A Linoleic Acid (8R)-Dioxygenase and Hydroperoxide Isomerase of the Fungus Gaeumannom yces graminis BIOSYNTHESIS OF (8R)-HYDROXYLINOLEIC ACID AND (7S78S)-DIHYDROXYLINOLEIC ACID FROM (8R)-HYDROPEROXYLINOLEIC ACID*

The fungus Gaeumannomyces graminis metabolized linoleic acid extensively to (8R)-hydroperoxylinoleic acid, (8R)-hydroxylinoleic acid, and threo-(7S,SS)-dihydroxylinoleic acid. When G. graminis was incu- bated with linoleic acid under an atmosphere of oxy-gen-18, the isotope was incorporated into (8R)-hy- droxylinoleic acid and 7,s-dihydroxylinoleic acid. The two hydroxyls of the latter contained either two oxy- gen-18 or two oxygen-16 atoms, whereas a molecular species that contained both oxygen isotopes was formed in negligible amounts. Glutathione peroxidase inhibited the biosynthesis of 7,s-dihydroxylinoleic acid. These findings demonstrated that the diol was formed from (8R)-hydroperoxylinoleic acid by intramolecular hydroxylation at carbon 7 , catalyzed by a hydroperoxide isomerase. The (8R)-dioxygenase appeared to metabolize substrates with a saturated carboxylic side chain and a 92-double bond. G. graminis also formed

in view of the many recently discovered transformations of linoleic acid to oxylipins in plants and algae, e.g. to hydroperoxides, epoxy alcohols, allene oxides, and the plant growth regulator jasmonic acid (3)(4)(5)(6).
The 100,000 x g supernatant of G. graminis was recently found to metabolize linoleic acid to (8R)-HODE,l 16-HODE, 17-HODE, and polar metabolites, which were not characterized (7). The hydroxylation of fatty acids at the wl-, w2-, or o3-position is catalyzed by cytochrome P-450 in mammals, microorganisms, and fungi. The enzymes have been wellcharacterized (8)(9)(10). In contrast, little is known about the biosynthesis of (8R)-HODE and its biological significance. CY-Linolenic acid was also metabolized to its 8-hydroxy metabolite, but arachidonic acid and eicosapentaenoic acid were not (7).
The fungus Lmtisaria arvalis was found by Bowers et al. (11) to secrete a substance that is fungicidal to some strains of pathogenic fungi. This substance was identified as 8-HODE, and it was confirmed that authentic (8R,8S)-HODE possesses antifungal activity (12).
The objective of this study was to determine the mechanism of biosynthesis of (8R)-HODE from 182n-6 by G. graminis and to determine the structure and biosynthesis of the polar metabolites.
Cultivation of Fungus-G. graminis was cultivated a t room temperature (22 "C) in culture broth, which contained, per liter, 20 g of dglucose and 5 g each of NaCl, K2HP04, Bacto-Soytone, and yeast extract. Casamino acid medium was prepared similarly, but 5 g of casamino acids replaced the yeast extract. The medium was stirred gently, and the fungus was typically grown for at least 5 days under aerobic conditions. The fungus was harvested by filtration through a nylon net (100 pm). The mycelia were washed with saline, weighed wet, and either used directly or stored at -80 "C.
Preparation of Supernatants-The mycelia of C. graminis were sonicated, and the 100,000 X g supernatant was obtained by differential centrifugation (7). Alternatively, the frozen mycelia were minced in a mortar (at +4 "C) with A1203 (half the weight of the fungus) (14) and -1.5 volumes of buffer (0.05 M Tris-HCI, pH 7.4, with 1 mM EDTA and 1 mM EGTA). The material was centrifuged (10,000 X g, 10 min at 4 "C), and the supernatant (0.65 mg of protein/ ml) was used. Incubation with Mycelia-The mycelia were incubated in 0.1 M sodium borate buffer, pH 8.2, with the fatty acids (added in a small volume of ethanol) for 2 h at room temperature (7). Mycelia (0.5-10 g , wet weight) were incubated with 7-50 mg of the fatty acid in 7-150 ml of borate buffer. In small-size experiments, only 30 mg of mycelia were incubated with [l-"C]182n-6 (0.4 pCi or 2 pg of exogenous 182n-6) in 0.8 ml of borate buffer (30 min was usually sufficient for extensive metabolism of the substrate). Incubations under an atmosphere of "0, were performed after repeated evacuation and purging with nitrogen. The incubations were terminated by filtration, and the medium was acidified to pH 3 or, in some experiments, to pH 5 and extracted either with ethyl acetate or on a cartridge of octadecasilane silica as described (7). Over 20 incubations were performed with 18:2n-6 and mycelia.
