Receptor-dependent formation of endogenous cannabinoids in cortical neurons

Abstract

Ž .E-mail address: piomelli@uci.eduD. Piomelli .  1 Present address: Department of Pharmacology, University of Washington, Seattle, WA 98195-7280, USA.they may act as endogenous modulatory substances Ž .Piomelli et al., 2000 .Unlike classical neurotransmitters, the endogenous Ž .cannabinoids endocannabinoids may be produced upon demand by enzymatic cleavage of membrane lipid precursors and immediately extruded from neurons without an Ž intermediate step of vesicle storage Piomelli et al., 2000; .Schmid, 2000 .This feature-unusual in a brain chemical transmitter, but reminiscent of other lipid-derived mediators-prompted the two questions addressed in the present study.The first is how formation of anandamide and 2-arachidonylglycerol is initiated: is it stimulated by action potentials invading the synaptic nerve endings or by receptor-activated mechanisms?Previous studies have shown that anandamide and 2-arachidonylglycerol can be produced during neural activity in vitro and in vivo, but they have not investigated the cellular mechanisms of this Ž response Di Marzo et al., 1994;Stella et al., 1997; .Giuffrida et al., 1999 .The second question is whether, regardless of the mechanism involved, anandamide and 2-arachidonylglycerol can be generated independently of each other.An affirmative answer to this question would 0014-2999r01r$ -see front matter q 2001 Elsevier Science B.V. All rights reserved.
Ž .PII: S 0 0 1 4 -2 9 9 9 0 1 0 1 1 8 2 -7 ( ) N. Stella, D. Piomellir European Journal of Pharmacology 425 2001 189-196 190 imply that these compounds may be released under different circumstances and, possibly, serve distinct functional roles in neuronal signaling.To begin to address these questions, we have investigated the mechanisms of anandamide and 2-arachidonylglycerol formation in primary cultures of rat brain cortical neurons.The results of our experiments show that anandamide and 2-arachidonylglycerol are produced in a receptor-dependent manner, and that segregated molecular mechanisms may underlie the formation of each of these compounds.

Drugs
All drugs were purchased from Research Biochemical Ž .Ž .Natick, MA or Sigma Saint Louis, MO .

Cell incubations
In preliminary experiments designed to study the kinetic of endogenous cannabinoid production, we labeled the w 14 x neurons by overnight incubation with C arachidonic acid w 14 x and determined the time-course of C 2-arachidonylg-Ž .lycerol formation in the presence of glutamate 300 mM Ž .Ž .andror carbachol 1 mM Stella et al., 1997 .At the end Ž . of various incubation periods 0.5 to 15 min , lipids were extracted and analyzed by thin-layer chromatography Ž .Stella et al., 1997 .Glutamate significantly increased radioactivity in the 2-arachidonylglycerol fraction at 2.5 Ž .min 130 " 6% of basal, n s 6 , whereas carbachol had no effect.Coapplication of glutamate and carbachol resulted w 14 x in a greater C 2-arachidonylglycerol accumulation, which reached a maximum of 192 " 6% of basal at 2.5 min and Ž .lasted for at least 15 min 146 " 16% of basal .Therefore, we selected a 2.5-min incubation time for subsequent gas chromatographyrmass spectrometry analyses.After 5-6 days in vitro, the incubation medium was removed and the neurons were washed twice in HEPES-bicarbonate buffer Ž .Ž .9 ml, 15 min each composed of mM : NaCl, 120; KCl, 5.0; CaCl , 2.0; MgSO , 1.0; NaH PO , 1.0; Glucose, 10; 2 4 2 4 Ž .NaHCO , 5.0; and HEPES, 20 pH 7.4 at 37 8C .Incuba-3 tions were started by adding 1 ml of buffer containing drugs at 10 times their final concentration, and by gently shaking the culture dishes for 10 s to allow the drugs to diffuse uniformly.Neurons from two culture dishes were Ž combined for each experimental point 1.17 " 0.03 mg of .proteinrdish, n s 33 .

