Oxidation of docosahexaenoic acid by rat liver microsomes.

[1-14C]Docosahexaenoic acid (n-3) was incubated at 37 degrees C for 30 min in the presence of rat liver microsomes and 1 mM NADPH. The products were isolated using organic solvent extractions, reverse phase, and normal phase high performance liquid chromatography. Isolates were identified using ultraviolet spectroscopy, capillary gas-liquid chromatography, and gas chromatography-mass spectrometry. The major metabolites were: 19,20-, 16,17-, 13,14-, 10,11-, and 7,8-dihydroxydocosapentaenoic acids, 22-hydroxydocosahexaenoic acid, and 21-hydroxydocosahexaenoic acid. The minor metabolites were 17-hydroxy-4,7,10,13,15,19-, 16-hydroxy-4,7,10,17,19-, 14-hydroxy-4,7,10,12,-16,19-, 13-hydroxy-4,7,10,14,16,19-, 11-hydroxy-4,7,9,13,16,19-, 10-hydroxy-4,7, 11,13,16,19-, 8-hydroxy-4,6,10,13,16,19-, and 7-hydroxy-4,8,10,13,16,19 -docosahexaenoic acids. These metabolites of docosahexaenoic acid resulted from four distinct classes of oxidation, omega-hydroxylations, (omega-1)-hydroxylations, epoxidations, and lipoxygenase-like hydroxylations. The similarity of these product profiles to those reported for comparable microsomal incubations with other essential fatty acids suggest that microsome cytochrome P-450 monooxygenases were involved.

Saturated fatty acids also undergo monooxygenase catalysis (w-1)-oxidations (9,12,14). The initial product, a secondary *This work was supported in part by Grants NS09199 and RR01152 from the National Institutes of Health. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Unsaturated fatty acids are converted to epoxides by P-450 monooxygenases in addition to the w -and (a-1)-oxidation products (25-29). The epoxide pathway appears to be a major microsomal oxidation route for polyunsaturated fatty acids pretreated with phenobarbital (27-30), which leads to the appearance in animals of a variety of vicinal diols via hydrolysis of epoxides by the microsomal and cytosolic hydrolases (27).
Docosahexaenoic acid is the most unsaturated fatty acid normally occurring in mammals and fish (35). It is found in highest concentration in brain, retina, and testis. The concentration ratios of docosahexaenoic acid to arachidonic acid are 1.4 in rat brain (36), 2.4 in rat retina (35), and 0.6 in human testis (37). The concentration of docosahexaenoic acid increases with age, while that of arachidonic acid decreases in gray matter of the adult human brain (38). The source of brain docosahexaenoic acid is either dietary or from docosahexaenoic acid synthesized in liver from dietary linolenic acid (39). Thus liver plays a central role in making docosahexaenoic acid available for peripheral tissues. In mammals the dietary requirement for n-3 essential fatty acids is very low (less than 40 mg/kg of food) in contrast to the relatively large dietary requirements for n-6 essential fatty acids (35). For this reason, it has been postulated that a metabolite of docosahexaenoic acid, required in only low concentrations, may be the basis for docosahexaenoic acid's dietary "essentiality."

Oxidation of Docosahexaenoic Acid by Hepatic Microsomes 5777
Docosahexaenoic acid is a potent inhibitor of mammalian cyclooxygenases, but a poor substrate for cyclooxygenase and lipoxygenase enzymes (40, 41); it is, moreover, a potent inducer of hepatic monooxygenase (34). The present study was designed to determine if liver microsomes could oxidize docosahexaenoic acid and to identify the metabolites produced.

MATERIALS AND METHODS
Preparation of Hepatic Microsomes-Three groups of 3 male Sprague-Dawley rats (240-270 g) were given daily intraperitoneal injections for 4 days of either 2.5 ml of trioctoylglycerol, p-naphthoflavone (12.0 mg dissolved in 2.5 ml of trioctoylglycerol), or phenobarbital (20.2 mg dissolved in 2.5 ml of trioctoylglycerol). After the fourth injection, each animal was deprived of food overnight.