Deriuatizations-Methyl esters were prepared by treatment with a n excess of ethereal diazomethane (7). Hydrogenation was performed CP-HPLC.
GC-MS Analyses-The GC-MS analyses were performed using a capillary GC column (Varian GC 3400 with a Varian 1075 split/ splitless injector) and a nonpolar capillary GC column (30-m DB-5, J & W Scientific, film thickness of 0.25 pm, inner diameter of 0.25 mm), which was connected to an ion trap mass spectrometer (ITS40, Finnigan MAT) with electron impact ionization and computer-supported evaluation. The carrier gas was helium. After splitless injection, the GC column was programed from 60 to 275 "C with 40-28 "C/ min and then kept at 275 "C.
Steric Analyses-To determine the relative configuration of the diol structure of 7&dihydroxylinoleic acid, an aliquot was converted into the cyclic carbonate derivative (17) and subjected to oxidative ozonolysis. The esterified product was analyzed by gas chromatography using a Hewlett-Packard Model 5890 gas chromatograph equipped with a methyl silicone capillary column (length of 25 m; film thickness of 0.33 pm; carrier gas, helium a t a flow rate of 25 cm/ s). The cyclic carbonate derivatives of dimethyl erythro-and threo-2,3-dihydroxyheptane-1,7-dioates and dimethyl erythro-and threo-2,3-dihydroxyundecane-l,ll-dioates (18) were used as references. Steric analysis of the methyl ester of 10-HODE was carried out by oxidative ozonolysis performed on the (-)-menthoxycarbonyl derivative (16), followed by capillary gas chromatography using a column of DB-210 (length of 15 m, film thickness of 0.25 pm, J & W Scientific). The (-)-menthoxycarbonyl derivatives of dimethyl ( R )and (S)-malates were used as references. The absolute configuration of 17-HODE methyl ester was determined after hydrogenation by GC-MS analysis of its (S)-phenylpropionic acid derivative as described (19). The enantiomers of 9-HODE and 13-HODE methyl esters were separated by CP-HPLC as described (20).
Other Analyses-Radioactivity was determined by liquid scintillation (Beckman LS2800) using ACS (Amersham International) as a scintillation mixture. Protein concentrations were determined as described by Bradford (21) with bovine albumin as a standard.

RESULTS
Metabolism of 18:2n-6 by G. graminis: Separation of Metabolites by RP-HPLC and TLC 182n-6 was extensively metabolized both by the cell-free supernatant and by mycelia to monohydroxy and dihydroxy compounds. In both cases, two of the main metabolites were (8R)-HODE and 7,8-DiHODE, which are identified below. Mycelia also converted 182n-6 to w2-and w3-hydroxy metabolites as well as to w2-and w3-hydroxy metabolites of (8R)-HODE. These metabolites were also formed by cell-free supernatants and NADPH.
On TLC analysis of methylated products formed by a largescale incubation of 18:2n-6 with mycelia showed six major bands in addition to the unmetabolized substrate. RF values are given in Table I. The most polar band contained 8,17-DiHODE methyl ester, the next 7,8-DiHODE and 8,16-Di-HODE methyl esters, the third 17-HODE methyl ester, the fourth 8-HODE methyl ester and small amounts of slightly less polar products (16-HODE, 10-HODE, and 11-HODE methyl esters), the fifth 9-HODE and 13-HODE methyl esters, and the sixth, least polar band 8-HPODE methyl ester, Bands 4-6 were only partly resolved. The monohydroxy metabolites of one representative incubation were roughly quantified by integration of a gas chromatogram. 8-HODE consti- tuted -35% of the material recovered, 10-HODE -6%, 11-HODE -3%, and 16-HODE and 17-HODE each -25%. The total yield of monohydroxy fatty acids was -15-20% as judged from TLC. In contrast to this complex mixture of metabolites formed by mycelia, incubation of the cell-free supernatant with ["C] 18:2n-6 (and without NADPH) resulted in only three bands of radiolabeled metabolites on TLC as shown in Identification of (BR)-Dioxygenase Metabolites of 18r2n-6 (8R)-HPODE Methyl Ester-This compound was identified in band 6 on TLC. Treatment with SnClz changed its polarity on TLC to that of (8R)-HODE methyl ester. GC-MS analysis of the same amount of material of band 6 before and after reduction with SnClz and derivatization is shown in Fig. 2. Chemical reduction with SnCI, increased the amount of (8R)-HODE methyl ester dramatically.