Lipid analyses
Incubations were stopped by adding 10 ml of ice-cold methanol.Cells were immediately scraped on ice and the Ž supernatants were transferred into ice-cold chloroform 20 .Ž ml that contained internal standards 0.6 nmol of each w 2 x w 2 x H acylethanolamide and 0.5 nmol of H 2-ara-4 8 . chidonylglycerol .This rapid procedure significantly reduced the non-enzymatic conversion of 2-arachidonylg-Ž .lycerol into 1 3 -arachidonylglycerol, yielding f 90% 2-arachidonylglycerol.Organic phases of duplicate samples were pooled, evaporated to dryness under N gas, 2 Ž .recovered in chloroform 2 ml , filtered through Ultra-free Ž .CL filters Waters , and evaporated again.Lipids were Ž .reconstituted in chloroform 100 ml for high-performance Ž .liquid chromatography HPLC fractionation and gas chromatographyrmass spectrometry analysis, which were car-Ž .ried out as described Giuffrida and Piomelli, 1998 .

Anandamide formation Õia coactiÕation of N-methyl-( ) D-aspartate NMDA and acetylcholine receptors
We used isotope dilution gas chromatographyrmass spectrometry to investigate anandamide biosynthesis in rat Ž cortical neurons in primary culture Giuffrida and Piomelli, .1998 .Fig. 1A shows the ion current trace obtained after analysis of a lipid fraction purified from these cells.A Ž .diagnostic component of mass-to-charge ratio mrz 404 Žthe fragment produced from the molecular ion by loss of .one methyl group , eluted from the column at the retention time of authentic standard, confirming that unstimulated Ž neurons contain detectable levels of anandamide Di Marzo . et al., 1994;Cadas et al., 1997 .Under basal conditions, we measured on average 0.3 " 0.1 pmol of anandamide per mg of protein, with values ranging from 0.1 to 1.35 Ž .pmol per mg of protein n s 17 .This variability appeared to reflect differences amongst culture preparations, rather than within a single preparation, and might result from subtle inequalities in neuronal maturation under the ) ) for 2.5 min with buffer basal or glutamate 300 mM plus carbachol 1 mM glut q Cch .P -0.serum-free culture conditions of these experiments.To limit the impact of variability on data interpretation, all treatment groups were compared to appropriate controls within the same cell culture preparation.
To determine whether activation of neurotransmitter receptors increased anandamide levels, we incubated the Ž .neurons for 2.5 min with glutamate 300 mM or carba-Ž .chol, an acetylcholine receptor agonist 1 mM .Although these agents had no effect separately, they stimulated anandamide formation more than 5 times when applied Ž .together Fig. 1B .Three pharmacological experiments suggest that anandamide biosynthesis resulted from the concomitant activation of glutamate NMDA and cholinergic a7 nicotinic receptors.First, the combination NMDA Ž .300 mM rcarbachol was more effective at stimulating Ž anandamide production than glutamatercarbachol Fig.
2q chloride 100 mM to block voltage-activated Ca Ž .channels had no effect Fig. 1D .Thus, NMDA and a7 nicotinic receptors may act synergistically to evoke anandamide biosynthesis via a tetrodotoxin-sensitive mechanism that requires mobilization of Ca 2q from intracellular stores.

