Identification of the 11,14,15- and 11,12,15-Trihydroxyeicosatrienoic Acids as Endothelium-derived Relaxing Factors of Rabbit Aorta*

A number of endothelium-derived relaxing factors have been identified including nitric oxide, prostacyclin, and the epoxyeicosatrienoic acids. Previous work showed that in rabbit aortic endothelial cells, arachidonic acid was metabolized by a lipoxygenase to vasodilatory eicosanoids. The identity was determined by the present study. Aortic homogenates were incubated in the presence of [U-14C]arachidonic acid, [U-14C]arachidonic acid plus 15-lipoxygenase (soybean lipoxidase), or [U-14C]15-hydroxyeicosatetraenoic acid (15-HPETE) and analyzed by reverse phase high pressure liquid chromatography (RP-HPLC). Under both experimental conditions, there was a radioactive metabolite that migrated at 17.5–18.5 min on RP-HPLC. When the metabolite was isolated from aortic homogenates, it relaxed precontracted aortas in a concentration-dependent manner. Gas chromatography/mass spectrometry (GC/MS) of the derivatized metabolite indicated the presence of two products; 11,12,15-trihydroxyeicosatrienoic acid (THETA) and 11,14,15-THETA. A variety of chemical modifications of the metabolite supported these structures and confirmed the presence of a carboxyl group, double bonds, and hydroxyl groups. With the combination of 15-lipoxygenase, arachidonic acid, and aortic homogenate, an additional major radioactive peak was observed. This fraction was analyzed by GC/MS. The mass spectrum was consistent with this peak, containing both the 11-hydroxy-14,15-epoxyeicosatrienoic acid (11-H-14,15-EETA) and 15-H-11,12-EETA. The hydroxyepoxyeicosatrienoic acid (HEETA) fraction also relaxed precontracted rabbit aorta. Microsomes derived from rabbit aortas also synthesized 11,12,15- and 11,14,15-THETAs from 15-HPETE, and pretreatment with the cyctochrome P450 inhibitor, miconazole, blocked the formation of these products. The present studies suggest that arachidonic acid is metabolized by 15-lipoxygenase to 15-HPETE, which undergoes an enzymatic rearrangement to 11-H-14,15-EETA and 15-H-11,12-EETA. Hydrolysis of the epoxy group results in the formation of 11,14,15- and 11,12,15-THETA, which relaxed rabbit aorta. Thus, the 15-series THETAs join prostacyclin, nitric oxide, and epoxyeicosatrienoic acids as new members of the family of endothelium-derived relaxing factors.

The vascular endothelium synthesizes and releases compounds that are involved in the regulation of vascular tone (1). These endothelial-derived vasoactive compounds include prostacyclin, endothelium-derived relaxing factor or nitric oxide, endothelium-derived hyperpolarizing factor, endothelium-derived contracting factor, and endothelin. These endothelial factors mediate the vasoactive effects of a number of hormones including acetylcholine, bradykinin, and ATP (1). Alterations in the production of these compounds may be associated with cardiovascular diseases, including atherosclerosis, coronary vasospasm, and hypertension.
Arachidonic acid is metabolized by the vascular endothelium to a variety of cyclooxygenase, lipoxygenase, and cytochrome P450 epoxygenase products (2). The identity and biological activity of some of these arachidonic acid metabolites have been determined; however, many metabolites have not been well characterized, either structurally or biologically. Singer and Peach (3) and DeMey et al. (4) first reported that arachidonic acid relaxed blood vessels and that arachidonic acid-induced relaxations were dependent on the presence of the endothelium (3)(4)(5)(6). Using inhibitors of arachidonic acid metabolism, Singer and Peach (6) concluded that a lipoxygenase metabolite mediated the effect. We have recently reported the ability of arachidonic acid to release a noncyclooxygenase metabolite from endothelial cells, which relaxes rabbit vascular smooth muscle (7). This relaxing factor is an endothelium-derived, lipoxygenase metabolite of arachidonic acid (8). In isolated aorta, the partially purified factor elicits relaxation of precontracted vessels.
