Evidence That the Cannabinoid CB1 Receptor Is a 2-Arachidonoylglycerol Receptor

An endogenous cannabimimetic molecule, 2-arachidonoylglycerol, induces a rapid, transient increase in intracellular free Ca2+ concentrations in NG108–15 cells through a cannabinoid CB1 receptor-dependent mechanism. We examined the activities of 24 relevant compounds (2-arachidonoylglycerol, its structural analogues, and several synthetic cannabinoids). We found that 2-arachidonoylglycerol is the most potent compound examined so far: its activity was detectable from as low as 0.3 nm, and the maximal response induced by 2-arachidonoylglycerol exceeded the responses induced by others. Activities of HU-210 and CP55940, potent cannabinoid receptor agonists, were also detectable from as low as 0.3 nm, whereas the maximal responses induced by these compounds were low compared with 2-arachidonoylglycerol. Anandamide was also found to act as a partial agonist in this assay system. We confirmed that free arachidonic acid failed to elicit a response. Furthermore, we found that a metabolically stable ether-linked analogue of 2-arachidonoylglycerol possesses appreciable agonistic activity, although its activity was apparently lower than that of 2-arachidonoylglycerol. We also confirmed that pretreating cells with various cannabinoid receptor agonists nullified the response induced by 2-arachidonoylglycerol, whereas pretreating cells with other neurotransmitters or neuromodulators did not affect the response. These results strongly suggested that the cannabinoid CB1 receptor is originally a 2-arachidonoylglycerol receptor, and 2-arachidonoylglycerol is the intrinsic physiological ligand for the cannabinoid CB1 receptor.

It is well known that ⌬ 9 -tetrahydrocannabinol (⌬ 9 -THC), 1 a psychoactive ingredient of marijuana, possesses a variety of pharmacological activities in vitro and in vivo (1), although, until recently, the mechanism of the action of ⌬ 9 -THC had long been unelucidated. In 1988, Devane et al. (2) provided evidence that a specific binding site(s) for cannabinoids is present in the brain. Soon after, Matsuda et al. (3) cloned a cDNA encoding a cannabinoid receptor (CB1) from a rat brain cDNA library. These findings raised the possibility that at least part of the action of ⌬ 9 -THC is mediated through such a specific receptor and prompted the search for endogenous cannabinoid receptor ligands in mammalian tissues.
In 1992, Devane et al. (4) isolated N-arachidonoylethanolamine (anandamide) from porcine brain as the first endogenous cannabinoid receptor ligand. They demonstrated that anandamide exhibits several cannabimimetic activities in vitro and in vivo (4,5). So far, a number of studies have been carried out on anandamide, and it has been assumed that anandamide is one of the important lipid mediators in the nervous system as well as in other systems (5). However, we (6,7) and others (8 -11) have found that the levels of anandamide are very low in several mammalian tissues. In addition, the biosynthetic pathways for anandamide, either the N-acylphosphatidylethanolamine pathway (6,7,(11)(12)(13)(14) or the condensation pathway (7,(15)(16)(17)(18), do not appear able to provide large amounts of anandamide, at least under normal conditions in living tissues, because the availabilities of the substrates are very low. Furthermore, several investigators demonstrated that anandamide is produced mainly in the post-mortem period in the brain (8,10). Thus, the physiological significance or meaning of anandamide, especially in the brain, has been questioned recently despite its high binding affinity toward the cannabinoid receptor(s).
