Arachidonic acid diols produced by cytochrome P-450 monooxygenases are incorporated into phospholipids of vascular endothelial cells.

Epoxyeicosatrienoic acids (EETs) are synthesized by cytochrome P-450 monooxygenases and released into the blood. When taken up by vascular endothelial and smooth muscle cells, the EETs are primarily esterified to phospholipids or converted to dihydroxyeicosatetraenoic acids (DHETs) and released. In the present studies, radiolabeled 8,9-, 11,12-, and 14,15-DHETs released into the medium from vascular smooth muscle cells were isolated and incubated for 4-16 h with cultured bovine aortic endothelial cells. The uptake ranged from 2 to 50% for the three regioisomers. Hydrolysis of the endothelial lipids and gas chromatographic-mass spectral analyses of the products indicated that all three DHET regioisomers were incorporated intact into phosphatidylcholine and phosphatidylinositol. Similar incubations with EETs confirmed that small amounts of DHETs were also esterified to endothelial phospholipids. These studies indicate that DHETs are incorporated into phospholipids either at the time of EET conversion to DHET or upon release and re-uptake of DHETs. Beside demonstrating for the first time that fatty acid diols are incorporated intact into endothelial lipids, these studies raise the possibility that both EETs and DHETs remain long enough in the vascular wall to produce chronic vasoactive effects.

The metabolism of arachidonic acid by cytochrome P-450 monooxygenases results in three major types of products, epoxyeicosatrienoic acids (14,15-, 11,12-, 8,9-, and 5,6-EETs), 1 20and 19-hydroxyarachidonic acid, and cis-trans-hydroxyeicosatetraenoic acids such as 12-R-HETE (1,2). Among the three types, EETs have attracted great interest as vasoactive components. Vasodilation has usually been reported, but the response depends upon the route of administration, the length of exposure, the regioisomer and stereoisomer, the dose range tested, and the particular vascular bed studied (3,4). In all cases, the acute response is transient and weak. One possible explanation for the transient vasoactivity is that epoxide hydrolases in vascular endothelial and smooth muscle cells rapidly convert EETs to dihydroxyeicosatetraenoic acids (DHETs) which are generally considered to be inactive (5)(6)(7)(8).
In contrast to EET vasoactivity, very little is known about the disposition of circulating EETs and their DHET products. The concentration of unesterified EETs in plasma from healthy humans and rats is about 1 nM (9). Recent evidence indicates that unesterified EETs are synthesized and released from endothelial cells (8) and activated human platelets (10,11). Upon release, the unesterified EET concentrations may approach 1 M (10). Both the released EETs and their DHET products appear to remain inside the vascular system, where they circulate for about 60 min (12). At least some of the circulating EETs are taken up in vivo into the vessel walls (3). There, the EETs may be incorporated into endothelial and vascular smooth muscle phospholipids or released as DHETs into the medium (6,7). What happens to the released DHETs is uncertain because if nM concentrations are infused, DHETs do not appear in the urine (12).
The concentration of circulating EETs and DHETs may be increased in certain diseases. Clinical studies demonstrate that an enhanced urinary excretion of DHETs accompanies hypertension (12,13) and atherosclerosis (14). Patients with coronary artery disease had elevated levels of urinary DHETs; moreover, coronary angioplasty doubled the urinary excretion of DHETs (14). Animal studies have confirmed that damage to the vascular endothelium by trauma (5) or hypercholesterolemic diets (15) results in an increased release of EETs and DHETs from the blood vessels. Atherogenic concentrations of low density lipoprotein also stimulate human endothelial cells to synthesize and release EETs (16). Together, these studies indicate that unesterified EETs and DHETs are present in low concentrations in the blood and that their concentrations may be increased during certain disease states.
In the present experiments, three radioactive regioisomers of DHETs and EETs were synthesized and their metabolism by endothelial cells compared. The objective was to determine whether DHETs are metabolized by endothelial cells, a mechanism that could be involved in the clearance of DHETs from the circulation.
