Enzymes and Receptors of Prostaglandin Pathways with Arachidonic Acid-derived Versus Eicosapentaenoic Acid-derived Substrates and Products*♦

Dietary fish oil containing ω3 highly unsaturated fatty acids has cardioprotective and anti-inflammatory effects. Prostaglandins (PGs) and thromboxanes are produced in vivo both from the ω6 fatty acid arachidonic acid (AA) and the ω3 fatty acid eicosapentaenoic acid (EPA). Certain beneficial effects of fish oil may result from altered PG metabolism resulting from increases in the EPA/AA ratios of precursor phospholipids. Here we report in vitro specificities of prostanoid enzymes and receptors toward EPA-derived, 3-series versus AA-derived, 2-series prostanoid substrates and products. The largest difference was seen with PG endoperoxide H synthase (PGHS)-1. Under optimal conditions purified PGHS-1 oxygenates EPA with only 10% of the efficiency of AA, and EPA significantly inhibits AA oxygenation by PGHS-1. Two- to 3-fold higher activities or potencies with 2-series versus 3-series compounds were observed with PGHS-2, PGD synthases, microsomal PGE synthase-1 and EP1, EP2, EP3, and FP receptors. Our most surprising observation was that AA oxygenation by PGHS-2 is only modestly inhibited by EPA (i.e. PGHS-2 exhibits a marked preference for AA when EPA and AA are tested together). Also unexpectedly, TxA3 is about equipotent to TxA2 at the TPα receptor. Our biochemical data predict that increasing phospholipid EPA/AA ratios in cells would dampen prostanoid signaling with the largest effects being on PGHS-1 pathways involving PGD, PGE, and PGF. Production of 2-series prostanoids from AA by PGHS-2 would be expected to decrease in proportion to the compensatory decrease in the AA content of phospholipids that would result from increased incorporation of ω3 fatty acids such as EPA.

element binding protein-1c, major transcription factors controlling lipid metabolism (2,23,24). Other proteins that can be activated directly by polyunsaturated fatty acids, and thus, whose activities might be altered by changes in EPA/AA ratios include protein kinase C (25), NADPH oxidase (26), and a twopore domain K ϩ channel (27). Polyunsaturated fatty acids such as AA can promote apoptosis but the mechanism is not known (28).
Finally, the eicosanoid pathways for lipid mediator formation; including the cyclooxygenase pathways, the 5-, 12-, and 15-lipoxygenase pathways, the P450 epoxygenase pathways, and non-enzymic oxidative pathways; are influenced by changes in EPA/AA ratios (29 -35). Anti-thrombotic, anti-inflammatory, and anti-carcinogenic effects of 3 fatty acids could result, at least in part, from their ability to attenuate the synthesis of specific eicosanoids and/or to alter the nature of the eicosanoid products formed or to serve as precursors of novel products such as isoprostanes and resolvins (32,33,(35)(36)(37)(38).
Prostanoids are synthesized via the cyclooxygenase pathway, most commonly from AA, in response to various hormones and physical stimuli (29). The pathway involves three stages: (a) mobilization of AA from membrane phospholipids by cytosolic phospholipase A 2 ␣ (cPLA 2 ␣) sometimes in conjunction with secretory sPLA 2 s; (b) conversion of AA to the prostaglandin endoperoxide PGH 2 by prostaglandin endoperoxide H synthase-1 or -2 (PGHS-1 or -2) also known as cyclooxygenase-1 or -2 (COX-1 or -2); and (c) isomerization of PGH 2 to a "2-series" product, PGD 2 , PGE 2 -, PGF 2␣ , PGI 2 , or thromboxane A 2 (TxA 2 ), by specific synthases. Newly formed PGs exit cells and function primarily through G-protein-coupled receptors on neighboring or parent cells to elicit responses. Because PGs act at or near their sites of synthesis and are rapidly metabolized, they are considered to be "local" hormones. Importantly, EPA can serve as a substrate for PG formation generating "3-series" PG products including PGD 3 , PGE 3 , PGF 3␣ , PGI 3 , and TxA 3 .
There is only limited biochemical information available on the specificities of the enzymes and receptors of the prostanoid pathways with EPA-derived versus AA-derived substrates and products. Here we report studies that address this topic.
Expression, Purification, and Assay of PGHSs-Hexahistidine-tagged (His 6 ) ovine (ov) PGHS-1, murine (mu) murine PGHS-2, and human (hu) PGHS-2 were expressed in S21 insect cells and purified through Ni-NTA chromatography essentially as described previously (39 -41). His 6 -muPGHS-1 was expressed also in insect cells but was unstable following Ni-NTA chromatography so in the experiment using this enzyme, the supernatant from centrifugation of solubilized cell pellets was used for COX assays. Oxygen electrode assays for COX activity were performed as detailed in previous reports (39 -41). COX assays of purified enzymes utilizing radio thin layer chromatography assays of PGH 2 or PGH 3 formation were performed using [1-14 C]AA and [1-14 C]EPA as described previously (41).
