a-Oxidation of Fatty Acids in Higher Plants IDENTIFICATION OF A PATHOGEN-INDUCIBLE OXYGENASE (PIOX) AS AN a-DIOXYGENASE AND BIOSYNTHESIS OF 2-HYDROPEROXYLINOLENIC ACID*

A pathogen-inducible oxygenase in tobacco leaves and a homologous enzyme from Arabidopsis were recently characterized (Sanz, A., Moreno, J. I., and Castresana, C. (1998) Plant Cell 10, 1523–1537). Linolenic acid incubated at 23 °C with preparations containing the recombinant enzymes underwent a-oxidation with the formation of a chain-shortened aldehyde, i.e., 8(Z),11(Z),14(Z)heptadecatrienal (83%), an a-hydroxy acid, 2(R)hydroxy-9(Z),12(Z),15(Z)-octadecatrienoic acid (15%), and a chain-shortened fatty acid, 8(Z),11(Z),14(Z)-heptadecatrienoic acid (2%). When incubations were performed at 0 °C, 2(R)-hydroperoxy-9(Z),12(Z),15(Z)-octadecatrienoic acid was obtained as the main product. An intermediary role of 2(R)-hydroperoxy-9(Z),12(Z),15(Z)octadecatrienoic acid in a-oxidation was demonstrated by re-incubation experiments, in which the hydroperoxide was converted into the same a-oxidation products as those formed from linolenic acid. 2(R)-Hydroperoxy9(Z),12(Z),15(Z)-octadecatrienoic acid was chemically unstable and had a half-life time in buffer of about 30 min at 23 °C. Extracts of cells expressing the recombinant oxygenases accelerated breakdown of the hydroperoxide (half-life time, about 3 min at 23 °C), however, this was not attributable to the recombinant enzymes since the same rate of hydroperoxide degradation was observed in the presence of control cells not expressing the enzymes. No significant discrimination between enantiomers was observed in the degradation of 2(R,S)-hydroperoxy-9(Z)-octadecenoic acid in the presence of recombinant oxygenases. A previously studied system for a-oxidation in cucumber was re-examined using the newly developed techniques and was found to catalyze the same conversions as those observed with the recombinant enzymes, i.e. enzymatic a-dioxygenation of fatty acids into 2(R)-hydroperoxides and a first order, non-stereoselective degradation of hydroperoxides into a-oxidation products. It was concluded that the recombinant enzymes from tobacco and Arabidopsis were both a-dioxygenases, and that members of this new class of enzymes catalyze the first step of a-oxidation in plant tissue. Fatty acid hydroperoxides serve as important intermediates in the oxylipin pathway of fatty acid oxygenation in plants and fungi (1–4). Further metabolism of the hydroperoxide derivatives of linoleic and linolenic acids results in the formation of fatty acid epoxides and epoxy alcohols (5–7), dihydroxy acids (8, 9), short-chain aldehydes (10, 11), and divinyl ethers (12–15). One specific hydroperoxide isomer, i.e. the 13(S)-hydroperoxide derivative of linolenic acid, is converted into jasmonic acid by a series of reactions catalyzed by allene oxide synthase, allene oxide cyclase, reductase, and b-oxidation enzymes (2). This pathway is of biological importance in plants because it produces compounds which are involved in defense reactions against insects and other phytopathogens (16), in mechanical responses such as tendril coiling (17), and pollen development (18). A variety of conditions, such as mechanical perturbation, osmotic stress, attack by plant pathogens and wounding, elicit increased formation of jasmonates and other biologically active oxylipins in plant leaves (19). This is partly a consequence of liberation of free linolenic acid from its esterified forms (20, 21) but may also depend on increased levels of enzymes catalyzing hydroperoxide formation and metabolism. In a recent study, tobacco leaves were found to accumulate a 75-kDa protein in response to bacterial infection (22). This protein, as well as a protein from Arabidopsis showing a 75% homology in amino acid sequence, were expressed in insect cells and found to cause uptake of molecular oxygen in the presence of polyunsaturated fatty acids such as linolenic acid, linoleic acid, and arachidonic acid. Interestingly, the tobacco enzyme, called “pathogen-inducible oxygenase” (PIOX), showed significant homology to prostaglandin-endoperoxide H synthases-1 and -2 present in animal tissue (22). The present study was carried out with the aim of identifying the catalytic function of the pathogen-induced oxygenase from tobacco leaves and its homologous enzyme from Arabidopsis. Evidence will be presented that both enzymes are fatty acid a-dioxygenases which catalyze conversion of linolenic acid and other fatty acids into their 2(R)-hydroperoxy derivatives. The mode of degradation of these unstable hydroperoxides into chain-shortened aldehydes and other a-oxidation products has also been studied. * This work was supported by Swedish Medical Research Council Project number 03X-5170, by Vesical Co., Stockholm, Sweden, and the Education and Science Ministry, Spain, Grant CICYT BIO97-0656. 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. § To whom correspondence should be addressed: Dept. of Medical Biochemistry and Biophysics, Div. of Physiological Chemistry II, Karolinska Institutet, S-171 77 Stockholm, Sweden. Tel.: 46-8-728-7640; Fax: 46-8-736-0439; E-mail: Mats.Hamberg@mbb.ki.se. 1 The abbreviations used are: PIOX, pathogen-inducible oxygenase; BHT, 2,6-di-tert-butyl-4-methylphenol; 9(S)-HPOD, 9(S)-hydroperoxy10(E),12(Z)-octadecadienoic acid; BSTFA, bis(trimethylsilyl)trifluoroacetamide; GLC, gas-liquid chromatography; GC-MS, gas-liquid chromatography-mass spectrometry; MC, (2)-menthoxycarbonyl; MO, O-methyloxime; Me3Si, trimethylsilyl; RP-HPLC, reversed phase high performance liquid chromatography. THE JOURNAL OF BIOLOGICAL CHEMISTRY Vol. 274, No. 35, Issue of August 27, pp. 24503–24513, 1999 © 1999 by The American Society for Biochemistry and Molecular Biology, Inc. Printed in U.S.A.