The rats were decapitated 24 h after the last treatment and livers were removed. All subsequent steps were performed at 0-4 "C. The livers were rinsed in isotonic saline, weighed, and minced. Each liver was homogenized (0.25 g/ml) in a buffered salt solution consisting of 150 mM KC1 and 50 mM Tris-HC1 (pH 7.5) using a loose fitting Teflon pestle. Combined homogenates from livers of the three animals in each treatment group were mixed and centrifuged at 10,000 X g for 10 min. The supernatant was then centrifuged at 100,000 X g for 60 min. The microsomal pellets were washed once, resuspended in buffer (the equivalent of 0.5 g of liver/ml), and used rapidly in the incubations.
Incubation Conditions-Microsomes (approximately 50 mg of protein/5 ml of buffer) were added to 45 ml of buffer being shaken at 37 "C. Three 50-p1 aliquots containing (a) 500 pmol of glucose 6phosphate, (b) 165 units of glucose-6-phosphate dehydrogenase (EC 1.1.1.49), and (e) 50 pmol of NADPH were added. After 1 min, 5 pmol of sodium ['4C]docosahexaenoate dissolved in 200 pl 0.1 M Tris-NaOH (pH 8.5) were added. After 30 min of shaking at 37 "C in an open 250-ml flask, each 50 ml of incubated microsomes were mixed with 20 ml of ice-cold methanol. The total number of 50-ml incubates was two, six, and six for C, BNF, and PB, respectively. Control incubations using boiled (3 min) microsomal fractions and incubations in which NADPH was excluded were carried out in 5-ml volumes in triplicate.
Extraction Procedures-Ethanol-quenched reaction mixtures were centrifuged at 1000 X g for 10 min at 4 "C. Pellets were suspended in 80% ethanol and recentrifuged. The combined supernatants were then concentrated to near dryness. Further sample concentration was achieved using several methanol and chloroform washes, a procedure also known to improve protein removal (43). Samples were transferred from round bottom flasks using water followed by ethyl acetate. After adjusting the aqueous phase to pH 3 with HCl, an equal volume of ethyl acetate was added. The isolated aqueous phase was reextracted with equal volumes of ethyl acetate. The combined ethyl acetate extracts were then backwashed with water (20% by volume) to remove residual HCl. Samples were concentrated, filtered (MF-1 Centrifugal Microfilter, Cole Scientific, Calabasas, CA), and stored in 1 ml of methanol at -80 "C until RP-HPLC could be performed.
For quantitative analysis of radioactivity distribution, each sample was injected in 50 pl of methanol and developed (1.6 ml/min) using an initial mobile phase of (43:57) acetonitri1e:aqueous phosphoric acid (pH 2.3). A linear gradient of 1.43%/min of acetonitrile was applied 85 min after injection. The HPLC effluent was sequentially monitored at 229 and 192 nm. Fractions were collected every 0.5 min and radioactivity measured by scintillation counting techniques. The Same RP-HPLC conditions described above were used for preparative runs. However, entire peaks containing radioactivity and absorbance at 192 nm were collected. The fractions were neutralized with excess NaHC03 and acetonitrile was removed in uocuo. The water residue was acidified to pH 3 with HCl, and 4 volumes of CHZCL were added. After centrifugation, the isolated organic extracts were concentrated and methylated using methanol-ethereal diazomethane (kg).
Derivatization for Gas Chromatography-Methyl esters, isolated following NP-HPLC, were either directly silylated or catalytically hydrogenated with 5% rhodium on alumina' and then silylated. Silylation was done by heating the sample 60 "C for 45 min after it was dissolved in 50 pl of acetonitrile plus 50 pl of bis(tri-methylsily1)trifluoroacetamide (Pierce Chemical Co.).