Preparation of Platelet-rich Plasma (PRP)-Platelets were obtained from normal human donors who had not taken medication during the 2 weeks prior to donation. Whole blood was drawn into 3.8% sodium citrate (1:9; citrate:blood). The blood was centrifuged at 180 ϫ g for 10 min at room temperature, and PRP was transferred to a new tube. The remaining blood was centrifuged at 1000 ϫ g for 10 min at room temperature to obtain PRP. For aggregation studies using PRP, the platelet count was determined on a Coulter counter (Model Z; Coulter Electronics, Hialeah, FL) and adjusted with HEPES-Tyrode's buffer (137 mM NaCl, 3 mM KCl, 12 mM NaHCO 3 , 0.34 mM Na 2 HPO 4 , 14.7 mM HEPES, 0.35% dextrose, and 0.35% bovine serum albumin, pH 7.4) to 2.2 to 2.5 ϫ 10 8 platelets/ml. For preparation of washed platelets, human platelets in PRP were separated from plasma by gel filtration over Sepharose 2B columns in HEPES-Tyrode's buffer. The peak tubes were pooled, and the platelet count was adjusted to 2.5 ϫ 10 8 platelets/ml before proceeding with platelet aggregation studies. Washed platelets (400 l) were placed in a cuvette in the aggregometer and stirred at 37°C. The integrity of the washed platelets was tested by their ability to be activated by collagen (1-5 mg/ml) and ␣-thrombin (3 nM).
Expression, Purification, and Assay of Human Microsomal PGE Synthase-1 (hu mPGES-1)-The human PGES cDNA (Invitrogen) was amplified using High Fidelity PCR kit (Invitrogen) with the 5Ј-primer (with BspHI) GAA TTC ATC ATG  ATC CCT GCC CAC AGC CTG GTG A and the 3Ј-primer  (with HindIII and His 6 -tag) CAT CCA AGC TTG TCA GTG  GTG GTG GTG GTG GTG CAG GTG GCG GGC CGC AAC. The PCR product was purified with a QIAquick PCR purification kit (Qiagen). The pRMGsp expression vector 4 was digested by AflIII and XhoI and the amplified PGES PCR product was digested with BspHI and HindIII. The digested DNA was isolated by electrophoresis on a 1% agarose gel, and the DNA band was purified with a QIAquick gel purification kit (Qiagen). The digested and purified expression plasmid and His 6 -hu mPGES-1 insert were ligated with T4 DNA ligase. The ligation sample was transformed into DH5-␣ competent cells, and plasmids from positive colonies were sequenced to confirm the expression construct pRMGsp-PGES-His 6 .
The pRMGsp-PGES-His 6 DNA with the aid DNA pAL9 (digested by PstI) were transformed into freshly made Schizosaccharomyces pombe competent cells. After selection of positive colonies on MAA plates, the potential transformants were screened twice on yeast extract/peptone/dextrose medium (YPD)/G418 plates (containing G418 at 20 g/ml). Positive single colonies were then grown in yeast extract-sodium lactate (YEL) medium (with 10 g/ml G418) to make glycerol stocks, which were stored at Ϫ80°C.
For expression of His 6 -hu mPGES-1 in S. pombe, 5 ml of YEL medium (with 10 g/ml G418) was inoculated with 200 l of a glycerol stock culture and shaken for 24 h at 32°C. The culture was transferred into 50 ml of fresh YEL/G418 medium and incubated at 32°C for 48 h with shaking until the A 600 was 10 -12. The culture was then transferred into 1 liter of YPD medium (with 100 g of G419/ml) and incubated at 32°C with shaking until the A 600 exceeds 20 (ϳ48 -60 h). The cells were harvested by centrifugation at 3000 ϫ g for 15 min at 4°C, and the cell pellet was stored at Ϫ80°C.
After completely thawing the cell paste in ice water, 5 ml of lysis buffer (15 mM Tris-HCl, 250 mM sucrose, 0.1 mM EDTA, 1 mM reduced glutathione, pH 8.0) was added for per gram of cell pellet. The resuspended cell pellet was lysed using an Emulsi-Flex-C3 (at 20,000 -25,000 p.s.i. with two passes). The cell lysate was centrifuged at 8000 ϫ g for 20 min at 4°C. The membrane-containing supernatant fraction was then centrifuged at ϳ200,000 ϫ g for 1 h at 4°C. The membrane pellet was resuspended in loading buffer (50 mM sodium phosphate, 300 mM NaCl, 10% glycerol, pH 8.0) and dodecyl maltoside (Anatrace) was added to a final concentration of 1%. The membrane fraction was stirred for 1 h at 4°C and then centrifuged at 200,000 ϫ g for 1 h at 4°C. Imidazole was added to the supernatant fraction to a final concentration of 10 mM, and the mixture was loaded onto an Ni-NTA (Qiagen) column equilibrated with loading buffer supplemented with 10 mM imidazole; all column buffers contained 0.05% dodecyl maltoside. After loading, the column was washed with loading buffer containing 22 mM imidazole. The bound His 6 -hu mPGES-1 was eluted using loading buffer containing 200 mM imidazole. Elution fractions containing His 6 -hu mPGES-1 were pooled and concentrated using a 30 molecular weight cutoff Centricon spin concentrator.
Microsomal Preparations of TxA Synthase (TxAS)-Mouse (mu) TXAS was expressed in Sf21 insect cells as described previously (44,45). The cells from a 250-ml culture were harvested after 4 days, collected, and washed twice with ice-cold phosphate-buffered saline (PBS) and stored at Ϫ80°C. Cell pellets were thawed on ice and resuspended in 100 mM Tris-HCl pH 7.4, 1 mM EDTA and 1ϫ Complete protease inhibitor. Cells were disrupted by sonication and centrifuged at 10,000 ϫ g for 10 min at 4°C. The supernatant was then centrifuged at 100,000 ϫ g for 1 h at 4°C. The resulting pellet was homogenized in 10 mM Tris-HCl, pH 7.4, 0.1 mM EDTA, and 20% glycerol using a Dounce homogenizer and the protein concentration measured. The protein was used immediately for in vitro synthesis of TxA 2 or TxA 3 .