Each arachidonic acid standard was methylated with diazomethane produced by a macro-generator (Aldrich). After reversed-phase HPLC to remove autoxidation and radiolysis contaminants (3), the methyl arachidonates were treated with 0.10 -0.25 eq of 3-chloroperoxybenzoic acid which converts cis double bonds into (Ϯ)-cis-epoxides. The products (four EET regioisomers) were resolved as methyl esters by normal phase and reversed-phase HPLC (3). Prior to experimental use, each standard was hydrolyzed by saponification and freshly isolated by normal phase HPLC. The structures of the synthesized products were identified by comparing their HPLC and capillary gas chromatographic retention times, as well as ultraviolet (UV) and electron ionization mass spectra, with those of unlabeled standards (Cayman Chemical Co., Ann Arbor, MI). The UV spectra of all regioisomers were identical and contained a single absorption maximum at 193 nm.
Incubation of Tritiated EETs/DHETs with Cultured Endothelial Cells-Primary cultures of bovine endothelial cells (BAEC) were prepared from the aortae of freshly slaughtered cows (18). In brief, endothelial cells were scraped from the aortae and suspended in Dulbecco's modified Eagle's medium (DMEM) supplemented with MEM Nonessential Amino Acids plus MEM Vitamin Solution (Life Technologies, Inc.), 10% heat-inactivated fetal bovine serum (HyClone Laboratories; Logan, UT), 15 mM HEPES, 2 mM L-glutamine (Sigma), and 50 mM gentamicin (Schering Corp.; Kenilworth, NJ). After being counted with a hemocytometer, 3 ϫ 10 5 BAEC were plated into 25-cm 2 flasks. Four hours later, the BAEC were rinsed to remove nonadherent contaminants. Fresh medium was added, and the BAEC were grown to confluency at 37°C under a humid atmosphere containing 5% CO 2 . Following trypsinization, BAEC stocks were subcultured weekly. The BAEC cultures have been characterized previously (18).
Confluent monolayers of BAEC at passages 9 -15 were used in these studies. The BAEC in six-well plates were washed twice with modified DMEM, and 2.0 M [ 3 H 8 ]EETs or [ 3 H 8 ]DHETs was applied in 0.8 ml of modified DMEM containing 0.1 M fatty acid-free bovine serum albumin (Miles Laboratories, Inc., Naperville, IL). After 4 or 16 h at 37°C in a humidified atmosphere of air containing 5% CO 2 , the medium was removed and the cells were washed twice with 1.0 ml of ice-cold isotonic solution containing 137 mM NaCl, 3 mM KCl, 1 mM CaCl 2 , 0.5 mM MgCl 2, 8 mM Na 2 HPO 4 , and 1.5 mM KH 2 PO 4 (pH 7.4). The cells were scraped off the bottom of the flask and suspended in 0.5 ml of fresh buffer solution. Previous studies with radioactive fatty acids demonstrated that scraping does not cause hydrolysis of the cell lipids (19).
Determination of Lipid Incorporation Patterns-To assess the amount of radioactivity remaining in the extracellular fluid at the end of the incubation, the medium was isolated by centrifugation (10,000 ϫ g for 3 min). About 4% of the medium was added to 10 ml of Budget Solve (Research Products International Corp., Mount Prospect, IL) and assayed to 3% precision (95% confidence) using a liquid scintillation spectrometer (Model LS3801; Beckman, Irvine, CA). Quenching was monitored using a 137 Cs external standard; counting efficiencies routinely were 42%. To determine the extent that radioactivity in the medium reflected lipid labeling, the pH of the remaining medium was adjusted to 8.0 with phosphate buffer, and the lipids were partitioned twice into 7.5 volumes of ice-cold ethyl acetate previously saturated with water. Unesterified fatty acids, epoxides, and diols are quantitatively recovered using these extraction conditions (3,20,21). The ethyl acetate was removed under a N 2 stream, and the products were resuspended in acetonitrile for analysis by reversed-phase radio-HPLC (6).