Production of other fatty acid ethanolamides
In cortical neurons, activity-dependent hydrolysis of saturated and monounsaturated species of N-acyl phosphatidylethanolamine generates several fatty acid etha-Ž nolamides in addition to anandamide Di Marzo et al., 1994;Cadas et al., 1997;Hansen et al., 1997

2-Arachidonylglycerol formation Õia NMDA receptor-mediated CA 2 q entry
To quantify 2-arachidonylglycerol in cortical neurons, we developed an isotope dilution gas chromatographyr mass spectrometry assay similar to that used for anandamide.The electron-impact mass spectra of synthetic w 2 x unlabeled and H -labeled 2-arachidonylglycerol 8 Ž .analyzed as bis trimethylsilylether derivatives included Žw x q .the molecular ions M , mrz 530 , and ions produced by Žw x q .loss of one methyl group M-15 , mrz 515 or one Žw x q .derivatizing trimethylsilenol group M-90 , mrz 440 .w x q We monitored the M-90 fragments, because of their reasonable abundance and high diagnostic value: rectilin-Ž 2 .ear mass spectrometry responses r ) 0.99 were obtained with these fragments when amounts of synthetic Ž .( ) N. Stella, D. Piomellir European Journal of Pharmacology 425 2001 189-196 194 compound that originates from the non-enzymatic isomer-Ž ization of 2-arachidonylglycerol Serdarevich, 1967;Mechoulam et al., 1995;Bisogno et al., 1997;Stella et  Three main features distinguished the mechanisms governing the receptor-dependent formation of 2-arachidonylglycerol from that of anandamide.First, activation of NMDA receptors was sufficient to cause a marked accu-Ž .mulation of 2-arachidonylglycerol in neurons Fig. 4B .The application of glutamate or NMDA increased 2-Ž .arachidonylglycerol production about three times Fig. 4B , Ž .and MK801 completely prevented this effect Fig. 4C .DNQX partially impaired 2-arachidonylglycerol production, indicating that AMPArkainate receptors may also be Ž .involved Fig. 4C .Second, carbachol had little effect on 2-arachidonylglycerol biosynthesis when applied alone or in combination with glutamate, albeit it enhanced the Ž .effect of NMDA Fig. 4B .The pharmacological basis of this enhancement was not investigated.Third, incubation with a medium containing the Ca 2q chelator EGTA abolished 2-arachidonylglycerol formation, whereas depleting 2q Ž .internal Ca with BAPTA-AM had no effect Fig. 4D .The results suggest that flux of external Ca 2q allowed by the activation of ionotropic NMDA receptor channels is sufficient to elicit 2-arachidonylglycerol formation in cortical neurons.

Discussion
Despite their similar chemical structures, anandamide and 2-arachidonylglycerol are produced through distinct biochemical pathways.Formation of anandamide may result from hydrolysis of the phospholipid precursor Narachidonyl phosphatidylethanolamine, catalyzed by a Ž phosphodiesterase such as phospholipase D Di Marzo et . al., 1994;Cadas et al., 1997;Schmid, 2000 .2-Arachidonylglycerol, on the other hand, may be produced by cleavage of 1,2-diacylglycerol generated by phospholipase Ž C acting on phosphatidylinositol bisphosphate Gammon et . al., 1989;Stella et al., 1997 , although the participation of Ž alternative pathways cannot be excluded Piomelli et al., . 2000 .The existence of different enzymatic routes for the formation of anandamide and 2-arachidonylglycerol suggests that under certain circumstances, these two endocannabinoid substances might operate independently of each other.This possibility is supported by two findings.In hippocampal slices, stimulation of glutamate-releasing fibers in the Schaffer collaterals increases the levels of Ž 2-arachidonylglycerol, but not those of anandamide Stella . et al., 1997 .Conversely, in vivo microdialysis experiments show that activation of striatal D -family dopamine 2 receptors enhances release of anandamide, but not of 2-Ž .arachidonylglycerol Giuffrida et al., 1999 .It is unclear, however, whether this discrepancy reflects different experimental conditions, regional segregation of the phospholipase C and phospholipase D pathways, or receptoractivated mechanisms linked to the generation of specific endocannabinoid lipids.To examine these different possibilities, in the present study we have investigated the receptor mechanisms underlying the formation of anandamide and 2-arachidonylglycerol in primary cultures of rat cortical neurons.
Our findings indicate that the excitatory neurotransmitter glutamate triggers 2-arachidonylglycerol biosynthesis by allowing external Ca 2q to enter cortical neurons through activated NMDA receptor channels.In these cells, 2-arachidonylglycerol formation is likely to be mediated by the phospholipase Crdiacylglycerol lipase pathway, because selective drug inhibitors of these enzyme activities prevent the accumulation of 2-arachidonylglycerol elicited 2q

