Formation of 9-Hydroxyoctadecadienoic Acid from Linoleic Acid in Endothelial Cells*

Human umbilical vein endothelial cells convert linoleic acid to two monohydroxyoctadecadienoic (HODE) acids, 9- and 13-HODE. More 9-HODE than 13-HODE is formed under most conditions. The pro- duction of these metabolites is reduced substantially by acetylsalicylic acid, ibuprofen, or arachidonic acid, suggesting that cyclooxygenase may be involved in endotheUal HODE synthesis. Incubations lasting up to 4 h indicate that the endothelial cells can convert [U- 14C]linoleic acid into at least four additional products, some of which may be derived from the HODE that is formed initially. Radioactive 9- and 13-HODE are produced when the endothelial cells are labeled with lin- oleic acid and then exposed to thrombin, suggesting that these metabolites also may be formed when the endothelium is activated. If endothelial monolayers grown on micropore filters are incubated with linoleic acid, a substantial amount of the HODE formed accu- mulates in the basolateral fluid. This suggests that HODE may have extracellular effects, especially within the vascular wall. Furthermore, when 9- or 13-HODE are added, endothelial cultures produce less prostaglandin 12 and convert less 12-hydroxyeicosa-tetraenoic acid to its main metabolite, 8-hydroxyhexa- decatrienoic acid. Therefore, in addition to extracellular actions, HODE also may have functional effects within the endothelium.

Endothelial cells also can directly oxygenate linoleic acid, but only a single product, 13-HODE, has been detected so far (8,9). The formation of 13-HODE in the endothelium is thought to be mediated by 15-lipoxygenase (8, 9).  has a number of biological actions. It maintains the water barrier in the skin (5), inhibits 5-lipoxygenase activity in leukocytes (IO), and modulates thromboxane AP . There is some uncertainty regarding the possible role of HODE in vascular processes, for some studies indicate that it remains entirely associated with the endothelium (8, 12), while others find that it is released into the medium (9). In an attempt to resolve this question and gain more insight into the vascular actions of linoleic acid, we have investigated HODE production by primary cultures of human umbilical vein endothelial cells.

EXPERIMENTAL. PROCEDURES
Materials-Tissue culture suppIies such as glutamine, basal media Eagle vitamins, minimum essential medium nonessential amino acids, and neomycin sulfate were obtained from Gibco. Collagenase was purchased from Worthington, and fetal bovine serum was obtained from HyClone Laboratories. [l-'*C]Linoleic acid (55.6 mCi/mmol) and [U-14C]linoleic acid (950 mCi/mmol) were obtained from Du Pont-New England Nuclear, and all other radioactive fatty acids were purchased from Amersham Corp. Nonradioactive fatty acids were obtained from NuChek Prep (Elysian, MN) and fatty acid-free albumin from Miles Laboratories, lnc. 9-and 13-HODE were obtained from Oxford Biomedical Research Inc. (Oxford, MI). Silica gel thinlayer chromatography plates were provided by Alltech Associates, Inc. (Deerfield, IL). Fatty acids were >98% pure as determined by TLC.
Tissue Culture-Human endothelial cells were obtained from umbilical veins (I3), and primary cultures were prepared according to a slight modification of the method of Jaffe et ~l . (14) as described previously (15). Briefly, the cells were suspended in modified Medium 199 containing 20% heat-inactivated fetal bovine serum, counted with a hemocytometer, and seeded in 10-cm2 wells at a concentration of 1.35 X lo6 cells/well. After incubation for 24 h at 37 "C in an atmosphere containing 5% COZ, this medium was replaced with 3 ml of Medium 199 containing 25 @M HEPES, pH 7.4, plus 20% fetal bovine serum. The confluent cultures were maintained for 3 days at 37 "C in the 5% COZ atmosphere.