To determine the amount of radioactivity present in endothelial lipids, the BAEC lipids were extracted using chloroform/methanol mixtures (22) except that slight alkalinity (pH 7.8) was ensured with a solution containing 8 mM Na 2 HPO 4 , 1.5 mM KH 2 PO 4 , and 137 mM NaCl. All solvents were of Optima or GC Resolve grade (Fisher). In brief, the BAEC suspensions were mixed vigorously for 5 min (Big Vortexer; GlassCol, Terre Haute, IN) with 20 volumes of chloroform/methanol (2:1), and the phases were separated by centrifugation (450 ϫ g for 30 min). The upper, aqueous phase was removed, mixed vigorously for 5 min with 5 volumes of a chloroform/methanol/phosphate buffer solution (86:14:1), and recentrifuged. The resulting organic phase was combined with the original chloroform extract. After 1 and 25% of the organic and aqueous phases, respectively, were removed for radioassay, the organic phase was placed under N 2, and the organic solvents were removed at 22°C using a flash evaporator; water was eliminated as an methanol azeotrope. The lipid extract was suspended in chloroform/methanol (2:1), and aliquots (5-10%) were analyzed using standard TLC systems for neutral lipids and phospholipids (6). The radio-distribution was analyzed using a gas-flow proportional scanner (Model R, Radiomatic).
Identification of Labeled Radyl Components Using Radio-HPLC-To assess whether neutral and phospholipid radyl moieties were radiolabeled, the total cellular lipids were dried under a N 2 stream, resuspended in 100 l of ethyl ether, and applied to a silicic acid column (6). Neutral lipids were eluted with chloroform/methanol (100:2) while phospholipids were eluted with methanol/water (100:2). The fractions were collected, dried under N 2 , and resuspended in 6 ml of chloroform/ methanol (2:1). Aliquots (1-2%) were dried under N 2 , mixed with 10 ml of Budget Solve, and assayed for 50 min in a liquid scintillation counter. The remainder of each column fraction was saponified. In brief, the neutral lipids and phospholipids were placed under an argon blanket and magnetically stirred with 5.0 N NaOH/methanol (1:10, v/v) for 16 h at 25°C (6). The products were transferred to a silylated glass centrifuge tube, and the pH was adjusted to 7.8 -8.0 with 0.5 M phosphate buffer (pH 6.6). This mixture was shaken with 10 volumes of ice-cold, water-saturated ethyl acetate and centrifuged (450 ϫ g for 10 min at 4°C). The samples were dried under vacuum and resuspended in methanol before being stored under argon at Ϫ80°C. Over 97% of saponified epoxide and diol methyl esters are recovered intact by this extraction procedure (3,20,21).
The identity of the hydrolysis products was assessed using reversedphase HPLC. A low pressure mixing system (Model 410, Perkin-Elmer) interfaced to a photodiode-array UV detector (Model 480, Perkin-Elmer) was employed. All the solvents were sparged with helium (99.9999% pure) and maintained under a 6 psig atmosphere of helium. The aqueous mobile phase was prepared and adjusted to pH 2.2 with phosphoric acid as described previously (23). After being dried under N 2 and mixed with 1.0 g of unlabeled EET and DHET for retention time markers, 8000 dpm of the saponified phospholipids was dissolved in 50 l of methanol and injected onto a guard and analytical column positioned in series (50 ϫ 4.6 mm (inner diameter) plus 250 ϫ 4.6 mm (inner diameter)) containing 5-m C18 particles (Ultremex 5C18 IP, Phenomenex, Rancho Palos Verdes. CA). Separations were done at 1.0 ml/min and 1440 psig using CH 3 CN/H 2 O (52:48). At 70 min following injection, the CH 3 CN concentration was increased linearly to 100% over 10 min. Fractions were collected every 15 s, mixed with 5 ml of Budget Solve, and counted for 15 min.
Identification of Phospholipid Radyl Components Using GC/MS-The saponified products were converted to pentafluorobenzyl esters using ␣-bromo-2,3,4,5,6-pentafluorotoluene (24). In turn, the pentafluorobenzyl esters were silylated with N-methyl-N-trimethylsilyltrifluoroacetamide as described (6). The derivatized compounds were analyzed using a gas chromatograph (Model 5980, Series II; Hewlett-Packard) interfaced to a quadrupole analyzer (Model 5989A, "Engine"; Hewlett-Packard). Each product was dissolved in 1.0 l of isooctane and injected via an on-column injector ("duckbill," Hewlett-Packard) into a wall-coated (0.25-m film of 5% diphenyldimethylpolysiloxane; DB5-MS; J&W, Rancho Cordova, CA) fused-silica column (0.25 mm (inner diameter) ϫ 29 m). At 1.0 min after injection, the oven temperature was ramped from 90 to 230°C at 70°C/min and then increased at 30°C/min to 280°C where it was maintained for 9.0 min. The transfer line was 285°C, whereas the injector temperature was kept 3°C above the oven temperature. The velocity of the helium (99.9999%) carrier gas was 66 cm/s throughout the analyses (17). The temperatures of the ion source and analyzer were set to 200 and 100°C, respectively. Analyses in the negative ion and positive ion chemical ionization (230 eV) modes were done with nominal methane pressures of 1.7 and 2.1 torr, respectively, in the ion source.