Ž
. by a Ca ionophore Stella et al., 1997 .This conclusion is also in agreement with previous work showing that NMDA receptor activation stimulates phospholipase C and Ž diacylglycerol lipase activities in cultured neurons Nico-. letti et al., 1986;Farooqui et al., 1993 .Noteworthy, we found that activation of acetylcholine receptors with carbachol had little or no effect, per se, on 2-arachidonylglycerol levels.This result, at variance with the ability of cholinergic agonists to stimulate phospholipase C activity in many tissues and to cause 2-arachidonylglycerol accu-Ž .mulation in vascular endothelium Mechoulam et al., 1998 , underscores the differences in 2-arachidonylglycerol biosynthesis between neurons and other non-neuronal cell types.
The functional role of 2-arachidonylglycerol formation in NMDA receptor signaling is unclear at present.However, experiments with superfused hippocampal slices suggest that 2-arachidonylglycerol may act as a localized Ž feedback signal within the hippocampus Stella et al., . 1997 .Electrical stimulation of the Schaffer's collaterals, a glutamatergic fiber tract that projects from CA3 to CA1 neurons, produced a marked increase in 2-arachidonylglycerol accumulation in the slices, which was prevented by the Na q channel blocker tetrodotoxin or by removal of 2q Ž .external Ca Stella et al., 1997 .In the same preparation, exogenous 2-arachidonylglycerol inhibited the induction of long-term potentiation at CA3-CA1 synapses by activating local cannabinoid CB receptors, whereas it had no 1 Ž .effect on basal synaptic transmission Stella et al., 1997 .The possible role of 2-arachidonylglycerol as a negative feedback regulator of N-methyl-D-aspartate-mediated responses, suggested by these results, is reinforced by the ability of cannabinoid drugs to reduce glutamate transmis-Ž .sion Shen et al., 1996;Shen and Thayer, 1999 , inhibit Ž long-term potentiation Collins et al., 1994;Terranova et . al., 1995;Misner and Sullivan, 1999 , and Stella, D. Piomellir European Journal of Pharmacology 425 2001 189-196 195 In the present experiments, the simultaneous application of glutamate and carbachol, but not of either agent alone, caused a marked stimulation of anandamide biosynthesis in cortical neurons.Pharmacological experiments suggest that this synergistic effect may result from the coactivation of NMDA and a7 nicotinic receptors, and may depend both on membrane depolarization and on mobilization of Ca 2q ions from intracellular stores.Though necessary for the response, membrane depolarization was insufficient per se to initiate anandamide biosynthesis.This observation appears to contradict previous results, which showed that membrane-depolarizing agents increase accumulation of radioactive fatty acid ethanolamides in neuronal cultures maintained in a serum-containing medium and labeled w 3 x Ž with H ethanolamine Di Marzo et al., 1994;Cadas et al., . 1996 .Differences in cell culture models may provide, however, a plausible explanation for this discrepancy: neurons cultured in the presence of serum actively release neurotransmitters when they are challenged with membrane-depolarizing agents, whereas neuronal cultures maintained in a defined medium, such as those used in the Ž present experiments, do not Pfrieger and Barres, 1997; . Evans et al., 1998 .Thus, fatty acid ethanolamide biosynthesis in the former cultures might result, not from a direct effect of membrane depolarization, but from the action of neurotransmitters released in the extracellular medium.A similar indirect mechanism may be responsible for the release of anandamide induced in vivo by a depolarizing concentration of KCl, administered in dorsal striatum by Ž .reverse dialysis Giuffrida et al., 1999 .Receptor-dependent anandamide formation may be distinguished pharmacologically from that of other fatty acid ethanolamides that do not activate cannabinoid receptors.Indeed, though biosynthesis of all fatty acid ethanolamides is contingent on NMDA receptor occupation, anandamide formation requires the coactivation of NMDA and a7 nicotinic receptors, while oleylethanolamide and palmitylethanolamide formation requires the coactivation of NMDA and muscarinic receptors.Thus, glutamate and acetylcholine may elicit the biosynthesis of different fatty acid ethanolamides, depending on the complement of acetylcholine receptors expressed in their target neurons.The finding that palmitylethanolamide and oleylethanolamide may be produced by dissimilar molecular mechanisms and may exert biological effects that are not medi-Ž ated by cannabinoid receptors Calignano et al., 1998;. Jaggar et al., 1998 raises the possibility that they might serve signaling functions independent from those of anandamide.Testing this possibility will require a thorough investigation of the pharmacological effects of palmitylethanolamide and oleylethanolamide in neurons, as well as the identification of the putative cellular targets that may mediate these effects.
In conclusion, biosynthesis of anandamide and 2arachidonylglycerol in cortical neurons may be triggered by activation of membrane receptors for the neurotrans-mitters, glutamate and acetylcholine.This suggests that a key function of the endocannabinoid system in the brain may be to modulate the effects of primary neurotransmitters by a localized feedback action.In agreement with this idea, recent electrophysiological experiments have suggested that endocannabinoid compounds may serve as retrograde messengers in both hippocampus and cerebel-Ž lum Kreitzer and Regerh, 2001;Ohno-Shosaku et al., . 2001;Wilson and Nicoll, 2001 .Finally, the finding that anandamide and 2-arachidonylglycerol may be produced through divergent molecular mechanisms indicates that these two endocannabinoid lipids may exert their modulatory effects in separate and possibly complementary ways.
Ž .Fig. 1. Coactivation of glutamate NMDA and acetylcholine a7 nicotinic receptors elicits anandamide formation in rat cortical neurons.A Identification of anandamide by selected ion monitoring gas chromatographyrmass spectrometry.Anandamide was purified from neurons by HPLC and analyzed by gas Ž .chromatographyrmass spectrometry as the trimethylsilylether TMS derivative.The fragment monitored was produced from the anandamide-TMS Žw x q .Ž .molecular ion by the loss of methyl group M-15 , mrz 404 .The arrow indicates the retention time of standard anandamide.B Effects of glutamate Ž .Ž .and carbachol on anandamide levels.Neurons were incubated for 2.5 min in the presence of glutamate glut, 300 mM , carbachol Cch, 1 mM , veratridine Ž .Ž Ž .10 mM , andror NMDA 300 mM; in the presence of 100 mM D-serine Johnson and Ascher, 1987 , and in the absence of extracellular magnesium Ž Nowak et al., 1984 .P -0.05 and P -0.01 significantly different from basal ANOVA followed by Dunnett's test .C Effects of various receptor Ž .antagonists on anandamide production.Neurons were pre-incubated for 30 min in buffer control , or in buffer containing one of the following drugs: , DNQX 10 mM , dihydro-b-erythroidine b-erythro, 10 mM , methyllycaconitine MLA, 30 nM or atropine 1 mM ; and then incubated Ž

