Metabolism of the endocannabinoids, 2-arachidonylglycerol and anandamide, into prostaglandin, thromboxane, and prostacyclin glycerol esters and ethanolamides.

Cyclooxygenase-2 (COX-2) action on the endocannabinoids, 2-arachidonylglycerol (2-AG) and anandamide (AEA), generates prostaglandin glycerol esters (PG-G) and ethanolamides (PG-EA), respectively. The diversity of PG-Gs and PG-EAs that can be formed enzymatically following COX-2 oxygenation of endocannabinoids was examined in cellular and subcellular systems. In cellular systems, glycerol esters and ethanolamides of PGE(2), PGD(2), and PGF(2alpha) were major products of the endocannabinoid-derived COX-2 products, PGH(2)-G and PGH(2)-EA. The sequential action of purified COX-2 and thromboxane synthase on AEA and 2-AG provided thromboxane A(2) ethanolamide and glycerol ester, respectively. Similarly, bovine prostacyclin synthase catalyzed the isomerization of the intermediate endoperoxides, PGH(2)-G and PGH(2)-EA, to the corresponding prostacyclin derivatives. Quantification of the efficiency of prostaglandin and thromboxane synthase-directed endoperoxide isomerization demonstrated that PGE, PGD, and PGI synthases catalyze the isomerization of PGH(2)-G at rates approaching those observed with PGH(2). In contrast, thromboxane synthase was far more efficient at catalyzing PGH(2) isomerization than at catalyzing the isomerization of PGH(2)-G. These results define the in vitro diversity of endocannabinoid-derived prostanoids and will permit focused investigations into their production and potential biological actions in vivo.

RAW 264.7 Cell Metabolism of AEA-Low passage number murine RAW 264.7 cells were grown to 30 -50% confluence and activated with IFN-␥ (10 units/ml) and LPS (1 g/ml) in serum-free DMEM for 7.5 h at 37°C. Following stimulation, medium was aspirated and replaced with fresh phosphate-buffered saline. Cells were treated with Me 2 SO vehicle, indomethacin (3 M), or indomethacin phenethylamide (3 M) for 20 min at 37°C followed by the addition of AEA (20 M). After 30 min, phosphate-buffered saline was removed and extracted twice with an equal volume of CHCl 3 :MeOH (2:1). The solvent was evaporated, and the residue was dissolved in H 2 O:MeCN (1:1) and subjected to liquid chromatography/mass spectrometry (LC/MS) analysis. Exogenous AEA metabolism by RAW 264.7 cells was evaluated in three independent experiments, and a representative result is displayed.
HCA-7 Cell Metabolism of 2-AG and AEA-HCA-7 cells were grown to near confluence (Ͼ80%) in DMEM containing 10% fetal bovine serum. Medium was aspirated and replaced with fresh Hanks' balanced salt solution. Cells were treated with Me 2 SO vehicle, indomethacin (3 M), or indomethacin phenethylamide (3 M) for 20 min at 37°C followed by the addition of 2-AG (20 M) or AEA (20 M). After 30 min, Hanks' balanced salt solution was removed and extracted twice with an equal volume of CHCl 3 :MeOH (2:1). The solvent was evaporated, and the residue was dissolved in H 2 O:MeCN (1:1) and subjected to LC/MS analysis. HCA-7 cell metabolism of each endocannabinoid was evaluated in three independent experiments, and representative results are displayed.
PGDS Assays-PGH 2 and PGH 2 -G isomerization by human, hematopoietic PGDS was quantitatively assessed by treating 50 l of buffer (100 mM Tris, 500 M phenol, 2 mM reduced glutathione, pH 8.0) containing 885 nM mCOX-2 and 19 nM PGDS with the indicated concentration of 2-AG or arachidonic acid. Following incubation at room temperature, reactions were quenched by the addition of 150 l of ice-cold MeOH. Quenched reaction mixtures were vigorously vortexed and placed on ice. The mixtures were then centrifuged (14,000 rpm, 10 min), and 150-l aliquots of the supernatants were transferred to clean microcentrifuge tubes. MeOH and water were removed in vacuo, and residues were taken up in 25 l of Me 2 SO and vigorously vortexed. Fresh, frozen rat plasma was then added to effect ester hydrolysis (final volume of 300 l) as previously described (17). Plasma-treated samples were incubated for 20 min at 37°C. PGD 2 quantification was conducted by gas chromatography/negative ion chemical ionization mass spectrometry (GC/NCI MS) as described previously (18). The assay conditions were developed using a standard TLC assay and 20 M [1-14 C]arachidonic acid with an initial incubation time of 5 min (19). Essentially complete consumption of [1-14 C]arachidonic acid was observed with 885 nM mCOX-2. Using 19 nM PGDS, [1-14 C]PGD 2 represented less than 50% of the total products.
