Omega-oxidation of 20-hydroxyeicosatetraenoic acid (20-HETE) in cerebral microvascular smooth muscle and endothelium by alcohol dehydrogenase 4.

20-Carboxyeicosatetraenoic acid (20-COOH-AA) is a bioactive metabolite of 20-hydroxyeicosatetraenoic acid (20-HETE), an eicosanoid that produces vasoconstriction in the cerebral circulation. We found that smooth muscle (MSMC) and endothelial (MEC) cultures obtained from mouse brain microvessels convert [3H]20-HETE to 20-COOH-AA, indicating that the cerebral vasculature can produce this metabolite. The [3H]20-COOH-AA accumulated primarily in the culture medium, together with additional radiolabeled metabolites identified as the chain-shortened dicarboxylic acids 18-COOH-18:4, 18-COOH-18:3, and 16-COOH-16:3. N-Heptylformamide, a potent inhibitor of alcohol dehydrogenase (ADH), decreased the conversion of [3H]20-HETE to 20-COOH-AA by the MSMC and MEC and also by isolated mouse brain microvessels. Purified mouse and human ADH4, human ADH3, and horse liver ADH1 efficiently oxidized 20-HETE, and ADH4 and ADH3 were detected in MSMC and MEC by Western blotting. N-Heptylformamide inhibited the oxidation of 20-HETE by mouse and human ADH4 but not by ADH3. These results demonstrated that cerebral microvessels convert 20-HETE to 20-COOH-AA and that ADH catalyzes the reaction. Although ADH4 and ADH3 are expressed in MSMC and MEC, the inhibition produced by N-heptylformamide suggests that ADH4 is primarily responsible for 20-COOH-AA formation in the cerebral microvasculature.

A major metabolite of 20-HETE, 20-carboxyeicosatetraenoic acid 3 (20-COOH-AA), was detected initially in rabbit kidney cells isolated from the thick ascending loop of Henle and in human polymorphonuclear leukocytes (7,8,18). The -oxidation of 20-HETE to 20-COOH-AA also was observed recently in porcine coronary artery endothelial cultures (19), and it may have been overlooked in other tissues because the difference in the retention times of 20-HETE and 20-COOH-AA in the commonly used reverse-phase high performance liquid chromatography (HPLC) separation systems is less than 2 min (7,19). There is increasing evidence that 20-COOH-AA is a bioactive metabolite rather than an inactivation product of 20-HETE. In the rabbit kidney, the potency of 20-COOH-AA in inhibiting the medullary Na ϩ -K ϩ -ATPase activity, the Na ϩ -K ϩ -2Cl Ϫ cotransporter, and the uptake of 86 Rb was similar to that of 20-HETE (7,8,20,21), and it was about 20% as potent as 20-HETE in inhibiting the 70-pS K ϩ channel in rat kidney (22). Furthermore, 20-COOH-AA at concentrations between 0.1 and 1 M produces vasodilation of porcine coronary arterioles preconstricted with endothelin, indicating that it also may have functional activity in the microvasculature (19). However, the metabolic pathways responsible for the conversion of 20-HETE to 20-COOH-AA in vascular cells are unknown. 20-HETE is synthesized by cerebral microvascular smooth muscle cells and produces vasoconstriction in the cerebral circulation (5,9,15,16,23). To investigate the metabolism of 20-HETE in the cerebral vasculature, we have examined the conversion of 20-HETE to 20-COOH-AA in smooth muscle cells (MSMC) and endothelial cells (MEC) cultured from mouse brain microvessels, as well as in the isolated microvessels, and we characterized the subsequent metabolism of 20-COOH-AA in the cell culture preparations. Because rat and human class IV alcohol dehydrogenases (ADH4) effectively oxidize 10 -16-carbon saturated -hydroxy fatty acids (24,25) and are expressed in rat and human blood vessels and rat microvascular endothelium (26,27), we tested the hypothesis that this form of ADH might be involved in 20-HETE metabolism in cerebrovascular tissue.
