Ca²⁺-free Ringer's solution was the same as control solution without CaCl₂ and with 0.5 mM EGTA. PTX and oxotremorine-M (Oxo-M) were dissolved in water. All other compounds were dissolved in DMSO. Final dilutions were made with Ringer's bath solution (DMSO <0.05%). One milligram per milliliter BSA was added as a carrier to all cannabinoid-containing solutions except those used in the receptor expression and IP measurements. After experiments, perfusion lines and tip were flushed with 100% ethanol followed by dH₂O. For the receptor expression and IP formation assays, siliconized tubes and pipette tips were used for drug preparation and delivery.

Results

WIN Evokes a Large Slow CB₁-Specific Increase in Intracellular Calcium Concentration Perfusion of 5 μM of the cannabimimetic AAI WIN raised intracellular calcium in CB₁-HEK293 cells by >500 nM (Fig. 1A). Perfusion of 5 μM WIN also increased intracellular calcium in HEK293 cells transiently expressing rCB₁ (790 ± 120 nM, n = 9). The CB₁ specificity of this response was shown in three ways. First, coapplication of 1 μM of the CB₁ antagonist SR with WIN markedly attenuated the calcium increase (Fig. 1 A). Second, perfusion of 5 μM WIN55,212-3, the inactive enantiomer of WIN (2), did not increase intracellular calcium (data not shown). Third, the transient was absent in nontransfected HEK293 cells perfused with WIN (Fig. 1 A). WIN caused a calcium transient in 100% of cells expressing CB₁ (121/121), although the amplitude and speed of the response varied among cells (Fig. 1B). Overall, the WIN-induced calcium increase was markedly slower than that caused by 10 μM of the muscarinic receptor agonist, Oxo-M, in HEK293 cells expressing M₁R. The M₁R-mediated calcium increase had a 10-90% rise time of 4 ± 1 s, significantly faster than the WIN-induced calcium rise time of 40 ± 8 s (P < 0.01) (Fig. 1 C and D).

Fig. 1.

Activation of CB₁ by WIN increased [Ca²⁺]_i to a variable extent and with varied kinetics. (A) The time course of changes in [Ca²⁺]_i in HEK293 cells loaded with fura-2 and perfused with 5 μM WIN. The lines above indicate drug application. The solid line indicates the WIN-induced response in CB₁-HEK293 (n = 8). A subset of CB₁-HEK293 were coperfused with 1 μM of the CB₁ antagonist SR (dashed line, n = 5). The change in [Ca²⁺]_i in response to WIN in nontransfected HEK293 cells is also shown (dotted line, n = 8). (B) Calcium elevations with WIN recorded in four different CB₁-HEK293 cells, tested on 4 different days. (C)CB₁-HEK293 were perfused with 5 μM WIN (solid line), and HEK293 cells transiently expressing M₁R were perfused with 10 μM Oxo-M (dashed line). These traces show the change in calcium in representative cells. (D) The black bar shows the 10-90% rise time for the calcium caused by 5 μM WIN in CB₁-HEK293 (n = 6). Open bar indicates the calcium rise time caused by a 40-s application of 10 μM Oxo-M in HEK293 cells transiently expressing M₁R (n = 7). The rise times were significantly different (P < 0.01, t test).

The effect of WIN is not restricted to expression systems. Five micromolar WIN caused a calcium rise in cultured mouse hippocampal neurons like that in CB₁-HEK293 (the F_340/380 ratio increased to 0.9, n = 9). This response also depended on CB₁, because 1 μM of the antagonist SR attenuated WIN-induced F_340/380 increase by 76% (P < 0.001). WIN had no effect on calcium levels in astrocytes, cells generally regarded to not express CB₁ (data not shown).