A variety of conditions, such as mechanical perturbation, osmotic stress, attack by plant pathogens and wounding, elicit increased formation of jasmonates and other biologically active oxylipins in plant leaves (19). This is partly a consequence of liberation of free linolenic acid from its esterified forms (20,21) but may also depend on increased levels of enzymes catalyzing hydroperoxide formation and metabolism. In a recent study, tobacco leaves were found to accumulate a 75-kDa protein in response to bacterial infection (22). This protein, as well as a protein from Arabidopsis showing a 75% homology in amino acid sequence, were expressed in insect cells and found to cause uptake of molecular oxygen in the presence of polyunsaturated fatty acids such as linolenic acid, linoleic acid, and arachidonic acid. Interestingly, the tobacco enzyme, called "pathogen-inducible oxygenase" (PIOX), 1 showed significant homology to prostaglandin-endoperoxide H synthases-1 and -2 present in animal tissue (22).
The present study was carried out with the aim of identifying the catalytic function of the pathogen-induced oxygenase from tobacco leaves and its homologous enzyme from Arabidopsis. Evidence will be presented that both enzymes are fatty acid ␣-dioxygenases which catalyze conversion of linolenic acid and other fatty acids into their 2(R)-hydroperoxy derivatives. The mode of degradation of these unstable hydroperoxides into chain-shortened aldehydes and other ␣-oxidation products has also been studied.
Incubations-Incubations of the tobacco and Arabidopsis oxygenases, and of whole homogenate of cucumber, were carried out with 50 -250 M fatty acid at 23 or 0°C for the times indicated. The mixtures were diluted with 1 volume of distilled water, acidified to pH 4, and extracted twice with diethyl ether. The combined ether phases were washed with water and taken to dryness in vacuo. In most incubations, the material obtained was immediately dissolved in HPLC mobile phase, centrifuged, and analyzed by RP-radio-HPLC. For incubations with peroxygenase, suspensions (0.5 ml) of the membrane fraction from V. faba seeds were preincubated at 23°C for 5 min with the lipoxygenase inhibitor 5,8,11,14-eicosatetraynoic acid (50 M). Subsequently, [9,10-3 H 2 ]oleic acid (100 M) and hydroperoxide (30 -264 M) were added and stirring continued for 15 min. The reaction products were extracted with diethyl ether, and the material obtained was analyzed by RP-radio-HPLC.