PGES Assays-PGH 2 and PGH 2 -G isomerization by human, microsomal PGES was quantitatively assessed by treating 100 l of buffer (100 mM Tris, 500 M phenol, 2.5 mM reduced glutathione, pH 8.0) containing 885 nM mCOX-2 and 140 nM PGES with the indicated concentration of 2-AG or arachidonic acid. Following incubation at room temperature, reactions were quenched by the addition of an equal volume of 4 N NaOH and incubated for 90 min at 37°C to effect ester hydrolysis and degradation to PGB 2 . Reaction mixtures were then neutralized with 4 N HCl and rapidly frozen (Ϫ80°C). PGB 2 quantification was conducted by GC/NCI MS as described previously (18). The assay conditions were developed using a standard TLC assay and 20 M [1-14 C]arachidonic acid with an initial incubation time of 5 min (19). Essentially complete consumption of [1][2][3][4][5][6][7][8][9][10][11][12][13][14]  PGH 2 and PGH 2 -G isomerization by TXAS was quantitatively assessed by treating 100 l of buffer (100 mM Tris, 100 M phenol, 0.05% Lubrol, pH 8.0) containing 885 nM mCOX-2 and 12.5 nM hTXAS with the indicated concentration of 2-AG or arachidonic acid. Following incubation at room temperature, reactions were quenched by the addition of an equal volume of 4 N NaOH and incubated for 90 min at 37°C to effect ester hydrolysis. Reaction mixtures were then neutralized with 4 N HCl and rapidly frozen (Ϫ80°C). TxB 2 quantification was conducted by GC/NCI MS as described previously (18). The assay conditions were developed using a standard TLC assay and 20 M [1-14 C]arachidonic acid with an initial incubation time of 5 min (19). Essentially complete consumption of [1-14 C]arachidonic acid was observed with 885 nM mCOX-2. Using 12.5 nM TXAS, [1-14 C]TxB 2 and [1-14 C]HHT represented less than 50% of the total products.
Prostacyclin Synthase Preparation and Assay-PGIS was prepared from bovine aorta essentially as described (20). Frozen aorta was chopped into small pieces (ϳ1 cm 3 ) and placed in liquid nitrogen. The frozen pieces were homogenized in 4 volumes of 100 mM Tris-Cl (pH 7.2) containing 100 M flurbiprofen with a Kinematica AG Polytron homogenizer at 4°C. The homogenate was centrifuged twice at 10,000 ϫ g for 15 min at 4°C. The resultant supernatant was centrifuged again at 150,000 ϫ g for 60 min at 4°C to collect the microsomal pellet. Microsomes from 10 -20 g of tissue were resuspended in 1 ml of 100 mM Tris-Cl (pH 7.2) containing 100 M flurbiprofen.