Capillary GC-The equivalent chain length of each metabolite was determined by measuring its retention time using capillary GC and relating the log of this retention time to the log retention times of 24:0, 260, and 280 fatty acid methyl esters. A nonpolar (DB-I, J & W, Rancho Cordova, CA) column (0.25 mm (inner diameter) X 30 m with 0.25 pm film thickness) and a "falling needle" all glass injector (R. A. Allen Co., Boulder, CO) were employed.
The carrier gas (helium) was adjusted to a linear velocity of 20 cm/s for the isothermal (240 "C) separations.
GC-Mass Spectrometry-Electron impact (70 eV) studies were conducted using a quadrupole mass spectrometer (Model 3200, Finnigan, Sunnyvale, CA) equipped with a Technivent data system (St. Louis, MO). The glass column (2.5 mm X 3 ft) used was silanized and then packed with 3% OV-101 on 80/lOO Supelcoport (Supelco). With the injector temperature at 260 "C, column temperature was increased at a rate of 2 "C/min from 215-255 "C. The flow rate of helium carrier gas was maintained at 30 ml/min. Spectroscopic Studies-Difference spectra (45) were determined using a Cary 18 spectrophotometer. UV spectra of isolated metabolites were examined using a Cary 14 instrument equipped with a Hamamatsu (Type R 456 HA) photomultiplier tube. Protein determinations were performed according to Lowry et al. (46). Cytochrome P-450 was quantitated using the method of Omura and Sat0 (47).

RESULTS
The average liver weights from C, BNF, and PB-treated rats were 9.30, 9.50, and 12.37 g, respectively. The specific contents of P-450 were 2.2 (C), 2.7 (BNF), Together the above findings indicated that the expected drug related inductions of the P-450 system had occurred and that 22:6 had a high affinity for the P-450 cytochrome system.
The average recovery of radioactivity from microsomes after ethanol precipitation and ethyl acetate extraction was 55% (C), 47% (BNF), and 46% (PB). The low recoveries of radioactivity presumably resulted from 22:6 incorporation into phospholipids which were not extracted into 80% ethanol.

I40
Similar 50% recoveries of radioactivity were reported for 204 when incubated and extracted in like manner (25). Boiled microsomes in which no phospholipid metabolism would be expected, resulted in over 94% recovery of the radioactivity added to the incubates. Table I summarizes the group distributions of radioactivity found with RP-HPLC when ['*C]22:6 was incubated with microsomes isolated from rats with different treatments. These numbers represent average values from 2-6 duplicate incubations. As expected from the above measurements on 226 oxidation rates, the amount of [14C122:6 remaining was highest in controls. Also, as reported previously for 204, Group B m e~~l i~s were highest for BNF-induced microsomes (27) and Group A and C metabolites were highest for PB-induced microsomes (27-33). The ratios of compounds I/ I1 + III/IV/V (see below) in Group A were 4:34:1 (C), 53:3:1 (BNF), and 5:54:1 (PB). Thus minor quantitative differences were suggested when 22:6 was metabolized by microsomes isolated from animals with different drug treatments.
However, drug treatments did not elicit qualitative differences in the way 22:6 was metabolized. The elution profiles from RP-HPLC of the ethyl acetate extracts, as monitored by radioactivity or UV absorption, were identical between all treatment groups. Therefore, ethyl acetate extracts from the three treatment groups were pooled prior to preparative HPLC in order to maximize product yield for identification purposes.
At least 23 radioactive peaks were resolved from 22:6 by RP-HPLC (Fig. 1). The amount of radioactivity in each individual peak was closely paralleled by the magnitude of UV absorption at 192 nm with two exceptions. For radioactivity:absorption at 192 nm, ratios for compounds with retention times of 54 min and 105 min were lower than the observed average. It is interesting that radioactive peaks eluting between 52-86 min and between 118-132 min revealed marked increases in the 229:192 absorption ratio (Fig. l), suggesting the possible existence of conjugated dienes in these moieties.