Cell Culture and Incubation-Microvessels were isolated from mouse brains (28), and MSMC and MEC cultured from the microvessels were isolated and grown in modified high glucose Dulbecco's minimum essential medium (DMEM) containing 10% fetal bovine serum (FBS), as described previously (29 -31). The cultures were maintained until confluent at 37°C in a humidified atmosphere containing 5% CO 2 . Stocks were subcultured weekly with 0.25% trypsin in 1 mM EDTA and maintained in modified high glucose DMEM supplemented with 10% FBS, 0.3 mg/ml L-glutamine, BME vitamin solution, and BME amino acid solution (Sigma), 10 mM HEPES (pH 7.4), and 50 g/ml gentamicin. Experiments were performed with cultures between passages 12 and 25.
Monolayers of MSMC and MEC that were 90% confluent were incubated at 37°C with 1 ml of modified DMEM or Medium 199 containing 0.1 M BSA and radiolabeled substrate in a humidified atmosphere containing 5% CO 2 . Radiolabeled substrates of the necessary concentration and specific activity for each experiment were prepared by mixing the [ 3 H]20-HETE with 20-HETE purchased from Cayman Chemical (Ann Arbor, MI) or [ 3 H]20-COOH-AA with 20-COOH-AA synthesized as described previously (19,32). Inhibitors were incubated with the cells for 20 min at 37°C before [ 3 H]20-HETE was added, and the incubation was continued in the presence of the inhibitor. Cell viability, monitored by determining the amount of radiolabeled ␤-oxidation products formed (19), cell morphology, and trypan blue staining indicated that none of the inhibitors produced cytotoxicity at the concentrations tested. The isolated mouse brain microvessels were incubated with [ 3 H]20-HETE under similar conditions (28).
Analysis of the Medium and Cell Lipids-After the medium was acidified to pH 4 with formic acid, the lipids were extracted with 4 volumes of ice-cold water-saturated ethyl acetate. The lipid-soluble radioactivity was measured by liquid scintillation counting using Budget Solve TM scintillation solution (Research Products International, Mount Prospect, IL) and a Beckman LS-3801 scintillation counter. Reverse-phase HPLC with an in-line flow scintillation counter was used to determine the distribution of the lipid radioactivity (19). Following removal of the ethyl acetate under N 2 , the lipids were dissolved in acetonitrile and separated by HPLC on a 4.6 ϫ 150 mm Discovery TM C 18 column (Supelco Inc., Bellefonte, PA) with a dual pump gradient system (Gilson Inc., Middleton, WI). The solvent system contained H 2 O adjusted to pH 4.0 with formic acid and an acetonitrile gradient that increased from 30 to 100% over 70 min at a flow rate of 0.7 ml/min. Radioactivity was measured by mixing the column effluent with Budget Solve TM scintillation solution at a 3:1 ratio and then passing the mixture through the flow scintillation detector (IN/US Systems, Inc., Tampa, FL).
The cells were washed with ice-cold PBS and harvested by scraping into ice-cold methanol/formic acid (100:1 v/v). Lipids then were extracted by addition of 2 volumes of chloroform, and the phases were separated by addition of 0.25 volume of 154 mM NaCl. The chloroform phase was dried under N 2 and dissolved in chloroform/methanol. An aliquot was dried and assayed for radioactivity in a liquid scintillation counter, and the lipids contained in another aliquot were separated by thin layer chromatography. The remainder was hydrolyzed by saponi-fication, and the distribution of radioactivity was determined by reverse-phase HPLC (19).