Incubations-Most of the experiments were done with a culture medium containing Medium 199,15 PM HEPES, and 0.1 PM albumin. These media were enriched with fatty acids by adding a warm solution of the sodium salt (16), and the pH was adjusted to 7.4 at 37 "C. After washing the cultures with Medium 199, 0.8 ml of the fatty acidsupplemented medium was added, and the incubations were carried out at 37 "C in a 5% COP atmosphere. The incubations were termi-6823 nated by removing the medium, the cells were washed twice with 1 ml of ice-cold Dulbecco's phosphate-buffered saline solution containing 137 mM NaCl, 3 mM KCl, 1 mM CaC12, 0.5 mM MgCI2, 8 mM Na2HP04, and 1.5 mM KH2P04, pH 7.4. After harvesting by scraping, the cells were suspended in 0.5 ml of fresh cold buffer. Previous studies with radioactive fatty acids indicated that this scraping procedure did not cause hydrolysis of fatty acids from phospholipids, as compared with other currently available methods for cell harvesting (17). A portion of the cell suspension was removed for determination of the protein content (18). The remainder was extracted with 20 volumes of ch1oroform:methanol:acetic acid (2:1:0.01, v/v/v) to isolate the cell lipids through a modification of the Folch method (19). Following separation and collection of the chloroform phase, the aqueous phase was washed with 5 volumes of chloroform:methanol:acid saline (4 mM HCl plus 155 mM NaCI) (86: 14:1, v/v/v) and the washings combined with the original extract. After the solvent was evaporated under N2 and the cell lipids resuspended in a known volume of ch1oroform:methanol (l:l, v/v), a portion of the sample was assayed for radioactivity by liquid scintillation counting (17).
The incubation media were centrifuged at 10,000 X g for 10 min and acidified to pH 3.4 using 2 N HCl. Aliquots of the supernatant solution were removed and assayed for radioactivity, and the remainder of the acidified medium was extracted 3 times with 0.5 ml of ethyl acetate (20) and the solvent evaporated under Nf. Lipids were resuspended in acetonitri1e:water (l:l, v/v) and stored at -20 "C until analyzed.
Chemical Analyses-Cell lipids were resuspended in ethanol and saponified by heating at 45 "C for 60 min with 1 M KOH. After acidification, the resulting fatty acids were extracted into n-heptane and the heptane removed under Nz. The fatty acids were resuspended in acetonitrile:HzO (1:2, v/v) and the radioactive lipid Separated by reverse phase HPLC. Similarly, lipid extracts of the medium were separated by reverse phase HPLC using a Beckman 332 gradient system (Palo Alto, CA) equipped with a 4.5 X 150-mm column packed with 3 pm spherical Adsorbosphere CIS (Alltech Associates, Deerfield, IL). The solvent system contained water adjusted to pH 3.4 with phosphoric acid and acetonitrile (21). An elution gradient starting with 27% acetonitrile and increasing to 100% acetonitrile over 42 min was used to separate the metabolites. 9-and 13-HODE were separated by normal phase HPLC using a 250 X 2.1-mm Absorbosphere 5 pm column from Alltech Associates (Deerfield, IL). An isocratic system containing n-heptane:isopropanol (10018, v/v) at a flow rate of 0.5 ml/min was used for this separation.
HPLC column effluents were mixed with Budget Solve scintillation fluid (RPI Corporation, Mount Prospect, IL) at a 0 5 3 ratio, and the radioactivity was detected and analyzed with a Radiomatic Flow 18 radioactivity flow detector (Radiomatic Instruments & Chemical Co., Tampa, FL).
Gas-Liquid Chromatography and Moss Spectrometry-The major metabolites of linoleic acid were isolated by HPLC and methylated with diazomethane in ether (22). The methyl ester of the metabolite was converted to the trimethylsilyl ether by incubation with bis(trimethy1silyl)trifluoroacetamide and 1% trimethylchlorosilane (Sylon BFT, Supelco, Inc., Bellefonte, PA) in pyridine (23). Catalytic hydrogenation was performed with 0.1 mg of platinum oxide added to the methyl ester, trimethylsilyl ether derivative of the metabolite in 0.3 ml of ethanol. After Hz was bubbled through the solution for 1 min, the mixture was diluted with 0.7 ml of water and extracted three times with 1 ml of ethyl acetate (24). Electron impact spectra of the hydrogenated and unhydrogenated methyl ester, trimethylsilyl ether derivatives of the linoleic acid metabolites were obtained with a Riber R 10-10 quadropole mass spectrometer containing a 25 m X 0.2-mm column packed with 5% phenylmethylsilicone and maintained at 195 "C. The energy of the electron beam was 22.5 eV.