The PCR was used to amplify the coding domain of the huEP1 receptor (nucleotides 1-1209; GenBank TM accession number L22647 (53)) from human kidney cDNA. The product encoding the huEP1 was purified by agarose gel electrophoresis and cloned into the EcoRV site of pcDNA3 to yield huEP1/pcDNA3. The sequence of huEP1 in huEP1/pcDNA3 was verified by DNA sequencing. huEP1/pcDNA3 encoding the huEP1 was transiently transfected into HEK293 cells using Lipofectamine 2000 according to the recommendation of the manufacturer, and cells were harvested 30 h post-transfection. Cell pellets were stored at Ϫ80°C and membranes prepared from these cells were used to determine the relative affinities of PGE 2 versus PGE 3 .
Membranes were prepared from HEK293 cells essentially as described by Ungrin et al. (54). Briefly, cell pellets were thawed on ice and resuspended in Buffer A (10 mM HEPES/KOH, pH 7.4, with 1 mM EDTA and 1ϫ Complete protease inhibitor), disrupted by sonication, and centrifuged at 10,000 ϫ g for 10 min at 4°C. The supernatant was centrifuged at 100,000 ϫ g for 1 h at 4°C. The pellet was homogenized in Buffer A, and aliquots of the suspended protein (50 -100 g) were used immediately for binding assays.
Binding assays were performed in 200 l of 10 mM MES, pH 6.0, 1 mM EDTA and 10 mM MgCl 2 . Binding isotherms were performed for [ 3  beled ligand corresponding to the K d value for each receptor was then used in competition binding assays with PGE 2 versus PGE 3 , PGF 2␣ versus PGF 3␣ , or U46619 versus ⌬17 U46619. Nonspecific binding was determined in the presence of 10 M unlabeled ligand. Samples were incubated at 30°C for 1 h and then filtered through Whatman GF/C glass filters. The filters were washed three times with cold MES buffer (without EDTA) and radioactivity measured by liquid scintillation counting. Receptor binding data were analyzed by nonlinear regression in Origin. Statistical analyses were performed using Student's t test and/or ANOVA.
cAMP Assays-HEK cell lines expressing EP2 and EP4 receptors that had been grown as described above in 6-well plates were treated for 15 min at 37°C with fresh DMEM containing 50 mM isobutylmethylxanthine. Cells were then treated with various concentrations of PGE 2 or PGE 3 for an hour. The treatments were terminated by scrapping cells into 0.5 ml of TE (50 mM Tris-HCl, pH 7.5, containing 4 mM EDTA) and boiling for 8 min. After centrifuging the lysates, 50 ml of the supernatant (from about 10 5 cells) was used for cAMP analysis using an Amersham Biosciences cAMP assay system kit following the instructions of the manufacturer. The samples were quantified by scintillation counting and values for cAMP were calculated from the cAMP standard curve.
Assays of Inositol Phosphates (IPs)-Assay of IPs was performed by measuring receptor induced production of [ 3 H] IPs as described previously (55). Briefly, HEK293 cells expressing EP3 or FP receptors were grown in 24-well plates as described above and labeled by incubating overnight with 1 Ci of [ 3 H]myoinositol (Amersham Biosciences) per ml of DMEM. The cells were treated with 10 mM LiCl for 15 min prior to adding various concentrations of PGE 2 /PGE 3 or PGF 2␣ /PGF 3␣ for 1 h. Assays were terminated by adding 3 ml of chloroform/ methanol/water (1:1:1) to each well. The whole cell lysates were collected and centrifuged, and the resulting aqueous phase was applied to a Dowex AG1-X8 anion exchange column (Bio-Rad) to remove unincorporated [ 3 H]myoinositol. The IPs were eluted with 0.2 M ammonium formate, 0.1 M formic acid and quantified by liquid scintillation counting.
Measurements of Intracellular Ca 2ϩ -Intracellular Ca 2ϩ concentrations were measured as described previously by Fisher et al. (56). Nontransfected HEK293 cells and huEP1 plasmid-transfected HEK293 cells were incubated with 1 M Fura-2/AM (Invitrogen) for 15 min before adding various concentrations of PGE 2 or PGE 3 . The Ca 2ϩ signals were measured using a Shimadzu RF-5301 PC spectrofluorometer.
Platelet Aggregation-Aggregation assays were performed on a Chronolog dual-channel aggregometer at 37°C. The assays were carried out with 360 l of undiluted human PRP, and the compounds to be tested were added to obtain a final volume of 400 l. The assay was measured for 3 min following platelet activation. Only PRP that was responsive to 20 -40 nM ␥IIathrombin was used for experiments. Variable amounts of each aggregatory compound were added to PRP to determine the threshold concentration for platelet aggregation: U46619 or ⌬17 U46619 (0.1 nM to 2 M) and collagen (1-2 g/ml). In some experiments, inhibition of aggregation induced by 2 M U46619 was examined using different amounts of anti-aggre-gatory compounds: iloprost (0.1-10 nM) and PGI 2 or PGI 3 (10 pM to 2 M); flurbiprofen was added at 100 M. All compounds were diluted from a stock solution in organic solvent into PBS, pH 7.5, prior to the collection of blood, except for PGI 2 and PGI 3 , which were prepared by removing a solid chemical stock vial from Ϫ80°C (stored for less than 1 week) and dissolving in PBS less than 15 min before adding to PRP. The stability of each prostacyclin was checked by comparing the anti-aggregatory activity of 1 nM at the end of the experiment to that of the beginning of the experiment.