On the other hand, several years ago, we found that 2-arachidonoylglycerol, an arachidonic acid-containing monoacylglycerol, possesses binding affinity toward the cannabinoid receptor in rat brain synaptosomes and that a rat brain contains a significant amount of arachidonoylglycerol (19,20). Indeed, the level of arachidonoylglycerol in the brain was found to be 800 times higher than that of anandamide present in the same tissue (7,20). We suggested the possibility that 2-arachidonoylglycerol is an endogenous ligand for the cannabinoid receptor in the brain (19,20). Mechoulam et al. (21) also isolated 2-arachidonoylglycerol from canine gut as another candidate for an endogenous cannabinoid receptor ligand; they demonstrated that 2-arachidonoylglycerol possesses binding activity toward cannabinoid receptors expressed on COS-7 cells transfected with cannabinoid receptor genes and induces the inhibition of adenylate cyclase in mouse spleen cells and twitch response in mouse vas deferens. Moreover, in recent studies, we found that 2-arachidonoylglycerol induces a rapid, transient elevation of intracellular free Ca 2ϩ concentration ([Ca 2ϩ ] i ) through a cannabinoid CB1 receptor-dependent mechanism and proposed that the cannabinoid CB1 receptor is originally a 2-arachidonoylglycerol receptor (22,23). Thus, the physiologi-cal significance of 2-arachidonoylglycerol came to receive increased attention (19 -33). Despite its possible physiological importance, however, the available information concerning this novel type of bioactive lipid is still limited. In addition, it is not clear whether or not 2-arachidonoylglycerol itself, but not its metabolites, is actually implicated in the observed biological effects, especially in cases of prolonged incubation of cells or in vivo experiments. In some cases, apparently, it is not easy to interpret the obtained experimental results because of the susceptibility of 2-arachidonoylglycerol to hydrolyzing enzymes, such as monoacylglycerol lipase.
In this study, we examined the activities of 2-arachidonoylglycerol and its ether-linked metabolically stable analogues, as well as other structurally related compounds, to stimulate NG108 -15 cells. We found that an ether-linked analogue of 2-arachidonoyl-glycerol, but not that of 1(3)-arachidonoylglycerol, exhibits appreciable cannabimimetic activity, although its activity was weak compared with that of 2-arachidonoylglycerol. The results obtained here provided strong evidence that the structure of 2-arachidonoylglycerol is strictly recognized by the cannabinoid CB1 receptor.
Synthesis of Ether-linked Analogues of 2-Arachidonoylglycerol and 1(3)-Arachidonoylglycerol-Arachidonic acid was dissolved in chilled, dehydrated diethyl ether and treated with LiAlH 4 while stirring for 1 h under an atmosphere of argon to yield eicosatetraenol. After adding ethyl acetate to remove excess LiAlH 4 , the mixture was filtered through Celite and then washed, first with distilled water and next with saturated NaCl aqueous solution. The organic layer was dried over MgSO 4 and evaporated to dryness. The residue was dissolved in dichloromethane and purified by silica gel column chromatography using hexane:ethyl acetate (5:1, v/v) as the elution solvent. Purified eicosatetraenol was then dissolved in dehydrated N,Ndimethylformamide and treated with methyltriphenoxyphosphonium iodide for 20 min to yield eicosatetraenyl iodide. Eicosatetraenyl iodide was purified by silica gel column chromatography (hexane:ethyl acetate, 20:2, v/v). 2-Eicosatetraenyl-1,3-benzylideneglycerol (an ether-linked analogue of 2-arachidonoyl-1,3-bezylideneglycerol) was obtained by condensing eicosatetraenyl iodide and 1,3-benzylideneglycerol using dimethylformamide as the solvent and Ag 2 O and tetra-n-butylammonium iodide as catalysts. After stirring at 90°C for 6 h under argon, the mixture was diluted with ethyl acetate and distilled water. The silver generated was removed by filtering through Celite, and the filtrate was washed, first with distilled water and then with saturated NaCl aqueous solution. The organic layer was dried over MgSO 4 and evaporated to dryness. The residue was dissolved in dichloromethane and purified by silica gel column chromatography using hexane: ethyl acetate (5:1, v/v) as the elution solvent. 2-Eicosatetraenylglycerol (an ether-linked analogue of 2-arachidonoylglycerol) was prepared from 2-eicosatetraenyl-1,3-benzylideneglycerol by treatment with boric acid and boric acid trimethyl ester, as in the preparation of 2-arachidonoylglycerol. 2-Eicosatetraenylglycerol was purified by silica gel column chromatography and then by TLC using ethyl acetate as the solvent (R F , 0.48). To prepare 1(3)-eicosatetraenylglycerol, rac-1,2-O-isopropylideneglycerol was employed instead of 1,3-benzylideneglycerol. The structures of 2-eicosatetraenylglycerol and 1(3)-eicosatetraenylglycerol were confirmed by NMR and electron impact mass spectrometry.