WIN Is the Only CB₁ Agonist to Increase Intracellular Calcium Robustly. We tested several classes of CB₁ agonists at concentrations at least 200 times the CB₁-binding K_d value [with the exception of 2-AG, which was used at 20 times its K_d (2)]. THC (10 μM) raised calcium by only 40 ± 15 nM (Fig. 2), which is well within the range seen for resting cells and was only 5% of the WIN-induced increase. Agonists structurally similar to THC, such as cannabidiol (CBD, 3 μM), also showed no significant effects, with the exception of 1 μM HU-210, which increased calcium slightly in CB₁-HEK293 cells (Fig. 2), but not in hippocampal neurons. Neither the nonclassical cannabinoid, CP55,940 (CP, 1 μM), nor the endogenous cannabinoid, 2-AG (1 μM), nor the metabolically stable anandamide analog, methanandamide (AM356, 10 μM), caused a significant calcium increase (Fig. 2). The increase in intracellular calcium by WIN was significantly more than for other agonists (ANOVA, 95% confidence interval).

Fig. 2.

Only the AAI agonist WIN caused a large increase in [Ca²⁺]_i in CB₁-HEK293. The WIN-induced change in [Ca²⁺]_i (black bar, n = 5) is significantly greater than that of other agonists (P < 0.05, ANOVA). Agonists tested: the classical cannabinoids, THC (10 μM, n = 9), HU-210 (1 μM, n = 4), the phytocannabinoid, cannabidiol (CBD, 3 μM, n = 4), the nonclassical cannabinoid, CP (n = 6), the endocannabinoid, 2-AG (n = 5), and the anandamide analog, AM356 (n = 9). All agonists were perfused for 150 s.

WIN Increases Intracellular Calcium via PTX-Insensitive G Proteins from the G_q/11 Family. To determine whether the WIN-evoked calcium increase was mediated by members of the G_i/o protein family, cells were pretreated for 16 h with 500 ng/ml PTX. Local perfusion of these cells with increasing concentrations of WIN revealed a concentration-dependent increase in intracellular calcium. Unexpectedly, the increase was augmented by PTX at all WIN concentrations (Fig. 3A). This augmentation was particularly dramatic for 1 μM WIN, where PTX increased the calcium rise by 1,200%. PTX increased the 3 μM WIN-evoked calcium rise by 220%, and the 5 μM WIN-evoked calcium rise by 120% (Fig. 3A). PTX pretreatment did not affect the 10-90% rise times for the WIN-evoked calcium increase (Fig. 3B). The calcium increase in PTX-treated cells was mediated by CB₁ receptors, because it was blocked by 1 μM SR (data not shown).

Fig. 3.

PTX pretreatment augmented the WIN-induced [Ca²⁺]_i increase in CB₁-HEK293. Black bars indicate the response in untreated cells, and open bars indicate overnight treatment with 500 ng/ml PTX. (A) Pretreatment with PTX enhanced the rise in [Ca²⁺]_i at all WIN concentrations. n = 5 for all conditions. (B) PTX pretreatment had no effect on the 10-90% rise time for the calcium rise following perfusion of 3 or 5 μM WIN.

In cells pretreated with heat-inactivated PTX, the WIN-increased calcium was like that in untreated cells (data not shown). The WIN-induced calcium rise in cultured hippocampal neurons was also PTX-insensitive. WIN increased the F_340/380 ratio to 0.5 ± 0.1 (n = 4). After PTX treatment, WIN increased the F_340/380 ratio to 0.4 ± 0.1 (n = 4). Activity of the PTX toward G_i/o was confirmed by its ability to block CP-stimulated activation of mitogen-activated protein kinase (data not shown). Together, these data show that the pathway we describe here is not inhibited by PTX, so it does not involve G_i/o G proteins.

PTX pretreatment enhanced the calcium rise in response to 1 μM CP to a much lesser extent. CP alone caused a negligible calcium increase of 10 ± 1 nM (n = 6). Overnight treatment with PTX increased the calcium rise in response to CP to 28 ± 5 nM (n = 5, P < 0.05). Unlike PTX, heat-inactivated PTX did not augment the response to CP (8 ± 3 nM, n = 5). PTX pretreatment had small and nonsignificant effects on the calcium rise following perfusion of 10 μM THC and 10 μM AM356 (data not shown).