To permit comparisons with literature values, a plot of carbon number versus log (retention time) was generated using 21:0, 22:0, 23:0, and 24:0 fatty acid pentafluorobenzyl esters (17). Individual equivalent chain length values were determined from the interpolated log 10 (retention times).

RESULTS
DHET Labeling of Cell Lipids-All of the radiolabeled DHETs were taken up by the BAEC, but the extent of uptake was regioisomer-specific (Table I). About 50% of the added 8,9-DHET was recovered in the endothelial cells at both 4 and 16 h. In contrast, the uptake of 11,12-DHET and 14,15-DHET was initially low (2-7%) but increased to 21% after 16 h. Thus, compared with 11,12-and 14,15-DHET, 8,9-DHET appeared to be rapidly and selectively taken up by endothelial cells.
For all three regioisomers, the partitioning of radioactivity between cell total lipids and nonlipids was essentially the same. Even after 16-h incubations, 98.3 Ϯ 0.2% (n ϭ 3) of the cellular radioactivity was present in the lipid extract. Following silicic acid chromatography of the total lipid extracts, 94 Ϯ 1% of the recovered radioactivity was in the phospholipids, whereas only 6 Ϯ 2% was in neutral lipids. Unesterified DHET regioisomers elute exclusively in the neutral lipid fraction. Thus, no more than 6% of the radioactivity in the endothelial cells represented unesterified DHETs.
Radiolabeled products more polar than the DHETs also appeared in the medium; their formation varied with time and depended upon the regioisomer (Table III). While one-half of the added 8,9-DHET was incorporated into BAEC, only onethird (53 ϫ 67%) remained in the medium after 4 h of incubation (Tables I and III). The remainder of radioactivity in the medium was contained in two products, compounds I and II (Fig. 1A). The distribution of 8,9-DHET radioactivity was similar after 16 h of incubation. These results suggest that a rapid equilibration of labeling occurred, perhaps due to cycling of 8,9-DHET between the cells and medium.
In contrast to 8,9-DHET, more than half of the added 11,12-DHET (72 ϫ 78%) remained in the medium at 4 h (Tables I and  III). At this time, compounds IЈ, IIЈ, and IIIЈ were readily evident and represented 22% of the medium radioactivity (Fig.  1B). After 16 h, only 16% of the added 11,12-DHET remained in the medium; 77% of radioactivity in the medium was contained in compounds IЈ, IIЈ, and IIIЈ. Thus, BAEC used 11,12-DHET primarily to form polar products that accumulate in the medium.
In contrast to 8,9-and 11,12-DHET, essentially all of the added 14,15-DHET (99 ϫ 98%) remained in the medium in the first 4 h of incubation (Tables I and III). Further incubation for 12 h increased cell labeling by only 19% and decreased the medium radioactivity by only 25% (Table I). Moreover, compounds IЉ, IIЉ, and IIIЉ (Fig. 1C) combined contained only 13% of the medium radioactivity after a 16-h incubation (Table III). Thus, compared with 8,9-and 11,12-DHET, 14,15-DHET was poorly taken up by the BAEC and poorly converted to polar products that accumulated in the medium. In corresponding incubations, no polar products were formed by the cell-free medium (data not shown). Thus, the formation of DHET products was dependent upon the presence of the BAEC. Taken together, the radiolabeling of endothelial cell lipids and the appearance of polar products in the medium indicated that DHETs were taken up by the BAEC. Furthermore, because DHETs remained in the medium for at least 16 h, it is possible that DHETs cycle between the medium and cells, thus prolonging the availability of DHETs in the extracellular fluid.
