Biosynthesis and Metabolism of 15-Hydroperoxy-5,8,11,13-eicosatetraenoic Acid by Human Umbilical Vein Endothelial Cells*

Incubation of cultured human umbilical vein endo- thelial cells with [1-“C]arachidonic acid, followed by reverse-phase high-pressure liquid chromatography analysis, results in the appearance of two principal radioactive products besides 6-keto-prostaglandin Fla. The first peak is 12-~-hydroxy-5,8,10-heptadecatri-enoic acid, a hydrolysis product of the prostaglandin endoperoxide. The second peak was esterified, converted to the trimethylsilyl ether derivative, and ana- lyzed by gas chromatography-mass spectrometry and shown to be the lipoxygenase product 15(S)-hydroxy-5,8,11,13-eicosatetraenoic acid (15-HETE). Incuba- tion of the 15-HETE precursor 15(S)-hydroperoxy-5,8,11,13-eicosatetraenoic acid (15-HPETE) with en- dothelial cells results in the formation of four distinct UV absorbing peaks. UV and gas chromatography- mass spectrometry analysis showed these peaks to be 8,15(S)-dihydroxy-5,8,11,13-eicosatetraenoic acids (8,15-diHETE) differing only in their hydroxyl config-uration and cis trans double-bond geometry. Formation of 8,15-diHETE

to remove any blood. The vein was then perfused with 10 ml of 0.1% collagenase in cord buffer and incubated at 37 "C in a bath of cord buffer for 10 min. The collagenase/cell mixture was flushed from the vein with 30 ml of buffer into a plastic centrifuge tube that contained 10 ml of medium 199 and then centrifuged at 200 x g for 5 min. The cell pellet was resuspended in 10 ml of fresh medium 199 and added to a T-75 flask. Cells were exposed to an atmosphere of 95% air, 5% CO, and were fed twice a week until subculturing. Confluency was usually reached in 4 to 5 days.
Purification of Arachidonic Acid and 15-HPETE Immediately prior to their use, 15-HPETE and arachidonic acid were repurified by HPLC to eliminate minor contaminants. Separations were achieved using a Varian 5000 liquid chromatograph equipped with a Varian UV-100 detector (Varian Associates, Inc., Walnut Creek, CA). 15-HPETE was chromatographed on a Partisil 10/25 PAC column (Whatman Inc., Clifton, NJ) using an isocratic mobile phase of hexane/2-propanol/glacial acetic acid (964:O.Ol) at a flow rate of 2 ml/min and UV detection at 235 nM. Under these conditions 15-HPETE had an uncorrected retention time of 19.1 min while 15-HETE eluted at 18.2 min. Collected fractions were taken to dryness under a stream of Nz and redissolved in methanol. The concentration of the purified material was determined on a Lambda 5 UV-Vis spectrophotometer (Perkin-Elmer Corp., Norwalk, CT) using an c235 of 29,500 (19). (1-"Clarachidonic acid (4.6 mCi/mmol) was purified on a 10 X 250 mm 5-pm Ultrasphere-ODS column (Altex Scientific Inc., a subsidiary of Beckman Instruments, Inc., Berkeley, CA) and eluted isocratically with methanol/water/glacial acetic acid (96.5:3.5:0.01) at a flow rate of 3.5 ml/min. Detection was at 210 nm.
Methanol was removed from collected fractions under a stream of Nz, and [l-'4C]arachidonic acid was extracted from the remaining acidified water with hexane. The purified sample was quantitated by liquid scintillation counting.

Incubation of Endothelial Cells with Arachidonic Acid or 15-HPETE
Prior to the incubation of cells with arachidonate or 15-HPETE, the confluent monolayers were harvested from the culture flasks by a 5-min incubation with a 1:l mixture of 0.2% collagenase, 0.5% BSA, 0.02% EDTA. After washing, the cells were resuspended in cord buffer to a concentration of 1-3 X 10' cells/ml and allowed to equilibrate at 37 "C for 10 min. In experiments utilizing aspirin, the cells were incubated an additional 30 min in the presence of 1 mM aspirin. The aspirin was washed from the cells before addition of substrates. When NDGA was used, it was added at a final concentration of 30 p~ 10 min before the addition of arachidonate. Heatinactivated cells were boiled for 10 min.
Reactions were initiated by transferring 10 ml of cells to a silanized vessel containing the desired amount of 15-HPETE or arachidonic acid. Normally, 100 pg of 15-HPETE or 150 p g of arachidonate was used per 10-ml incubation (final concentrations equaling 30 and 50 FM, respectively). Incubations were carried out at 37 "C with stirring.
The reactions were terminated by the addition of two volumes of cold acetone. Internal standard (0.5 fig of LTB,) was added to some samples and the reaction mixture was left at 4 "C for 30 min. The precipitate was removed by centrifugation and the acetone was removed in vacuo. The residual aqueous phase was acidified to pH 4.0 with 1 N HsP0, and applied to a CI8 Sep-Pak (Waters Associates, Milford, MA). Hydroxylated fatty acids were eluted with ethyl acetate according to the method of Powell (11). The ethyl acetate was removed under a stream of nitrogen and the samples were redissolved in MeOH/water (6040) for injection on RP-HPLC.
Although all of the experiments in this paper were done with suspensions of endothelial cells to make subsequent extractions of lipid easier, we have found exactly the same patterns of arachidonate and 15-HPETE metabolism in monolayer cultures of endothelial cells.