Preparation of 2(R)-Hydroperoxy-9(Z),12(Z),15(Z)-octadecatrienoic Acid-Linolenic acid (5.9 mg; concentration, 152 M) was stirred at 0°C for 30 min with a suspension (140 ml) of the 48,000 ϫ g particle fraction of homogenate of cucumber. The mixture was acidified to pH 4 and rapidly extracted with 2 volumes of diethyl ether. The material ob-tained following evaporation of the solvent was suspended in HPLC mobile phase (0.4 ml). After centrifugation, aliquots of 0.1 ml were subjected to RP-HPLC using solvent system I at a flow rate of 2 ml/min. Effluent containing the hydroperoxide (37.0 -39.4 ml) was immediately extracted with diethyl ether and the solution dried over MgSO 4 . Hydroperoxide obtained from several such incubations was dissolved in 0.5 ml of dry acetone (concentration, 10 mM) and stored at Ϫ25°C. The yield of hydroperoxide from the incubated linolenic acid was 5-10% and the radiochemical purity was in excess of 95%. The identity of the hydroperoxide with 2(R)-hydroperoxy-9(Z),12(Z),15(Z)-octadecatrienoic acid was based on chemical and spectral analyses as described under "Results." Methods for Estimation of the Stability of 2-Hydroperoxides-The rate of breakdown of 2-hydroperoxides in enzyme preparations or in buffer was determined by using one of two methods. In method A, tritium-labeled 2(R)-hydroperoxylinolenic acid (35 M) was stirred at 23°C with enzyme preparation or buffer (0.8 or 1 ml). At different times of stirring, the sample was directly subjected to RP-radio-HPLC using a column protected with a pre-column and a solvent system consisting of acetonitrile/water (80:20, v/v) at a flow rate of 2 ml/min. Remaining hydroperoxide was eluted as its salt (3.6 -6.0 ml effluent), well separated from the main product of hydroperoxide breakdown, i.e. 8(Z),11(Z),14(Z)-heptadecatrienal (16.8 -19.2 ml effluent). The rate of hydroperoxide breakdown was estimated by plotting the integrated radioactivity associated with the peak of unconverted hydroperoxide versus time. When 2-hydroperoxylinolenic acid was allowed to degrade in the presence of enzyme preparations, a portion of the product consisted of 2-hydroxylinolenic acid and nor-linolenic acid. These compounds are expected to elute as their salts together with the hydroperoxide, however, because of the small and variable amounts of these decomposition products (together 10% or less), no attempt was made to correct for their presence. In method B, 2(R,S)-hydroperoxyoleic acid (30 M) was stirred with the test preparation (6 ml) at 23°C. Aliquots of 1 ml were removed at different times and added to 5 ml of ethanol containing 25 mg of stannous chloride. After 20 min at 23°C, an internal standard of tetracosanoic acid (70 nmol) was added and the mixtures were extracted with diethyl ether. Aliquots of the methylesterified product was subjected to GLC (column temperature, 270°C) and the peak areas of methyl 2-hydroxyoleate (retention time, 3.6 min) formed by reduction of 2-hydroperoxyoleic acid remaining in the incubation mixture, and of methyl tetracosanoate (retention time, 7.8 min) due to the added internal standard were determined. The rate of hydroperoxide breakdown was calculated from plots of the ratio between the peak areas of methyl 2-hydroxyoleate and methyl tetracosanoate versus time. As with method A, no attempt was made to correct for the small amount of 2-hydroxy acid produced from the hydroperoxide during the incubation period. In some experiments, the reduced samples containing methyl 2-hydroxyoleate were derivatized with (Ϫ)-menthoxycarbonyl chloride, purified by TLC, and subjected to GC-MS operated in the selected ion monitoring mode using the ions m/z 294 and 262. By combining the peak areas of the MC derivatives of methyl 2(S)-hydroxyoleate (retention time, 13.15 min) and methyl 2(R)-hydroxyoleate (13.37 min) with the half-life data, it was possible to separately monitor breakdown of the "R" and "S" enantiomers of 2-hydroperoxyoleate.
Chemical Methods-Configurational determination of 2-hydroxy acids were performed by analysis of MC derivatives by GLC or GC-MS (30). MO derivatives of carbonyl compounds and Me 3 Si ethers of hydroxy compounds were prepared as described previously (31). [ 2 H 9 ]Me 3 Si derivatives, occasionally needed to verify correct interpretation of mass spectra, were prepared by derivatization with [ 2 H 18 ]N,Obis(trimethylsilyl)acetamide (98%, Cambridge Isotope Laboratories, Andover, MA) at 23°C for 30 min. For analysis of 2-hydroperoxy acids by GC-MS, the hydroperoxides (10 -50 g) were derivatized with BSTFA (0.1 ml) and an aliquot of 1-2 l containing the Me 3 Si peroxide/ Me 3 Si ester was directly injected onto the column. Hydroperoxides were reduced into alcohols by treatment with SnCl 2 in ethanol (5 mg/ml) at room temperature for 10 min, or with triphenyl phosphine in diethyl ether (10 mg/ml) at room temperature for 1 h. Catalytic hydrogenation was performed with platinum catalyst (3 mg) and methanol (1 ml) as the solvent. Oxidative ozonolysis was carried out as described (30) using an ozone generator model T-12 purchased from TriO3 Industries, Fort Pierce, FL. Incubations under 18 O gas were conducted in an all-glass apparatus attached to a high vacuum line. 18 O 2 (isotopic purity, 96%) was obtained from Larodan AB, Malmö, Sweden.