EET Labeling of Endothelial Lipids-14,15-DHET was the major metabolite when 14,15-EET was incubated with BAEC for 2 h (6). In the present study, 16-h incubations were undertaken to determine the extent that the generated DHETs remained intact or were metabolized further to lipid-and watersoluble products. Data represent the mean value Ϯ standard deviation of three incubations with 2 M EETs. After 16 h, most of the cellular radioactivity was lipid-soluble and represented from 50% (14,15-EET) to 75% (8,9-EET) of the radioactivity initially added to the medium (Table IV). Lipid labeling by the three EETs averaged 62 Ϯ 11% of the applied radioactivity, which was about twice the 30 Ϯ 16% observed for DHETs (Table I), indicating that endothelial lipids are more efficiently labeled by EETs than by DHETs. As with the DHETs (Table I), h. The medium was removed and assayed using liquid scintillation techniques. After the cells were washed and scraped from the plates, the lipids were extracted using chloroform/methanol mixtures and similarly radioassayed. The amount of radioactivity in cellular nonlipids (Folch upper phase) averaged 0.58 Ϯ 0.06% of the radioactivity added to the cultures for the three regioisomers. Data represent the mean percent of radioactivity added to the cultures in duplicate incubations. Two different batches of BAEC were utilized for the 16-h incubations, and the values represent the mean percent Ϯ half the range of added radioactivity. It should be noted that the small excess of radioactivity recovered over that added to the incubation was within experimental error. the 8,9-epoxide was the most effective regioisomer in labeling lipids (Table IV). Moreover, as with the DHETs, the EETs preferentially labeled phospholipids, i.e. 98.1 Ϯ 0.6% of the lipid radioactivity from the three regioisomers was in the phospholipid fraction whereas only 2.2 Ϯ 0.3% was in the neutral lipid fraction. Thus, no more than 2% of the lipid radioactivity in the endothelial cells remained as unesterified EETs, which elute only in the neutral lipid fraction.
No intact EET was left in the medium by 16 h (Fig. 2). Only products more polar than EETs were detectable, and the types present in the medium depended upon the regioisomer. For example, almost all of the radioactivity present in the medium after incubation with 8,9-EET was present as 8,9-DHET ( Fig.  2A). In contrast, after 16-h incubations with radiolabeled 11,12-and 14,15-EET, only about 23 and 46% of the medium radioactivity was present as 11,12-and 14,15-DHET, respectively. The remaining radioactivity was associated with unidentified polar compounds aЈ-cЈ and aЉ-cЉ (Fig. 2, B and C).
These results indicate that DHETs are available for continued uptake by BAEC, even after 16 h. In the above studies, the EETs generated no polar products when incubated in cell-free medium (data not shown). Thus, BAEC were essential for both the conversion of EETs to DHETs and the formation of products more polar than DHETs. Taken together, these results raise the possibility that EETs are converted to DHETs, which cycle between medium and cells or are further metabolized.
To determine whether DHETs derived from EETs were incorporated into endothelial phospholipids, the BAEC phospholipids were hydrolyzed by saponification and the radiolabeled products analyzed using reversed-phase HPLC. About 68 -79% of the radioactivity coeluted with EET standards; however, 5-31% coeluted with DHET standards (Fig. 3). These results confirmed that most of the EET labeling of phospholipids reflects the incorporation of intact EETs (6). However, as noted earlier for incubations of radiolabeled 14,15-EET with endothelial cells from porcine aorta, a small percentage of the phospholipid radioactivity derived from each of the EET regioisomers also co-elutes with DHETs (6).