Polarity of HODE and PG12 Formation-The procedure for growing endothelial cells on micropore filters and measuring the polarity of radioactive metabolite release has been described (25)(26)(27). Briefly, 24-mm, 0.4-pm pore Transwell Cell Culture Inserts (Costar, Cambridge, MA) were coated with 16 pg/ml human fibronectin (Collaborative Research, Inc., Bedford, MA) and suspended in Costar 6-well tissue culture plates. The filters were seeded for 4 h with 1.4-1.5 X lo6 endothelial cells in 1.5 ml of modified Medium 199 containing 20% heat-inactivated fetal bovine serum and then maintained in fresh medium for 3-4 days until intact monolayers formed. The integrity of each monolayer was tested prior to use by adding 50 p M albumin to the fluid above the cells contained in the Transwell insert (apical fluid) and measuring the amount transferred to the tissue culture well (basolateral fluid) (26). Only monolayers that transferred less than 2% of the albumin in 30 min were utilized for the experiments. The albumin solutions were removed, both surfaces of the filter washed with Dulbecco's buffer solution, and fresh modified Medium 199 containing 0.1 y~ albumin was added to the apical compartment (0.5 ml) and basolateral compartment (1.6 ml). Radioactive fatty acid was added to either the apical or basolateral fluid, either 7.5 p M [1-"C]linoleic acid or 7.5 p~ [l-14CC]arachidonic acid. After a 20-min incubation at 37 "C, these media were collected and the radioactive products separated by reverse phase HPLC (25)(26)(27). Previous measurements with [3H]-6-keto-PGF1, and [3H]prostaglandin EZ added to either the apical or basolateral fluid indicated that in 30 min under these conditions only 7-10% of the radioactivity was transferred from the apical to basolateral fluid, and 2-3% from the basolateral to apical fluid (27).
PGIz Assay-The amount of PGIZ formed was measured by radioimmunoassay of the stable inactivation product, 6-keto-PGF1, (15,28). The cultures were incubated for 30 min with fatty acids or inhibitors and then for 20 min with either 7.5 p~ arachidonic acid or 2 units/ml thrombin. A 100-p1 aliquot of the medium was removed and incubated for 4 h at 4 "C with 50 pl of anti-6-keto-PGF1, antibody (Seragen, Boston, MA) and 15,000 cpm of [3H]6-keto-PGF1,. After addition of 400 pl of 1% dextran-coated charcoal and centrifugation, a 500-pl aliquot of the supernatant was added to 5 ml of Budget Solve scintillation solution. Radioactivity was measured in a liquid scintillation spectrometer, and quenching was estimated by channels ratio. A complete standard curve was run with each assay. The antibody has 7.8% cross-reactivity with PGF1,, 6.8% with 6-keto-prostaglandin Et, 2.2% with PGFz,, and 4 % with other prostaglandins.

HODE Formation-When primary cultures of human umbilical vein endothelial cells were incubated with [l-14C
]linoleic acid, a major metabolite with a retention time of 32 min was detected in the medium by reverse phase HPLC (Fig. 1,  toppanel). Standards of two linoleic acid derivatives, 9-HODE and 13-HODE, comigrated with this product. The unmodified linoleic acid and HODE were extractable into ethyl acetate, but the radioactivity eluting with the solvent front did not enter the ethyl acetate phase. Appreciable amounts of HODE were not formed when [l-'4C]linoleic acid was incubated with the medium alone.
As opposed to the accumulation in the medium, no HODE radioactivity was detected in the cell lipids. Only unmodified linoleic acid and several small components eluting in proximity to linoleic acid were observed when a cellular chloroform:methanol extract was saponified prior to separation by reverse phase HPLC (bottom panel). Detectable amounts of HODE radioactivity also were not present in a methanol extract of the cells. Furthermore, no clearly discernible HODE radioactivity was detected in the cells, either in saponified ch1oroform:methanol or methanol extracts, at any of the incubation times tested, 5-90 min.