To assay PGIS activity, resuspended microsomes were placed in 100 l of 100 mM Tris-Cl (pH 7.4) containing 100 M flurbiprofen. To this solution was added [1-14 C]PGH 2 , PGH 2 -G, or PGH 2 -EA in 100 l of 100 mM Tris-Cl (pH 8.0) containing 500 M phenol. Reactions were allowed to proceed at room temperature for 5 min prior to termination with 300 l of ice-cold EtOAc. In reactions with [1-14 C]PGH 2 , 6 l of 1 N HCl was added after termination to acidify the aqueous layer and improve product extraction. For [1-14 C]arachidonic acid analysis, products were analyzed by TLC as previously described using an eluant of EtOAc:2,2,4trimethylpentane:HOAc:H 2 O (110:50:20:100, v/v, upper phase) (21). PGH 2 and PGH 2 -G isomerization by PGIS was quantitatively assessed by treating 100 l of buffer (100 mM Tris, 500 M phenol, pH 8.0) containing 885 nM mCOX-2 with the indicated concentration of 2-AG or arachidonic acid. Following incubation at room temperature for 1 min, reactions were treated with 14.5 g of the PGIS microsomal protein suspension. Following incubation at room temperature for the indicated time, reactions were quenched by the addition of an equal volume of 4 N NaOH and incubated for 90 min at 37°C to effect ester hydrolysis. Reaction mixtures were then neutralized with 4 N HCl and rapidly frozen (Ϫ80°C). 6-Keto-PGF 1␣ quantification was conducted by GC/ NCI MS as described previously (18). The assay conditions were developed using a standard TLC assay and 20 M [1-14 C]arachidonic acid (21). Essentially complete consumption of [1-14 C]arachidonic acid was observed with 885 nM mCOX-2. Using 14.5 g of microsomal protein, [1-14 C]6-keto-PGF 1␣ represented less than 50% of the total products.
Mass Spectrometry-LC/MS was conducted with a Waters 2690 Separations Module and a Zorbax RX-C18 narrow bore column (15 cm ϫ 2.1 mm, 5 M) interfaced to a Finnigan TSQ-7000 triple quadrupole mass spectrometer as previously described (10). Sodiated analytes were eluted with increasing concentrations of acetonitrile in 0.001% aqueous sodium acetate and detected as positive ions. The presence of sodium in analytes was confirmed by detecting a fragment of 23 atomic mass units following collision-induced dissociation.

Endocannabinoid-derived Prostanoids of the D-series-PGD
synthase catalyzes the isomerization of 2-AG-derived PGH 2 -G to PGD 2 -G (10). To determine if this enzyme can catalyze the analogous reaction with AEA-derived PGH 2 -EA, LPS/IFN-␥activated RAW 264.7 cells were treated with exogenous AEA and the PG-EA products were analyzed by LC/MS. Only one PG-EA product was observed in the medium of AEA-treated, activated RAW 264.7 macrophages. This PG-EA product coeluted with the minor PG-EA product with m/z 418 produced by COX-2 oxygenation of AEA as well as with synthetic PGD 2 -EA ( Fig. 1). PGD 2 -EA was not detected in the medium of unactivated macrophages or LPS/IFN-␥-activated macrophages that were not treated with AEA (data not shown). Finally, PGD 2 -EA synthesis was inhibited by indomethacin and the selective COX-2 inhibitor, indomethacin phenethylamide ( Fig. 1). Thus, cellular PGD synthase can direct the isomerization of PGH 2 -EA. The absence of the nonenzymatic, PGH 2 -EA isomerization product, PGE 2 -EA, further suggests that PGD synthase efficiently catalyzes this reaction.
To address, quantitatively, the capacity of PGDS to metabolize endocannabinoid-derived endoperoxides, a coupled assay was developed to compare the rates of PGH 2 and PGH 2 -G isomerization by human, hematopoietic PGDS. Reaction mix-tures containing both COX-2 and PGDS were treated with either arachidonic acid or 2-AG at concentrations ranging from 0 to 20 M. Reactions were incubated at room temperature for 0 -5 min. To obtain a common chemical moiety for mass spectrometric analysis, namely the free acid PGD 2 , reaction products were treated with rat plasma, which contains a very active glycerol ester hydrolase (17). PGDS catalyzed the isomerization of both PGH 2 and PGH 2 -G to D-series prostanoids in a time-and concentration-dependent manner (Fig. 2). At all time points and concentrations, PGD 2 formation from arachidonic acid was ϳ2-fold greater than that observed when 2-AG was used as substrate in this coupled assay.