Chromatographic and Instrumental Methods-RP-radio-HPLC was performed with columns of Nucleosil 100-5 C 18 (250 ϫ 4.6 mm) purchased from Macherey-Nagel (Dü ren, Germany). The solvent systems consisted of mixtures of acetonitrile, water, 2 M hydrochloric acid in volume proportions 55:45:0.013 (system I), 60:40:0.013 (system II), 65: 35:0.013 (system III), or 80:20:0.013 (system IV). The absorbance (210 nm) and radioactivity of HPLC effluents were determined on-line using a Spectromonitor III ultraviolet detector (Laboratory Data Control, Riviera Beach, FL) and a liquid scintillation counter (IN/US Systems, Tampa, FL), respectively. GLC was performed with a Hewlett-Packard (Avondale, PA) model 5890 gas chromatograph equipped with a methylsilicone capillary column (length, 25 m; film thickness, 0.33 m) and a flame ionization detector. Helium at a flow rate of 25 cm/s was used as the carrier gas. Retention times found on GLC were converted into C-values as described (31). GC-MS was carried out with a Hewlett-Packard model 5970B mass selective detector connected to a Hewlett-Packard model 5890 gas chromatograph fitted with a 5% phenylmethylsilicone capillary column (length, 12 m; film thickness, 0.33 m). In most runs the initial column temperature was 120°C and raised at 10°C/min until 240°C. Ultraviolet spectra were recorded with a Hitachi (Tokyo, Japan) model U-2000 UV/VIS spectrophotometer. Infrared spectrometry was carried out on films using a Perkin-Elmer (Norwalk, CT) model 1650 Fourier transform-infrared spectrophotometer. Radioactivity was determined with a Packard Tri-Carb model 4450 liquid scintillation counter (Packard Instruments, Downer's Grove, IL).

Oxygenation of Fatty Acids by Recombinant Oxygenases
Incubation of Linolenic Acid with Recombinant Enzymes from Tobacco Leaves or Arabidopsis-A preparation of the tobacco leaf oxygenase obtained from insect cells infected with baculovirus carrying pFASTBAC-tob.A5.2 was stirred for 20 min at 23°C with 100 M [9,10,12,13,15,16-3 H 6 ]linolenic acid. The product isolated by extraction with diethyl ether (recovery of radioactivity, 87%) was analyzed by RP-radio-HPLC. Three peaks of radioactive compounds appeared in addition to the peak of linolenic acid remaining unconverted (Fig. 1A). Compounds 1 (7% of the recovered product, 22.9 ml effluent), 2 (1%, 38.4 ml effluent), and 3 (29%, 85.8 ml effluent) were collected for structural determination. As seen in Fig. 1B, a corresponding incubation using cells infected with virus carrying the pFASTBAC vector only gave undetectable conversion of the added tritium-labeled linolenic acid. Incubation of preparations from cells infected with virus carrying pFASTBAC-ara.N38086 (Arabidopsis oxygenase) gave rise to the same products as those formed by the tobacco leaf enzyme, i.e. compound 1 (4%), compound 2 (1%), and compound 3 (34%) ( Fig. 2A). As seen in Fig. 2B, the same products, and an additional one (compound 4) eluting just after compound 1, were produced upon incubation of linolenic acid with a preparation of cucumber (see below).
Identification of Compound 1-The UV spectrum of compound 1 was featureless, indicating the absence of conjugated double bonds. On RP-HPLC, compound 1 co-chromatographed with authentic 2-hydroxylinolenic acid. Considerable chromatographic tailing was observed for both compounds. The retention time found on GLC (C-value, 18.96) and the mass spectrum of the methyl ester of compound 1, were identical to those of methyl 2-hydroxylinolenate (see "Experimental Procedures"). Results obtained upon GC-MS analysis of the Me 3 Si derivatives of the methyl esters of compound 1 and 2-hydroxylinolenic acid were identical. Catalytic hydrogenation of compound 1 followed by esterification produced methyl 2-hydroxystearate. Analysis by GLC of the MC derivative of the methyl ester of compound 1 showed that the absolute configuration at C-2 was R (less than 1% of the S enantiomer). Degradation of the MC derivative of the methyl ester of compound 1 by oxidative ozonolysis produced the MC derivative of methyl hydrogen 2(R)-hydroxyazelate, thus confirming the R configuration at C-2 as well as the presence of a double bond in the ⌬ 9 position. Based on these results, compound 1 was identified as 2(R)hydroxy-9(Z),12(Z),15(Z)-octadecatrienoic acid (2(R)-hydroxylinolenic acid) (Fig. 3).
Identification of Compound 2-Compound 2 co-chromato-graphed with authentic 8(Z),11(Z),14(Z)-heptadecatrienoic acid on RP-HPLC. Furthermore, the retention time found on GLC (C-value, 16.76) and the mass spectrum of the methyl ester were identical to those of methyl 8(Z),11(Z),14(Z)-heptadecatrienoate. Catalytic hydrogenation of compound 2 followed by methyl-esterification afforded methyl heptadecanoate, whereas oxidative ozonolysis of compound 2 yielded suberic acid as the main non-volatile fragment. On the basis of these results, compound 2 was identified as 8(Z),11(Z),14(Z)-heptadecatrienoic acid (nor-linolenic acid) (Fig. 3). Identification of Compound 3-Compound 3 co-chromatographed with authentic 8(Z),11(Z),14(Z)-heptadecatrienal on RP-HPLC. Both compounds gave an unusually broad but symmetric peak (cf. Figs. 1 and 2). The C-value found on GLC (15.78) and the mass spectrum were identical to those of 8(Z),11(Z),14(Z)-heptadecatrienal. Furthermore, the MO derivatives of compound 3 and the authentic reference had identical C-values and mass spectra. Reduction of compound 3 with  Table I). The same set of compounds was produced from linoleic acid (100 M) upon incubation with a whole homogenate preparation of cucumber.