Results of an incubation lasting up to 4 h are shown in Fig.  2. This study was done with [U-'*C]linoleic acid, and the medium was extracted with ethyl acetate prior to analysis by reverse phase HPLC in order to remove the polar radioactivity that otherwise eluted with the solvent front. HODE, unmodified linoleic acid, and a small amount of radioactivity eluting just ahead of linoleic acid were detected in the incubation medium after 20 min (top panel). After 2 h, at least four additional metabolites were observed (middle panel). The additional metabolites became more prominent after 4 h, and the amount of HODE and unmodified linoleic acid remaining in the medium decreased considerably. These metabolites were not identified, but they may be similar or identical to the epoxyhydroxy and trihydroxy derivatives formed by leukocytes, skin, and aortic slices (4, 5, 7).
Further studies indicated that a metabolite similar to 229. They are formed by cleavage on both sides of the trimethylsilyl group and indicate that the hydroxyl group is present at carbon 9. These ions are the same as those reported for the reduced derivative of 9-HODE (2, 3). Taken together, these spectra are consistent with the interpretation that the main metabolite produced when human endothelial cells are incubated with linoleic acid is 9-HODE.
GC/MS analysis of a methyl ester, trimethylsilyl ether derivative of the less abundant metabolite (Fig. 5 ) contains the m/z 382, 311, 292, and 225 ions reported for 13-HODE (2, 3). A similar spectrum was obtained with the 13-HODE standard. The mlz 311 ion has a greater relative abundance than the m / z 225 ion, the reverse of what is observed with 9-HODE. This difference has been reported previously (2, 3). In most of our studies, 13-HODE accounted for only 20-30% of the total product, and appreciable losses occurred during hydrogenation. Therefore, we could not prepare sufficient amounts of the reduced metabolite to obtain a suitable spec-trum and conclusively demonstrate the position of the hydroxyl group.
Factors Affecting HODE Formation-The dependence of HODE formation on the time of incubation and linoleic acid concentration was investigated. About 75% of the total HODE accumulation occurred during the first 5 min of incubation (Fig. 6, top panel). The amount continued to increase up to 40 min and then decreased substantially. HODE formation also increased as the concentration of linoleic acid in the medium was raised (bottom panel).
To determine the distribution of the radioactivity between 9and 13-HODE, similar experiments were done in which the lipid extracts of the media were separated by normal phase HPLC. As shown in Table I

TABLE I Distribution of HODE radioactivity
Following incubation of the endothelial cultures with [l-"C]linoleic acid, the medium was collected and the radioactive products separated by normal phase HPLC as indicated in Fig. 3. In this and other tables where HODE production was measured with radioactive isotopes, the amount formed was calculated based on the specific activity of the [1-"Cjlinoleic acid added to the medium. Each value is the mean f S.E. of data from three separate cultures.

Linoleic Acid Oxygenation 6827
However, the proportion of 9-HODE decreased at the longer incubation times and when the concentration of linoleic acid was low.
To determine whether the presence of other fatty acids may decrease the capacity of the endothelium to convert linoleic acid to HODE, either palmitic or oleic acid was added to incubation media containing 15 FM [l-'4C]linoleic acid. The presence of 15 or 30 FM concentrations of these fatty acids did not reduce radioactive HODE formation; in fact, 40% more HODE was produced when 30 p~ oleic acid was added. Therefore, the simultaneous availability of other physiologically abundant fatty acids does not interfere with the capacity of the endothelial cells to produce HODE.
Effects of Metabolic Inhibitors- Table I1 shows the effects of several metabolic inhibitors on HODE formation. Although previous reports indicated that HODE synthesis in endothelium and aortic slices is mediated by a lipoxygenase (4, 8,9), we find that nordihydroguaiaretic acid has little effect on 9or 13-HODE formation. By contrast, acetylsalicylic acid reduced HODE formation by 80-90%, suggesting involvement of cyclooxygenase. To determine whether the residual HODE formation might be due to a failure of acetylsalicylic acid to exert a complete effect in this system, PG12 formation also was measured in a second experiment. Again, acetylsalicylic acid reduced HODE formation by 70-80%, and ibuprofen, another cyclooxygenase inhibitor, produced a 75-80% reduction. By contrast, these inhibitors decreased PG12 formation by more than 95%, indicating that some HODE production persists in the presence of an almost complete cyclooxygenase block.