A Hewlett-Packard 1100 MSD liquid chromatography and mass spectrometry (LC/MS) system was used to separate and identify the metabolites formed by the cells from either 20-HETE or [16,16,17,17,18, H]20-HETE ([ 2 H 6 ]20-HETE) purchased from Cayman Chemical (19). The HPLC separation used a Supelco C 18 5 m 4.6 ϫ 150 mm Discovery TM column and a solvent system containing H 2 O and acetonitrile, both acidified with formic acid (100:0.03 v/v), at a flow rate of 0.7 ml/min. The gradient began with 0% acetonitrile and was increased to 95% acetonitrile over 25 min. Negative ion electrospray was used with the fragmenter voltage set at 110 V; the N 2 nebulizing gas was maintained at 60 bars, and the N 2 drying gas was set to a flow rate of 10 liters/min at 350°C. A Hewlett-Packard Chemstation TM software program was employed to process the data.
Western Blot Analysis of ADH-MSMC and MEC were washed twice with PBS and then scraped into a lysis buffer containing 150 mM NaCl, 50 mM Tris-HCl (pH 7.5), 100 g/ml phenylmethylsulfonyl fluoride, 1 g/ml aprotinin, 1 g/ml leupeptin, 1 mmol/liter diethyldithiocarbamic acid, 1% Nonidet P-40, and 1% sodium deoxycholate. Homogenates were prepared by sonication on ice for 3 cycles of 10 -20 s, with intermittent cooling, and the protein content was determined with the protein assay reagent (Bio Rad) using bovine serum albumin as a standard. Proteins (50 g) were separated in 10% SDS-PAGE under reducing conditions and then transferred to nitrocellulose membranes by electroblotting at 100 V for 75 min. Blots were blocked by incubation with 0.05% (v/v) Tween 20 and 5% (w/v) solution of dried skimmed milk dissolved in NaCl/Tris-HCl (pH 8.0), at room temperature for 70 min and then incubated with a 1:500 dilution of rabbit antiserum raised against mouse ADH4, ADH3, or ADH1 (33). The blot was then incubated with a horseradish peroxidase-conjugated donkey anti-rabbit antibody (1: 4,000). Antibody labeling was detected by enhanced chemiluminescence with an ECL detection system (Pierce). The blots were then stripped and reprobed for ␤-actin with a mouse monoclonal IgG (1:2,500) and anti-mouse IgG-horseradish peroxidase (1:4,000). ␤-Actin was used as the control for protein loading and transfer (30,31). To control for the specificity of the ADH4 antibody, homogenates of stomachs from normal and Adh4 gene-deleted mice also were prepared and included in the assay. The mutant mice lack ADH4 in stomach and other tissues as demonstrated by Western blots and have impaired metabolism of retinol (34).
Preparation of Purified ADH-Mouse ADH4 was expressed in Escherichia coli M15 (pREP4) from the full-length cDNA (35) that was linked to a 6ϫ His tag in the pQE-30 plasmid (Qiagen, Valencia, CA). The ADH4 was purified as described previously (36) and gave a single band with a relative molecular mass of about 43 kDa, approximating the expected 42-kDa size, upon separation by SDS-PAGE and staining with Coomassie Brilliant Blue. The enzymatic activity was controlled before each experiment by a standard assay (37). Crystalline horse liver ADH1 (EE isozyme) was obtained from Roche Applied Science. Human ADH4 was expressed in E. coli and purified and characterized according to published procedures (38,39). Dr. Paresh Sanghani (Indiana University School of Medicine, Indianapolis) kindly provided us with purified human ADH3 (40).
Detection and Analysis of ADH Products-The isolated mouse and human ADH4, human ADH3, and horse liver ADH1 were incubated at 37°C with either 1 or 2 M [ 3 H]20-HETE with 1 mM NAD ϩ , and the radiolabeled products were separated by HPLC and detected with the in-line flow scintillation counter. In additional studies, mouse ADH4 was incubated with 20-HETE and NAD ϩ , and the products were analyzed by LC/MS with the Hewlett-Packard 1100 MSD LC/MS system as described previously (19). The data were processed with the Hewlett-Packard Chemstation TM software program.
Statistical Analysis-Data were expressed as mean Ϯ S.E. Analysis of variance, followed by the Student-Newman-Keuls method, was used to compare the conversion of 20-HETE to 20-COOH-AA under control and treatment conditions. Differences with p Ͻ 0.05 are considered significant.