Endocannabinoid-derived Prostanoids of the E-and Fseries-To determine if endocannabinoid-derived, PGH 2 -like lipids can be metabolized into PG-Gs and PG-EAs of the E-and F-series, a cell-based enzymatic assay was employed. The human colon adenocarcinoma cell line, HCA-7, is known to express COX-2 constitutively (18). In addition, the primary prostaglandins produced by arachidonic acid-treated HCA-7 cells are PGE 2 and PGF 2␣ (18). Treatment of HCA-7 cells with 2-AG or AEA resulted in the appearance of PG-Gs or PG-EAs in the cell medium. Cells treated with 2-AG produced a PG-G with m/z 449 that coeluted with synthetic PGE 2 -G (Fig. 3A). HCA-7 cell production of PGE 2 -G was inhibited by indomethacin and indomethacin phenethylamide and, under the conditions em- ployed, required exogenous 2-AG ( Fig. 3A and data not shown). Similarly, 2-AG-treated HCA-7 cells produced a PG-G with m/z 451 that coeluted with synthetic PGF 2␣ -G (Fig. 3B). HCA-7 cell production of PGF 2␣ -G was inhibited by indomethacin and indomethacin phenethylamide and, under the conditions employed, required exogenous 2-AG ( Fig. 3B and data not shown).
AEA treatment of HCA-7 cells resulted in the appearance of PG-EA products in the medium. As expected, PGE 2 -EA and PGF 2␣ -EA were detected in selected ion mass chromatograms at m/z 418 and 420, respectively (Fig. 3C). A relatively small amount of PGD 2 -EA was also observed (Fig. 3C). Formation of these products required exogenous AEA and could be inhibited by nonselective COX or selective COX-2 inhibitors (data not shown). Confirmation of the identity of the PG-EA products was achieved by coelution with synthetically and enzymatically generated standards. In addition to PGE 2 -EA, a prominent, less polar, indomethacin-sensitive product appeared with m/z 418 (Fig. 3C, top panel). We hypothesized that this product may represent a metabolite of PGE 2 -EA with an identical mass. To test this hypothesis, tetradeuterated 13,14-dihydro-15-keto-PGE 2 -EA was synthesized by coupling the free acid to d4- ethanolamine. As suspected, the synthetic standard coeluted with the unknown m/z 418 species, identifying it as 13,14dihydro-15-keto-PGE 2 -EA (Fig. 3C, lower panel). Thus, HCA-7 cells appear to express active 15-hydroxyprostaglandin dehydrogenase (15-HPGDH) and ⌬ 13 -15-keto-prostaglandin reductase. The significant conversion of PGE 2 -EA to 13,14-dihydro-15-keto-PGE 2 -EA contrasts with little observed conversion of PGF 2␣ -EA to less polar metabolites of the same mass (Fig. 3C). This observation is consistent with previous reports that PGE 2 is a much better substrate than PGF 2␣ for 15-HPGDH (22)(23)(24)(25).
To address, quantitatively, the capacity of PGES to metabolize endocannabinoid-derived endoperoxides, a coupled assay, similar to the one described above for PGDS, was developed to compare the rates of PGH 2 and PGH 2 -G isomerization by human, microsomal PGES. Reaction mixtures containing both COX-2 and PGES were treated with either arachidonic acid or 2-AG at concentrations ranging from 0 to 20 M and incubated at room temperature for 0 -5 min. PGE 2 and PGE 2 -G were then converted to the free acid, PGB 2 , by treatment with sodium hydroxide thus permitting mass spectrometric analysis of a single, common product. PGES catalyzed the isomerization of both PGH 2 and PGH 2 -G to E-series prostanoids as evidenced by the time-and concentration-dependent production of PGB 2 in this assay (Fig. 4). At all time points and concentrations, the sequential actions of COX-2, PGES, and sodium hydroxide on 2-AG provided slightly lower levels of PGB 2 than those ob-served when arachidonic acid was used as substrate (60 -75%).
Endocannabinoid-derived Thromboxanes-TXAS catalyzes the isomerization of PGH 2 to thromboxane A 2 (TxA 2 ) and concurrently directs the scission of PGH 2 into HHT and malondialdehyde. TxA 2 and HHT are formed in roughly equal amounts by TXAS (26). In aqueous solutions, TxA 2 is rapidly hydrolyzed

FIG. 8. Concentration-and time-dependent isomerization of PGH 2 and PGH 2 -G by human TXAS.