Synthesis of a Methylene-linked Analogue of 2-Arachidonoylglycerol-A methylene-linked analogue of 2-arachidonoylglycerol (2-hydroxymethyl-(all Z)-7,10,13,16-docosatetraen-1-ol) was prepared as follows: ethyl arachidonate was reduced with LiAlH 4 in dry diethyl ether under the usual conditions in the presence of butylated hydroxytoluene as an antioxidant and in a nitrogen atmosphere. The product eicosatetraenol was isolated by silica gel column chromatography and converted to its p-toluenesulfonyl ester. The product, without purification, was iodinated with lithium iodide in dry acetone in the presence of butylated hydroxytoluene in the dark and under nitrogen (reflux for 14 h). The iodide was isolated by silica gel column chromatography using hexane: ethyl acetate (98:2, v/v) as the elution solvent and converted to a diethyl malonate derivative as follows: to a suspension of oil-free sodium hydride in a mixture of dry tetrahydrofuran and dry N,N-dimethylformamide, a solution of dry diethyl malonate was added dropwise in the same solvent mixture under nitrogen. After stirring for 15 min, the iodide solution previously described, containing a trace of butylated hydroxytoluene, was added dropwise under nitrogen. This reaction mixture was refluxed for 4 h. The diethyl malonate derivative was isolated by silica gel column chromatography (hexane:ethyl acetate, 98:2, v/v) and reduced in dry diethyl ether with LiAlH 4 in a nitrogen atmosphere. The methylene-linked analogue of 2-arachidonoylglycerol was obtained in a pure state after silica gel column chromatography (hexane:ethyl acetate, 7:3, v/v). The R F value for the methylene-linked analogue of 2-arachidonoylglycerol in the TLC system using hexane: ethyl acetate (6:4, v/v) as the developing solvent was 0.15. The structure of the methylene-linked analogue of 2-arachidonoylglycerol was confirmed by NMR.
Cells-NG108 -15 cells were kindly donated by Prof. H. Higashida (Kanazawa University School of Medicine, Kanazawa, Japan). NG108 -15 cells were grown at 37°C in Dulbecco's modified Eagle's minimum essential medium supplemented with 10% fetal bovine serum and a mixture of hypoxantine, aminopterine, and thymidine under an atmosphere of 90% air and 10% CO 2 .
Measurement of [Ca 2ϩ ] i -Subconfluent cells were further incubated in fresh medium without fetal bovine serum for 24 h. The cells were next suspended by gentle pipetting in Hepes-Tyrode's solution (ϪCa 2ϩ ) containing 3 M Fura-2/AM and further incubated at 37°C for 45 min. The cells were then centrifuged (180 ϫ g for 5 min), washed twice with Hepes-Tyrode's solution (ϪCa 2ϩ ), and resuspended in Hepes-Tyrode's solution (ϪCa 2ϩ ) containing 0.025% bovine serum albumin. [Ca 2ϩ ] i was estimated using a CAF-100 Ca 2ϩ analyzer (JASCO, Tokyo, Japan) as described previously (22,23). CaCl 2 was added 4 -5 min before the measurement (final Ca 2ϩ concentration in the cuvette, 1 mM). Conformational Modeling-Conformational modeling was performed on a single molecule under vacuum with the QUANTA/CHARMm system (Micron Separations) using a dielectric constant, ⑀ ϭ 4r. Fig. 1 illustrates the chemical structures of 2-arachidonoylglycerol, its analogues, and related compounds, the biological activities of which were examined using NG108 -15 neuroblastoma ϫ glioma hybrid cells. Here, we examined whether or not these compounds are capable of inducing rapid transient increases in [Ca 2ϩ ] i . The dose response curves of the 24 compounds tested are depicted in Fig. 2. First, we examined the activity of 2-arachidonoylglycerol and confirmed that low concentrations of 2-arachidonoylglycerol elicit rapid transient increases in [Ca 2ϩ ] i (Fig. 2A). The response was detectable from as low as 0.3 nM and was augmented with increased concentrations of 2-arachidonoylglycerol. We also found that 1-arachidonoylglycerol and 3-arachidonoylglycerol, isomers of 2-arachidonoylglycerol, exhibited substantial activities (Fig. 2, B and C), although their activities were apparently lower than that of 2-arachidonoylglycerol. We confirmed that free arachidonic acid failed to induce a similar response, even at high concentrations (Fig. 2D).