20-HETE Conversion to 20-COOH-AA by Cultured Cells-
We first determined whether 20-HETE contained in MSMC can be converted to 20-COOH-AA. It was necessary to load the MSMC with [ 3 H]20-HETE to obtain enough intracellular radioactivity for metabolic studies. In preliminary experiments, we found that MSMC incubated with [ 3 H]20-HETE took up increasing amounts of radioactivity during a 4-h incubation. The distribution of the radioactivity in the cell lipid extract as determined by thin layer chromatography was 75% in phospholipids, 16% in neutral lipid esters, and 9% as fatty acid after 1 h of incubation. This distribution did not change appreciably as the incubation continued. HPLC analysis of the hydrolyzed cell lipid extract showed that 99% of the radioactivity incorporated into the cells after 1 h remained as 20-HETE. After 2 h, 94% of the radioactivity still was present as 20-HETE. This distribution did not change appreciably when the incubation was extended to 4 h. The retention and metabolism of the incorporated [ 3 H]20-HETE were determined during subsequent incubation of the MSMC. Fig. 1 shows the changes in distribution of lipid radioactivity in the cells and medium when MSMC containing [ 3 H]20-HETE were washed and then incubated for 1-8 h in fresh medium. Although no agonist was added to the cultures to stimulate the mobilization of endogenous lipids, the radiolabeled lipid content of the cells declined 40%, and a similar amount of radiolabeled lipid accumulated in the medium over the course of the incubation. Fig. 2 shows the HPLC analysis of the radiolabeled lipids recovered from the medium during the 8-h incubation. A metabolite with the same retention time as [ 3 H]20-COOH-AA (19) accounted for 83% of the lipid radioactivity after 1 h ( Fig. 2A) and 80% after 4 h (Fig. 2B). In addition, 6% of the radioactivity present in the medium after 4 h eluted with a retention time of 26.3 min, designated as X, and 11% eluted with a retention time of 22 min, designated as Y. Fig.  2 shows that these products continued to build up and accounted for 6 and 17%, respectively, at the end of the 8-h incubation.
Additional studies demonstrated that MEC also converted [ 3 H]20-HETE to 20-COOH-AA. During a continuous incubation with 2 M [ 3 H]20-HETE for 4 h, 12% of the lipid radioactivity in the medium was present as 20-COOH-AA and 39% as other products, and this increased to 17 and 58%, respectively, after 8 h (data not shown).
Identification of the 20-HETE Metabolites-MSMC cultures were incubated with either 20-HETE or [ 2 H 6 ]20-HETE, and the products contained in the lipid extract of the media were separated by liquid chromatography and identified by negative ion electrospray mass spectrometry at 110 V. Fig. 3, A-D, shows the mass spectra obtained from the incubation with 20-HETE, and Fig. 3, E-H, shows the corresponding spectra from the incubation with [ 2 H 6 ]20-HETE. The most abundant product obtained from the incubation with 20-HETE, which had a retention time of 36.9 min in this liquid chromatography gradient system, contained an ion (M Ϫ H) Ϫ1 m/z 333 (Fig. 3A) and fragmented with loss of H 2 O (Ϫ18), CO 2 (Ϫ44), and H 2 O plus CO 2 (Ϫ62). The corresponding product in the incubation with [ 2 H 6 ]20-HETE contained (M Ϫ H) Ϫ1 m/z 339 and was fragmented as indicated above (Fig. 3E). These data are consistent with the structure of 20-COOH-AA and its hexa-deutero analogue (19). Na ϩ adduct ions (m/z 355 and m/z 361) are present in these and the other spectra shown in Fig. 3.
The material designated as X in Fig. 2 separated into two components in this liquid chromatography gradient. One of the components from the 20-HETE incubation (M Ϫ H) Ϫ1 m/z 305 (Fig. 3B) had a retention  indicates that the six 2 H were retained in this 16-carbon product, implying that it also was formed by ␤-oxidation from the original COOH-end of the substrate.