Because inhibition of G_i/o G proteins did not reduce the WIN-induced calcium increase, we investigated the importance of the PTX-resistant G_q/11 family of G proteins. Transient expression of a dominant-negative G_αq (DN G_αq, Q209L/D277N) (24) suppressed the WIN-induced transient by 70% (P < 0.001, Fig. 4A). M₁R activation increases intracellular calcium by G_q G proteins (25). Thus, as a positive control for DN G_αq, we checked its action on the M₁R-mediated calcium transient. In cells transiently expressing M₁R, overexpression of DN G_αq reduced the 10 μM Oxo-M calcium increase by 40% (P < 0.05, Fig. 4A). Thus, G_q/11 G proteins are involved in the WIN-induced calcium increase.

Fig. 4.

G_q/11 is involved in the WIN-induced increase in [Ca²⁺]_i.(A) The rise in [Ca²⁺]_i during perfusion of either 5 μM WIN (left three bars) or 10 μM Oxo-M (right two bars) is plotted. Control cells were not treated with PTX and did not express dominant negative G_αq (black bars, n = 10 for WIN control, n = 4 for Oxo-M control). The rise in [Ca²⁺]_i in HEK293 cells expressing DN G_αq was less than control (P < 0.001 for WIN-treated and P < 0.05 for Oxo-M treated, t test). PTX pretreatment (open bar, n = 5) augments the rise in calcium (P < 0.05, t test). (B) The change in calcium upon WIN application in a representative cell for each treatment. Solid line indicates WIN only, dashed line is WIN+DN G_αq, and dotted line is WIN+PTX.

The WIN-Induced Calcium Rise Depends on PLC Activation. The major downstream effector for G_αq is PLCβ (26). PLCβ is a likely candidate for the pathway identified in this study, because it cleaves the membrane lipid phosphatidylinositiol-4,5-bisphosphate into the second messengers inositol 1,4,5-trisphosphate (IP₃) and diacylglycerol, both of which modulate intracellular calcium. To test for the involvement of PLC, CB₁-HEK293 cells were pretreated for 160 s with 3 μM of the PLC inhibitor U73122 followed immediately by 5 μM WIN. This concentration of U73122 has been shown to inhibit PLC (27) yet not increase intracellular calcium on its own (Fig. 5B). The latter is an important point, because U73122 at concentrations >3 μM can increase intracellular calcium concentration by releasing calcium from intracellular stores (28). U73122 (3 μM) suppressed the 5 μM WIN-evoked calcium transient by 80% (P < 0.01, Fig. 5A). The same concentration of its inactive analog, U73343, affected neither the calcium rise in response to 5 μM WIN (n = 4, Fig. 5A) nor intracellular calcium by itself (Fig. 5B). As a positive control for U73122, we verified that it inhibited the M₁R-induced calcium rise, which requires PLCβ (29). In cells transiently expressing M₁R, 3 μM U73122 decreased the calcium rise in response to 5 μM Oxo-M by 70%. The phosphatidylinositol-specific PLC inhibitor, ET-18-OCH₃ (30), also attenuated the WIN-induced calcium rise. A 20-min pretreatment with 15 μM ET-18-OCH₃ reduced the calcium increase by 70% (180 ± 60 nM; n = 6) compared with that in control cells treated with 5 μM WIN (650 ± 145 nM; n = 12; P < 0.001; t test).

Fig. 5.

PLC mediates the increase in [Ca²⁺]_i by WIN. (A) CB₁-HEK293s were either untreated (black bar, n = 12) or pretreated for 160 s with either 3 μM of the PLC inhibitor U73122 (n = 5) or 3 μM of its inactive analog U73343 (n = 4) followed immediately by perfusion of 5 μM WIN for 150 s. U73122 strongly suppressed the calcium rise (P < 0.01, t test), whereas U73343 had no significant effect. (B) Neither 3 μM U73122 (solid line) nor 3 μM U73343 (dashed line) increased [Ca²⁺]_i on its own.