Identification of DHET-labeled Phospholipid Radyl Groups by Reversed-phase HPLC-In the next set of experiments, the BAEC cultures were incubated with DHETs, and after the cells were washed, the lipids were extracted and separated. Although the amount of radioactivity in the neutral lipid fraction was inadequate for more detailed studies, the phospholipid fraction contained enough radioactivity in the 16-h studies to . After the medium was collected, the DHET products were partitioned into ethyl acetate (pH 7.8), dried under a nitrogen stream, and resuspended in methanol. Each sample (ϳ25,000 dpm) was injected into a C18 reversed-phase column and resolved in 50 min by linearly increasing the acetonitrile concentration from 30 to 100%. The flow rate was 0.9 ml/min, and radioactivity was monitored using an on-line flow scintillation counter. The results shown here are from single incubations, but similar chromatograms were obtained from duplicate cultures. Because the product retention times varied relative to that of the parent DHET, compounds I-III are accented to emphasize their uncertain homology.  Table I, and the values represent the mean percent and standard deviation (n ϭ 3) of applied radioactivity. In addition, the total lipid fractions from two of the three incubations were pooled and separated into neutral lipids and phospholipids using silicic acid column chromatography. The radioactive content of these two lipid subfractions was measured using liquid scintillation techniques. permit further analyses. After saponification, the radyl products were extracted into ethyl acetate, and aliquots were mixed with unlabeled DHET standards before injection onto a reversed-phase, C18 HPLC column (Fig. 4). Most of the radioactivity from the three DHET regioisomers migrated slightly in front of the corresponding unlabeled DHET standards. In the case of 8,9-and 11,12-DHETs, two minor unidentified polar products (X and Y, respectively) were also detected. No radiolabeled EETs or fatty acids were evident in these chromatograms, i.e. the DHETs were not converted to EETs or reduced to long-chain fatty acids before being esterified into endothelial phospholipids. In summary, the HPLC data suggested that almost all of the BAEC labeling was due to the incorporation of intact DHETs into phospholipids.

Identification of DHET-labeled Phospholipid Radyl Components
Using GC/MS-The major products of hydrolyzed BAEC phospholipids were isolated by reversed-phase HPLC, silylated with N-methyl-N-trimethylsilylfluoroacetamide, and esterified with pentafluorobenzyl bromide for GC/MS analysis. The equivalent chain lengths (carbon numbers) of the derivatized products were 22.3, 22.3 and 22.5 and matched those of 8,9-11,12-, and 14,15-DHET standards, respectively, when chromatographed on nonpolar capillary columns. The measured equivalent chain lengths were also very close to published values (17).
The negative ionization spectra of all three compounds were similar and contained m/z 481 (M Ϫ CH 2 C 6 F 5 , 100%) and 409 (M Ϫ (CH 2 C 6 F 5 ϩ CH 2 ϭSi(CH 3 ) 2 ); 6.3 Ϫ 13.1%); they also closely resembled the spectra of corresponding DHET standards run under identical conditions. Fig. 5A illustrates the spectrum of the 14,15-DHET isolated from the cells, and the spectrum of the 14,15-DHET standard is shown in Fig. 5B. Likewise, the spectra resembled published spectra where the interpretations had been validated using synthetic, deuterated analogs (17). Thus, the negative ion spectra suggested that the derivatives had a molecular mass of 662 Da.
The above interpretations on diol and double bond positions were based on data derived from electron ionization studies (17). The fragments assumed to identify double bond positions required careful validation, particularly since only small amounts of each compound were recovered, and the observed spectra did not exactly match those of the DHET standards ("fingerprint matches"). Thus, deuterated DHET standards were synthesized and their spectra compared with those of the nondeuterated DHET standards generated under identical GC/MS conditions.
trimethylsilyl groups at C 8 and C 9 .
FIG. 3. HPLC radiochromatograms of EET products released from hydrolyzed BAEC phospholipids. BAEC were incubated 16 h with EETs as described in Fig. 2. After the cells were collected and washed, the total phospholipids were isolated by organic extraction plus silicic acid column chromatography and hydrolyzed by saponification. The released radyl moieties were partitioned into ethyl acetate (pH 7.8), dried under a nitrogen stream, and mixed with 1.0 g of unlabeled EET and DHET. The resulting mixture (6000 -8000 dpm/50 l) was injected into a C18 reversed-phase HPLC column through which acetonitrile/water (pH 2.2), 48:52, flowed at 1.0 ml/min. Absorption at 195 nm was monitored (upper A-C tracings), and fractions were collected every 15 s for radioactivity determinations (lower AЈ-CЈ tracings). Note the presence of an UV chromophore which eluted at 17.2 min in the upper tracings. This unlabeled compound was a contaminant from the medium and possessed a wavelength maximum of 237 nm.