Metabolism of 20-COOH-AA-Further studies were done to determine whether the three dicarboxylic acid products that accumulated in the medium are formed directly from 20-COOH-AA. Fig. 4A shows HPLC analyses of the radiolabeled lipids in the medium during incuba-tion of MSMC with [ 3 H]20-COOH-AA. Two radiolabeled components with the same HPLC retention times as X and Y were detected after 4 h, and the amounts of radioactivity present in these components increased to 3 and 7%, respectively, of the lipid-soluble radioactivity in the medium after 24 h. Less than 5% of the [ 3 H]20-COOH-AA added to the MSMC cultures was recovered in the cells at any of the times tested, and essentially all of it was recovered as 20-COOH-AA (data not shown).
Larger amounts of the radiolabeled components X and Y were formed when the MEC were incubated with [ 3 H]20-COOH-AA under the same conditions. The two components accounted for 20 and 37%, respectively, of the radiolabeled lipid in the medium at the end of the 24-h incubation (Fig. 4B).
The compounds produced from 20-COOH-AA by the MSMC cultures also were identified by LC/MS. As noted in the studies with the 20-HETE metabolites, the material designated as X in Fig. 4    hyde dehydrogenases, can potentially metabolize 20-HETE. In addition, 20-HETE may undergo free radical-mediated peroxidation. Therefore, we incubated the MSMC with several pharmacological agents in an attempt to gain some insight into the pathways responsible for 20-COOH-AA formation. There was no appreciable effect on [ 3 H]20-HETE metabolism when the incubation contained 5 M miconazole or 10 -40 M 17-octadecynoic acid (inhibitors of cytochrome P450 enzymes), 2 mM 4-methylpyrazole (inhibitor of some forms of ADH), or 50 mM ethanol (a mouse ADH1 substrate), 50 M cyanamide (aldehyde dehydrogenase inhibitor), or 1 mM tiron (free radical scavenger). However, as seen in Fig. 6, conversion of [ 3 H]20-HETE to 20-COOH-AA by MSMC and MEC cultures was reduced by N-heptylformamide, a potent inhibitor of ADH1 and ADH4 (36,41,42). Fig. 6A shows that there was a 70% decrease in radiolabeled 20-COOH-AA production when 20 M N-heptylformamide was added to the MSMC incubations, and the decrease reached a limiting level of about 90% when the concentration was raised to 100 M. Likewise, as shown in Fig. 6B, the addition of 10 M N-heptylformamide to MEC cultures decreased radiolabeled 20-COOH-AA production by 72%, and by 94% when the concentration was raised to 50 M. Addition of N-heptylformamide did not decrease [ 3 H]20-HETE uptake or the formation of ␤-oxidation products by the cells (data not shown), suggesting that the decreases in 20-COOH-AA production were not because of either a reduction in the intracellular availability of [ 3 H]20-HETE or cytotoxicity. The production of radiolabeled 20-COOH-AA also was decreased by 55-65% when either 30 M all-trans-retinol (an ADH substrate) or 30 M all-trans-retinoic acid (an ADH4 inhibitor) was added to MSMC cultures during a 4-h incubation with 2 M [ 3 H]20-HETE (data not shown).
N-Heptylformamide also decreased the oxidation of 20-HETE to 20-COOH-AA by isolated mouse cerebral microvessels. HPLC analysis demonstrated that the microvessels converted [ 3 H]20-HETE to a radiolabeled product with a retention time of 32 min (Fig. 6C), the same as a [ 3 H]20-COOH-AA standard. The amount of [ 3 H]20-COOH-AA formed decreased by 65% when 20 M N-heptylformamide was added to the incubation (Fig. 6D).