RESULTS
The activities of metabolically stable ether-linked analogues of arachidonoylglycerols were next examined. We found that 2-eicosatetraenylglycerol, an ether-linked analogue of 2-arachidonoylglycerol, possesses some agonistic activity (Fig. 2E), although its activity was markedly lower than that of 2-arachidonoylglycerol. The response was detectable from 30 nM, and the response induced by 10 M 2-eicosatetraenylglycerol was about 50% of that induced by 10 M 2-arachidonoylglycerol. On the other hand, the activity of 1(3)-eicosatetraenylglycerol, an ether-linked analogue of 1(3)-arachidonoylglycerol, was very weak, compared with that of 2-eicosatetraenylglycerol, even at high doses, such as 10 M (Fig. 2F). Such preference of the 2-isomer over the 1(3)-isomer coincides well with the results of ester-linked compounds previously described (Fig. 2, A-C). We also found that another metabolically stable analogue of 2-arachidonoylglycerol, 2-hydroxymethyl-(all Z)-7,10,13,16-docosatetraen-1-ol (a methylene-linked analogue of 2-arachidonoylglycerol), lacked appreciable activity (Fig. 2G); this indicates that having an oxygen atom present in the linkage is important for agonistic activity.
We then examined the activities of propanediol-type analogues of arachidonoylglycerols. We found that 2-hydroxypropyl arachidonate, one of two hydroxy groups lacking analogues of 1(3)-arachidonoylglycerol, possesses substantial agonistic activity (Fig. 2H); its activity was almost comparable to that of 1(3)-arachidonoylglycerol. Interestingly, the activity of 3-hydroxypropyl arachidonate (Fig. 2I), another analogue of 1(3)arachidonoylglycerol, was found to be much weaker than that of 2-hydroxypropyl arachidonate, suggesting that the presence of a free hydroxy group adjacent to the ester linkage is essential in exhibiting strong agonistic activity.
Next, we examined the activities of anandamide and its analogues. As shown in Fig. 2J, we confirmed that anandamide induces the elevation of [Ca 2ϩ ] i . The response was detectable from as low as 3 nM, whereas the magnitude of the response induced by anandamide was not as pronounced as that induced by 2-arachidonoylglycerol even at high concentrations. We also found that (R)-1-methanandamide (Fig. 2K) exhibits agonistic activity comparable to that of anandamide, although the activity of (S)-1-methanandamide (Fig. 2L) appears to be somewhat lower than those of anandamide and (R)-1-methanandamide.
The activities of 2-monoacylglycerols containing various species of saturated and unsaturated fatty acids were next examined. As shown in Fig. 2M, 2-eicosatetraynoylglycerol, an eicosatetraynoic acid-containing analogue of 2-arachidonoylglycerol, failed to exhibit appreciable agonistic activity even at high concentrations, indicating that the presence of double bonds, but not triple bonds, is indispensable for agonistic activity. We also found that saturated or monoenoic 2-monoacylglycerols, such as 2-palmitoylglycerol (data not shown) and 2-oleoylglycerol (Fig. 2N), were inactive. The activities of 2-linoleoylglycerol (Fig. 2O) and 2-␥-linolenoylglycerol (Fig. 2P) were also very low, suggesting that the acyl moiety of 2-arachidonoylglycerol is strictly recognized by the receptor.
We then examined the activities of several synthetic cannabinoids and compared their activities to that of 2-arachidonoylglycerol. We found that HU-210 and CP55940 exhibited appreciable agonistic activities (Fig. 2, W and X). The response was detectable from as low as 0.3 nM, with a plateau at around 100 nM in each case. However, the maximal responses induced by these cannabinoids are apparently smaller than that induced by 2-arachidonoylglycerol.