Lipoxygenases are a family of enzymes that convert arachidonic acid into monohydroperoxyeicosatetraenoic acids (HPETEs) 1 (9). Lipoxygenases differ in their specificity for placing the hydroperoxy group in arachidonic acid. 12-and 15-lipoxygenase activity are present in the endothelium of the rabbit aorta (10). The HPETEs are metabolized by peroxidases to the monohydroxyeicosatetraenoic acids (HETEs). There is evidence that the HPETEs and to a lesser extent, the HETEs elicit vasorelaxation in certain vascular tissues (11)(12)(13). However, in rabbit aorta, 12-and 15-HETE are inactive (10). This finding suggests that the HPETE must be metabolized to some product other than a HETE to cause vasodilation. In various tissues and cells, both 12-and 15-HPETEs can undergo additional transformations to leukotrienes, lipoxins, long chain al-dehydes, trihydroxyeicosatrienoic acid, and hydroxyepoxyeicosatrienoic acids (14 -16). The hydroxyepoxyeicosatrienoic acids and trihydroxyeicosatrienoic acids of 12-lipoxygenase have been called hepoxilins and trioxilins, respectively (14). The hepoxilins have been shown to release insulin and enhance the vasoconstrictor effects of norepinephrine (14,29). There is no information on the biological effects of the hydroxyepoxy and trihydroxy products derived from 15-lipoxygenase. Our previous studies (7,8) and the studies of others (14 -16) support the hypothesis that arachidonic acid is metabolized to a potentially important lipoxygenase product(s) involved in the regulation of vascular tone. The purpose of the present study was to chemically identify the biologically active lipoxygenase metabolite(s) synthesized from arachidonic acid by rabbit aorta.

Materials
Norepinephrine, arachidonic acid, soybean lipoxidase (type I-B), and indomethacin were all from Sigma. [U-14 C]Arachidonic acid (specific activity 920 mCi/mmol) and [1-14 C]arachidonic acid (53 mCi/mmol) were obtained from NEN Life Science Products. All solvents were high pressure liquid chromatography (HPLC) grade and purchased from Burdick and Jackson, Muskegan, MI. 15-Lipoxygenase standards were obtained by incubating with [U-14 C]arachidonic acid (0.05 Ci, 10 Ϫ5 M) in buffer containing hematin (1 M) and soybean lipoxidase (0.2 mg/ml) for 10 min at room temperature or chemically as described by Falck et al. (17). 15-HPETE was synthesized by incubating arachidonic acid (216 mg) in 400 ml of borate buffer, pH 9.0, with soybean lipoxidase under a stream of oxygen for 30 min at room temperature. The reaction was acidified to pH 5 with 5% hydrochloric acid and extracted with diethyl ether. The combined extracts were washed with saturated NaCl, dried with MgSO 4 , filtered, and evaporated to dryness. The extract was purified by silica gel column chromatography using hexane/diethyl ether/acetic acid (2.5/10/0.25) at 4°C. The yield was approximately 50%. A similar approach was used to synthesize [U-14 C]15-HPETE.

Tissue Preparation and Incubation
Aortas (either freshly isolated from 1-2 month old New Zealand White rabbits or purchased from Pel-Frez Biologicals, Rogers, AR) were obtained and cleaned of adhering connective tissue and fat. The vessels were rinsed in Tris buffer (0.05 M, pH 7.5) and then cut into small pieces. Vessel segments were homogenized (5 passes with a mechanical homogenizer) in fresh buffer (500 mg of tissue/10 ml of buffer). The homogenate was centrifuged at 750 ϫ g for 15 min, and the supernatant was used. Aliquots (5 mg/ml) of the supernatant were incubated for 20 min in Tris-HCl buffer containing indomethacin (10 Ϫ5 M) and [U-14 C]arachidonic acid (0.05 Ci, 10 Ϫ7 M). In an additional study, aortic homogenates were incubated with [1-14 C]arachidonic acid (0.05 Ci, 10 Ϫ7 M) instead of uniformly labeled arachidonic acid. In some incubations, 15-lipoxygenase (soybean lipoxidase, 0.2 mg/ml) was included. Alternatively, the homogenate was incubated with [U-14 C]15-HPETE instead of [U-14 C]arachidonic acid. For analysis of metabolites by gas chromatography-mass spectrometry (GC/MS), aortic homogenates (50 mg/ml) were incubated with [U-14 C]arachidonic acid (0.05 Ci, 5 ϫ 10 Ϫ5 M) or arachidonic acid plus 15-lipoxygenase (0.2 mg/ml/incubation).
All reactions were stopped by adding ethanol to a final concentration of 15%. The samples were acidified (pH Ͻ 3.5) and extracted using octadecylsilyl extraction columns as described previously (8). The extracted metabolites were evaporated to dryness under a stream of nitrogen and stored at Ϫ40°C until analysis by HPLC.

Purification of Unknown Factor by HPLC
The biological samples were first resolved by reverse phase (Nucleosil-C18 column, 5 m, 4.6 ϫ 250 mm) HPLC using solvent system I. Solvent A was water, and solvent B was acetonitrile containing 0.1% glacial acetic acid. The program was a 40-min linear gradient from 50% solvent B in A to 100% solvent B. Flow rate was 1 ml/min. The fractions corresponding to the unknown factor (fractions [27][28][29][30][31][32][33][34][35] were pooled, acidified, extracted with cyclohexane/ethyl acetate (50/50), and rechromatographed on reverse phase HPLC using solvent system II. In solvent system II, solvent A was water containing 0.1% glacial acetic acid, and solvent B was acetonitrile. The program consisted of a 5-min isocratic phase with 35% B in A, followed by a 35-min linear gradient to 85% B. The flow rate was 1 ml/min. Radioactivity of the column eluate was collected in 0.2-ml aliquots and measured by liquid scintillation spectrometry. In some cases, fractions 15-50 from system I were analyzed by reverse phase HPLC using solvent system III. The solvents were identical to those used for solvent system II, but the program consisted of a 40-min isocratic phase with 31% B in A followed by a 20-min linear gradient to 100% B. The flow rate was 1 ml/min.