q
Veratridine 10 mM , a Na channel activator, did not Ž .increase anandamide levels in neurons Fig.1Balthough it elevated intracellular Ca 2q in a cadmium-sensitive man-2q Ž ner, as assessed by Ca imaging N. Stella and D. .Piomelli, unpublished results .However, receptor-dependent anandamide production was significantly reduced by q Ž ; Schmid, .2000 .These compounds, of which palmitylethanolamide and oleylethanolamide are two prominent examples, do not activate cannabinoid receptors and have no established Ž .signaling function Piomelli et al., 1998 .Incubating the neurons with glutamate plus carbachol, not glutamate or carbachol alone, stimulated the formation of both Ž palmitylethanolamide and oleylethanolamide Figs.2A,B .and 3A,B .This response was mediated in all likelihood by co-activation of NMDA receptors and muscarinic re-Ž .ceptors because, i NMDA was as effective as glutamate in eliciting palmitylethanolamide and oleylethanolamide Ž .Ž. production Figs.2B and 3B ; ii MK801 and atropine Ž .Ž. blocked the response Figs.2C and 3C , whereas iii methyllylcaconitine and dihydro-b-erythroidine had no ef-Ž .fect Figs.2C and 3C .Pertussis toxin had no significant effect on palmitylethanolamide and oleylethanolamide for-Ž .mation Figs.2D and 3D .Additional experiments will be necessary to determine the G protein coupling of the putative muscarinic receptor involved in this response.Receptor-dependent biosynthesis of palmitylethanolamide and oleylethanolamide was not affected by tetrodotoxin or by removal of external Ca 2q with EGTA, but was blocked 2q Ž by chelating intracellular Ca with BAPTA-AM Figs. .2D and 3D .The results indicate that coactivation of NMDA and muscarinic receptors stimulates palmitylethanolamide and oleylethanolamide formation by mobilizing Ca 2q ions from intracellular stores.