HPLC Analysis of Incubation Products
Separation of Momhydroxyhted Fatty Acids Obtained from the Incubation of [l-"C]Arachidonic Acid with Endothelial Cells-Monohydroxy fatty acids were separated on a 5-pm 10 X 250 mm Ultrasphere-ODS column at 3.5 ml/min isocratically with methanol/ water/glacial acetic acid (8020:O.Ol) and p-hydroxyheptyl benzoate was used as an internal standard with UV detection at 235 nm. The system resolved 15-HETE, lZ-HETE, and 5-HETE (Rt = 14, 16.2, and 19 min, respectively). Radioactivitiy in the peaks was detected using a Flo-One HS radioactive flow detector (post UV) and Flo Scint 11 scintillant (Radiomatic Instruments and Chemical Co., Inc., Tampa, FL). Peaks of interest were eluted from the mobile phase as described above for [l-"Clarachidonic acid and redissolved in methanol, and their UV spectra were recorded.
Separation of Dihydroxy Fatty Acids Obtained from the Incubation of 15-HPETE with Endothelial Cells-Dihydroxy fatty acids synthesized from 15-HPETE were separated as above except that the mobile phase was methanol/water/glacial acetic acid (73.75:26.25:.01) and detection was at 270 nm. Peaks of interest were extracted as above and the UV spectra were recorded in methanol. Peaks showing triene spectra were treated as below.
Purification of Reverse-phase Peaks by Straight-phase HPLC Monohydroxy and dihydroxy fatty acids collected from reversephase HPLC were methylated with etheral diazomethane, and the resulting methyl esters were dissolved in hexane with sonication. Purification of dihydroxy-and monohydroxy fatty acids was achieved on a Bakerbond chiral column (4.6 mm X 25 cm) (J. T. Baker Research Products, Phillipsburg, NJ). Monohydroxy fatty acids were eluted isocratically in hexane/2-propanol (982) at a flow rate of 1.5 ml/min with UV detection at 235 nm. Dihydroxy fatty acids were isocratically eluted at 2.0 ml/min with hexane /2-propanol (96:4) with UV detection at 270 nm.

Gas Chromatography-Mass Spectrometry of Purified Products
The purified methyl esters of the monohydroxy-and dihydroxy fatty acids were derivatized with bis-(trimethylsily1)rifluoroacetamide, 1% trimethylchlorosilane (Pierce Chemical Company, Rockford, IL) and analyzed by GC/MS (12). Separations were achieved on a 2-foot, 1% SE-30 column heated initially to 170 "C; the temperature was raised linearly at 4 "C/min up to 250 "C. The injection port was 180 "C, and the helium flow was 25 ml/min. Separations and detections were performed on a Hewlett Packard model 5992 spectrometer (Hewlett-Packard Co., Palo Alto, CAI. The energy of the electron beam was set at 20 eV. Labeling Endothelial Cells with PHlArachidonnte Suspended human endothelial cells were incubated for 2 hr at 37 "C with [3H]arachidonate (1 pCi/106 cells). After 2 h, the cells were washed three times with ice-cold Hanks' buffer containing 0.1% fatty acid-free BSA. The labeled cells were then resuspended in Hanks' buffer with 0.1% BSA and stimulated with 2 units of bovine thrombin/106 cells. After a 10-min exposure to thrombin, the samples were analyzed by HPLC exactly as described for the incubations with arachidonate except that a 5-pm analytical Ultrasphere-ODS column was used with a flow rate of 1.2 ml/min. Exogenous 15-HETE (100 ng) was added to each sample to improve recoveries.