To confirm that the response induced by 2-arachidonoylglycerol is mediated through the cannabinoid CB1 receptor, we examined the effects of cell pretreatment with SR141716A, a cannabinoid CB1 receptor-specific antagonist (or an inverse agonist). As shown in Fig. 3 Table I, the cell pretreatment with various cannabinoid receptor agonists, such as ⌬ 9 -THC, HU-210, WIN55212-2, and CP55940, completely blocked the response induced by 2-arachidonoylglycerol, probably through the desensitization of the receptor molecule. On the other hand, cell pretreatment with various compounds other than cannabinoid receptor agonists did not affect the response induced by 2-arachidonoylglycerol. DISCUSSION In preceding studies, we investigated the mechanism and physiological meaning of 2-arachidonoylglycerol-induced transient increases in [Ca 2ϩ ] i in NG108 -15 cells (22,23). We established that the response induced by adding 2-arachidonoylglycerol to the cells is mediated through the cannabinoid CB1 receptor and G i or G o (22,23), yet the detailed mechanism is not yet fully understood. One important issue to be verified is whether or not 2-arachidonoylglycerol itself, but not its metabolites, is actually involved in this rapid cellular response. Here, we provided clear evidence to show that this is the case. First, a metabolically stable analogue of 2-arachidonoylglycerol, 2-eicosatetraenylglycerol, was found to exhibit appreciable agonistic activity (Fig. 2E). Second, free arachidonic acid is not capable of inducing the response (Fig. 2D). In addition, we already confirmed that cell pretreatment with either indomethacin or nordihydroguaiaretic acid does not affect the cellular response (22). Thus, it is evident that the structure of 2-arachidonoylglycerol itself is recognized by the cannabinoid CB1 receptor as a receptor agonist.
Importantly, among various cannabimimetic molecules, 2-arachidonoylglycerol exhibited the highest agonistic activity. We detected its activity from as low as 0.3 nM, and the magnitude of the response induced by 2-arachidonoylglycerol was the most prominent, compared with the responses of the other compounds (Fig. 2). HU-210 and CP55940 also exhibited appreciable agonistic activities from as low as 0.3 nM (Fig. 2, W and X), whereas these compounds were shown to act as partial agonists similar to ⌬ 9 -THC (23) and anandamide (Fig. 2J). WIN55212-2 has also been shown to exhibit strong agonistic activity, whereas its activity was detectable only above 10 nM (22). These findings that the various cannabimimetic molecules are less active, compared with 2-arachidonoylglycerol, strongly support that the cannabinoid CB1 receptor is originally a 2-arachidonoylglycerol receptor and that 2-arachidonoylglycerol is the intrinsic physiological ligand. Table II summarizes the preferred structures of monoacylglycerols and their structural analogues as cannabinoid receptor agonists. Glycerol is the most suitable head group, and the 2-isomer is preferable over the 1(3)-isomer. As for the fatty acyl moiety, arachidonic acid is the most preferred fatty acid constituent of monoacylglycerols, among those examined, although the activity of eicosatrienoic acid (n-9)-containing species was almost comparable to that of arachidonic acid-containing species. Because the activities of 2-eicosatrienoyl(n-6)glycerol, 2-eicosatrienoyl(n-3)glycerol and 2-docosatetraenoyl(n-6)glycerol are apparently lower than those of 2-eicosatrienoyl(n-9)glycerol and 2-eicosapentaenoyl(n-3)glycerol (Fig. 2), we assume that the structure near the ester linkage, such as the presence of the double bond at the ⌬5 position rather than the structure near the methyl end, is crucially important, probably for some characteristic conformation of the agonistic molecules.
Another important feature of the structural requirement is that the presence of an ester linkage, especially one that is adjacent to a free hydroxy group, is essential for strong agonistic activity. A methylene-linked analogue of 2-arachidonoyl-  glycerol possesses only weak agonistic activities (Fig. 2G). Furthermore, the activity of an ether-linked analogue of 2-arachidonoylglycerol was also considerably lower than that of 2-arachidonoylglycerol (Fig. 2E). The activity of 3-hydroxypropyl arachidonate, which lacks an adjacent free hydroxy group, was considerably lower than that of an adjacent free hydroxy group-containing analogue, 2-hydroxypropyl arachidonate (Fig. 2, H and I). It is possible, therefore, that either the oxygen atoms in the ester bond or the adjacent free hydroxy group may have close interaction with the receptor molecule. Another possibility may be that the carbonyl group of the ester bond and an adjacent free hydroxy group of the glycerol moiety are linked by a hydrogen bond under some conditions; this possibility is predicted by computer analysis evaluating the minimum energy conformation of a single molecule of 2-arachidonoylglycerol in a vacuum: one of the most stable conformations of 2000 conformations has an intramolecular hydrogen bond (OϪH ϭ 2.1 Å). Such a linkage, possible in the ester-linked compound but not in ether-linked analogues and methylene-linked analogues, is effective in fixing the glycerol-head group and yields a ring structure consisting of three carbon atoms, three oxygen atoms, and a hydrogen atom. It will be of value to examine whether or not such a ring structure can be formed in 2-arachidonoylglycerol molecules under physiological conditions, especially at the receptor site, and whether or not such a possible ring structure is implicated in the induction of strong agonistic activities.