Chemical Modification
To aid in the identification of the unknown factor isolated from rabbit aorta, a number of chemical modifications were performed on column fractions containing the factor. After incubation of aortic homogenate as described above, the incubation buffer was extracted and purified by reverse phase HPLC using solvent system I. The fractions (27)(28)(29)(30)(31)(32)(33)(34)(35) corresponding to the unknown factor were collected and extracted into cyclohexane/ethyl acetate. The unknown factor was divided into two parts with one-half being treated with various reagents as described below and the other half retained for comparison. The resulting products were resolved by reverse phase HPLC using solvent system III as described above. The column eluate was collected in 0.5-ml aliquots, and the radioactivity was measured by liquid scintillation spectrometry. A change in migration time suggests a modification of the molecule. In each case, eicosanoids with known functional groups were treated with the same reagents and analyzed in a similar manner by reverse phase HPLC.
The chemical modification procedures of the unknown factor included: 1. Pentafluorobenzyl Ester (PFB)-The sample was dissolved in 100 l of acetonitrile, 5 l of dimethylformamide, 5 l of diisopropylethylamine, and 3 l of PFB bromide. After a 40-min incubation at 40°C, the sample was partitioned between 6 ml of hexane and 4 ml of water. The organic layer was separated, and the extraction procedure was repeated. The organic fractions were pooled, evaporated to dryness under a nitrogen stream, and analyzed by HPLC.
2. Methyl Ester-The sample was dissolved in 500 l of methanol. After treatment with 1 ml of ethereal-diazomethane (CH 2 N 2 ), the sample was incubated for 30 min at 0°C. The sample was evaporated to dryness under a nitrogen stream and analyzed by HPLC.
3. Perchloric acid (HClO 4 )-The sample was dissolved in 600 l of tetrahydrofuran, 300 l of water, and 10 l of 70% perchloric acid. After an overnight incubation at 4°C, the sample was extracted 3 times with cyclohexane:ethyl acetate (50:50). The organic fractions were pooled, evaporated to dryness under a nitrogen stream, and analyzed by HPLC.
4. Hydrogen/Platinum Oxide (H 2 /PtO 2 )-The sample was dissolved in 600 l of methanol previously saturated with hydrogen. A small amount of platinum oxide was added to the solution. After a 6-min incubation, the sample was filtered through 500 mg of silica gel in a pasteur pipette. The sample was collected, evaporated to dryness under a nitrogen stream, and analyzed by HPLC.
5. Sodium Borohydride (NaBH 4 )-The sample was dissolved in 500 l of methanol. Sodium borohydride (20 mg) was added, and the sample was incubated for 40 min at 0°C. After the incubation, the sample was acidified and extracted with cyclohexane:ethyl acetate. The organic fractions were pooled, evaporated to dryness under a nitrogen stream, and analyzed by HPLC.
6. Sodium Hydroxide (NaOH)-The sample was dissolved in 500 l of methanol and 500 l of 0.1 M NaOH. Following a 4-h incubation at room temperature, the sample was acidified and extracted with cyclohexane: ethyl acetate. The organic fractions were pooled, evaporated to dryness under a nitrogen stream, and analyzed by HPLC.
7. t-Butyldimethylsilyl Ether/Methyl Ester-The methyl ester was made as described above. Then, 1 ml of t-butyldimethylsilylimidazoledimethylformamide solution (Supelco) was added, and the sample was incubated for 30 min at room temperature. After the incubation, 5 ml of cyclohexane:ethyl acetate and 5 ml of water were added. The organic fraction was removed, and extraction was repeated. The organic fractions were pooled, evaporated to dryness under a nitrogen stream, and analyzed by HPLC.
8. Ozonolysis-The sample was dissolved in 3 ml of methanol and cooled to Ϫ78°C. After 30 min of bubbling with ozone, the reaction was quenched with sodium borohydride. The sample was acidified and extracted with cyclohexane:ethyl acetate. The organic fractions were pooled, evaporated to dryness under a nitrogen stream, and analyzed by HPLC.