Platelet Aggregation by TxA 2 Versus TxA 3 -An aliquot of PRP (370 l) was placed into a cuvette in the aggregometer while stirring at 37°C. The TxA 2 and TxA 3 were prepared just before adding to PRP. In an Eppendorf tube, AA or EPA was added to a final concentration of 5 M in reaction buffer (100 mM Tris-HCl, pH 7.4, 1 mM phenol, and 10 M heme) containing 260 units of COX-2 to initiate the synthesis of PGH 2 or PGH 3 . The sample was vortexed for 20 s. Then, 30 l of TxA microsomal protein (ϳ500 g) was added to produce TxA 2 or TxA 3 in a final volume of 100 l. The mixture was vortexed for 10 s, at which time 30 l was removed and added to the 370 ml of PRP. Immediately, prior to the experiments using PRP, parallel reactions were performed with 14 C-labeled AA or EPA and the products quantified by radio thin layer chromatography as described above. This provided an accurate estimate of the various products being added in each case to the PRP.

RESULTS
Specificities of PGHS-1 and PGHS-2 with AA and EPA-As shown in Fig. 1, purified ovPGHS-1 and muPGHS-2 oxygenate AA with comparable catalytic efficiencies at concentrations of Ն1 M AA where reasonably precise O 2 electrode measurements of enzyme activity can be performed. ovPGHS-1 is essentially inactive with EPA while muPGHS-2 can use EPA with about 30% of the efficiency of AA in the range of 1-100 mM. 5 A solubilized preparation of muPGHS-1 expressed in baculovirus showed qualitatively similar results to those shown for ovPGHS-1 (data not shown); muPGHS-1 was unstable in our hands, and so we were unable to analyze purified enzyme. Although ovPGHS-1 was not active with low concentrations of EPA, significant activity (ϳ10% of that with AA) was observed when 15 M 15-hydroperoxyeicosatetraenoic acid was added to the reaction mixtures (data not shown). Purified huPGHS-2 also showed results very similar to those in Fig. 1 for muPGHS-2 (data not shown) when tested under essentially identical enzyme and substrate conditions.
Results similar to those illustrated in Fig. 1 have been reported by Kulmacz and co-workers using ovPGHS-1 (57,58). Moreover, the results with purified and semipurified enzymes are consistent with studies comparing the utilization of AA versus EPA by microsomal huPGHS-1 and huPGHS-2 (59), where, under optimal conditions, huPGHS-1 is 5% as active with EPA as with AA while huPGHS-2 is 25-30% as active with EPA as AA. It also is clear that EPA can be oxygenated by PGHS-1 in intact cells in a manner that is peroxide dependent (60).
EPA and AA have similar K m values with PGHS-1 and PGHS-2 ( Fig. 1 and (57, 59)), and so EPA would be expected to compete with AA for oxygenation. EPA/AA competition has been shown previously with PGHS-1 (61), and the results in Fig.  2 confirm these findings. Half-maximal inhibition occurs with equimolar AA versus EPA. Essentially identical results were also obtained with a solubilized preparation of His 6 -muPGHS-1. AA is about a 10 times better substrate than EPA for ovPGHS-1 in vitro, and as shown in Fig. 3, inhibition of oxygenation reflects primarily inhibition of AA oxygenation. A 5-fold excess of EPA caused 40% inhibition of [1-14 C]AA oxygenation by ovPGHS-1. This result is similar but not identical to that of Fig. 2, which shows about 75% inhibition at these concentrations of AA plus EPA. As expected, [1-14 C]EPA was a poor substrate for PGHS-1; however, EPA oxygenation was augmented slightly by the presence of AA. Again, this is probably because hydroperoxide is being generated when AA is present along with EPA in the reaction mixtures and hydroperoxides potentiate EPA oxygenation (59,60,62).
In contrast to the results obtained with PGHS-1, EPA was a relatively poor inhibitor of AA oxygenation by PGHS-2 (Fig. 2). For example, at equimolar AA and EPA concentrations, the first point at which there was a statistically significant decrease in the rate of oxygenation with muPGHS-2, there was only a 10% decrease in O 2 consumption and even with a 5-fold excess of EPA there was less than a 20% decrease in oxygenase activity. Based on the kinetic constants for muPGHS-2 ( Fig. 1) and huPGHS-2 for AA and EPA tested individually, one would expect about a 35% lower oxygenation rate with 20 M AA plus 20 M EPA and a 60% decrease in the rate with 20 M AA plus 100 M EPA. To examine this inconsistency, we incubated purified enzymes with [1-14 C]AA or [1-14 C]EPA with and without unlabeled competing substrate and measured the formation of radioactive PGH 2 or PGH 3 (plus HHTE (12(S)-hydroxy-5,8,10-heptadecatetraenoic acid) a degradation product of PGH 3 ) using radio thin layer chromatography (Fig. 3). When 20 M unlabeled AA was added to reaction mixtures containing 20 M [1-14 C]EPA, oxygenation of EPA by PGHS-2 was inhibited by 70 -90% (Fig. 3). In contrast, with 20 M [1-14 C]AA and 100 M unlabeled EPA there was only a modest inhibition (ϳ10%) of AA oxygenation (Fig. 3, lower panel).