Products Formed from Oleic Acid-Incubation of 70 M oleic acid with the tobacco leaf or Arabidopsis enzyme preparations carried out as described for linoleic acid provided three compounds which were identified by their C-values and mass spectra as 8(Z)-heptadecenal, 8(Z)-heptadecenoic acid, and 2-hydroxy-9(Z)-octadecenoic acid ( Table I). The mass spectrum of the aldehyde showed a molecular ion at m/z 252, which was shifted to m/z 281 upon treatment with O-methylhydroxylamine. The methyl ester of the 2-hydroxy acid produced the expected molecular ion at m/z 312 (4%) and a prominent ion at m/z 253 (24%) formed by elimination of the carbomethoxy group. Incubation of oleic acid (100 M) with the cucumber preparation resulted in the formation of an identical set of products.
Products Formed from Palmitic Acid-Palmitic acid (70 M) incubated with the tobacco leaf, Arabidopsis, or cucumber enzyme preparations produced three compounds which were identified as pentadecanal, pentadecanoic acid, and 2-hydroxypalmitic acid. The two first mentioned compounds gave C-values and mass spectra which were identical to those of the authentic compounds, and the 2-hydroxy acid gave identical data as those of the 2-hydroxy acid formed upon stannous chloride reduction of 2-hydroperoxypalmitic acid.
Kinetic Constants of Recombinant Oxygenases-Fatty acids were stirred with preparations of the tobacco leaf and Arabidopsis enzymes (protein, 0.13 and 0.07 mg/ml, respectively) at 30°C in 1.5 ml of 0.1 M Tris buffer, pH 8.0, and the rate of oxygen uptake was monitored using a Clark oxygen electrode. The K m and V max values were determined from double-reciprocal plots of the maximum velocity of oxygen uptake and substrate concentration. The results are given in Table II. As seen, linolenic, linoleic, and oleic acids were all good substrates for the two enzymes and had K m values of 1-2 M (tobacco leaf oxygenase) and 13-18 M (Arabidopsis oxygenase). In agreement with previous results (22), arachidonic acid was a less effective substrate compared with the 18-carbon fatty acids.

Biosynthesis of 2(R)-Hydroperoxylinolenic Acid
Formation of 2(R)-Hydroperoxylinolenic Acid by Recombinant Enzymes-Arabidopsis enzyme in 0.1 M potassium phosphate buffer, pH 7.4 (5 ml; protein, 0.2 mg/ml), was stirred at  0°C for 20 min with 50 M [9, 10, 12,13,15,16-3 H 6 ]linolenic acid. The product was rapidly extracted with diethyl ether and subjected to RP-radio-HPLC. As seen in Fig. 4, in addition to compounds 1-3 (3, 2, and 8%, respectively, of the recovered radioactivity), an additional peak of radioactivity appeared immediately after compound 1. This material, i.e. compound 4 (24% of the recovered radioactivity), was converted into compound 1 (2-hydroxylinolenic acid) upon treatment with mild reducing agents such as SnCl 2 or triphenylphosphine, suggesting that it was due to the 2-hydroperoxy derivative of linolenic acid. Reduction of compound 4 with sodium borodeuteride led to the formation of 2-hydroxylinolenic acid with no detectable incorporation of deuterium, thus excluding the presence of a keto function. Analysis of compound 4 by GC-MS without derivatization resulted in thermally induced decarboxylation and the appearance of a single peak giving the same mass spectrum as 8(Z),11(Z),14(Z)-heptadecatrienal. A similar analysis carried out following trimethylsilylation using BSTFA reagent showed peaks due to 8 1, 2, 3, 4, compounds 1-4; 18:3, linolenic acid. The peak of UV absorption due to BHT present in the diethyl ether used has been substracted.  (1), 73 (0) (Fig. 5B).
On the basis of the data described, compound 4 was identified as 2-hydroperoxy-9(Z),12(Z),15(Z)-octadecatrienoic acid (2-hydroperoxylinolenic acid). The absolute configuration at C-2 was R as shown by GLC analysis of the MC derivative of 2-hydroxylinolenic acid prepared by reduction with SnCl 2 .