Since the discovery of the cannabinoid receptor gene and an endogenous cannabinoid receptor ligand, anandamide, numerous studies have been conducted on cannabinoid receptors and their endogenous ligands. It is becoming evident that the cannabinoid receptor-endogenous ligand system plays important physiological roles in the nervous system. Herkenham (35) reported that whole brain cannabinoid receptor density is similar to the whole brain densities of receptors for amino acid transmitters, such as glutamic acid and GABA; this observation leads us to postulate that the endogenous cannabinoid receptor ligand(s) is also a common molecule and abundantly present in the brain, like glutamic acid and GABA. 2-Arachidonoylglycerol, but not anandamide, meets this requirement. In previous studies, we (19,20) and others (28) found that the level of 2-arachidonoylglycerol in the brain is on the order of nmol/g of tissue, which is several hundred times higher than that of anandamide in the same tissue. Indeed, 2-arachidonoylglycerol is one of the major species of monoacylglycerols in the brain and other tissues (31). Furthermore, we recently found that a significant amount of 2-arachidonoylglycerol was generated in a rat brain homogenate during the incubation, especially in the presence of Ca 2ϩ (31). Stimulus-induced rapid formation of 2-arachidonoylglycerol occurred in neuroblastoma cells (25), hippocampal slices (28), and dorsal root ganglion neurons (36) besides vascular endothelial cells (30), fibroblasts (37), and platelets (38). In contrast to 2-arachidonoylglycerol, the level of anandamide in the brain was very low (7)(8)(9)(10)(11), and only small amounts of anandamide, if any, were produced in stimulated tissues and cells (12,14,39,40). All these observations validate our hypothesis: the cannabinoid CB1 receptor is originally a 2-arachidonoylglycerol receptor.
The most striking and noticeable issue concerning 2-arachidonoylglycerol is that this unique lipid molecule links enhanced inositol phospholipid metabolism in stimulated neuronal cells with the function of cannabinoid receptors expressed mainly on presynaptic membranes (35,41,42). As described previously, 2-arachidonoylglycerol is formed in neuronal cells upon stimulation (25,28) and in a brain homogenate upon the addition of Ca 2ϩ (31), and the cannabinoid CB1 receptor is known to participate in the attenuation of neurotransmission (41,42). Such a linkage should be effective in calming some neurons after excitation, which leads to negative feedback control of neurotransmission in some synapses in which the cannabinoid CB1/2-arachidonoylglycerol receptor is present. Similar negative feedback control mechanism may operate in the case of adenosine, which is assumed to be derived principally from ATP. In any case, the cannabinoid CB1/2-arachidonoylglycerol receptor-dependent negative feedback control system appears to be of great physiological significance, because sustained activation of neuronal cells is known to cause cell exhaustion and may lead to neuronal cell death. In relation to this, we recently found that 2-arachidonoylglycerol suppresses the activation of differentiated NG108 -15 cells upon depolarization (29). Furthermore, Stella et al. (28) reported that 2-arachidonoylglycerol reduces long term potentiation in hippocampal slices. In vivo administration of 2-arachidonoylglycerol has been shown to induce analgesia, immobility, and reduction of spontaneous activity in mice (21). Thus, there is increasing evidence that 2-arachidonoylglycerol plays important physiological roles in the attenuation of neurotransmission and the protection of neuronal cells presumably with the cooperation of other inhibitory neurotransmitters or neuromodulators, such as GABA and adenosine.