Detection of ADH in Cerebral Microvascular Cells-
The inhibition of 20-HETE metabolism by N-heptylformamide suggested that an ADH might be involved in the conversion of 20-HETE to 20-COOH-AA. ADH4 is the principal ADH isoform present in rodent blood vessels (26,27). Moreover, the effects produced by the retinoids in the MSMC, and the fact that the concentrations of ethanol (a good substrate for ADH1) and 4-methylpyrazole (a potent inhibitor of ADH1) tested did not decrease radiolabeled 20-COOH-AA formation, suggested that ADH4 was involved in the conversion of 20-HETE to 20-COOH-AA in the microvascular cells. Therefore, the expression of ADH4 was investigated in MSMC and MEC cell extracts. The Western blots shown in Fig.  7 demonstrate that ADH4 is expressed in both cell types. Blots obtained from homogenates of stomach from normal mice and mice with deletion of the gene (33) are included to demonstrate the specificity of this polyclonal antibody for ADH4. Additional blots demonstrated that ADH3 also is expressed in the MSMC and MEC extracts (Fig. 7). However, a distinct ADH1 band was not detected in these cell extracts (data not shown).
Studies with Purified Alcohol Dehydrogenases-Purified recombinant mouse ADH4, human ADH4, and human ADH3 were tested to determine whether 20-HETE was a substrate for these enzymes. The steadystate kinetic constants that were obtained are shown in TABLE ONE. Each of the enzymes oxidized 20-HETE, and the highest catalytic efficiency was obtained with human ADH4. Although N-heptylformamide  inhibited mouse and human ADH4, it did not inhibit the oxidation by human ADH3. Fig. 8 shows an analysis of the time-dependent changes in radiolabeled lipids present during incubation of purified recombinant mouse ADH4 with [ 3 H]20-HETE. Two radiolabeled compounds were detected by HPLC, 20-HETE and a product with a retention time of 32 min. This product, which comigrated with synthetic [ 3 H]20-COOH-AA, increased from 24% of the radioactivity at 30 min (Fig. 8A) to 44% at 60 min (Fig. 8B) and 52% at the end of the 90-min incubation (Fig. 8C). The product was not detected when [ 3 H]20-HETE was incubated without enzyme under the same conditions (Fig. 8D).
LC/MS was used to confirm the identities of the products formed by the mouse ADH4. Fig. 9 shows the total ion chromatogram, selective ion scans, and negative ion electrospray mass spectra obtained from an incubation of 20-HETE with mouse ADH4. The liquid chromatography gradient began with 0% acetonitrile and was increased to 95% acetonitrile over 25 min at a flow rate of 0.7 ml/min. Two major components with retention times of 17.4 and 18.5 min, and a minor component with a retention time of 20.8 min, were detected in the total ion chromatogram (Fig. 9A). A selective ion scan for (M Ϫ H) Ϫ1 m/z 333 had a retention time of 17.4 min (Fig. 9B). The fragmentation pattern of this substance was identical to that shown in Fig. 3A, consistent with the structure of 20-COOH-AA. Fig. 9C demonstrates that the selective ion scan for (M Ϫ H) Ϫ1 m/z 319 had a retention time of 18.5 min, and the negative ion electrospray mass spectrum of this component was identical to that obtained previously for 20-HETE (19). As shown in Fig. 9D Horse liver ADH1 also converted [ 3 H]20-HETE to 20-COOH-AA in a sodium glycine buffer at pH 10. The reaction went to completion under these conditions, and HPLC analysis demonstrated that 20-COOH-AA was the only distinct radiolabeled product (data not shown).