Gas Chromatography/Mass Spectrometry
The pooled fractions were isolated from solvent system II (fractions 88 -94), which corresponded to the unknown factor previously described (8). The pooled fractions were extracted with cyclohexane/ethyl acetate,

15-Lipoxygenase Arachidonic Acid Metabolites as Relaxing Factors
evaporated to dryness under nitrogen, and derivatized for GC-MS as described previously (10,18). For electron impact (EI) or positive ion chemical ionization (PCI) MS analysis, the sample was dissolved in 120 l of acetonitrile and then esterified with ethereal diazomethane for 6 min at room temperature. The reaction mixture was evaporated to dryness under nitrogen, and the hydroxyl groups were then silylated with 15 l of bis-TMS-trifluoroacetamide for 60 min at 37°C. For negative ion chemical ionization MS, the sample was dissolved in 80 l of acetonitrile, 20 l of diisopropylethylamine, and 10 l of PFB and allowed to react for 15 min at room temperature. The sample was extracted twice with water/ethyl acetate (1:1), and the organic fractions were pooled and evaporated to dryness under nitrogen. The PFB ester was then treated with bis-TMS-trifluoroacetamide, as described above. GC-MS was performed with a Hewlett Packard 5989A mass spectrometer coupled with a 5890 series 2 gas chromatograph. Ionization of the samples was done by electron impact at 65-70 eV or collisionally using methane as the reagent gas. The derivatized factor was resolved using a 14-m capillary DB-5 column (J & W Scientific Inc, Folsom, CA) with a linear gradient from 100 to 300°C. Standards were derivatized and analyzed by GC-MS using the identical methods described for the biological samples. In an additional experiment, the 11.5-14-min fraction isolated from solvent system I was pooled, extracted, derivatized, and analyzed by PCI MS.

Vascular Reactivity
Rabbit aortic homogenates were incubated with [U-14 C]arachidonic acid or 15-lipoxygenase and [U-14 C]arachidonic acid as described above. Identical control incubations without tissue (referred to as cell-free) were carried out in parallel. Incubations were also performed with [U-14 C]arachidonic acid, 15-lipoxygenase, and hematin. After incubation and extraction, the samples were chromatographed on reverse phase HPLC using solvent system I. The fractions 27-35, corresponding to the unknown factor, were collected, extracted, and rechromatographed on reverse phase HPLC using solvent system II. This system further purified the unknown factor (fractions 88 -94), which was then tested for biological activity. For these biological activity experiments, thoracic aortas were obtained from 1-2-month-old New Zealand White rabbits and placed in Krebs bicarbonate buffer (118 mM NaCl, 4 mM KCl, 3.3 mM CaCl 2 , 24 mM NaHCO 3 , 1.2 mM KH 2 PO 4 , 1.2 mM MgSO 4 , and 11 mM glucose) as described previously (10). The tissue was carefully cleaned of adhering fat and connective tissue and cut into rings (3-mm thick) taking care not to damage the endothelium. Aortic rings were suspended in 6-ml tissue baths containing Krebs bicarbonate buffer maintained at 37°C and continuously bubbled with 95% O 2 , 5% CO 2 . Isometric tension was measured with force-displacement transducers (Grass) and recorded with a Grass polygraph model 7D. Resting tension was adjusted to its length tension maximum of 2 g, and vessels were allowed to equilibrate for 1 h. Contractions were produced by increasing the KCl concentration of the baths to 40 mM. After the vessels reached peak contraction, tissue baths were rinsed, and vessels were allowed to return to resting tension. Once the aortic rings had reproducible, stable responses to KCl, the tissue was contracted with norepinephrine (10 Ϫ7 M). The effect of the unknown factor derived under the various incubation conditions was tested. The samples were suspended in a known volume of ethanol, and 10 l/6-ml bath was the maximal ethanol concentration administered. Aortic incubations and cell free fractions were always analyzed in parallel and compared with its own control response. Fractions 57-70 obtained from aortic homogenates incubated with [U-14 C]arachidonic acid and 15-lipoxygenase and isolated from reverse phase HPLC using solvent system I were tested for biological activity in a similar manner.

Preparation of Microsomes
Rabbit aortas were isolated and homogenized in cold 0.25 M sucrose solution using a Potter-Elvehjem tissue grinder with a Teflon pestle. Subcellular fractions were prepared by differential centrifugation (19). The homogenate was centrifuged at 5000 ϫ g for 20 min to remove cellular debris, mitochondria, and nuclei. The supernatant was further centrifuged at 100, 000 ϫ g for 60 min. The microsomal pellet was resuspended in 1.15% KCl and homogenized. The homogenate was centrifuged at 100,000 ϫ g for 60 min. The pelleted microsomes were resuspended in microsome incubation buffer (50 mM Tris, 150 mM KCl, 10 mM MgCl 2 , pH 7.5) and stored at Ϫ80°C. Protein concentrations were determined using the Bradford technique (Bio-Rad) with IgG as the standard. Aliquots of microsomal protein (2 mg/ml) were incubated with [U-14 C]arachidonic acid and 15-lipoxygenase as described above. Alternatively, incubations were performed with [U-14 C]15-HPETE for 10 min at 37°C. In a separate experiment, microsomes were pretreated with the cytochrome P450 inhibitor, miconazole (10 Ϫ5 M), before the addition of [U-14 C]15-HPETE. Metabolites were extracted and chromatographed as described above using solvent system I then II.