FIG. 4. HPLC radiochromatograms of DHET products released from hydrolyzed BAEC phospholipids. BAEC were incubated 16 h with DHETs as described in Fig. 1. The cell phospholipids were isolated, saponified, and analyzed by reversed-phase HPLC as described in Fig.  3. The results shown are from single incubations, but similar chromatograms were obtained from duplicate cultures.
In summary, analysis of the pentafluorobenzyl esters in the negative ionization mode identified a common molecular weight of 662 for the derivatized products. Analysis of the pentafluorobenzyl esters in the positive ionization mode confirmed this molecular weight and unequivocally established vicinal diol positions as well as the number of double bonds located on either side of the diols. When coupled with close matches of retention times and peak shapes (widths and symmetry) with standards during reversed-phase HPLC (underivatized) and capillary GC, the data indicated the DHET products did not contain conjugated double bonds or keto-enol groups. Thus, the spectral and chromatographic data demonstrated that 8,9-DHET, 11,12-DHET and 14,15-DHET were incorporated intact into BAEC phospholipids. DISCUSSION Since the initial work of Stenson and Parker (26), monohydroxy regioisomers of arachidonic acid (HETEs) have been reported to be taken up by multiple cell types and incorporated into phospholipids (10,27). Recently, Fitzpatrick and colleagues (28) suggested that small amounts of the vicinal diol, 14,15-DHET, are also taken up by mastocytoma cells; however, they were unable to find any 14,15-DHET esterified to phospholipids when they were analyzed by tandem mass spectrometry. The present study demonstrates that DHET regioisomers are taken up by vascular endothelial cells and incorporated intact into membrane phospholipids. Lacking conjugated double bonds, these DHET vicinal diols differ fundamentally from diols produced by lipoxygenases like DiHETEs and leukotriene B 4 , which are not incorporated into phospholipids (27,29).
In previous studies, radioactive compounds with the same retention times as DHETs were found to be released in small amounts after hydrolysis of phospholipids isolated from vascular endothelial and smooth muscle cells incubated with radiolabeled EETs (6,7). The saponification procedure used to hydrolyze the phospholipids does not artifactually generate DHETs from EETs (6). In the present study, the amount of DHET released was equivalent to 4 -32% of the EETs initially incorporated into lipids. The DHETs generated were identified by mass spectrometry as well as by reversed-phase HPLC and capillary gas-chromatographic techniques. In both the present and prior studies (6), the BAEC rapidly converted EETs to DHETs which were primarily secreted into the medium. However, the present results further suggest that some of the DHET formed in the process is incorporated into phospholipids. In addition, a basal release of EETs from BAEC phospholipids would provide substrates for DHET formation by the epoxide hydrolases long after EETs cease to be available in the medium (30).
The accumulation of DHET intracellularly may affect endothelial function. DHETs may compete with endogenous arachidonate for activation and incorporation into phospholipids. Alternatively, as found for HETEs (31), the DHETs may reduce arachidonate availability, either by displacing it from phospholipids or by being preferentially released from phospholipids by certain stimuli. Receptor ligands like thrombin and platelet activating factor transiently elevate EET intracellular concentrations up to 1 M by stimulating the release of EETs from phospholipids (10). In addition, a transient increase in EET synthesis from arachidonate may follow receptor activation (32). Such studies suggest that the intracellular concentration of unesterified EET is variable and may even approach that of unesterified arachidonate. In light of the multiple biological effects of EETs, the rapid conversion to DHET and subsequent esterification of DHETs to phospholipids may provide a mechanism whereby the vascular wall can regulate the bioactivity of EETs and their metabolites.
The amount of DHET accumulated by BAEC depended upon the regioisomer. The relative order of BAEC labeling by DHETs was 8,9 Ͼ 11,12 ϭ 14,15 and paralleled that found for EETs incubated with endothelial cells from bovine and porcine aorta (6), as well as from microvessels of rat, mouse, and human brain. 2 Similarly, the order for HETE incorporation into endothelial cells is 5-Ͼ Ͼ 12-Ն 15-HETE (27,33). Thus, the net uptake of regioisomers from the three eicosanoid families is lowest when the oxygen moiety is furthest removed from the carboxyl group. Because the uptake patterns are the reverse of the elution order for reversed phase HPLC (3,34), the hydrophobicity of each eicosanoid may be a major factor in determining EET and DHET accumulation within endothelial cells.