Thus, EPA acts as an effective inhibitor of AA oxygenation by PGHS-1 but not PGHS-2, and indeed PGHS-2 shows a marked and unanticipated preference for AA when presented with a mixture of AA and EPA. The basis for the uneven competition between AA and EPA with PGHS-2 is not clear. It may involve  half of sites activity with PGHS-2 (40). In the case of PGHS-2 but not PGHS-1, binding of certain fatty acids to one COX site may facilitate oxygenation of AA bound to the other site.
In brief, our results with PGHSs show that (a) AA is an equally good substrate for PGHS-1 and PGHS-2; (b) EPA is a poorer substrate than AA for both PGHS-1 and PGHS-2 and a particularly poor substrate for PGHS-1; (c) EPA is an efficient inhibitor of AA oxygenation by PGHS-1 but not PGHS-2; and (d) in the presence of EPA, PGHS-2 shows a marked preference for AA.
Specificities of Lipocalin PGDS, Hematopoietic PGD Synthase (hPGDS), and Microsomal PGES-1 (mPGES-1) toward PGH 2 Versus PGH 3 - Table 1 shows data obtained in estimating the specificities of hPGDS, lPGDS, and mPGES-1 with PGH 2 versus PGH 3 . PGH 2 and particularly PGH 3 are unstable, and so they were generated in situ quantitatively from AA or EPA using an excess of purified PGHS-2 and then a PGDS or PGES was added immediately and the reactions continued for 20 -40 s. PGH 3 was found to be less stable than PGH 2 , so it was necessary to add more EPA than AA in generating the endoperoxides so that the PGH 2 and PGH 3 concentrations were about the same when a PGDS or PGES was added. The reactions were terminated before 20% of the PGH was consumed, and the reactions were performed with amounts of enzyme that provided approximately linear product formation with time. Somewhat different V max and K m values have been reported for each of the various PGD and PGE synthases we tested (63)(64)(65)(66)(67)(68). Because of this and the technical difficulties associated with multiple assays with unstable substrates and limited amounts of enzymes, we elected to use an endoperoxide substrate concentration in the range of 5 M for all of our assays, because as noted earlier, 5 5 M PGH 2 or PGH 3 would likely be as high a concentration as would be encountered by a PGDS or PGES in an intact cell. With all these provisos, the human versions of H-PGDS, L-PGDS, and mPGES-1 were all more than 3-fold less active with PGH 3 than with PGH 2 .
Quantitative data comparing the specificities of various prostanoid biosynthetic enzymes with AA versus EPA derived substrates is summarized in Table 2.
PGE and PGF Receptor Specificities-Membranes from HEK cell lines expressing various PGE (EP) and PGF (FP) receptors were used to determine the relative affinities of 2versus 3-series PGE and PGF (Table 3). Except for the EP4 receptor, the affinity of each receptor was significantly greater for the 2-series than the 3-series PGs. The most dramatic difference was with the FP receptor, which had a 78-fold higher affinity for PGF 2␣ than PGF 3␣ . Fig. 4 shows the potencies of 2versus 3-series PGs in eliciting second messenger formation via the EP and FP receptors. Differences were significant with EP1, EP2, EP3, and FP receptors as determined by ANOVA for EP1 (p Ͻ 0.001), EP2 (p Ͻ Radio thin layer chromatography assays were performed as described under "Experimental Procedures" (41). The indicated substrates were mixed with 0.5 g (ϳ12 units) of the purified His 6 -tagged PGHSs and the reactions continued for 30 s. Products were extracted, separated, and visualized by autoradiography. The thin layer plates were subsequently scraped and the amounts of radioactivity associated with the substrates and products determined by scintillation counting and used to compute the relative rates indicated in the figure.

TABLE 1 Specificities of human hematopoietic and lipocalin PGD synthases and microsomal PGE synthase-1 toward PGH 2 versus PGH 3
Human hematopoietic PGD synthase (H-PGDS; 0.04 unit (1 g/assay)) and lipocalin PGDS (L-PGDS; 0.04 unit (18 g/assay)) PGD synthases were from Cayman Chemical Co. Purified, solubilized His 6 mPGES-1 was expressed and purified as indicated under "Experimental Procedures" and 0.67-1 g used for the assays presented in the table. ͓1-14 C͔PGH 2 or ͓1-14 C͔PGH 3 was prepared by incubation of 18 M ͓1-14 C͔AA or ͓1-14 C͔EPA for 20 or 40 s. Hematopoietic or lipocalin PGDS or mPGES-1 was then added and the incubations performed for 40 s for PGDSs or 20 sec for mPGES-1 under conditions in which the rate of conversion to product (PGD or PGE) was approximately linear with time and added enzyme. ͓1-14 C͔PGH 2 and ͓1-14 C͔PGH 3 were generated in situ with purified muPGHS-2 or huPGHS-2 and H-PGDS, L-PGDS, or PGES was added to initiate the reactions in a final volume of 0.1 ml. Products were extracted, separated at 4°C by thin-layer chromatography, and quantified by scintillation counting. Values in parentheses are numbers normalized for the indicated starting PGH 2 concentration. 0.005), EP3 (p Ͻ 0.0014), and FP (p Ͻ 0.0001); ANOVA indicated no difference with the EP4 receptor (p Ͻ 0.054). In all cases the differences in potencies were less than the differences in binding affinities. However, it is important to note that, with the possible exception of the EP4 receptor, 3-series PGs were partial agonists. Quantitative data on receptor potencies are summarized in Table 4. TP␣ Receptor Specificity toward U46619 and ⌬17 U46619-Using membranes from HEK293 cells expressing the huTP␣ receptor, we determined the equilibrium dissociation constant for the TP␣ antagonist [ 3 H]SQ29548 to be 9 M. Equilibrium competition binding assays with the PGH 2 /TxA 2 and PGH 3 / TxA 3 analogues U46619 and ⌬17 U46619, respectively, were used to measure displacement of 5 M [ 3 H]SQ29548 (Fig. 5). The IC 50 values for U46619 and ⌬17 U46619 were identical (i.e. 200 and 210 nM, respectively).