(8R)-HODE Methyl Ester-This compound was identified in band 4. The mass spectrum (Me3Si ether methyl ester derivative) was as previously reported (7). In addition, some GC-MS analysis showed the presence of an isomer with a slightly larger C-value (Fig. 2 4 ) . The mass spectrum of this compound and (8R)-HODE methyl ester was almost identical, and it was tentatively identified as a 2,E-isomer of 8-HODE methyl ester. Incubation under an atmosphere of oxygen-18 led to the incorporation of this isotope into 8-HODE as shown by GC-MS analysis (Fig. 3A). The incorporation of oxygen-18 into (8R)-HODE was -48% (average value calculated from the intensity of several structural ions). It is noteworthy that (8R)-HODE also is an endogenous constituent of the fungus (see below). 7,8-DiHODE Methyl Ester-The material in band 2 was derivatized to its Me& ether methyl ester and analyzed by mass spectrometry. The major metabolite of band 2 was identified as 7,8-DiHODE methyl ester, but this band often also contained smaller amounts of 8,16-DiHODE methyl ester (see below); the two compounds were separated by capillary gas chromatography. The mass spectrum (Fig. 4A)  When 7,8-DiHODE was isolated during an experiment under oxygen-18, its hydroxyls contained either two atoms of l60 (56%) or two atoms of "0 (44%), but hardly any species with both isotopes could be detected (Fig. 3B). This finding indicated that the two hydroxyls were derived from the same molecule of oxygen, either from 1 6 0 2 or from Furthermore, the glutathione peroxidase concentration dependably inhibited the biosynthesis of 7,8-[14C]DiHODE from ["C] 18:2n-6 with an ICbo of -0.03 unit/ml (Fig. 1B).
Both the oxygen-18 experiment and the effect of glutathione peroxidase were consistent with 7,8-DiHODE being derived from the 8-hydroperoxide of 18:2n-6, which also is the precursor of (8R)-HODE.
Oxidative ozonolysis performed on the cyclic carbonate derivative of the methyl ester of 7,8-DiHODE followed by esterification resulted in the formation of the cyclic carbonate derivative of dimethyl 2,3-dihydroxynonane-1,9-dioate. Identification was based on GC-MS using the cyclic carbonate derivatives of dimethyl erythro-and threo-2,3-dihydroxyhep-  (Table I)  IO-HODE Methyl Ester-This compound was one of the many compounds in the upper half of band 4, and it could be purified from (8R)-HODE methyl ester (but not from 16-HODE methyl ester) by repeated TLC. The mass spectrum is shown in Fig. 5A (left). The hydroxyl oxygen was derived from the atmosphere as shown by the incorporation of 43% oxygen-18 (Fig. 5A, right). Hydrogenation yielded 10-hydroxyoctadecanoic acid (Fig. 5B). Ozonolysis of the (-)-menthoxycarbonyl derivative of 10-HODE methyl ester yielded the (-) -menthoxycarbonyl derivative of malate. The absolute configuration was 65% @)-malate and 35% (S)-malate. Ozonolysis also determined the position of the double bonds, and the metabolite was thus identified as 10-hydroxyoctadeca-8,12-dienoic acid.
11-HODE Methyl Ester-This compound could be isolated from incubations, which were extracted at pH 5-6. At pH 3, 11-HODE is rearranged to racemic 13-HODE and 9-HODE.' The structure of 11-HODE was determined from GC-MS analysis (Me3Si ether methyl ester derivative) before and after hydrogenation (  18)). These data were consistent with those from the original description of 13-HODE and 9-HODE Methyl Esters-13-HODE methyl ester and its E,E-isomer were the main constituents of band 5 , which also contained smaller amounts of 9-HODE methyl ester (and in some experiments, its E,E-isomer). 9-HODE and 13-HODE methyl esters were not separated by capillary gas chromatography, but were resolved by SP-HPLC. CP-HPLC showed that these compounds were nearly racemic. Furthermore, relatively little oxygen-18 (<20%) was incorporated into 13-HODE, its E,E-isomer, and 9-HODE. This was in contrast to 43-47% oxygen-18, which was incorporated into (8R)-HODE, 10-HODE, or 7,8-DiHODE in the same experiment. This indicated that 9-HODE and 13-HODE could have been formed by autoxidation of 18%-6 during the isolation procedure or by another mechanism. TLC analysis of 18:2n-6 showed no evidence of autoxidation products, but small amounts cannot be excluded (13) and could also be formed after termination of the experiment. Nevertheless, this finding led to the attempts described above to isolate 11-HODE, which can be rearranged nonenzymatically to 9-HODE and 13-HODE as discussed above.