Individual fractions separated by RP-HPLC were further resolved using NP-HPLC. Here again changes in radioactivity and UV absorption at 192 nm revealed a close parallelism. However, sensitivity to absorption at 192 nm was dramatically decreased with NP-HPLC, presumably because of the high absorptivity of isopropyl dcohol at 192 nm.  Group A Metabolites-NP-HPLC using isopropyl alcohokhexane (1:lOO) was performed to further resolve RP-HPLC isolates which eluted between 18-34 min (Fig. 1). The five components in Group A, together representing 8.5% of the radioactivity recovered after RP-HPLC (Table 11), have an identical UV spectrum ( Fig. 2A). However, the mass spectra of the trimethylsilyl ether methyl esters of these five compounds and their hydrogenated derivatives were distinctive. The electron impact spectrum of C o~~ I (Fig. 3A) revealed   * ~d i~c t i v i t y is presented as percentage of total radioactivity recovered from ODS columns.
These were determined on methyl ester trimethylsilyl ether derivatives. Numbers in parentheses represent values for corresponding hydrogenated derivatives. e Percentage of radioactivity was calculated from data in Fig. 1 and mass ratios obtained by capillary GC.   Fig. 1.
Radioactivity is presented as percentage of total r a d i o a~i~t y recovered from ODS columns. The distribution of radioactivity between the two peaks is determined by using the radioactivity measured after NP-HPLC.
These were determined on methyl ester tnmethylsilyl ether derivatives. Numbers in parentheses represent values for corresponding hydrogenated derivatives.
In summary, Group A metabolites consisted of five of the six possible diols which would arise if oxygen attacked each of the double bonds in 226. The retention times for these vicinal diols (Table 11) decreased with RP-HPLC and increased with capillary GC as the pair of hydroxyl groups became located closer to the w end of the molecule. Surprisingly with NP-HPLC, the retention times for the paired hydroxyl moieties increased if the pair of hydroxyl groups were positioned very close to either the w or the carbomethoxy end of the molecule.
Group B Metabolites-The Group B metabolites (42-50 min, Fig. 1) consisted of two compounds which together represented 6.4% of the radioactivity recovered after RP-HPLC (Table 111). Compounds VI and VI1 were base-line separated from each other by NP-HPLC using isopropyl alcohokhexane (0.5:lOO). After isolation it was found that they had an identical UV spectrum (Fig. 2B) which was different from that of Group A metabolites, which contained a shoulder at 215 nm (Fig. 24).
The electron impact mass spectrum of Compound VI (  Compound VI1 as 21-hydroxy-4,7,10,13,16,19docosahexaenoic acid was confirmed when the spectrum of the hydrogenated derivative was examined (Miniprint). Group C Metabolites-Individual peaks in Group C (58-86 min, Fig. 1) were further resolved by NP-HPLC using isopropyl alcohokhexane (0.21:lOO). As with the acetonitri1e:water (43:57) mixtures used for RP-HPLC, these conditions were employed because they gave optimal separations for the 10 standards of hydroxylated conjugated dienes derived from 226.5 Group C had eight metabolites which together repre-  66.4 Experimental conditions and peak numbering are described in Fig. 1. 'Radioactivity is shown as percentage of total radioactivity recovered from ODS columns. For Peaks VI11 and X the distribution of radioactivity between the A and B components was determined after NP-HPLC.
Separations were achieved isocratically using isopropyl alcohokhexane (0.21:lOO).  sented 1.2% of the radioactivity recovered from ODS columns (Table IV). NP-and RP-HPLC retention times of these eight metabolites matched those of standards generated by autoxidation? The UV spectra of the major metabolite in Group C, Compound X-B, is given in Fig. 2C. An absorption maximum at 235 confirmed the presence of a conjugated diene. The mass spectra of the hydrogenated Compound X-B also showed as the major fragments 271 and 273 verifying the presence of a trimethylsilyl group at carbon IO2. Thus Compound X -3 was identified as 10-hydroxy-4,7,11,13,16,19-d~os~exaenoic acid. The other seven metabolites (Table IV) were identified as: Cornpound VIII-A, 17-hydroxy-4,7,10,13 (Fig. l), represented 3.5% of the total radioactivity recovered after RP-HPLC (Table v). Radioactive peaks XIV-XVIII (86-114 min) absorbed only at 192 nm. In contrast, peaks XIX-XXIII (114-132 min) also revealed significant absorption at 229 nm. No further analyses were carried out on these metabolites.