A, concentration-dependent conversion of arachidonic acid (AA) and 2-AG to TxB 2 by the sequential actions of COX-2, TXAS, and sodium hydroxide. Incubations with COX-2 and TXAS were conducted at room temperature for 5 min (n ϭ 3, mean Ϯ S.E.). B, time-dependent conversion of AA and 2-AG to TxB 2 by the sequential actions of COX-2, TXAS, and sodium hydroxide. All incubations contained 20 M substrate and were conducted at room temperature (n ϭ 3, mean Ϯ S.E.). to thromboxane B 2 (TxB 2 ) (t1 ⁄2 ϳ 30 s) (27). To determine if TXAS can catalyze the isomerization of endocannabinoid derived PGH 2 -like lipids to the corresponding thromboxanes, solutions containing recombinant human COX-2 were treated with 2-AG or AEA in the presence or absence of recombinant TXAS. PGH 2 -G spontaneously isomerizes to, primarily, PGE 2and PGD 2 -G (Fig. 5, m/z 449, top panel). Inclusion of TXAS in these reactions led to a reduction in PGE 2 -and PGD 2 -G formation suggesting that TXAS metabolizes PGH 2 -G (Fig. 5, m/z  449, bottom panel). In addition, inclusion of TXAS in incubations led to the concurrent generation of new products with masses corresponding to HHT-G (Fig. 5, m/z 377) and TxB 2 -G (Fig. 5, m/z 467). The identification of HHT-G was accomplished by coelution of the TXAS-generated product with m/z 377 with authentic HHT-G formed as a minor side product in COX-2 oxygenations of 2-AG. Identification of TxB 2 -G was accomplished by coelution, under multiple conditions, of the TXAS-generated product with m/z 467 with d5-TxB 2 -G synthesized by coupling the free acid with deuterated glycerol (Fig. 6). The appearance of two peaks in chromatograms of both enzymatically and chemically produced TxB 2 -G may reflect the reversibility of hemiacetal formation. Similar results were obtained in AEA experiments (Fig. 7). Although a synthetic standard of TxB 2 -EA was not generated, the identity of this metabolite is strongly suggested by its mass, polarity, and chromatographic properties that closely resemble those observed for TxB 2 -G.

FIG. 11. Concentration-and time-dependent isomerization of PGH 2 and PGH 2 -G by bovine aorta PGIS.
A, concentration-dependent conversion of arachidonic acid (AA) and 2-AG to 6-keto-PGF 1␣ by the sequential actions of COX-2, PGIS, and sodium hydroxide. PGH 2 and PGH 2 -G were generated in a 1-min, room temperature preincubation of AA or 2-AG, respectively, with an excess of COX-2. PGIS-catalyzed isomerization of the resultant endoperoxide substrates was accomplished by treating the COX-2 products with bovine aorta microsomal protein for 5 min at room temperature. Treatment with sodium hydroxide provided the final common analyte, 6-keto-PGF 1␣ (n ϭ 3, mean Ϯ S.E.). B, time-dependent conversion of AA and 2-AG to 6-keto-PGF 1␣ by the sequential actions of COX-2, PGIS, and sodium hydroxide. Oneminute room temperature preincubations with excess COX-2 and 20 M substrate were followed by room temperature incubations with bovine aorta microsomal protein for the indicated times. Reaction termination and 6-keto-PGF 1␣ generation were accomplished by treatment with sodium hydroxide (n ϭ 3, mean Ϯ S.E.).
A coupled assay was employed to quantitatively compare the rates of PGH 2 and PGH 2 -G isomerization by human TXAS. Reaction mixtures containing both COX-2 and TXAS were treated with either arachidonic acid or 2-AG at concentrations ranging from 0 to 20 M and incubated at room temperature for 0 -5 min. The free acid, TxB 2 , was generated by treatment of the reaction products with sodium hydroxide to effect glycerol ester hydrolysis, thus providing a common analyte for mass spectrometric quantitation. TXAS catalyzed the isomerization of both PGH 2 and PGH 2 -G to thromboxane-like lipids in a timeand concentration-dependent manner (Fig. 8). In contrast to the comparable rates of PGH 2 and PGH 2 -G isomerization observed with both PGDS and PGES, TxB 2 production was markedly reduced (ϳ20-fold) when 2-AG was employed in this assay. In addition to quantifying the relative efficiency of TXAS metabolism of PGH 2 and PGH 2 -G, these studies offer strong confirmatory evidence that the product of TXAS action on PGH 2 -G, following spontaneous hydrolysis, is TxB 2 -G; base treatment of 2-AG/COX-2/TXAS incubations yielded a product with an identical mass and chromatographic behavior as enzymatically generated, as well as commercially obtained, TxB 2 .