ll-HODE.* Substrate Requirements of 18:2n-6 (8R)-Dioxygenase and Hydroperoxide Isomerase The substrate requirements were assessed by incubation of mycelia with a series of CIS and Czo fatty acids. The products were first analyzed by TLC and then by GC-MS. 18%-6 appeared to be more efficiently metabolized by the (8R)dioxygenases than by the other substrates as judged from TLC. Practically all fatty acids were metabolized by w2-and/ or w3-hydroxylation, and these metabolites were identified by 9Z-I8:ln-9-Metabolites of oleic acid, which could be separated by TLC into three bands, were identified by their mass spectra as 7,8-DiHOME (RF value of 0.12, C-value of 20.8), 17-HOME (&value of 0.31), and 8-HOME (RF value of 0.37, C-value of 19.0). 7,8-DiHOME (Me3Si ether methyl ester derivative) had a mass spectrum similar to that of 7,8-Di-HODE, but some ions were increased by 2 mass units, e.g. at  /z 73, 113, 147, 171, 217, and 269).
7-Hydroxyricinoleic acid methyl ester was tentatively identified as a less polar metabolite than 8- The endogenous hydroxy fatty acids were not studied systematically, but large amounts of (8R)-HODE and (7S,8S)-DiHODE were present in the fungus (extracted mycelia or the 100,000 X g supernatant) and were determined by GC-MS. Smaller amounts of 8-HOME, 7,8-DiHOME, and 7,8-DiHOTrE were also found to be endogenous products of the fungus and could even be isolated from the growth medium after harvesting of the fungus. Our culture broth, based on yeast extract and hydrolyzed soy meal, presumably provided the fungus with 18ln-9, 18%-6, and 183n-3.

Reproducibility
The enzymatic activity of the two strains of G. graminis (CBS 903 and 904) were compared for (8R)-dioxygenase activity with 18:2n-6 as a substrate. As judged from TLC, no qualitative difference between them could be noticed. (8R)-HODE and (7S,8S)-DiHODE were formed by both strains.
The w2-and w3-hydroxylase activities of the same strain of G. graminis (CBS 903) varied considerably with time. Similar observations with cytochrome P-450 in fungi have been reported by many investigators (10,14). This could be due to variations in oxygen tension, in the composition of the culture broth, and of the growth phase of the fungus (10). In one experiment with mycelia, 17-HODE and 16-HODE were very prominent relative to the formation of (8R)-HODE, possibly due to a technical mistake in the preparation of the culture broth (discolored due to a Maillard reaction). G. graminis was also grown on a more defined medium that was based on casamino acids instead of yeast extract. Under these conditions, G. graminis metabolized 18:2n-6 to (8R)-HODE and (7S,8S)-DiHODE as the main metabolites.
The formation of 11-HODE appeared to be variable and so was the formation of 13-HODE and 9-HODE. 8-HPODE, as TLC band 6, was noticed in the majority of incubations of mycelia with 18:2n-6.

DISCUSSION
The main finding of this study is that (8R)-HODE is formed from an intermediate hydroperoxide, (8R)-HPODE. This hydroperoxide may also be isomerized enzymatically to (7S,8S)-DiHODE. This metabolism of 18%-6 by G. graminis is summarized in Fig. 6. The sequence was deduced from Racemic 8-HODE was reported to possess antifungal activity, and the fungicidal effect of L. arvalis was attributed to its biosynthesis of 8-HODE (11). The mechanism of biosynthesis of 8-HODE by L. arvalis was not determined, but it seems possible that it is also derived from (8R)-HPODE. This study suggests that (8R)-HPODE, and possibly also (7S,8S)-DiHODE, should be analyzed for fungicidal and other biological activities. G. graminis may produce (8R)-HPODE for the same objective as L. arvalis, namely as a defense against certain fungi or other microorganisms.
The (8R)-dioxygenase differs from lipoxygenases in its mechanism of oxygenation. Lipoxygenases abstract a hydrogen from one carbon atom and insert molecular oxygen after radical migration, whereas (8R)-dioxygenase catalyses both reactions at the same carbon. Lipoxygenases have different substrate requirements. In short, lipoxygenases require a fatty acid with one or more 1,4-2,2-pentadiene systems as a substrate (23). The first step is hydrogen abstraction from the methylene carbon atom, followed by the formation of conjugated 2,E-double bonds, insertion of molecular oxygen, and the formation of a hydroperoxide. In contrast, (8R)-dioxygenase metabolizes fatty acids with a 92-double bond and a saturated carboxyl side chain. Molecular oxygen is apparently inserted at carbon 8 after abstraction of a hydrogen, and this appears to take place without any change in the position or configuration of the adjacent double bond. This mechanism of oxygenation shows a superficial similarity to hydroxylations catalyzed by cytochrome P-450, but with the noticeable exception that the cytochrome P-450 hydroxylases are monooxygenases and not dioxygenases (8,24). It would be interesting to investigate whether the amino acid sequence of linoleic-acid (8R)-dioxygenase shows any homology to lipoxygenases or cytochrome P-450 enzymes.