RESULTS
Formation of Metabolites-When aortic homogenate was incubated with [U-14 C]arachidonic acid and extracted, and metabolites were resolved by reverse-phase HPLC using solvent system I, there was the synthesis of radioactive products that migrated with the prostaglandins, the dihydroxyeicosatrienoicdihydroxyeicosatetranoic acids, and the HETEs (Fig. 1A). Fractions 27-35 (5-7.5 min) were further analyzed by reversephase HPLC using solvent system II (Fig. 1B). A number of radioactive peaks were observed and were labeled peaks 1-4. The major peak migrated at 17.5-18.5 min and was peak 2. Peak 2 represented 1.4 Ϯ 0.2% of the added arachidonic acid (n ϭ 7). Our previous results indicated that only peak 2 elicited relaxation of precontracted rabbit aorta (8). Because the previous results also suggested that peak 2 was derived from lipoxy- genase-mediated metabolism of arachidonic acid, the next experiments investigated the effect of the addition of 15lipoxygenase to the aortic homogenate incubations. 14 C-Arachidonic acid was incubated with 15-lipoxygenase plus the aortic homogenate or 15-lipoxygenase alone. The only product synthesized by 15-lipoxygenase alone comigrated with 15-HETE on HPLC using solvent system I (data not shown). With the combination of 15-lipoxygenase and the homogenate, three major radioactive peaks were observed with solvent system I ( Fig. 2A). One peak eluted with the unknown (fractions 27-35, 5-7.5 min), the second less polar peak eluted at 11.5-14 min (fractions 57.5-70), and the third major radioactive peak eluted with 15-HETE (fractions 100 -120, 20 -22 min). Analysis of the 5-7.5-min fraction using solvent system II indicated a pattern identical to the pattern observed with arachidonic acid and the homogenate, i.e. peak 2 was the major product (Fig. 2B). The 11.5-14-min fraction comigrates with the hydroxyepoxyeicosatrienoic acids (HEETAs) according to previous reports (15,20). With equal amounts of tissue incubated, the synthesis of peak 2 was much greater when 15-lipoxygenase was added (Figs. 1 and 2).
Additionally, aortic microsomes were incubated with [U-14 C]arachidonic acid and 15-lipoxygenase, and extracted metabolites were analyzed by reverse phase HPLC using solvent system I followed by solvent system II. The major radioactive peak also migrated at 17.5-18.5 min with peak 2 (Fig.  3A). When aortic microsomes were incubated with [U-14 C]15-HPETE, the major metabolite again comigrated with peak 2 (17.5-18.5 min) on reverse phase HPLC using solvent system II min, and metabolites were separated using reverse phase HPLC with use of solvent system I. B, metabolites in fractions 27-35 were collected, extracted using cyclohexane/ethyl acetate, and rechromatographed using solvent system II. The migration times of known standard eicosanoids are shown above the chromatogram and are not meant to identify the corresponding radioactive peaks. DHET, dihydroxyeicosatrienoic; AA, arachidonic acid; PGF 1␣ , prostaglandin F 1␣ . (Fig. 3B). Pretreatment with miconazole (10 Ϫ5 M) blocked the formation of peak 2 from [U-14 C]-15-HPETE (Fig. 3C). These data suggest that the metabolite in peak 2 is derived from 15-HPETE and that the aortic homogenate and aortic micro-somes transform 15-HPETE to the unknown metabolite.
Chemical Modification of Metabolites-The next series of experiments were designed to obtain structural information about the unknown factor. Table I summarizes the results of a number of chemical modifications performed on the unknown fraction isolated from reverse phase HPLC using solvent system I. The unknown peak (fraction 27-35 from solvent system I) was divided in two parts, with one-half being treated with various reagents and the other half retained for comparison. After treatment, the samples were extracted and analyzed by reverse phase-HPLC using solvent system III. The unmodified unknown eluted at 52 min, and a change in migration time from 52 min suggests a chemical modification in the molecule. As a positive control, eicosanoids with known functional groups were treated with the same reagents and analyzed in a similar manner by reverse phase-HPLC. If aortic homogenates were incubated with [1-14 C]arachidonic acid, the production of a radioactive product comigrating with the unknown factor derived from incubations with uniformly labeled arachidonic acid was seen, indicating that the metabolite retains the C-1 carboxyl group. Treatment of the unknown factor with NaBH 4 did not change the retention time of the peak as analyzed by reverse phase HPLC using solvent system III. This result demonstrated that a keto group was not present in the compound. When the unknown factor was treated with NaOH, there was no change in polarity of the metabolite, indicating that there FIG. 4. Mass spectrum of peak 2 (17.5-18.5 min) collected from reverse phase HPLC with use of solvent system II. It was derivatized to the methyl ester-TMS ether and analyzed by PCI GC/ MS. Panel A is a representative reconstructed ion chromatogram from the gas chromatographic analysis. Panels B and C show the PCI mass spectra for compound A and B, respectively. The proposed structure of each derivatized compound is given, and its prominent ions are indicated.