Conversions of 2-Hydroperoxy Fatty Acids
Degradation of 2(R)-Hydroperoxylinolenic Acid in Buffer-Tritium-labeled 2(R)-hydroperoxylinolenic acid (35 M) was added to 0.1 M potassium phosphate buffer, pH 7.4, and kept at 23°C for 30 min. The product was analyzed by RP-radio-HPLC and found to consist of unchanged hydroperoxide (48%), 8(Z),11(Z),14(Z)-heptadecatrienal (46%), and 8(Z),11(Z),14(Z)heptadecatrienoic acid (6%). Longer times of treatment led to further loss of hydroperoxide with concomitantly increased formation of heptadecatrienal and heptadecatrienoic acid. An increased ratio of heptadecatrienoic acid/heptadecatrienal was observed with longer times of treatment, indicating that the heptadecatrienoic acid was formed from the aldehyde by air oxidation. Analysis of hydroperoxide degradation using method A demonstrated a first order decay with a rate constant of 0.0234 min Ϫ1 corresponding to a half-life time of 30 min. 2-Hydroxylinolenic acid was not detectable in these experiments.
Degradation of 2(R)-Hydroperoxylinolenic Acid in Enzyme Preparations-Tritium-labeled 2(R)-hydroperoxylinolenic acid (25 M) was added to preparations of cells expressing the tobacco leaf or Arabidopsis enzymes, or none of these enzymes, and kept at 23°C for 20 min. Analysis by RP-radio-HPLC showed that only trace amounts of hydroperoxide remained. The product compositions were similar in the three incubations and consisted of 8(Z),11(Z),14(Z)-heptadecatrienal (about 90%), 2-hydroxy-9(Z),12(Z),15(Z)-octadecatrienoic acid (about 7%), and 8(Z),11(Z),14(Z)-heptadecatrienoic acid (about 3%). Degradation of the hydroperoxide followed first-order kinetics with k ϭ 0.221-0.245 min Ϫ1 corresponding to half-life times ranging from 2.8 to 3.1 min. No significant difference in the rates of degradation of hydroperoxide in the presence of tobacco leaf or Arabidopsis enzymes or in their absence was noticeable (Fig. 7).

Degradation of 2(R,S)-Hydroperoxyoleic Acid in Enzyme
Preparations-Tritium-labeled 2(R,S)-hydroperoxyoleic acid (30 M) was stirred at 23°C with preparations of recombinant oxygenases or with the 48,000 ϫ g particle fraction of homogenate of cucumber. Analysis by RP-radio-HPLC demonstrated a rapid degradation of the hydroperoxide with concomitant formation of 8(Z)-heptadecenal, 2-hydroxy-9(Z)-octadecenoic acid, and 8(Z)-heptadecenoic acid. The rate of hydroperoxide degradation as determined by method B followed first-order kinetics with half-life times of 5.9 min (cucumber preparation) or 2.4 -2.7 min (preparations of insect cells expressing the tobacco leaf or Arabidopsis enzymes). In order to determine whether degradation of the hydroperoxide was associated with chiral discrimination, samples removed at the different time points were derivatized with (Ϫ)-menthoxycarbonyl chloride and subjected to steric analysis. As shown in Fig. 8, no significant difference in the rates of degradation of the 2(R)-and 2(S)-enantiomers of the hydroperoxide was noticeable (half-life times for 2(R)-and 2(S)-enantiomers in the presence of tobacco leaf enzyme, 2.7 and 2.5 min, respectively; half-life times for 2(R)-and 2(S)enantiomers in the presence of Arabidopsis enzyme, 2.4 and 2.5 min, respectively). Similar results were obtained in incubations with preparations of cells not expressing oxygenases.
2(R)-Hydroperoxylinolenic Acid as a Substrate for Peroxygenase-A particulate fraction from homogenate of V. faba seeds containing peroxygenase (6) was suspended in buffer (0.5 ml; 0.3 mg of protein). The preparation was stirred with 5,8,11,14eicosatetraynoic acid (50 M) at 23°C for 5 min in order to block lipoxygenase activity and subsequently treated with [9,10-3 H 2 ]oleic acid (100 M) and 2-hydroperoxylinolenic acid at 23°C for 15 min. Control incubations performed in the absence of 2-hydroperoxylinolenic acid, or in the presence of 9(S)-HPOD, were also performed. Conversion of the tritium-labeled oleic acid into cis-9,10-epoxyoctadecanoic acid (6) was monitored by RP-radio-HPLC. As seen in Fig. 9, 2-hydroperoxylinolenic acid supported epoxidation of oleic acid into 9,10-ep-FIG. 6. Reversed phase HPLC radiochromatogram of product formed from linolenic acid incubated with a cucumber preparation at 0°C. [9,10,12,13,15,16-3 H 6 ]Linolenic acid (152 M) was stirred with the 48,000 ϫ g particle fraction of homogenate of cucumber at 0°C for 30 min and the product was isolated by extraction with diethyl ether. Solvent system I (0 -25 min) followed by system III (25-55 min) at a flow rate of 2 ml/min was used. 1, 2, 3, 4, compounds 