As found for EETs 2 (6) and HETEs (31), DHETs labeled predominantly phospholipids rather than neutral lipids in resting endothelial cells. Moreover, analogous regioisomers from the three eicosanoid families showed similar preferences for incorporation into phospholipid classes. As with EETs, the 8,9and 11,12-DHET regioisomers preferentially labeled phosphatidylcholine, whereas the 14,15-DHET regioisomer labeled mostly phosphatidylinositol. Similarly, 5-and 12-HETE label primarily BAEC phosphatidylcholine, whereas 15-HETE labels phosphatidylinositol much more (35). Thus, the labeling of phosphatidylinositol is highest for those eicosanoids that have the oxygen-containing moiety located furthest from the carboxyl group. Such labeling patterns raise the possibility that analogous regioisomers produced by epoxygenases and lipoxygenases may share acyltransferases specific for phosphatidylcholine and phosphatidylinositol, respectively.
Depending upon the regioisomer, varying proportions of unidentified polar metabolites of DHETs appeared in the medium. Based on studies with other eicosanoids, a number of oxidative routes for DHETs are possible. Some of the DHET metabolites may represent ␤-oxidation products. Monools like HETEs commonly undergo several cycles of ␤-oxidation in resting endothelial cells (35). Another possibility is that certain DHET regioisomers act as substrates for cyclooxygenases and lipoxygenases, as found for HETEs and EETs (1, 36 -38). In addition, enzyme-catalyzed transformations with regioisomeric specificity may be accelerated by ligand binding (39). The presence of such a variety of DHET metabolites raises the possibility that biologically active products are also being formed as the result of smooth muscle-endothelial cell interactions.
Minor, unidentified polar metabolites of DHETs were also found esterified to BAEC phospholipids (Fig. 4). Because the amounts present resembled that of DHET accumulations (8,9 Ͼ Ͼ 11,12, and 14,15; Table I), the phospholipid levels of the esterified DHET metabolites may primarily reflect the intracellular concentrations of DHETs. ␤-Oxidation products of monools like HETEs are esterified to glycerides (31,40); thus, one possibility is that these esterified compounds represent ␤-oxidation products of DHETs. Together, these and the above results suggest that EETs can be metabolized by multiple routes in vascular endothelial cells, as summarized in Fig. 7.
As with the isoprostane triols (41) and the HETE monools (27), fatty acid diols like DHETs may perturb the normal structure of membranes when esterified to phospholipids. It is thus important to assess the concentration of DHETs in endothelial membranes. In the present study, both the 8,9-and 11,12-DHETs were primarily found in phosphatidylcholine, suggesting that they may be localized to particular domains in lipid bilayers. Physical measurements indicate that 15-HETE alters the fluidity of lipid bilayers when present at only 3.5 mol % in phosphatidylcholine (42). Because diols are expected to be more disruptive than monools, it seems possible that their presence in phospholipids may also perturb membrane function, particularly in disease states where vascular EET concentrations may be chronically elevated, e.g. atherosclerosis (14), eclampsia (12), or hypertension (43,44). In addition, the presence of DHETs in endothelial phosphatidylcholine and phosphatidylinositol may lead to the formation of novel second messengers. For example, bradykinin stimulates the release of the monool 15-HETE from endothelial phosphatidylinositol and also generates diglyceride species rich in 15-HETE (45). Whether comparable unesterified DHETs and DHET-enriched diglycerides are generated and whether DHET-enriched diacylglycerides can alter protein kinase C activity is not known.
In summary, this is the first demonstration that the DHETs are not irreversibly excreted but can be taken up and incorporated intact into endothelial phospholipids. Because each DHET regioisomer was less efficiently taken up by endothelial cells than the parent EET, DHETs released into the vessel lumen may be preferentially carried away. However, the results raise the possibility that DHETs released into the subintimal space will be taken up by contiguous endothelial cells and perturb their function. The results further suggest that the vascular effects of EETs may also be prolonged due to EET and DHET incorporation, possibly resulting in alterations in membrane integrity, the formation of novel second messengers, or conversions to biologically active metabolites.