Accumulation of IP derivatives is a measure of PLC activity. In CB₁-HEK293 cells, 5 μM WIN (n = 6) increased IP levels by 90% over basal amounts (measured in untreated cells, n = 9, P < 0.01; t test), as well as 80% over amounts measured in nontransfected cells (n = 6, P < 0.01; t test) stimulated with the same concentration of WIN (Fig. 6). Incubation with 3 μM WIN also increased IP levels (Fig. 6). As a positive control, 10 μM Oxo-M increased IP accumulation by 210% over basal amounts in HEK293 cells transiently expressing M₁R (n = 9, P < 0.002; t test), and by 150% over levels in nontransfected cells also treated with 10 μM Oxo-M (n = 6, P = 0.006; t test, Fig. 6). IP production was PLC-specific: 3 μM U73122 suppressed both the 3 μM and 5 μM WIN-mediated increases in IP accumulation by 70%, and the 10 μM Oxo-M-mediated increase by 80% (data not shown).

Fig. 6.

WIN increases IP levels. CB₁-HEK293 transiently transfected with M₁R (solid bars, n = 9) and nontransfected HEK293 cells (open bars, n = 6) were stimulated with agonist for 10 min and then assayed for IP production. IP levels increased in CB₁-HEK293 treated with either 3 or 5 μM WIN compared with basal levels or levels in nontransfected controls. In cells expressing M₁R, 10 μM Oxo-M robustly increased IP accumulation. *, comparison with basal control; †, comparison with nontransfected HEK293 cells treated with the same agonist concentration. * and †, P < 0.05; ** and ††, P < 0.01.

The WIN-Evoked Calcium Rise Comes from IP₃- and Ryanodine-Sensitive Intracellular Calcium Stores Two main candidate calcium sources could contribute to the intracellular calcium increase. First, extracellular calcium might enter through plasmalemmal ion channels. Second, calcium might be released from the endoplasmic reticulum (ER) or another intracellular pool. A role for extracellular calcium was assessed by using a calcium-free medium (0 Ca Ringer's). External calcium entry did not appear to be important, because the calcium transient was unaffected when WIN was delivered in 0 Ca Ringer's (Fig. 7). The latency to onset and the time course of the calcium increase were unchanged (Fig. 7 Inset).

Fig. 7.

WIN releases calcium from IP₃- and Ry-sensitive intracellular stores. In CB₁-HEK293, the 5 μM WIN-induced rise in [Ca²⁺]_i did not require extracellular calcium (“0 Ca,” n = 12 vs. “Control,” black bar, n = 16). One micromolar TG (n = 13), the IP₃R inhibitor XeD, 1 μM, n = 10), the RyR inhibitors dantrolene (Dan, 10 μM, n = 11) and 1,1′-diheptyl-4,4′-bipyridinium dibromide (DHBP) (50 μM, n = 6) all reduced calcium rises in response to WIN (P < 0.05, ANOVA), as did coapplied XeD and Dan (n = 4). The reversible SERCA pump inhibitor 2,5-di-t-butyl-1,4-benzohydroquinone (BHQ) in the absence of WIN, modestly increased calcium (100 μM, n = 5, P < 0.05, ANOVA). (Inset) Removal of extracellular calcium (“0 Ca,” open bar, n = 12) affected neither the latency nor rise time of the calcium transient (“Control,” black bar, n = 16). Latency was calculated as the time from start of drug application until the [Ca²⁺]_i rose to 10% of the maximum value.

Because CB₁ activation did not promote calcium entry, it was likely that calcium was being released from intracellular stores. The sarco/endoplasmic reticulum Ca²⁺ (SERCA) pump is an ATP-dependent pump located on the ER surface that maintains and replenishes agonist-sensitive calcium stores. If the SERCA pump is blocked, the stores empty as ER calcium leaks out and is not replaced. CB₁-HEK cells transiently transfected with M₁R were incubated with 1 μM of the SERCA pump inhibitor thapsigargin (TG) in 0 Ca Ringer's for 20 min before the experiment, a treatment sufficient to empty TG-sensitive intracellular stores (31). TG pretreatment reduced the 5 μM WIN calcium transient by 90% (P < 0.05 by ANOVA, Fig. 7). As a positive control, the same TG treatment almost abolished the calcium increase from M₁R activation, known to be mediated by TG- and IP₃-sensitive stores (29) (data not shown). These observations suggest that either the calcium is released from TG-sensitive intracellular calcium stores or that WIN inhibits the SERCA pump. In the latter case, the rise in intracellular calcium concentration would be due to leakage of calcium from SERCA pump-dependent stores. To test this possibility, we applied 100 μM 2,5-di-t-butyl-1,4-benzohydroquinone (BHQ), a fast and reversible SERCA pump inhibitor, and measured the change in intracellular calcium. We reasoned that if BHQ perfusion increased calcium to a lesser extent than WIN, we could discount the possibility that the actions of WIN were solely due to SERCA pump inhibition. BHQ treatment elicited a calcium increase only 10% of that seen with WIN (P < 0.05 by ANOVA, Fig. 7). This suggests that inhibition of SERCA pump activity is not the major cause of the WIN-evoked calcium transient.