PG
Comparison of U46619 and ⌬17 U46619 Activation of Platelet Aggregation-PRP from human donors was treated with U46619 or ⌬17 U46619 in amounts ranging from 0.1 to 2 M to measure the potency of each compound in platelet aggregation (Fig. 6). The threshold concentrations ranged from 0.5 to 0.8 M for U46619 and from 0.7 to 1 M for ⌬17 U46619 for platelets from four donors. For individual donors, the relative potencies of the compounds were similar, with the threshold concentration of ⌬17 U46619 consistently around 1.2-fold greater than that of U46619. To confirm the specificities of the analogues for the TP receptor, the platelets were incubated first with TP antagonist SQ29548 (1 M) before addition of the diene or triene analogue (2 M) or of collagen (2 g/ml). SQ29548 completely blocked aggregation by either U46619 or ⌬17 U46619 and inhibited collagen-induced aggregation by ϳ50% (data not shown).
Effects of TxA 2 Versus TxA 3 on Platelet Aggregation-Previous studies had suggested that TxA 3 was essentially inactive in platelet aggregation (69,70), while our results with ⌬17 U46619 suggested that TxA 3 would be pro-aggregatory. This assessment raised the possibility that the ⌬17 U46619 analogue behaves differently than authentic TxA 3 . Because TxAs have very short half lives, we developed a system for synthesizing TxA 2 or TxA 3 , which could then be added immediately to platelets. In brief, AA or EPA were treated with excess huCOX-2 to convert the fatty acids quantitatively to their respective endoperoxides. Excess microsomal TxAS was then added to quantitatively convert PGH 2 or PGH 3 to TxA 2 or TxA 3 , respectively; TxAS is reported not to discriminate between PGH 2 and PGH 3 (71). An aliquot of the reaction mixture was immediately added to PRP and platelet aggregation was monitored. The amounts of the various products formed from AA and EPA by huPGHS-2 and TxAS were monitored in parallel reactions using [1-14 C]AA or [1-14 C]EPA.
A representative experiment is shown in Fig. 7. Reaction 2, with a concentration of 78 nM TxA 2 , induced irreversible aggregation. However, Reaction 4, which contained 45 nM TxA 3 , caused only a small reversible aggregation. When added to  human PRP, PGH 3 was found to isomerize to PGD 3 with a t1 ⁄ 2 of 90 s, unlike PGH 2 which has a t1 ⁄ 2 of 350 s. PGD 3 is also reported to be somewhat more potent at inhibiting aggregation than PGD 2 (70), both of which cause an increase in cyclic AMP. To block the putative inhibitory effects of any PGD 3 , 1 M adenylate cyclase inhibitor SQ22536 was added to the PRP prior to the addition of Reaction 5. This unmasked an aggregatory effect equal to that of the TxA 2 reaction (Reaction 2). Both PGH 2 and PGH 3 are ligands of the TP receptor. However, neither the diene nor the triene endoperoxide produced in the PGHS-2-only reactions 1 and 3 (Fig. 7) was generated at sufficient concentrations to induce irreversible aggregation. Thus, residual endoperoxide in the reactions containing TxAS could not be responsible for the irreversible aggregation seen in Reactions 2 or 5.
Thromboxane and HHTrE are reported to be produced in equimolar amounts by TxAS (71). However, in our in vitro reaction, HHTrE and HHTE were produced at approximately twice the concentration of their respective thromboxanes. The effect of HHTrE on platelet aggregation was investigated by adding 1 M HHTrE to PRP alone or prior to the addition of either 2 M U46619 or 2 g/ml collagen. HHTrE neither induced nor inhibited platelet aggregation (results not shown). Likewise, up to 100 M malondialdehyde, another side product of the TxAS reaction, had no effect on platelet aggregation (results not shown). The results of studies with the TPa receptor and platelet aggregation suggest that TxA 2 or TxA 3 are approximately equipotent.
Comparison of PGI 2 and PGI 3 as Inhibitors of Platelet Aggregation-PGI 2 or PGI 3 (0.1-2 M) was added to PRP, immediately followed by 2 M U46619. Preliminary experiments with several donors were performed to optimize experimental conditions, including the stabilization of the prostacyclins and determination of the approximate threshold concentrations of each compound, before proceeding to perform dose response measurements with three donors (Fig. 8). The initial slope of each curve in Fig. 8 was measured and expressed as the percent inhibition of aggregation versus PGI concentration. The average IC 50 values were 0.92 Ϯ 0.28 nM and 1.30 Ϯ 0.18 nM for PGI 2 and PGI 3 , respectively. Thus, the potencies of PGI 2 and PGI 3 in inhibiting platelet aggregation are approximately the same confirming earlier results (69).