RESULTS
Incubation of human endothelial cells (30 x lo6) with 50 p~ [l-14C]arachidonic acid for 10 min at 37 "C results in the formation of two major radioactive UV absorbing peaks on RP-HPLC ( Fig. 1). Radioactive peaks are also evident at the origin, but were identified as prostaglandin Ez and 6-ketoprostaglandin Fla, known products of arachidonate metabolism in endothelial cells (2-4). Peak I is also known from previous work to be the prostaglandin Hz hydrolysis product HHT. Peak I1 is a new product with a retention time of 20.5 min that chromatographs in the monohydroxy arachidonate region (Fig. L4). Boiling the endothelial cells for 10 min completely inhibited the synthesis of both peaks I and 11, (Fig. 1B). Preincubation of the endothelial cells with 1 mM acetylsalicylic acid, a cyclo-oxygenase inhibitor, inhibited peak I formation, but did not significantly influence peak 11 synthesis (Fig. IC). Incubation with the lipoxygenase inhibitor, NDGA at a concentation of 30 p~ inhibited peak I1 formation approximately 85%, and peak I synthesis about 30% (Fig. 1D). Approximately 0.13% of the total counts on the chromatogram were found in peak 11.
This mass spectrum is identical to the published mass spectrum of the methyl ester, trimethylsilyl ether derivative of . Incubation of [l-'4C]arachidonic acid with endothelial cells for extended periods of time (60 min) revealed increased 15-HETE synthesis and the presence of additional radioactive products in addition to 15-HETE (Fig. 3A). Peak I is HHT, peak I1 is 15-HETE, and the new peaks I11 and IV cochromatograph with 12-HETE and 5-HETE standards, respectively. NDGA (30 p M ) completely inhibits the synthesis of peaks 11,111, and IV (data not shown). The LTB, peak is an exogenously added internal standard. Incubation of heatinactivated endothelial cells with [ l-'4C]arachidonic acid for 60 min did not show any evidence of hydroxy fatty acid synthesis (Fig. 3B). Subsequent experiments have shown that maximal 15-HETE synthesis occurs with 100 p~ arachidonate, and that half-maximal stimulation is achieved with 32 p~ arachidonate.
15-HETE is also formed from endogenous sources of arachidonate. Incubation of [3H]arachidonate with endothelial cells results in labeling of several phospholipid classes (16). After extensive washing, labeled endothelial cells were challenged with bovine thrombin for 10 min (2 units of thrombin/106 cells). Subsequent RP-HPLC with radiodetection showed a profile similar to that observed with exogenous arachidonate. Again, peak I (HHT) and peak I1 (15-HETE) radioactive zones were detected (Fig. 4). In this experiment, the 15-HETE peak represents 0.19% of the total radioactivity.

Since endothelial cells can synthesize 15-HETE (by way of a 15-hydroperoxy arachidonate intermediate), we evaluated 15-HPETE metabolism in endothelial cells. Incubation of 15
x lo6 endothelial (1.5 X 106/ml) cells with 30 pM 15-HPETE for 15 min at 37 "C results in the formation of four major UV absorbing peaks on RP-HPLC (Fig. 5A). Peaks I, 11,111, and IV have RP-HPLC retention times (14.5-21.0 min) analogous to dihydroxy arachidonate molecules. Incubation of endothelial cells at 100 "C for 10 min completely inhibits the formation of peaks I-IV (Fig. 5B). Ultraviolet spectrophotometry revealed that each peak has a conjugated triene structure with identical UV maxima at 258, 268, and 279 nm (Fig.  6). Esterification and derivitization of the four unknown peaks followed by GC/MS showed that the four molecules had C numbers of 24.6 to 25.2 min and gave essentially identical mass spectra (Fig. 7). The prominent ions were observed at to the trimethylsilyl ether, and analyzed by GC/MS. All four molecules gave an essentially identical mass spectrum. This spectrum is identical to the published mass spectrum for esterified and derivatized 8, heat-inactivated endothelial cells resulted in essentially the same profile, with large peaks for I and 111, and small I1 and IV peaks (Fig. 8). This is very different from the chromatogram obtained with 15-HPETE and endothelial cells where nearly equal amounts of peaks I, 111, and IV were observed with a somewhat smaller peak I1 (Fig. 5). These data suggest that the 8,15-diHETE peaks I1 and IV are enzymatically formed from endogenously produced 14,15-LTA4, but that exogenously added 14,15-LT& is nonenzymatically hydrolyzed before it can be metabolized by the endothelial cell.