IP₃-sensitive ER calcium stores were a likely source contributing to our calcium increase because IP₃ is produced by PLCβ, which is central in the pathway (Figs. 5 and 6). A 30-s pretreatment with 1 μM of the IP₃R inhibitor xestospongin D (XeD) in 0 Ca Ringer's immediately followed by perfusion of 5 μM WIN reduced the calcium rise by 75% (P < 0.05 by ANOVA, Fig. 7). The ryanodine receptor (RyR) also plays a role in calcium release, because pretreatment for 20-25 min with 10 μM of the RyR inhibitor dantrolene in 0 Ca Ringer's reduced the calcium rise by 50% (P < 0.05 by ANOVA, Fig. 7). As further evidence for RyR involvement, treatment with its inhibitor, 1,1′-diheptyl-4,4′-bipyridinium dibromide (DHBP, 50 μM) attenuated the WIN-evoked calcium increase by 70% (P < 0.05 by ANOVA, Fig. 7). Pretreatment with both XeD and dantrolene decreased calcium transients to 15% of that seen in untreated cells perfused with WIN (P < 0.05 by ANOVA, Fig. 7). These data suggest that the calcium transient is the result of calcium release from IP₃- and Ry-sensitive intracellular stores.

Discussion

We have found that WIN, a cannabinoid AAI agonist, can increase intracellular calcium in CB₁-HEK293 cells, as well as in cultured hippocampal neurons. Because the WIN-induced calcium increase was absent in untransfected HEK293 cells and was blocked by a CB₁ antagonist, it is specific. The large slow WIN-induced rise in intracellular calcium was PTX-insensitive and required G_q/11 G proteins and PLC. WIN increased phosphatidyl inositide turnover, as measured by IP production, and the IP₃ released calcium from IP₃-sensitive ER stores. RyRs also contributed to the calcium release.

Other groups have reported instances where CB₁ activation increases intracellular calcium (3, 4, 32-35). Generally, the reported calcium rises were more modest than those we report here. Unlike our results, the calcium rise in the earlier studies was mediated by PTX-sensitive G_i/o βγ acting via PLC to release calcium from TG-sensitive intracellular stores (3, 4, 32, 33, 35). The one exception was a report that the nonclassical cannabinoid agonist, DALN, increases intracellular calcium by enhancing the entry of calcium ions through l-type calcium channels by a PTX-insensitive mechanism (34). Because HEK293 cells do not have l-type voltage-sensitive calcium channels, this could not be the mechanism for the calcium increase we report here. Thus, our study provides evidence that CB₁ functionally couples to G proteins from the G_q/11 family to increase intracellular calcium.

The WIN-induced calcium increase identified here is agonist-specific. For example, WIN was more potent and efficacious in increasing intracellular calcium than CP. In contrast, WIN is less potent than CP at inhibiting cAMP production, a G_i-coupled pathway (36). As a working hypothesis, we propose that CB₁ exists in several active states; the relative ratio of each is a function of the ligand. WIN likely stabilizes a conformation of CB₁ that couples more readily to G_q/11. The reversal of agonist potency in the inhibition of cAMP production (36) supports this hypothesis. Similarly, accumulating evidence suggests that the binding site for WIN only partially overlaps that for other CB₁ agonists. Although WIN displaces CP from CB₁ in radioligand-binding assays (2), mutagenesis studies uncovered an amino acid residue (K192) in helix three of CB₁ that, when mutated to alanine, abolished CP binding with little effect on binding or receptor activation by WIN (37). A second mutation, where V282 in helix five was mutated to phenylalanine, increased the affinity of CB₁ for WIN with no effect on affinity for CP (38). Favored interactions with distinct residues may result in WIN preferentially stabilizing a specific active conformation of CB₁ that couples not only to G_i/o but also to G_q/11. Other structurally related AAIs may also enhance CB₁ coupling to G_q/11.