DISCUSSION
These findings demonstrate that cerebral microvessels, as well as smooth muscle and endothelial cells cultured from the microvessels, can oxidize 20-HETE to 20-COOH-AA. Although the cells also converted 20-COOH-AA to chain-shortened dicarboxylic acids, most likely through ␤-oxidation (19), this was a relatively slow process. As a result, much of the 20-COOH-AA accumulated in the extracellular fluid where it presumably would modulate endothelial and smooth muscle vasoregulatory signaling pathways. Previous work demonstrated that 20-COOH-AA and 20-HETE have similar effects on ion

Kinetic constants for purified alcohol dehydrogenases acting on 20-HETE
The steady-state kinetic constants were determined with 83 mM potassium phosphate, 40 mM KCl, and 0.25 mM EDTA buffer (pH 7.3), 37°C, with 1.0 mM NAD ϩ and varied 20-HETE concentrations by using an SLM 4800 fluorimeter to detect production of NADH. For the mouse enzyme, 0.5 mM NAD ϩ was used, and the buffer contained 0.02% acetone and 0.02% Tween 80. The ethanol contained in the commercial 20-HETE was removed by evaporation under nitrogen, dissolving the 20-HETE in acetonitrile, and re-evaporation under nitrogen. Initial velocities were fitted with the HYPER program (55). Kinetic constants are apparent for the ADH4 enzymes because the coenzyme concentration was not fully saturating. The turnover numbers, V/E t , are based on active site concentrations, which were calculated for ADH4 by reference to a standard assay (36,39) and for ADH3 by using the turnover number of 0.22 s Ϫ1 for 12-hydroxydodecanoic acid (40). Two NADHs should be produced for each 20-HETE oxidized to 20-COOH-AA, but the velocity is based on moles of NADH produced.
Enzyme  transport in the rabbit kidney loop of Henle (7), implying that 20-COOH-AA might augment some of the functional effects of 20-HETE. In this context, the 20-COOH-AA formed in cerebral microvascular tissue might potentiate the vasoconstriction initiated by 20-HETE in the cerebral circulation (5,23). An alternative possibility, suggested by the fact that 20-COOH-AA produces relaxation of porcine coronary microvessels (19), is that its function in the cerebral circulation is to modulate or terminate the vasoconstriction produced by 20-HETE. Because of the potential role of 20-COOH-AA in vascular regulation, it is important to determine the enzymatic pathway that converts 20-HETE to 20-COOH-AA. Several of our findings suggest that ADH4 may play a major role in this process in the cerebral microvasculature, in agreement with the observation that ADH4 oxidizes medium and long chain -hydroxy fatty acids (24,25). ADH4 was detected in the MSMC and MEC cultures by Western blotting, a finding that is in agreement with reports that ADH4 is the principal ADH isoform present in rat blood vessels and microvascular endothelium (26,27). The conversion of 20-HETE to 20-COOH-AA was inhibited in both cells by relatively low concentrations of N-heptylformamide, an analogue of the aldehyde substrate of ADH4 and a potent uncompetitive inhibitor of this enzyme (36,41,42). Furthermore, all-trans-retinoic acid, an ADH4 inhibitor, and all-trans-retinol, a good substrate for mouse, rat, and human ADH4 (24,25,41,43,44), decreased the conversion of 20-HETE to 20-COOH-AA in the MSMC. Finally, purified recombinant mouse ADH4 oxidized 20-HETE, producing both the intermediate aldehyde and the end product, 20-COOH-AA. This finding is consistent with the fact that ADH not only reversibly oxidizes alcohols to aldehydes, but also can readily oxidize a hydrated aldehyde to the acid with NAD ϩ (45,46). Although an aldehyde dehydrogenase may participate in the conversion of 20-HETE to 20-COOH-AA under physiological conditions, our data indicate that it is not required and that mouse and human ADH4, human ADH3, and horse liver ADH1 can catalyze the two-step reaction.
The kinetic parameters suggest that mouse ADH4 efficiently oxidizes 20-HETE, at least to the aldehyde, even though the apparent K m value of 35 M for 20-HETE calculated from the kinetic analysis is probably far above the physiological concentration range. Because of the high 20-HETE concentrations used for the kinetic studies, Tween 80 was included to ensure solubility. Tween 80 is a competitive inhibitor of the oxidation of retinoids, raising the apparent K m value by a factor of 10 at 0.02% (47). Nevertheless, the catalytic efficiency of ADH4 acting on 20-HETE (V/K m ϭ 21 mM Ϫ1 s Ϫ1 ) is 50 times higher than with all-transretinol and 2600 times higher than with ethanol (36), suggesting that the activity of ADH4 on 20-HETE may be physiologically relevant. Furthermore, the kinetic data indicate that human ADH4 has a high catalytic efficiency on 20-HETE, similar to its efficiency on -hydroxy acids and retinoids (25,48).