1-4; 18:3, linolenic acid. The peak of UV absorption due to BHT present in the diethyl ether used has been substracted. oxyoctadecanoic acid. In contrast to epoxidations carried out with 9(S)-HPOD and other lipoxygenase-generated hydroperoxides, the 2-hydroperoxylinolenic acid-supported epoxidation appeared to plateau at a hydroperoxide concentration of about 100 M. Whether this phenomenon was due to enhanced rate of inactivation of peroxygenase (cf. Ref. 32) by the 2-hydroperoxylinolenic acid remains to be determined. The absolute configuration of the cis-9,10-epoxyoctadecanoic acid produced from oleic acid in the presence of peroxygenase and 2-hydroperoxylinolenic acid was determined (33) and found to be 9(R),10(S) (81%) and 9(S),10(R) (19%). This result was similar to that earlier found with other hydroperoxides, strengthening the notion that the stereochemistry of peroxygenase-catalyzed epoxidation is solely dictated by the enzyme and is not related to the stereochemistry of the hydroperoxide co-substrate (6,32). In another experiment, 2-hydroperoxyoleic acid (17 M) was incubated with the peroxygenase preparation in the absence of oxidizable co-substrate. The product was methyl-esterified and subjected to TLC using a solvent system of ethyl acetate:hexane, 20:80 (v/v). Two bands appeared, the less polar of which (R F ϭ 0.59) was due to methyl 2-hydroxy-9(Z)-octadecenoate as judged by GC-MS analysis. The more polar material (R F ϭ 0.32) was analyzed as its . Treatment of the material with perchloric acid in aqueous tetrahydrofuran afforded methyl 2,9,10-trihydroxyoctadecanoate, thus confirming the identity of the compound produced from 2-hydroperoxyoleic acid in the presence of peroxygenase as 9,10-epoxy-2-hydroxyoctadecanoic acid.

DISCUSSION
A pathogen-inducible oxygenase (PIOX) in tobacco leaves and a homologous enzyme from Arabidopsis were identified in recent work (22). Sequence and functional analysis of the piox cDNA-encoded protein showed significant homology with prostaglandin-endoperoxide H synthase types-1 and -2, key enzymes in the synthesis of lipid signal molecules in vertebrates. Endoperoxide synthases are dual function enzymes possessing cyclooxygenase and peroxidase activities (34). The recombinant PIOX proteins from tobacco and Arabidopsis possessed oxygenase activity toward several polyunsaturated fatty acids, however, peroxidase activity could not be demonstrated (22).
In order to establish the catalytic function of the oxygenase from tobacco leaves and its homologue from Arabidopsis, incubations with linolenic acid were performed and the isolated products were characterized by chemical and spectral methods. With both enzymes, the major compound consisted of a C 17 unsaturated aldehyde, which was identified as 8(Z),11(Z),14(Z)heptadecatrienal by comparison with authentic material. In addition, small amounts of 2(R)-hydroxylinolenic acid and the C 17 homologue of linolenic acid, i.e. 8(Z),11(Z),14(Z)-heptadecatrienoic acid, were produced. Other fatty acids, including linoleic, oleic, and palmitic acids, were metabolized in an analogous way (Tables I and II). The product profile observed, consisting of a chain-shortened aldehyde, a 2-hydroxy acid, and a chain-shortened fatty acid, was the same as the profile encountered previously in studies of ␣-oxidation in plant tissues. This metabolic pathway was characterized by Stumpf (35), who found that a preparation from peanut cotelydons catalyzed the oxidation of palmitic acid into a long chain fatty aldehyde with concomitant liberation of CO 2 . In subsequent work, ␣-oxidation of various C n fatty acids into C n-1 aldehydes together with varying amounts of C n -hydroxy acids and C n-1 fatty acids has been studied in preparations of pea leaves (36), cucumber (27)(28)(29), potato (37), and the green alga Ulva pertusa (38). The FIG. 9. Peroxygenase-catalyzed epoxidations of oleic acid. Suspensions of the 105,000 ϫ g particle fraction of homogenate of V. faba (0.5 ml) were stirred with 50 M 5,8,11,14-eicosatetraynoic acid at 23°C for 5 min and subsequently treated with 100 M tritium-labeled oleic acid and hydroperoxide at 23°C for 15 min. Formation of tritiumlabeled 9,10-epoxyoctadecanoic acid was monitored by RP-radio-HPLC. q, incubations with 2(R)-hydroperoxylinolenic acid; E, incubations with 9(S)-HPOD.
␣-oxidation enzymes have been suggested to operate together with aldehyde dehydrogenase and NAD ϩ and thus to provide a pathway for stepwise degradation of fatty acids into shorter chain homologues (for review, see Ref. 39).