DISCUSSION
The essential fatty acid docosahexaenoic acid was oxidized using NADPH-fortified microsomes from rat liver. Over 23 metabolites were found of which the first 13 were identified. The identified monools and diols represent approximately 10% of the docosahexaenoic acid added to microsomes. It is unlikely that these metabolites resulted from autoxidation considering the observation that only 0.5% of the docosahexaenoic acid incubated with boiled microsomes was oxygenated. In fact, less than one-fifth of even these autoxidation products comigrated with identified monools and diols. These metabolites were not due to autoxidation since they were almost absent when boiled microsomes were employed. The major products (Fig. 6) resulted from metabolism along three pathways: 1) w-hydroxylations, 2) (a-1)-hydroxylations, and 3) (w-3)-, (~-6 ) -, (w-9)-, (w-12)-, and (w-l5)-epoxidations, The minor products, conjugated dienes with allylic hydroxyl groups, could result from li~xygenase-like activity, but required NADPH addition for their formation, suggesting that docosahexaenoic acid was oxidized via four distinct routes in hepatic microsomes fortified with NADPH.
The microsomal cytochrome P-450 enzyme system was probably responsible for all four pathways since (a) the kinds of metabolites produced have been reported for other polyunsaturated fatty acids exposed to P-450 monooxygenases, (b) docosahexaenoic acid appeared to bind to P-450 with high affinity, and (c) drugs known to induce the P-450 system increased the amount of d~osahexaenoic acid oxidized; moreover as reported for arachidonic acid (27), pretreatment of rats with BNF appeared to stimulate the production of w-and (w-1)-hydroxylations (Group B) more than that of diols (Group A). It is interesting that pretreatment of rats with PB not only appeared to increase the production of vicinal diols but also resulted in doubling the amounts of conjugated dienes produced (Group G, Table I). The number of monooxygenases required for all the observed oxidations is unknown but it is believed that the liver contains at least 10 (50).
Microsomal epoxidases act preferentially toward the w end of unsaturated fatty acids (25,(27)(28)(29). This also held true for docosahexaenoic acid as judged by the proportions of diols formed. A 5,6-epoxide was reported to be formed in microsomal incubations with arachidonic acid (28). However in the present study no comparable 4,5-diol from docosahexaenoic acid was isolated. It is possible that the 4,5-diol of docosahexaenoic acid was converted to a y-or &lactone, and such lactonization should chromatographically remove the 4,5-diol from Group A and place it with Group D metabolites which were not analyzed.
Recently human platelets were reported to hy~oxylate docosahexaenoic acid at positions 11 and 14 (41). Rat basophils also produce trace amounts of the 4-and 7-hydroxy isomers from docosahexaenoic acid (40). Although the biological actions of these compounds are not yet established, it is known t h a t p a t h o~o~e a~ states which increase free arachidonic acid concentrations also elevate free docosahexaenoic acid concentrations (51, 52). Thus increased enzymatic oxidations of docosahexaenoic acid might be expected to accompany the increased hydroxylation of arachidonic acid typically found with increased levels of free arach~donic acid (53). The biological actions of arachidonic acid HETEs, their precursors, and products, assuming different shapes, are currently the subject of intense investigations (54 Arachidonic acid epoxides and diols, the production of which is mediated by microsomal mixed function oxidases, have recently been shown to be potent agonists for the release of luteinizing hormone from anterior pituitary cells (55). Furthermore, factors which enhance P-450 oxidations increase irreversible binding of arachidonic acid metabolites to proteins with possible carcinogen implications (26). Thus, examination of the 15 novel metabolites derived from docosahexaenoic acid for biological activity should be of interest, especially in tissues rich in docosahexaenoic acid such as brain, retina, and testis. Perhaps the "essentiality" of the n-3 fatty acids may be due to one or more of these metabolites.