Fig. 2 .
Formation of palmitylethanolamide in cortical neurons.Experiments were carried out as described in Fig. 1 legend.A Identification of Ž w x q .palmitylethanolamide by selected ion monitoring gas chromatographyrmass spectrometry mrz 356, M-15 .The arrow indicates the retention time of Ž .Ž .Ž .standard palmitylethanolamide PEA .B Effects of glutamate and carbachol on palmitylethanolamide levels.C Effects of various receptor antagonists Ž .D Effects of tetrodotoxin TTX , pertussis toxin PTX and treatments that affect intracellular Ca on palmitylethanolamide. , D. Piomellir European Journal of Pharmacology 425 2001 189-196 193 Ž .Fig. 3. Formation of oleylethanolamide in cortical neurons.Experiments were carried out as described in Fig. 1 legend.A Identification of Ž w x q .oleylethanolamide by selected-ion monitoring gas chromatographyrmass spectrometry mrz 382, M-15 .The arrow indicates the retention time of Ž .Ž .Ž .standard oleylethanolamide OEA .B Effects of glutamate and carbachol on oleylethanolamide levels.C Effects of various receptor antagonists on Ž D Effects of tetrodotoxin TTX , pertussis toxin PTX and treatments that affect intracellular Ca on oleylethanolamide production.2-arachidonylglycerolranging from 0 to 1 nmol were injected into gas chromatographyrmass spectrometry tow 2 x gether with a fixed amount of H 2, on average, 16.5 " 3.3 pmol Ž . of 2-arachidonylglycerol per mg of protein n s 17 .In these analyses, only a minor peak was detectable at the Ž .Ž .retention time of 1 3 -arachidonylglycerol Fig.4A , aŽ .Fig. 4. NMDA receptor activation stimulates 2-arachidonylglycerol formation.Experiments were carried out as described in Fig. 1 legend.A Ž w x q .Identification of 2-arachidonylglycerol by selected ion monitoring gas chromatographyrmass spectrometry mrz 432, M-90 .The arrows indicate the Ž .Ž .Ž Ž . .Ž .retention times of standard 2-arachidonylglycerol 2-AG and 1 3 -arachidonilglycerol 1 3 -AG .B Effects of glutamate and carbachol on 2-Ž .Ž .Ž .arachidonylglycerol levels.C Effects of various receptor antagonists on 2-arachidonylglycerol production.D Effects of tetrodotoxin TTX , pertussis Ž .2q toxin PTX and treatments that affect intracellular Ca on 2-arachidonylglycerol production.