Oxygenation of the Endocannabinoid, 2-Arachidonylglycerol, to Glyceryl Prostaglandins by Cyclooxygenase-2*

Cyclooxygenases (COX) play an important role in lipid signaling by oxygenating arachidonic acid to endoperoxide precursors of prostaglandins and thromboxane. Two cyclooxygenases exist which differ in tissue distribution and regulation but otherwise carry out identical chemical functions. The neutral arachidonate derivative, 2-arachidonylglycerol (2-AG), is one of two described endocannabinoids and appears to be a ligand for both the central (CB1) and peripheral (CB2) cannabinoid receptors. Here we report that 2-AG is a substrate for COX-2 and that it is metabolized as effectively as arachidonic acid. COX-2-mediated 2-AG oxygenation provides the novel lipid, prostaglandin H2 glycerol ester (PGH2-G), in vitro and in cultured macrophages. PGH2-G produced by macrophages is a substrate for cellular PGD synthase, affording PGD2-G. Pharmacological studies reveal that macrophage production of PGD2-G from endogenous sources of 2-AG is calcium-dependent and mediated by diacylglycerol lipase and COX-2. These results identify a distinct function for COX-2 in endocannabinoid metabolism and in the generation of a new family of prostaglandins derived from diacylglycerol and 2-AG.

Site-directed mutagenesis of murine COX-2 was performed as described (23). COX-2 enzymes were expressed in SF-9 insect cells by using the pVL1393 expression vector (PharMingen, San Diego, CA) and purified by ion-exchange chromatography and gel filtration as described (23). Apoenzymes were reconstituted with hematin prior to activity assays. COX activity was quantified as described (24). Enzyme kinetics were analyzed by nonlinear regression using the computer program, Enzyme Kinetics 1.5 (Trinity Software, Campton, NH). Initial reaction velocity data were obtained from the linear portion of oxygen uptake curves.

Oxygenation of 2-AG by Purified Human and Mouse COX-2. Incubation of 2-AG
with purified recombinant human COX-2 triggered O 2 uptake comparable in rate and extent to that observed with arachidonic acid (Figure 1). In contrast, relatively little O 2 uptake was observed following addition of sheep COX-1 to 2-AG. Steady-state kinetic analysis revealed that both human COX-2 and mouse COX-2 oxygenated 2-AG with apparent k cat /K M values similar to those determined for arachidonic acid ( Table 1). The k cat /K M of human COX-2 for 2-AG is 60-fold higher than the value of 0.065 s reported for anandamide (27). Among a series of arachidonyl esters, 2-AG was clearly the preferred substrate ( Figure 2). The initial rate of 2-AG oxygenation displayed in Figure 2 is approximately half the initial rate of arachidonic acid oxygenation in agreement with the k cat values determined for both substrates with murine COX-2 (Table   1). Of particular interest, COX-2 oxygenated the more stable arachidonylglycerol regioisomer, 1-AG, at a markedly reduced rate. In addition, DAG, the biosynthetic precursor for 2-AG, and the related esters, 1,2-diarachidonylglycerol and 1,3diarachidonylglycerol, were very poor substrates for COX-2 ( Figure 2). suggests a role for phospholipase C in DAG generation leading to PGD 2 -G biosynthesis but we cannot rule out a contribution from phospholipase D (43). Thus, PGD 2 -G biosynthesis is calcium-dependent and results from the sequential actions of DAG lipase, COX-2, and PGD synthase (eq 2).

Discussion
The present findings establish a function for COX-2 in the oxidation of the endogenous cannabinoid, 2-AG. The striking catalytic efficiency of COX-2 mediated 2-AG oxygenation coupled with evidence that 2-AG is the preferred ester substrate, even when compared to 1-AG, suggests that this endocannabinoid is a natural COX-2 substrate.
Comparable oxygenation of 2-AG was observed with human and mouse COX-2's whereas 2-AG was a very poor substrate for sheep COX-1. This suggests that 2-AG oxygenation is a selective function of COX-2. Similar selectivity is observed in the COX-2-dependent oxygenation of anandamide (11). Indeed, in the present study, no glyceryl prostaglandins were detected following incubation of 2-AG with unactivated RAW264.7 cells, which contain detectable levels of COX-1 ( Figure 6).
The identification of active site residues that promote efficient metabolism of 2-AG provides a molecular basis for isoform selectivity and suggests a raison d'etre for the highly conserved side pocket in COX-2 enzymes. Thus, in addition to differential regulation and tissue distribution, altered substrate specificity may represent an evolutionary impetus for the existence of two COX isoforms. Our initial studies only evaluated the role of residues that represent conserved differences between COX-2 and COX-1 in the side pocket region. There are other conserved differences between the two enzymes in the lobby region located below Arg-120 and their importance in the differential oxygenation of 2-AG is under evaluation. The ability of the endoperoxide product of COX-2 action, PGH 2 -G, to serve as a substrate for PGD synthase raises the possibility that 2-AG may be the precursor to a family of eicosanoids as diverse as those produced from arachidonic acid. Our finding that PGD 2 -G is synthesized intracellularly and released extracellularly suggests that glyceryl prostaglandins are sufficiently stable to serve as intracellular or intercellular mediators.
Evaluation of the binding of glyceryl prostaglandins to cell surface or nuclear receptors may provide insight into the mechanisms by which COX-2 produces manifold responses in both the central nervous and immune systems.
Finally, these findings establish a role for COX-2 in endogenous cannabinoid metabolism. Although anandamide oxygenation by COX-2 has been reported, previous investigations produced varied results, employed anandamide concentrations significantly higher than those present in vivo, and did not demonstrate oxygenation of anandamide released from endogenous cellular stores (11,27,44). The present findings establish that endogenously released 2-AG is oxygenated by COX-2 in agonist-stimulated macrophages. This suggests that COX-2 may play a regulatory role in endocannabinoid signaling by decreasing 2-AG levels and reducing tone at CB receptors (45,46).