A reviewer made us aware of a mechanistically interesting analogy between the (8R)-dioxygenase and prostaglandin H synthase. The first step in prostaglandin biosynthesis from 20:3n-6 is abstraction of the pro-S (and bisally1ic)-hydrogen from the fatty acid (25). Mead acid, 20:3n-9, is not converted t o prostaglandins by prostaglandin H synthase, but it is slowly transformed to the 13-hydroxy metabolite of 20:3n-9 in the presence of calcium (26). This reaction has not been investigated in detail, but it presumably involves abstraction of the allylic hydrogen at C-13, dioxygenation, and then reduction to an alcohol.
The (8R)-dioxygenase showed a rather strict substrate requirement. First, 18:ln-9, 18:2n-6, and 18:3n-3 were substrates provided that their carboxyl groups were not esterified. Stearic acid and elaidic acid were not substrates. Double bonds distal to the 92-double bond did not inhibit the enzyme, but proximal double bonds appeared to do so. 18:3n-6 was not a substrate, and neither was 20:3n-6, 20:4n-6, or 20:5n-3. Ricinoleic acid was metabolized to 8-hydroxyricinoleic acid, indicating that its (12R)-hydroxy group did not inhibit the (8R)dioxygenase. The position of the 92-double bond appeared not to be critical since cis-vaccenic acid (18ln-7) was metabolized to 10-hydroxy-cis-vaccenic acid by G. graminis, presumably by the (8R)-dioxygenase. However, neither the 10-hydroperoxide of cis-vaccenic acid nor the 8-hydroperoxide of ricinoleic acid appeared to be a substrate for the hydroperoxide isomerase as the corresponding diols could not be detected. This could indicate that (8R)-dioxygenase and hydroperoxide isomerase have slightly different substrate requirements, a point that will need further evaluation.
In agreement with previous reports, G. graminis also contained prominent w2-(R)-and w3-(R)-hydroxylase activities ( 2 , 7, 22), and all fatty acids of this study were metabolized by these hydroxylases. These products are likely to be formed by cytochrome P-450 and have been described in a large number of fungi (10). Certain oxygenated metabolites of polyunsaturated fatty acids can be formed by autoxidation, including (8R)-HODE. Autoxidation of 18:2n-6 will result in formation of small amounts of racemic 8-H(P)ODE (and 14-H(P)ODE), -1% of the amount of 2,Eand E,E-isomers of 13-H(P)ODE and 9-H(P)ODE (6,27). Other oxygenated products of 18:2n-6, which are not formed by lipid peroxidation in significant amounts, are 10-HODE and 11-HODE. These two compounds were isolated during incubations with G. graminis and were therefore likely to be formed enzymatically, although in relatively small amounts.
10-HODE was found to contain atmospheric oxygen in its hydroxyl. Biosynthesis of 10-HPODE as well as enzymatic cleavage products of the hydroperoxide has been described in the mushroom Psalliota bispora (28,29), but these metabolites could not be detected in G. graminis. The mechanism of biosynthesis of 11-HODE in G. graminis is unknown. (11R)-HODE was originally described as a metabolite of 18:Zn-6 in a red alga, Lithothamnion corallioides,' which does not contain (8R)-dioxygenase. Nevertheless, it is possible that this unstable metabolite in G. graminis might be a minor side product of the (8R)-dioxygenase that does not show an absolute substrate and position specificity as discussed above. However, 11-HODE was of particular interest for another reason. 11-HODE is unstable at an acidic pH, where it is rearranged to racemic 9-HODE and 13-HODE.' The biosynthesis of 11-HODE could therefore to some extent explain why 13-HODE and 9-HODE, which were isolated from an incubation under oxygen-18, contained less oxygen-18 than the other oxygenated metabolites and were nearly racemic.
Lipoxygenase metabolites of polyunsaturated fatty acids are conveniently monitored during chromatography by the typical UV absorption of conjugated double bonds (23). 8-HPODE and its metabolites lack a conjugated double bond and are therefore more difficult to detect. Whether (8R)dioxygenase or related dioxygenases are present in higher organisms cannot be determined at the present stage and is worthy of future investigations.