Effects of Fatty Acids-Studies with fatty acids also are consistent with involvement of cyclooxygenase in endothelial HODE formation. As shown in Table 111, HODE formation was reduced by 90% if the endothelial cultures were initially incubated with arachidonic acid. Previous studies have demonstrated that exposure to high concentrations of arachidonic acid inactivates cyclooxygenase (32,33). An initial incubation with linoleic acid also reduced the capacity of the endothelial

TABLE I1
Effect of inhibitors on HODE and PGI, production Endothelial cultures were incubated with the inhibitors for 30 min prior to the addition of 15 PM [l-14C]linoleic acid. After an additional 20-min incubation, the 9-and 13-HODE content of the medium was assayed by normal phase HPLC. PGI, was assayed by reverse phase HPLC as 6-keto-PGFI,. The amounts of HODE and PGI, formed are calculated based on the specific activities of the radioactive fatty acid substrates added to the medium. Each value is the mean f S.E. of results obtained from three seuarate cultures.  14C]linoleic acid. HODE was determined by reverse phase HPLC, the quantity being calculated from the specific radioactivity of the added linoleic acid. In Experiment 2, the second incubation was with 7.5 p~ arachidonic acid, and PGI, was measured by radioimmunoassay for 6-keto-PGF1,. Each value is the mean & S.E. of results obtained from three separate cultures. cells to form PG12, but higher concentrations of linoleic than arachidonic acid were required to produce equivalent reductions. Effect of Albumin-As shown in Fig. 7, HODE formation was reduced when increasing amounts of defatted serum albumin were added to the incubation medium. However, some HODE continued to be formed even when the albumin concentration was 30 PM and the molar ratio of linoleic acid to albumin was only 0.5.
Polarized Formation-The ability of endothelial cells grown on micropore filters to produce HODE was examined. [ 1-"CC] Linoleic acid was added to either the apical or basolateral fluid, and the formation of radioactive metabolites was measured by reverse phase HPLC after a 20-min incubation. The results are presented in Table IV. HODE was recovered primarily in the basolateral fluid, independently of where the linoleic acid was added. Furthermore, 3.2 times more total HODE was produced when the linoleic acid initially was available in the basolateral fluid, as compared with the apical fluid. In additional experiments (data not shown), the radioactive HODE contained in the apical and basolateral fluid TABLE IV Polarity o/ HODE and PGIz formation by endothelial cells grown on micropore filters Endothelial monolayers maintained in a confluent state on micropore filters suspended in tissue culture wells were incubated with either 7.5 pM [l-'4C]linoleic acid or 7.5 p~ [l-14C]arachidonic acid for 20 min. The radioactive fatty acid was added to either the cylindrical compartment above the filter (apical fluid), or to the tissue culture compartment below the filter (basolateral fluid). The main radioactive products present in the apical and basolateral fluid at the end of the 20-min incubation were assayed by reverse phase HPLC.
PGIz was detected as 6-keto-PGF1,. Each value is the mean -+ S.E. of results obtained from three separate cultures. was separated by normal phase HPLC. The ratio of 9-to 13-HODE in the apical fluid was between 2.0 and 2.4; it was higher, between 4.8 and 7.7, in the basolateral fluid.
The results concerning the polarity of HODE formation are different from those obtained with PGIZ. The endothelial monolayers produced PGIz, detected by reverse phase HPLC as 6-keto-PGF1,, when [l-'4C]arachidonic acid was added to either the apical or basolateral fluid. However, more total PG12 was formed when the arachidonic acid was added to the apical fluid. In addition, PGIz accumulated to a greater extent in the apical fluid, even when the arachidonic acid was added to the basolateral compartment.