DISCUSSION
The goal of the studies reported here was to compare the specificities and potencies of PG biosynthetic enzymes and receptors toward AA-derived, 2-series versus EPA-derived, 3-series substrates and products. We reason that this new information will contribute to understanding whether any of the reported beneficial health effects of dietary 3 fish oil fatty acids are mediated through PG pathways. Our biochemical results along with those of others are summarized in Fig. 9 and Tables 2 and 4. Phospholipases-cPLA 2 ␣ is the key phospholipase involved in AA release in most PG forming cells (72). Previous studies have shown that cPLA 2 ␣ exhibits specificity toward AA and EPA esterified at the 2-position of phospholipids in comparison to other 2-position acids such as linoleate and oleate (73)(74)(75)(76). Moreover, although EPA-and AA-containing phospholipids are equally good substrates for cPLA 2 ␣, DHA-containing phospholipids are essentially inactive with cPLA 2 ␣ (76 -78). Certain sPLA 2 forms can also participate in PG biosynthesis (79,80). Because of the nature of the interaction with its substrates, sPLA 2 does not discriminate among 2-position acyl groups (81). In short, neither cPLA 2 nor sPLA 2 appears to differentiate between the acyl chains of AA versus EPA.
PGHS-1 and PGHS-2-PGHS-1 and PGHS-2 both exhibit specificity toward AA versus EPA. As reported previously by others and us, both enzymes have very similar K m and V max values with AA (57,59). PGHS-2 oxygenates EPA at about 30% of the rate of AA. Purified PGHS-1 is not active with EPA unless an exogenous hydroperoxide is added to the reaction mixture. This is a consequence of the higher hydroperoxide requirement of PGHS-1 versus PGHS-2 (57,82,83). PGHS-1 present in platelets cells does oxygenate exogenously supplied EPA albeit at a low rate in the presence of alkyl hydroperoxides (60). It is not clear whether the hydroperoxide concentration in cells is usually sufficient to support EPA oxygenation or whether there are differences in hydroperoxide concentrations among cell types (57,84).
The behavior of PGHS-1 with AA plus EPA is consistent with the kinetic properties of the enzyme determined with AA and EPA individually; thus, EPA is a reasonably good inhibitor of AA oxygenation by PGHS-1 as was originally reported by Lands and co-workers (61). A comparison of the crystal structures of EPA and AA with PGHS-1 suggests that EPA prefers to bind in a catalytically incompetent conformation in the PGHS-1 cyclooxygenase site and competes with AA for binding (85)(86)(87).
PGHS-1 mediated biological events include platelet aggregation and parturition (88,89) and certain types of acute inflammation (79,90,91). Cellular events involving PGHS-1 may be dampened when EPA/AA ratios in phospholipids are increased. At an EPA/AA ratio of 1.0, one would expect that there would be 50% less AA to be mobilized from phospholipids by cPLA 2 and that PGHS-1 would function at only 50% of maximal efficiency because of inhibition of AA oxygenation by EPA; however, this may well be an oversimplification because the concentrations of enzymes, receptors, and substrates in intact cells are unknown.
One of our most surprising observations was that PGHS-2 preferentially oxygenates AA when EPA and AA are tested together. The results observed when PGHS-2 is mixed with EPA plus AA cannot be explained based on the simple kinetic properties of PGHS-2 with AA or EPA individually. The biochemical basis for the selectivity of PGHS-2 for EPA versus AA when the substrates are together may relate to the half of sites activity of the enzyme (40). One possibility is that EPA binds one of the two cyclooxygenase sites of the PGHS-2 dimer and elicits an allosteric effect on the other cyclooxygenase site causing it to preferentially bind and oxygenate AA. If this is true and also applicable to any fatty acid, it could explain why PGHS-2 can preferentially oxygenate AA at low substrate concentrations when AA represents a small part of the available fatty acid pool in cells (92)(93)(94). A situation like this could occur in socalled late phase PG synthesis when an sPLA 2 is the operative phospholipase (95).
PGs are importantly involved in inflammation (79, 91, 96 -98), and in this context PGHS-2 is the most important PGHS isoform (99,100). Based on our biochemical studies, a decrease in the formation of 2-series PGs via PGHS-2 would be  expected to occur only to the extent that AA levels in phospholipids were decreased by 3 fatty acids supplanting AA. PGD, PGE, PGF, PGI, and TxA Synthases-There are seven different synthases reportedly involved in the conversion of PG endoperoxides to what are considered to be the biologically active PGs. Including the results from the present studies, there are now data on the specificities toward PGH 2 versus PGH 3 for H-PGDS, L-PGDS, mPGES-1, PGIS, and TXAS ( Fig. 9 and Table 2). Still lacking is information on cPGES (101) and mPGES-2 (101, 102) and various putative NADPH-dependent and GSH-dependent PGFSs (103, 104).
PGDSs and mPGES-1 are about one third as active with PGH 3 as PGH 2 . mPGES-1 does play a role in inflammation (67,96,97,105,106), and in principle, elevated levels of EPA could increase PGH 3 production and decrease PGH 2 formation, and the net effect would be decreased formation of PGE 2 with less than a corresponding increase in PGE 3 . However, PGHS-2 appears to be the most relevant enzyme in inflammation and, as discussed above, increases in EPA have a relatively modest effect on the formation of PGH 2 . PGE formation can occur via PGHS-1 and mPGES-1 (107). There are functions such as salt and water metabolism in the kidney that involve these two enzymes, and renal PGE 2 synthesis is diminished with no detectable production of PGE 3 in rats fed diets having elevated levels of fish oil (108).