DISCUSSION
Several laboratories have shown that human umbilical vein endothelial cells actively metabolize arachidonic acid to form prostaglandin Ez and 1 2 (3, 4). Endogenous arachidonate is released from endothelial cell phospholipids by phospholipase activity (16) and converted into eicosanoids following stimulation by various stimuli (1)(2)(3)(4)(5)(6)(7)(8). The prostaglandins can then either escape and inhibit platelet or neutrophil aggregation and adhesion, or feedback inhibit their own synthesis in endothelial cells (17)(18)(19)(20). Although the cyclooxygenase pathway in endothelial cells is well-characterized, we were interested in the possibility that endothelial cells might have both a cyclooxygenase and lipoxygenase system analogous to that observed in vascular smooth muscle (21-23).
Incubation of human umbilical vein endothelial cells with [ l-'4C]arachidonic acid followed by extraction and appropriate RP-HPLC showed two prominent radioactive peaks with UV absorbance at 235 nm. Peak I was identified as the cyclooxygenase product HHT, while peak I1 eluted in the monohydroxy arachidonate region. GC/MS analysis of peak I1 indicated that the unknown compound was 15-HETE. These metabolism experiments displayed a classical inhibitor profile. The cyclooxygenase inhibitor acetylsalicylic acid blocked HHT synthesis, but 15-HETE synthesis was not significantly changed. In a few experiments aspirin did slightly decrease the 15-HETE peak, suggesting some contribution of the cyclooxygenase or that aspirin can have a negative influence on 15-lipoxygenase activity. NDGA, a general lipoxygenase inhibitor, only slightly reduced HHT synthesis, but markedly inhibited 15-HETE synthesis. The precursor of 15-HETE biosynthesis is 15-HPETE. Incubation of 15-HPETE with endothelial cells results in the formation of 4 isomers of 8,15-diHETE. The four isomers have essentially identical mass spectra. The UV absorbance pattern of the four 8,15-diHETE molecules are also identical and reveal the presence of a conjugated triene system with absorption max-ima at 258, 268, and 279 nm. Compounds I and I11 have identical retention times on RP-HPLC and UV spectra as the diastereomeric (8R,15S)-and (8S-l5S)-dihydroxy acids formed from the hydrolysis of synthetic 14,15-LTA4. These compounds were detected by Maas et al. (14) in incubations of 15-HPETE with porcine leukocytes, and are thought to be formed by the incorporation of oxygen from water at C-8. Peak I was identified as (8R,15S)-dihydroxy-5-cis-9,11,13trans-eicosatetraenoic acid, and peak I11 as (8S-l5S)-dihydroxy-5-cis-9,11,13-trans-eicosatetraenoic acid (14). These hydrolysis products correspond to the 5,12-dihydroxy acids formed nonenzymatically from LTA, via acid-catalyzed formation of a carbonium ion (24). The mechanism of formation of compounds I1 and IV is not clear. Originally, an attack on the 14,15-LTA4 epoxide by either hydroxyl radical or superoxide anion by direct nucleophilic attack or by electron transfer was proposed (14). More recently, a reaction scheme was proposed that has many mechanistic features in common with the 12-lipoxygenase (25). The mechanism involves the removal of the pro(S) hydrogen at C-10 and migration of the radical to c-14. This radical centered on C-14 then reacts with the carbon-bound hydroperoxy oxygen with homolytic cleavage of the oxygen-oxygen bond to form 14,15-LTA4. In addition, migration of a radical centered on C-10 in the opposite direction was envisaged to give antarafacial addition of oxygen at C-8, leading to the formation of compounds I1 and IV (25). Since we have indirect HPLC evidence of 12-HETE formation, a similar mechanism in endothelial cells is possible. It is believed that compound I1 is (8S, 15s)-dihydroxy-5-cis-9-trans-ll-cis-13-trans-eicosatetraenoic acid, and compound IV is (8R-15S)-dihydroxy-5-cis-9-tram-ll-cis-13trans-eicosatetraenoic acid (14). Although the absolute stereochemistry of the 8,15-diHETE molecules produced by endothelial cells is not known, our data suggeits that 14,15-L T L is produced from 15-HPETE and that two enzymatic (peaks I1 and IV) and two nonenzymatic (peaks I and 111) are synthesized from 15-HPETE by human endothelial cells. At the present time we do not know if these molecules are produced by a single enzyme or a sequence of enzymes. It is possible that 14,15-LTA, could be formed by interaction with hemoproteins (Cyt P-450 and Cyt c) instead of a specific 14,29). However, the high concentrations of these proteins required to catalyze 14,15-LTA4 synthesis from 15-HPETE make this an unlikely alternative.
The possible physiological significance of 15-HPETE metabolism in endothelial cells is not clear. Buchanan et al. (26) reported that lipoxygenase inhibitors could attenuate neutrophil adhesion to endothelium, but it is not established whether or not this was a direct effect on the neutrophil, the endothelium, or both.
The presence of 11-, 12-, and 15-OH arachidonate derivatives in vascular smooth muscle cells is well-documented, and a recent report showed that 12-HETE was chemotactic for smooth muscle cells (27). With both smooth muscle and endothelial cells displaying hydroxy fatty acid metabolism, the interesting possibility of transcellular metabolism between these two vascular cells as well as circulating cells cannot be ignored. Although the physiological significance of diHETE in endothelial cells is not known, the presence of these molecules in our experiments suggest that the involvement of 15-lipoxygenase products in neutrophil-endothelial cell and platelet-endothelial cell interaction warrants further study.