A relatively high concentration of WIN (5 μM) was required for a substantial calcium increase (>400 nM); however, many published studies use concentrations in this range or higher. WIN-activated CB₁ appears to couple preferentially to G_i/o, and only when receptor occupancy is very high or when the ratio of G_αq to functional G_i/o increases will coupling to less-favored G proteins, such as G_q/11, be evident. This could explain how PTX pretreatment enhanced the potency of WIN. By inhibiting the G_i/o α subunit, PTX prevents G_i/o heterotrimeric G proteins from interacting with the receptor (39). This increases the likelihood of CB₁ coupling to G_q/11, leading to an enhanced calcium response at the same WIN concentration. PTX pretreatment also unmasked modest coupling between G_q/11 and CB₁ activated by CP and THC, but even after PTX, CP and THC were still far less efficacious than WIN.

Our study is not the first example of agonist-driven coupling of CB₁ to a less-favored G protein. Bonhaus et al. (40) reported a difference in the ability of cannabinoid agonists to activate G_s-coupled pathways. They showed that WIN was more potent and efficacious than CP in increasing forskolin-stimulated cAMP accumulation, yet CP was more potent than WIN in inhibiting adenylyl cyclase (40).

There are other examples of promiscuous receptors activating different signal transduction pathways in an agonist-dependent manner that may be explained by different receptor conformations coupling to different G proteins. Rey et al. (18) have studied agonist-specific signal trafficking of the Ca²⁺-sensing receptor (CaR) (18). They find that the CaR activated by external calcium stimulates G_q/PLC, increasing [Ca²⁺]_i in an oscillatory fashion. However, if the aromatic amino acid l-phenylalanine activates CaR, the receptor no longer couples to G_q and instead transiently increases [Ca²⁺]_i through the activation of the small GTPase Rho and G proteins of the G₁₂ subfamily. A second example is from Cordeaux et al. (41), who show that the A₁ adenosine receptor, traditionally described as G_i-coupled, can also couple to G_s and G_q. The efficacy of this coupling depends on the agonist. The two agonists N(6)-cyclopentyladenosine (CPA) and 5′-(N-ethylcarboxamido)adenosine (NECA) are equally efficacious in G_i coupling. After PTX treatment in cells expressing a low level of A₁R, NECA, but not CPA, promoted A₁R-G_s interactions. After PTX treatment in cells expressing a high level of A₁R, receptors activated by NECA coupled to G_q more efficaciously than CPA-activated receptors.

CB₁ generally signals more slowly then other GPCRs, even when modulating the same response. For example, Mackie and Hille (42) found that WIN-activated CB₁ inhibited N-type calcium channels 15-fold more slowly than the norepinephrine-activated α₂-adrenergic receptor, even at supramaximal agonist concentrations. In the current study, the difference was just as pronounced. The WIN-induced calcium increase was 10-fold slower than the Oxo-M-induced calcium increase (Fig. 1D). The reason for this marked kinetic difference is unknown. It may be that CB₁ assumes its activated conformation relatively slowly. Another possibility is that CB₁ activates G proteins less efficiently compared with other GPCRs (43). In this case, it would take longer to reach the threshold concentration of downstream effectors necessary to initiate a response.

Our data indicate that unique among CB₁ agonists, WIN releases calcium from intracellular stores. WIN does so by coupling to G_q/11 and enhancing PLC activity. Because PLCβ initiates the production of the endocannabinoid 2-AG, this suggests the intriguing possibility that WIN activation of CB₁ might increase 2-AG levels. These results also suggest that caution should be taken when extending observations made with high levels of WIN to other drugs activating CB₁. Although WIN exhibits the full range of pharmacological and behavioral effects seen with THC, there are some differences. Among these, WIN is more potent in drug discrimination trials and in producing antinociception and less potent in the production of hypothermia (9). The origin of these differences is unknown, but they may be due to high concentrations of WIN increasing intracellular calcium.