TABLE I Effect of chemical modifications on HPLC retention time of the unknown factor
Rabbit aortas were incubated with [U-14 C]arachidonic acid, and the incubation buffer was extracted and chromatographed by reverse phase HPLC using solvent system I. The fractions corresponding to the unknown factor were collected (15-50) and extracted into cyclohexane/ ethyl acetate. The unknown factor was treated with various reagents as described under "Experimental Procedures." The resulting products were resolved by reverse phase HPLC using solvent system III. A change in migration time suggests a modification of the molecule. TBDMS/Cl, t-butyldimethylsilylimidazole chloride. were no alkali labile groups present. Alternatively, treatment with HClO 4 resulted in a number of products with decreased polarity, as shown by an increased retention time of radioactive products on reverse phase HPLC. This treatment suggested the presence of acid labile groups. Treatment of the unknown factor with hydrogen resulted in an increased retention time, indicating the presence of double bonds in the metabolite. Treatment with ozone to fragment the molecule at double bonds resulted in the formation of a series of polar fragments that all eluted at 10 min. This also indicated the presence of several double bonds. When the compound was treated with PFB bromide or diazomethane to form the PFB or methyl ester, there was an increase in retention time of the factor. This further indicated that a carboxylic acid was present. Additional treatment of the methyl ester with t-butyldimethylsilylimidazole increased the retention time of the unknown factor to an even greater extent than the methyl ester treatment alone. These results suggested the presence of a hydroxyl group or groups in the metabolite. Based on the large increase in retention time of the methyl ester-t-butyldimethylsilylimidazole derivative, it seemed likely that more than one hydroxyl group was present. Gas Chromatography-Mass Spectrometry Analysis-To identify the metabolite associated with peak 2, further characterization of peak 2 from the aortic homogenate incubations with  6% for compound B). The spectra indicated a molecular weight of 584 for the methyl ester-TMS ether of compound A and B. Based on these mass spectra, peak 2, isolated from reverse phase-HPLC using solvent system II, contains two metabolites identified as the methyl-ester TMS ether derivatives of 11,12,15-trihydroxyeicosatrienoic acid (11,12,15-THETA) and 11,14,15-trihydroxyeicosatrienoic acid (11,14,15-THETA). The proposed structures for each of the derivatized forms of peak 2 are shown as an inset of Fig. 4, B and C. Cleavage resulting in the prominent ions is also indicated. Finally, the EI mass spectrum of the methyl ester-TMS ether derivative is also consistent with peak 2 being a mixture of 11,12,15-THETA and 11,14,15-THETA (Fig. 5). Analysis of peak 2 by EI GC/MS showed two major GC peaks (Fig. 5A)

15-Lipoxygenase Arachidonic Acid Metabolites as Relaxing Factors
of CH 3 ) (Fig. 5C). Based on the relative intensity of mass fragment ions 173 and 283, compound A is identified as 11,12,15-THETA, and compound B is identified as 11,14,15-THETA. It should be noted that in the EI mass spectra of compound A, the base fragment ion is 161, which represents a further fragmentation of the 283 ion (283 and (CH 3 ) 3 SiOH and OCH 3 ). Similar mass spectra have been published for these compounds by others (15,20). We have chemically synthesized 11,14,15-THETA as described by Falck et al. (17) and by incubating 15-lipoxygenase and hematin with arachidonic acid (15). The derivatized 11,14,15-THETA standard comigrates with the derivatized unknown on GC (13.87 min) and gives a similar mass spectrum (data not shown). This structure is also consistent with chemical modification studies and synthesis of the unknown from 15-HPETE.