When the recombinant enzymes from tobacco and Arabidopsis were incubated with substrate at 0°C rather than at room temperature, formation of aldehyde was suppressed and a new main product was formed, i.e. compound 4 (Fig. 4). Compound 4 was converted into 2(R)-hydroxylinolenic acid upon treatment with mild chemical reductants and underwent thermal decarboxylation into 8(Z),11(Z),14(Z)-heptadecatrienal. These results indicated that compound 4 was identical to 2(R)-hydroperoxylinolenic acid, a new member of the oxylipin family of compounds. The trimethylsilyl peroxide/ester derivative of 2-hydroperoxylinolenic acid was sufficiently stable to be analyzed by gas chromatography-mass spectrometry ( Fig. 5A; cf. Refs. 40 and 41). As expected, an incubation carried out under 18 O gas resulted in incorporation of 18 O 2 and formation of doubly 18 O-labeled hydroperoxide (Fig. 5B). Isolation of 2(R)hydroperoxylinolenic as the main product of oxygenation of linolenic acid defined the tobacco and Arabidopsis enzymes as fatty acid ␣-dioxygenases (Fig. 10). The product profile observed following incubation of linolenic acid with the well studied ␣-oxidation system in cucumber (27)(28)(29) (Figs. 2B and 6) was similar to that observed in the corresponding incubations with the recombinant ␣-dioxygenases (Figs. 1A, 2A, and 4), thus suggesting the general involvement of ␣-dioxygenase in plant ␣-oxidation.
Isolation and characterization of 2(R)-hydroperoxylinolenic acid was of interest in relation to the mechanism of ␣-oxidation. Already in 1974, Shine and Stumpf (42) found that inclusion of glutathione and glutathione peroxidase to incubations of palmitic acid with systems for ␣-oxidation resulted in increased formation of 2-hydroxypalmitic acid and a concomitant decrease in the formation of aldehyde and CO 2 . On the basis of this result, a 2-hydroperoxide was proposed as an intermediate in ␣-oxidation (42). This hypothesis was recently supported by the finding that incubations of fatty acids with a preparation from pea leaves carried out in the presence of stannous chloride afforded enantiomerically pure 2(R)-hydroxy acids at the ex-pense of aldehydes (43), and very recently by the isolation of 2(R)-hydroperoxypalmitic acid in incubations of palmitic acid with the ␣-oxidation system from the green alga U. pertusa (44).
2(R)-Hydroperoxylinolenic acid isolated in the present work was considerably less stable than the lipoxygenase-type of fatty acid hydroperoxides. The products formed upon nonenzymatic decomposition consisted of heptadecatrienal (about 90%) accompanied by small and variable amounts of 2-hydroxylinolenic acid and heptadecatrienoic acid. Methodology for estimation of the rate of breakdown of 2-hydroperoxylinolenic acid was devised. Chemical degradation of the hydroperoxide in aqueous buffer, pH 7.4, at 23°C followed first-order kinetics with a half-life time of about 30 min. The rate of decomposition was increased about 10-fold in the presence of preparations of cells expressing the recombinant ␣-dioxygenases. The proximal and distal heme-binding histidines of prostaglandin-endoperoxide H synthase-1 (His 388 and His 207 , respectively) as well as the distal glutamine (Gln 203 ) (34, 45) are conserved in the ␣-dioxygenases from tobacco and Arabidopsis (22), thus indicating that these enzymes are heme proteins capable of further transformation of fatty acid hydroperoxides. However, no specific hydroperoxide degrading activity could be detected for the recombinant ␣-dioxygenases, since the increased rate of hydroperoxide degradation observed with preparations of cells expressing ␣-dioxygenases was observed also with control cells not expressing ␣-dioxygenase (Fig. 7). Furthermore, degradation of a racemic hydroperoxide, i.e. 2(R,S)-hydroperoxyoleic acid, in the presence of ␣-dioxygenases proceeded in a nonstereoselective way (Fig. 8). Although it is conceivable that ␣-dioxygenases, when tested in purified form (cf. Ref. 29), will promote degradation of 2-hydroperoxides, the available data suggest that other tissue-derived factors will prove more important in this respect.
The ␣-oxidation pathway in mammals is of critical importance for degradation of phytanic acid and other ␤-methyl branched fatty acids (46), however, the function of the corresponding pathway in plants is not fully understood. The fact that PIOX, now established as a fatty acid ␣-dioxygenase involved in ␣-oxidation, is pathogen-inducible, suggests that the importance of the ␣-oxidation pathway in plants may be related to plant-pathogen interactions and defense reactions rather than to serve as a pathway for stepwise degradation of fatty acids. Possibly, the 2-hydroperoxides generated by action of ␣-dioxygenases can act as signaling compounds for inductions of genes and enzymes of importance for plant's defense against pathogens (cf. Ref. 47). A direct toxic effect of the hydroperoxide or its degradation products on the invading pathogen is also conceivable. Finally, because 2-hydroperoxides support peroxygenase-catalyzed epoxidation (Fig. 9), biosynthesis of fungitoxic fatty acid epoxides (48) may take place by coupling of the ␣-dioxygenase and peroxygenase pathways.