ADH3 also was expressed in the MSMC and MEC, and the kinetic studies with purified human ADH3 indicated that the catalytic efficiency for 20-HETE was similar to that of mouse ADH4 but less than that of human ADH4. Because the sequences of human and mouse ADH3 are 92.8% identical and the substrate-binding sites are almost identical (49), these enzymes should have similar activities on 20-HETE. The oxidation of 20-HETE by ADH3 was not inhibited by N-heptylformamide. This finding is opposite that obtained with N-heptylformamide in the cultured cells, suggesting that ADH4 rather than ADH3 probably is primarily responsible for 20-HETE oxidation in the cerebral microvasculature.
The finding that 20-HETE is also a substrate for horse liver ADH1 (a class I enzyme closely similar to mouse ADH1) indicates that 20-HETE can be oxidized in the liver, which has very high levels of ADH1. However, ADH1 accounts for only 4% of total extrahepatic ADH activity in rats (50), and it seems unlikely that it would have a major role in 20-HETE metabolism in the vasculature. In support of this conclusion, 2 mM 4-methylpyrazole, a potent inhibitor of mouse ADH1 (K i ϭ 0.15 M, Ref. 41), did not decrease the conversion of 20-HETE to 20-COOH-AA in the MSMC. Because the K i for 4-methylpyrazole is 10 mM for rat ADH4 (50), little inhibition of the homologous mouse ADH4 Human ADH4 (4 g/ml) and human ADH3 (100 g/ml) were incubated with 2 M [ 3 H]20-HETE and 1 mM NAD ϩ in 50 mM potassium phosphate, 0.12 mM EDTA, buffer, pH 7.4. After incubation for 1 h at 37°C, the amount of radiolabeled 20-COOH-AA formed was determined by HPLC using the gradient system described in Fig. 9. Single chromatograms are shown. A, ADH4; B, ADH3, but similar results were obtained from a second chromatogram in each case. would be expected. Furthermore, the formation of 20-COOH-AA by the MSMC also was not decreased by 50 mM ethanol, an excellent substrate for ADH1 (K m ϭ 1.4 mM, k cat ϭ 0.65 s Ϫ1 , Ref. 24) but a very poor substrate for mouse ADH4 (K m ϭ 1.1 M, k cat ϭ 22 s Ϫ1 , Ref. 36). This difference arises in part because of the substitution of Val-294 in ADH1 with Ala in the rat and mouse ADH4 (51). Finally, ADH1 was not detected by Western blotting in either the MSMC or MEC.
A physiologically relevant concentration of ethanol has been shown to inhibit the oxidation of 20-hydroxyleukotriene (LT) B 4 to 20-carboxy-LTB 4 , and -hydroxy-N-acetyl-LTE 4 to the -carboxy-N-acetyl-LTE 4 , by ADH in rat liver and isolated rat hepatocytes (52,53). These findings establish a precedent for an ADH-dependent interaction between ethanol and eicosanoid metabolism, and the present observation that 20-HETE is a substrate for ADH1 suggests the possibility of a similar interaction between ethanol and 20-HETE oxidation in the liver. Ethanol, a poor substrate for mouse ADH4 (36), did not inhibit the conversion of 20-HETE to 20-COOH-AA in the MSMC. However, ethanol is a reasonably good substrate for human ADH4 (39,48,51), and low concentrations of ethanol could inhibit 20-HETE oxidation or facilitate the reduction of the aldehyde intermediate back to 20-HETE, as has been demonstrated with human ADH4 acting on retinol or retinal (39,54). This suggests the possibility that ethanol may decrease catabolism of 20-HETE in the human vasculature and thereby perturb vascular reactivity. In view of these possibilities, the role of ADH4 and ethanol in the metabolism of 20-HETE in human blood vessels should be assessed.