Thrombin-stimulated HODE Release-To determine whether HODE release occurred when the endothelium was activated, cultures were labeled with radioactive linoleic acid and then exposed to thrombin. The initial studies indicated that very little radioactivity was released from the cells, and it was necessary to label the cells with [U-'4C]linoleic acid to obtain enough radioactivity to detect by HPLC. As shown in Fig. 8 (top panel), the bulk of the released radioactivity was in the form of unmodified linoleic acid. However, a small amount of radioactive HODE was detected. Substantially less HODE and unmodified linoleic acid were released in a corresponding incubation without thrombin (bottom panel).
To determine the distribution of the released HODE, the radioactivity contained in the medium was separated by normal phase HPLC. Both forms of HODE were detected (Fig.   9). However, as opposed to the results obtained when linoleic acid was present in the medium (Fig. 3 and Table I), equivalent amounts of radioactivity were present in 9-and 13-HODE when the cells were exposed to thrombin.
Thrombin did not increase radioactive HODE formation when it was added together with 15 PM [l-14C]linoleic acid during a 20-min incubation. Furthermore, the addition of thrombin to incubations with [l-'4C]linoleic acid did not appreciably affect either the distribution of HODE between the apical and basolateral media or the ratio of 9-to 13-HODE when the cells were grown on micropore filters. This suggests that thrombin acts by stimulating linoleic acid release from endothelial lipids rather than by directly affecting the conversion of linoleic acid to HODE.
Effects of HODE on Eicosanoid Metabolism-Analysis by reverse phase HPLC indicates that endothelial cells can oxidize 12-HETE to several metabolites (31). The main product is 16:3(8-OH) (29, 31). As seen in Table V,  FIG. 9. Separation of the lipid radioactivity released by the cells exposed to thrombin. The procedure was the same as described in Fig. 8, except that the radioactivity released into the medium was separated by normal phase HPLC.
,.LM HODE reduced the conversion of 12-HETE to  by 24-52%. The extent of the reduction was about the same with 9or 13-HODE. By contrast, 1 pM linoleic acid did not appreciably reduce  formation. Table VI shows that thrombin-stimulated PGIz production also was decreased following addition of HODE. While 13-HODE produced more suppression at 1 PM, it was much less effective than 9-HODE at the higher concentrations.

DISCUSSION
These findings demonstrate that human endothelial cells produce both 9-and 13-HODE and that 9-HODE, not 9), is the major product under most conditions. With the exception of neutrophils which form only 7), the other tissues that synthesize HODE also produce both isomers (1-5). As in the case of endothelial cells, 9-

TABLE VI
Effect of HODE on endothelial PG12forrnation Endothelial cultures were incubated with 1 ml of serum-free medium containing 0.1 W M albumin and the indicated amount of 9-or 13-HODE. After 30 min, 300 pl of this medium was removed for PGL measurement by radioimmunoassay for 6-keto-PGF1,. Subsequent analysis revealed that these media contained only low levels of PGI2, 8-17 pmol/ml. 2 IU/ml thrombin was added, and after 20 min the medium was collected and the PGIP content measured. Thrombinstimulated PGI, release was calculated by subtracting the value obtained after the initial 30-min incubation from that found after incubation with thrombin. Each value is the mean f S.E. of results obtained from three separate cultures, except in the case of 5 PM 13-HODE where one of the samples was lost. There was a single set of three control cultures incubated without HODE. The value is listed in the column for 9-HODE. * Average of values from two cultures. e No thrombin-stimulated increment over the value present after the initial 30-min incubation, 12 pmol, was observed.
HODE is the major product formed by the vesicular gland, VX2 carcinoma, and aorta (1, 2, 4). Our results indicate that cyclooxygenase is involved in the formation of both 9-and 13-HODE. However, the reductions in HODE formation produced by the cyclooxygenase inhibitors and by exposure to high concentrations of arachidonic acid were not complete, and some decrease in 13-HODE formation occurred when nordihydroguaiaretic acid was added. 13-HODE synthesis in leukocytes and skin is mediated by 15-lipoxygenase (5-7), and endothelial cells are reported to contain this enzyme (3435). Therefore, while the endothelial cyclooxygenase appears to be primarily responsible for HODE production, it is possible that 15-lipoxygenase may mediate some 13-HODE formation in this tissue.