Biological Activity-Peak 2 (fractions 88 -94) was isolated from aortic homogenates incubated with arachidonic acid and tested for activity on isolated rings of rabbit aorta. In precontracted vessels, peak 2 derived from the arachidonic acid and aortic homogenates elicited a concentration-dependent relaxation response (Fig. 7A). The cell-free incubation with arachidonic acid did not relax the rabbit aorta (data not shown). In an additional experiment, the 11.5-14-min fraction isolated from reverse phase HPLC using solvent system I and that corre-sponds to the HEETAs was tested for activity. This fraction relaxed the precontracted rabbit aorta by 51 Ϯ 5% (Fig. 7B). DISCUSSION Many studies have documented the importance of the vascular endothelium in the regulation of vascular tone. A number of endothelium-derived relaxing factors have been identified including nitric oxide (1), prostacyclin (21), and the epoxyeicosatrienoicacids (22).Vesselsfromtherabbitexhibitendotheliumdependent relaxations to acetylcholine (23)(24)(25). In the abdominal aorta, mesenteric artery, carotid artery, and femoral artery, these relaxations are only partially inhibited by L-nitroarginine and indomethacin, inhibitors of the synthesis of nitric oxide and prostacyclin. In addition, the rabbit aorta does not synthesize epoxyeicosatrienoic acids (10). These data indicate that some factor other than nitric oxide, prostacyclin, and epoxyeicosatrienoic acids must mediate a portion of the relaxations to acetylcholine. Arachidonic acid also causes relaxation of blood vessels, which is dependent on the presence of the endothelium (3,4,23,24,26). For example, Vanhoutte and co-workers (4) reported that arachidonic acid relaxed canine femoral arteries and provided evidence that this response was mediated by the endothelial production of the cyclooxygenase metabolite, prostacyclin. In canine coronary artery, arachidonic acid-induced relaxations were mediated by an noncyclooxygenase metabolite and were blocked by ouabain (27). Singer and Peach (3,6) showed that arachidonic acid mediated relaxations of rabbit aorta via a mechanism that was endothe- lium-dependent and provided evidence that the relaxations involved a noncyclooxygenase metabolite of arachidonic acid. Our previous work supported this observation and described an endothelium-dependent, lipoxygenase metabolite of arachidonic acid that elicited relaxation of rabbit aorta (8). Arachidonic acid-induced relaxations were enhanced by treatment with the cyclooxygenase inhibitor, indomethacin, and inhibited by lipoxygenase inhibitors, including nordihydroguaiaretic acid and cinnamyl-3,4-dihydroxy-␣-cyanocinnamate. The cytochrome P450 epoxygenase inhibitor metyrapone had no effect on arachidonic acid-induced relaxations.
Our previous work indicated that incubation of rabbit aorta with [ 14 C]arachidonic acid resulted in the synthesis of a previously unidentified 14 C-labeled metabolite and was called the unknown factor. The production of the unknown factor was enhanced by indomethacin and decreased by lipoxygenase inhibitors. Both the production of the unknown factor and arachidonic acid-induced relaxations were dependent on an intact endothelium, indicating that the cellular source of the unknown relaxant factor was the endothelial cell. Thus, the present paper has further characterized this metabolite and found the presence of two different THETAs, 11,12,15-and 11,14, 15-THETA. The 15-lipoxygenase product, 15-HPETE, has been shown to be converted to a series of hydroxyepoxy and trihydroxy products by rat liver microsomes (20) and rabbit peritoneal polymorphonuclear leukocytes (28). Until the present report, there was no evidence that these metabolites were made in vascular tissue or endothelial cells.
The structures of these metabolites is based on a number of observations. First, we performed a number of chemical manipulations on the unknown metabolite to gain some insight regarding the functional groups present in the molecule. Results indicated that the unknown contained the C-1 carbon of arachidonic acid, a carboxyl group(s), a hydroxyl group(s), and double bonds. Incubation of vessels in the presence of a 15(S)-HPETE generating system or [ 14 C]15(S)-HPETE indicated that the unknown metabolite was produced. In the presence of 15lipoxygenase, the synthesis of the unknown was greatly increased. The incubation of 15-lipoxygenase alone with arachidonic acid did not produce the unknown, indicating that an additional component or enzyme was in the homogenate. Similarly, incubation of [ 14 C]15-HPETE with aortic homogenates resulted in the formation of the unknown metabolite. Similar results were obtained when aortic microsomes were substituted for the aortic homogenate. Thus, the additional component or enzyme that converted 15-HPETE to the unknown metabolite was microsomal. These data suggests that the unknown represents a metabolite of 15(S)-HPETE and suggests the presence of a 15-hydroxyl group in the molecule. Results also showed that when arachidonic acid was incubated with aortic homogenate plus 15-lipoxygenase, a less polar metabolite was observed that eluted at 11.5-14 min on reverse phase HPLC using solvent system I. This fraction comigrated with the HEETAs, and GC-MS analysis of the methyl ester-TMS ether derivative determined its structure to be 11-H-14,15-EETA and 15-H-11,12-EETA.
There is no information on the biological importance of the hydroxyepoxy and trihydroxy products derived from 15-lipoxygenase. However, the present results indicated that peak 2, which contains both 11,12,15-and 11,14,15-THETA isolated from rabbit aortic incubations or synthesized from arachidonic acid, 15-lipoxygenase, and hematin, relaxed rabbit aorta. The corresponding HEETA isolated from rabbit aorta incubations also relaxed precontracted rabbit aorta. It is not known if the HEETAs are vasodilators or whether HEETAs are metabolized by the aorta to 11,12,15-and/or 11,14,15-THETA, which then mediate the relaxation response.