PGIS and TXAS are reported to be similarly reactive with PGH 2 and PGH 3 (45,71). This suggests that any effects of changes in tissue EPA/AA levels on PGI and TxA formation would occur primarily at the level of PGHSs and not PGIS or TXAS.
Prostanoid Receptors-There are nine G-protein-linked PG receptors. Previous comparisons of receptor specificities for the 2versus 3-series PGs had been performed for the DP1 (70), DP2 (110,111), EP1 (54), TP (69), and IP (69,112) receptors. In all cases except for the TP receptor, there was little or no difference in the potencies of the 2versus 3-series PGs. We performed both binding measurements and measurements of receptor potencies for all of the human receptors except the IP and DP receptors. In the case of the IP receptor, we analyzed potencies of purified PGI 2 versus PGI 3 using human platelets.
The EP1, EP2, and EP3 receptors bound less well and were less responsive to PGE 3 than PGE 2 . As recently reviewed by Narumiya and coworkers (113), each of these receptor subtypes participate in a large number of functions each of which has the potential to be affected by increased tissue EPA/AA levels. However again, it should be noted that functions most likely to be affected are those that would be mediated via PGHS-1 and mPGES-1.
The FP receptor is known to be involved in parturition. Mice lacking cPLA 2 (114,115), PGHS-1 (88,89), or the FP receptor (116) have failures of parturition. Interestingly, this is also a characteristic of essential fatty acid deficiency that can be overcome with omega-6 but not omega-3 fatty acids (117)(118)(119). This could be accounted for by the low activity of PGHS-1 with EPA and the low potency of PGF 3␣ with the FP receptor. It is not clear what enzyme is responsible for PGF 2␣ formation in vivo, so we did not examine the PGF synthases that have been described (103) for their specificities toward PGH 2 versus PGH 3 .
An unexpected observation in our studies of PG receptors was that TxA 3 is almost as active as TxA 2 with the TP receptor. Earlier studies indicating that TxA 3 is inactive in platelet aggre-FIGURE 6. Comparison of potencies of U46619 and ⌬17 U46619 for platelet aggregation. Human PRP (2.25 ϫ 10 8 platelets per 0.4 ml) was treated with various concentrations of either U46619 or ⌬17 U46619 as indicated, and platelet aggregation as indicated by the change in light transmission was recorded on an aggregometer. Shown is a representative result of four different donors. The concentration at which irreversible aggregation occurred for ⌬17 U46619 was 1.2 times higher than for U46619 with platelets from each donor. FIGURE 7. Platelet aggregatory properties of AA-versus EPA-derived COX-2 and TXAS products. Enzyme reactions were initiated by adding 5 M AA (Reactions 1 and 2) or EPA (Reactions 3, 4, and 5) to a reaction mixture containing 750 units of purified His 6 -tagged huPGHS-2. A microsomal huTXAS preparation (540 mg of protein) (Reactions 2, 4, and 5) or microsomal buffer (Reactions 1 and 3) was added and the sample vortexed for an additional 10 s. An aliquot (30 l) of each reaction mixture was immediately added to PRP and platelet aggregation was measured. The adenylate cyclase inhibitor SQ22536 was added to the PRP 1 min prior to addition of the reaction mixture for Reaction 5. To calculate the concentrations of products from the COX-2/TXAS reactions that were added to PRP, [1-14 C]AA, or [1-14 C]EPA was used in place of the unlabeled substrate. The radiolabeled products were separated by TLC and the bands corresponding to fatty acid, HETE, (12-hydroxyheptadecatrienoic acid), HHT, PGH, and TxB were scraped and quantified by liquid scintillation counting and the final concentration of each product (nM) in the PRP was calculated. The concentration of TxA 2 or TxA 3 was corrected for degradation to TxB 2 or TxB 3 during the 10 s incubation as described under "Experimental Procedures." gation (69) were probably compromised by the formation of PGD 3 from PGH 3 and PGD 3 being a potent anti-aggregatory compound (70). In general, PGH 3 appears to be significantly less stable than PGH 2 in the aqueous systems used for our enzyme assays; PGH 3 is rapidly converted to HHTE and malon-dialdehyde whereas spontaneous conversion of PGH 2 to the homologous products is relatively slow. This is apparent in Fig. 3 where there is an accumulation of HHTE but not HHTrE.
To the extent that we have discussed our biochemical data in the context of the biological changes seen with dietary fish oil, we have assumed simple linear relationships based on K m , V max , and EC 50 values for the various enzymes and receptors. All of these values were obtained under optimal in vitro conditions. Obviously, what occurs in vivo cannot yet be predicted with any certainty because the ratios of enzymes and receptors to substrates and agonists involved in PG signaling may well be different in vivo (89). There may also be other eicosanoid mediators, including those derived from omega-3 fatty acids that are importantly involved in PG signaling (34,35). FIGURE 8. Comparison of anti-aggregatory potencies of PGI 2 and PGI 3 . Human PRP from three donors was treated with the indicated concentrations of either PGI 2 or PGI 3 for 10 s, followed by treatment with 2 M U46619 to induce aggregation. Aggregation was measured as an increase in light transmission using a platelet aggregometer. Upper panels are PGI 2 , and lower panels are PGI 3 . Agonist was introduced at points indicated by the arrows.