Endocannabinoid-derived Prostacyclins-PGIS catalyzes the isomerization of PGH 2 to prostacyclin. In solution, prostacyclin is rapidly hydrolyzed to 6-keto-PGF 1␣ (28). To determine if PGIS can catalyze the isomerization of endocannabinoidderived PGH 2 -like lipids to the corresponding prostacyclins, microsomal PGIS was prepared from mature bovine aorta. PGH 2 -G and PGH 2 -EA were prepared by briefly incubating the respective endocannabinoid with purified mCOX-2. When PGH 2 -G was treated with heat-inactivated bovine aorta microsomes containing PGIS or was left untreated, nonenzymatic isomerization provided PGE 2 -G and, to a lesser extent, PGD 2 -G (Fig. 9A, top panel). When freshly resuspended microsomes were included in incubations, a reduction in PGE 2 -and PGD 2 -G formation was observed suggesting that PGH 2 -G was metabolized by a microsomal enzyme (Fig. 9A, bottom panel). In addition, the inclusion of freshly resuspended microsomes in incubations led to the generation of one or more new, more polar products corresponding in mass to 6-keto-PGF 1␣ -G (Fig.  9B). Coelution of synthetic d5-6-keto-PGF 1␣ -G generated by coupling the free acid with deuterated glycerol, under multiple conditions, suggested that the new product was 6-keto-PGF 1␣ -G (Fig. 10). The poor chromatographic behavior of 6-keto-PGF 1␣ -G likely results from the existence of multiple tautomeric forms as observed with 6-keto-PGF 1␣ (21,29). The enzymatic production of 6-keto-PGF 1␣ -G was inhibited by pretreatment of COX-2 with indomethacin and required the addition of 2-AG to incubation mixtures (data not shown). Similar results were obtained in AEA experiments (Fig. 9, C and D). Although a synthetic standard of 6-keto-PGF 1␣ -EA was not generated, the identity of this metabolite is suggested by its mass, polarity, and chromatographic properties that closely resemble those observed for 6-keto-PGF 1␣ -G.
To quantify the ability of PGIS to metabolize PGH 2 and PGH 2 -G, these endoperoxides were generated by treating 0 -20 M arachidonic acid or 2-AG with a large excess of COX-2. SCHEME 1 Bovine aorta microsomes were then added, and the mixture was incubated at room temperature for 0 -5 min. To obtain a common chemical moiety for mass spectrometric analysis, the free acid, 6-keto-PGF 1␣ , was generated by treating the enzymatic products with sodium hydroxide to effect glycerol ester hydrolysis. PGIS catalyzed the isomerization of both PGH 2 and PGH 2 -G to prostacyclin-like lipids as evidenced by the timeand concentration-dependent production of 6-keto-PGF 1␣ when arachidonic acid or 2-AG was used as substrate in this assay (Fig. 11). As with PGDS and PGES, and in stark contrast to TXAS, PGIS appeared relatively efficient at catalyzing PGH 2 -G isomerization. Production of 6-keto-PGF 1␣ in this assay was comparable with either arachidonic acid or 2-AG as substrate with relatively modest reductions observed with the endocannabinoid. In addition to quantifying the relative efficiency of PGIS metabolism of PGH 2 and PGH 2 -G, these studies offer strong confirmatory evidence that the product of PGIS action on PGH 2 -G, following spontaneous hydrolysis, is 6-keto-PGF 1␣ -G; base treatment of 2-AG/COX-2/PGIS incubations yielded a product with an identical mass and chromatographic behavior as enzymatically generated, as well as commercially obtained, 6-keto-PGF 1␣ .