There is substantial precedent for the conclusion that HODE formation is mediated at least in part by the endothelial cyclooxygenase. For example, HODE synthesis in the vesicular gland, VX2 carcinoma, and peritoneum also is catalyzed by cyclooxygenase (1-3). In addition, cyclooxygenase causes the lipoxygenation of arachidonic acid at carbons 11 and 15 in umbilical arteries and smooth muscle cultures, forming 11-HETE and 15-HETE (36,371. Likewise, a purified cyclooxygenase preparation catalyzes the lipoxygenation of eicosadienoic acid, the elongation product of linoleic acid (38). The structures of arachidonic acid (20:4n-6) and eicosadienoic acid (202n-6) between their methyl termini and carbon 11 are identical, including the positions of the two double bonds contained in this segment of the acyl chains. Moreover, these structures are the same as the structure of linoleic acid (18:Zn-6) from the methyl terminus to carbon 9, with carbons 11 and 15 of arachidonate and eicosadienoate being equivalent to carbons 9 and 13 of linoleate. Since cyclooxygenase can oxygenate arachidonic and eicosadienoic acids, it is understandable that this enzyme might also catalyze the partial oxygenation of linoleic acid at these positions. By contrast, linolenic acid (183n-3), which has an additional unsaturation in this segment of the chain, was not hydroxylated by the endothelial cells. This is consistent with the finding that eicosapentaenoic acid (20:5n-3), which also has an n-3 unsaturation, is a poor cyclooxygenase substrate in endothelial cultures (39,40). Because HODE is produced from extracellular linoleic acid without the need for any additional stimulus, it is thought to exert its effects under basal conditions (8,12). Many of our findings are consistent with this interpretation. However, we find that some HODE also can be formed from intracellular linoleic acid when the endothelial cells are exposed to thrombin. This suggests that in addition to the possibility of exerting effects in the basal state, 9-and 13-HODE also may play a role when the endothelium is activated.
Although HODE has been observed to have a number of biological actions (5, 8, [10][11][12], its function is not completely understood. A possibility suggested by previous work is that HODE acts within the endothelium, modifying the adherence of platelets and leukocytes to the endothelial surface (8, 12). Our finding that HODE reduces endothelial PGIz production is consistent with such an effect on surface adhesive properties. Some decrease in PG12 formation occurred at HODE concentrations as low as 1 p~. At 5 and 10 p~, 9-HODE, the isomer produced in larger amounts, was more effective in reducing PG12 formation than 13-HODE. Linoleic acid also decreased PG12 formation, but concentrations of 20-30 PM were necessary to obtain this effect in 30 min. A substantial amount of HODE is formed during the first 20 min when the endothelial cells are exposed to these concentrations of linoleic acid. Therefore, the reduction in PGI, formation that occurs when linoleic acid is added may be mediated by the HODE that is produced.
Another effect of HODE on the endothelium appears to involve HETE metabolism. The endothelial cells converted less 12-HETE to , the main metabolic product (29-31), when either 9-or 13-HODE was present. This was observed at a concentration of 1 p~ HODE. Linoleic acid was ineffective at this concentration, a finding consistent with the fact that very little HODE is formed when the linoleic acid concentration is 2.5 p~ or less (Fig. 6). Endothelial cells can take up 41), and this reduces their capacity to form PGI, (31,42). Thus, HODE formation may reduce the capacity of the endothelium to produce PG12 through an indirect mechanism, by decreasing the catabolism of any 12-HETE that accumulates within the cells.
The present findings demonstrate that most of the 9-and 13-HODE formed by the endothelial cells is released into the extracellular fluid. This agrees with results obtained in other tissues (4-7), including one of the previous studies with endothelium where 13-HODE was recovered in the medium (9). Because 9-and 13-HODE rapidly accumulate in the extracellular fluid, it is likely that some of their effects may be directed at surrounding tissues. In this regard, 13-HODE has been shown to modulate thromboxane and 12-HETE synthesis when it is added to platelets (11). Our studies with endothelial monolayers grown on micropore filters indicate that HODE can be released into the basolateral fluid, even when linoleic acid is availabIe only in the medium bathing the apical surface of the endothelium. This suggests that in addition to acting on platelets, HODE released by the endothelium may have effects on cells contained in the vascular wall.