Dho et al. (30) reported that the rat lung converts the 12lipoxygenase product, 12-HPETE, to both the 8-H-11,12-EETA and 10-H-11,12-EETA (hepoxilin A 3 and B 3 , respectively). These compounds are converted by epoxide hydrolase into the corresponding 8,11,12-THETA and 10,11,12-THETA (trioxilin A 3 and B 3 , respectively). It is now known that the hepoxilins are formed in a number of different tissues, including platelets and rat aorta and possess a variety of different biological actions (29 -31). We found no indication of the synthesis of these compounds by rabbit aorta. The effect of the hepoxilins on vascular tone has been studied. Hepoxilin A 3 potentiates the contractile response to norepinephrine (29). This effect is the opposite of the observed effect of the unknown factor isolated from rabbit aorta and characterized by the present study. However, if the biosynthetic pathway for HEETAs and THETAs in rabbit aorta is similar to the synthesis of hepoxilins and trioxilins (14), our results indicate that 15-HPETE is converted to 15-H-11,12-EETA and 11-H-14,15-EETA by a hydroperoxide isomerase (Fig. 8). Little is known about this enzyme. Several heme-containing enzymes, cytochrome P450, hemoglobin, and hematin can function as hydroperoxide isomerases and catalyze this rearrangement (15,20). In analogous manner, prostacyclin synthase is a cytochrome P450 enzyme that catalyzes the rearrangement of prostaglandin H 2 to prostacyclin (32). These rearrangements do not require cofactors such as O 2 or NADPH or reductases. The endothelial cell has several enzymes that contain heme and may serve as a hydroperoxide isomerase. These include cytochrome P450, nitric oxide synthase, prostacyclin synthase, and cyclooxygenase. Because the synthesis of 11,12,15-and 11,14,15-THETA were increased rather than decreased by cyclooxygenase inhibitors (8), it is unlikely that cyclooxygenase is involved. Weiss et al. (20) showed that rat liver microsomal cytochrome P450 catalyzed the transformation of 15-HPETE into hydroxyepoxy-and trihydroxyeicosatrienoic acids. We have confirmed these findings. The identity of the trihydroxy compounds corresponded to 11,12,15-and 11,14,15-THETAs. It is not known if cytochrome P450 catalyzes the reaction in the rabbit aorta; however, the experiments that demonstrated that microsomes derived from rabbit aortas synthesized the THETAs from 15-HPETE are supportive for this mechanism. In addition, the cytochrome P450 inhibitor, miconazole, attenuated the production of the [ 14 C]THETAs in aortic microsomes incubated with [ 14 C]15-HPETE. We have previously shown that in rabbits fed a 2% cholesterol diet for 2 weeks, there is an enhanced synthesis of the cytochrome P450 epoxygenase metabolites of arachidonic acid in aortas with an intact endothelium (10). Removal of the endothelium suppressed the formation of cytochrome P450 metabolites, suggesting that the enzyme is found in the endothelial but not vascular smooth muscle cells. We and others (10,33,34) have confirmed the presence of the cytochrome P450 epoxygenase in endothelial cells. Because cholesterol feeding increases the production of the unknown metabolite and the EETs (8,10), these results support the role of cytochrome P450 in the synthesis of the THETAs. The mechanism whereby the THETAs cause relaxation is not known at the present time.
In summary, we have identified the factor(s) that mediates arachidonic acid-induced relaxations of rabbit aorta as metabolite(s) of 15-lipoxygenase, 11,12,15-and 11,14,15-THETA using bioassay, HPLC, chemical modification, and GC/MS. Although the physiological importance of these THETAs is not clearly known, our previous work showed that the calcium ionophore, A23187, stimulated their release from rabbit aorta (8). Additionally, the activity and synthesis of the THETAs were enhanced in vessels obtained from cholesterol-fed rabbits (8). It is of interest that 15-lipoxygenase activity is increased in aortas obtained from hypercholesterolemic rabbits (35). The present studies suggest that arachidonic acid is metabolized by 15-lipoxygenase to 15-HPETE, which undergoes an enzymatic rearrangement to an HEETA(s), either 11-H-14,15-EETA, 15-H-11,12-EETA, or both. Hydrolysis of the epoxy group results in the formation of 11,12,15-and 11,14,15-THETA. The HEE-TAs and THETAs relax the rabbit aorta, and one or both may represent the active metabolite of arachidonic acid. At the present time, it is not known which isomer is the active metabolite. Thus, 15-series THETAs represents new members of the family of endothelium-derived relaxing factors, which also includes prostacyclin, nitric oxide, and epoxyeicosatrienoic acids.