DISCUSSION
Metabolism of COX-2-generated, endocannabinoid-derived endoperoxides was investigated in the present study. The demonstration that activated RAW 264.7 cells metabolize exogenous AEA to a single ethanolamide PG, PGD 2 -EA, indicates that PGH 2 -EA is short-lived in this cellular environment; rapid PGDS-directed isomerization of PGH 2 -EA prevents nonenzymatic isomerization of the unstable hydroxyendoperoxide that would result in the formation of predominantly PGE 2 -EA (Fig.  1). This result conforms to the metabolism of the arachidonatederived endoperoxide, PGH 2 , in this murine macrophage cell line. IFN-␥/LPS-activated RAW 264.7 cells secrete roughly 100fold more PGD 2 than PGE 2 (30). Coupled with previous results from studies of PGH 2 -G, we conclude that hematopoietic PGD synthase efficiently catalyzes the formation of PGD 2 -like lipids from COX-2-oxygenated endocannabinoids in a cellular environment (10). To more rigorously characterize the efficiency of PGDS-catalyzed isomerization of PGH 2 -G and PGH 2 , a cellfree, enzyme-coupled assay was developed. As shown in Fig. 2, the sequential actions of COX-2 and PGDS provide PGD-like products more efficiently with arachidonic acid than with 2-AG as substrate. However, relatively modest differences in efficiency (ϳ2-fold) were observed. Taken together, these data suggest that PGH 2 -G may represent a biologically relevant PGDS substrate.
Investigations using HCA-7 human colon adenocarcinoma cells that constitutively express COX-2 demonstrated that 2-AG and AEA can be metabolized to PGE 2 -G and PGF 2␣ -G (2-AG) or the related ethanolamide species (AEA) (Fig. 3). The requirement for a two-electron reduction of PGH 2 -G or PGH 2 -EA to provide the F 2␣ metabolites suggests the involvement of a PGFS. The involvement of PGES in the production of E-series PG-Gs or -EAs is less clear, because these lipids represent the major nonenzymatic isomerization products of the respective hydroxyendoperoxides. Results displayed in Fig. 3C suggest that PGES is involved in directing PGH 2 -EA isomerization in HCA-7 cells. The combined E-series PG-EAs (PGE 2 -EA and 13,14-dihydro-15-keto-PGE 2 -EA) are produced in much higher quantities than PGD 2 -EA (Fig. 3C). Typical ratios of PGE 2 -EA to PGD 2 -EA formation following nonenzymatic isomerization of PGH 2 -EA are much lower (see Figs. 7A and 9C, top panels). Although E 2 to D 2 ratios are sensitive to the environment, it seems most likely that PGES directs the isomerization of PGH 2 -EA in this cellular context (31). In ad-dition, these studies uncovered an intact PG catabolic pathway in HCA-7 cells comprised of 15-HPGDH and ⌬ 13 -15-keto-prostaglandin reductase (Fig. 3C). This result highlights the complexities of cellular model systems used in the study of PG biology and indicates that PG catabolism must be considered when using the HCA-7 cell line and, potentially, colon cancer cell lines in general. The colon tumor-enhancing effects of the isoflavanoid, genistein, in azoxymethane-treated rats has been attributed, in part, to inhibition of 15-HPGDH activity (32). Our results identify the HCA-7 cell line as a potentially useful tool in studying the role of 15-HPGDH in colon carcinogenesis and in the identification of novel 15-HPGDH modulators.
The capacity of PGES to catalyze the isomerization of PGH 2 -G was assessed more quantitatively in an enzyme-coupled assay. The sequential actions of COX-2 and PGES on arachidonic acid and 2-AG provided PGE 2 and PGE 2 -G, respectively, in a time-and concentration-dependent manner (Fig. 4). As evidenced by the comparable concentrations of PGB 2 produced, the efficiency of endocannabinoid catabolism by the COX-2/PGES pathway approaches that observed with the presumed natural substrate, arachidonic acid, suggesting that 2-AG-derived, E-series prostanoids may represent biologically meaningful natural products. This possibility is particularly interesting in light of the observation that the inflammatory cytokine, interleukin-1␤, induces both COX-2 and microsomal PGES expression (33).
TXAS is generally considered to have strict substrate selectivity (34 -38). Due to the difficulties associated with endoperoxide synthesis, this consideration is based on the examination of only a few substrates. TXAS generates thromboxanes from PGH 2 , PGH 3 , PGG 2 , and C 21 -PGH 2 analogs (34 -39). In contrast, TXAS action on PGH 1 , ⌬ 4 -PGH 1 , 8-iso-PGH 2 , 13(S)-hydroxy-PGH 2 , and 15-keto-PGH 2 yields, almost exclusively, the corresponding HHT derivative (34 -39). These results have been interpreted to indicate that the position of the ⌬ 5 -double bond with respect to the cyclopentyl moiety and a PGH 2 -like structure between carbons 12 and 15 is essential for thromboxane generation. The observation that PGH 2 -G and PGH 2 -EA serve as substrates for human TXAS is consistent with this structure-activity relationship, because both endoperoxides possess these structural features (Figs. 5 and 7). However, quantitative analysis of synthase-directed PGH 2 -G isomerization demonstrated that, although TXAS acts on the endoperoxide glycerol ester to provide TxA 2 -G, this reaction proceeds more than an order of magnitude less efficiently than the analogous reaction with the free acid substrate, PGH 2 (Fig. 8). This finding is consistent with site-directed mutagenesis studies and computer modeling that have shown that TXAS interactions with the carboxylate moiety of PGH 2 -like compounds are critical in thromboxane production (40). In addition, it is noteworthy that platelets, expressing only the COX-1 isoform, are the predominant source of thromboxanes in vivo (41,42). Because both AEA and 2-AG are poor substrates for COX-1, it is unlikely these endocannabinoids serve as significant endoperoxide precursors in platelets (10,11). Thus, it is not surprising that the endocannabinoid-derived product, PGH 2 -G, is a poor TXAS substrate.
PGIS has a less restricted substrate profile than TXAS and has been reported to efficiently isomerize PGH 2 , PGG 2 , PGH 3 , 13(S)-hydroxy-PGH 2 , and 15-keto-PGH 2 to the corresponding prostacyclin derivatives (38). In addition, COX-2, rather than COX-1, appears to be the main source of PGIS substrate in vivo, and PGIS preferentially couples with COX-2 in coexpression systems (42)(43)(44). Thus, from both enzymologic and evolutionary perspectives, one anticipates that the COX-2-specific products, PGH 2 -G and PGH 2 -EA, would serve as PGIS sub-strates. As shown in Fig. 9, PGH 2 -G and PGH 2 -EA should be added to the list of substrates that undergo isomerization to prostacyclins when acted upon by PGIS. In addition, quantitative characterization of the COX-2/PGIS metabolic pathway demonstrated that the addition of the glycerol moiety to arachidonic acid did not result in dramatic reductions in the efficiency of prostacyclin generation (Fig. 11). Given the recent demonstration that PGI 2 activates the nuclear receptor, PPAR␦, and that elevations of both COX-2 and PPAR␦, but not COX-1, are found in human colon cancers, it will be interesting to investigate whether endocannabinoid-derived prostacyclins represent PPAR␦ signal mediators (45). Such studies are particularly compelling given the recent demonstration that 15-hydroxyeicosatetraenoic acid glycerol ester displays markedly different PPAR activity than the corresponding free acid (46).
As shown in Scheme 1, the endocannabinoid, 2-AG, is the lipid precursor to a family of glycerol esters that is as diverse as the known prostaglandins. A similar outline can be offered for an AEA-derived family of ethanolamides. The demonstration that this series of novel lipids can be formed enzymatically following COX-2 oxygenation of endocannabinoids provides a framework for further investigations into the production and action of PG glycerol esters and ethanolamides in vivo. Our results demonstrate that 2-AG-derived thromboxanes are produced by COX-2/TXAS at dramatically reduced levels when compared with arachidonic acid. Thus, endocannabinoid-derived thromboxanes are unlikely to be biologically significant mediators. In contrast, endocannabinoid-derived prostanoids of the D-, E-, and I-series are generated by the sequential actions of COX-2 and the corresponding prostaglandin synthase at rates comparable to those observed with the presumed natural substrate, arachidonic acid. It is tempting to speculate on the potential biological activities of these novel lipids given their structural similarity to the PGs and the widespread coincident expression of COX-2, PG synthases, and 2-AG biosynthetic enzymes. However, we anticipate that further chemical and biological experimentation should abrogate the need for speculation. Nonetheless, it is possible that the selective biosynthesis of these two classes of lipids by COX-2 may represent an isoform-specific function for this enzyme (47).