Purification and Characterization of Linoleate 8-Dioxygenase from the Fungus Gaeumannomyces graminis as a Novel Hemoprotein*

The fungus Gaeumannomyces graminis , which causes the major root disease of wheat known as “take-all,” can metabolize linoleic acid to (8 R )-hydroperoxylinoleic acid. The enzyme linoleate 8-dioxygenase abstracts hydrogen and introduces molecular oxygen in an antarafa- cial way at C-8. We have now purified the enzyme 1000-fold to a specific activity of 1.8 (cid:109) mol/min/mg of protein. Acetone powder of mycelia of G. graminis was subjected to extraction and ammonium sulfate precipitation with solubilization. The 8-dioxygenase was purified by hy- drophobic interaction chromatography, size-exclusion chromatography, anion-exchange chromatography, and immobilized metal ion affinity chromatography. The active enzyme appeared to consist of four subunits since the active enzyme had an apparent molecular mass of 520 kDa determined by gel filtration, while SDS-poly- acrylamide gel electrophoresis showed a protein band of 130 kDa. Spectroscopy indicated the presence of heme. The characteristic pyridine ferrohemochrome (cid:97) -band was found at 557 nm and the (cid:98) -band at 525 nm. The purified protein showed an absorption maximum at 408 nm ( (cid:103) , Soret). The absorption maximum shifted to 429 nm after reduction with dithionite and to 421 nm after treatment of the reduced enzyme with carbon monox-ide. BW A4C, a hydroxamic acid derivative, inhibited the enzyme by > 90% at 10 (cid:109) M . The pH optimum was 7.2–7.4,

Dioxygenases oxygenate aromatic rings or aliphatic structures by inserting molecular oxygen (Webb, 1992). Almost all dioxygenases contain loosely bound ferrous or ferric iron, and only a few dioxygenases are hemoproteins.
Polyunsaturated fatty acid can be oxygenated by two major classes of dioxygenases, namely lipoxygenases, which contain non-heme iron, and PGH 1 synthases, which contain heme. Lipoxygenases occur in mammals and plants and catalyze the first step in biosynthesis of biologically active products, e.g. leukotrienes from arachidonic acid in animals or jasmonic acid from linoleic acid in plants (Gardner, 1991;Siedow, 1991;Yamamoto, 1992). Many lipoxygenases have been purified, cloned, and sequenced, and their reaction mechanism is known in detail. Lipoxygenases abstract a bisallylic hydrogen of the 1Z,4Z-pentadienyl group of the polyunsaturated fatty acid and insert molecular oxygen in an antarafacial way at C-1 or C-5, with formation of a cis,trans-conjugated double bond. The three-dimensional structure of soybean lipoxygenase-1 has been determined, and non-heme iron has been unequivocally demonstrated at the active site (Boyington et al., 1993;Minor et al., 1993). PGH synthases catalyze the double dioxygenation of arachidonic acid to PGG 2 (the cyclooxygenase activity) and the reduction of PGG 2 to PGH 2 (the peroxidase activity) (Smith and Marnett, 1991;Hla and Neilson, 1992). The first step in the biosynthesis of PGG 2 is abstraction of the pro-S-hydrogen at C-13 of arachidonic acid and antarafacial insertion of oxygen at C-11 to generate the 11R-peroxy radical (Hamberg and Samuelsson, 1967;Smith and Marnett, 1991;Oliw et al., 1993b). The three-dimensional structure of PGH synthase-1 has been determined (Picot et al., 1994). The enzyme is dimeric and requires heme for both its cyclooxygenase and peroxidase activities.
The mechanism of interaction between the enzyme, its substrate, and molecular oxygen at the active site of linoleate 8-dioxygenase is uncharacterized, and only little is known about the biological function of the enzyme. Its product, 8-HPODE, can be further metabolized. Reduction yields the alcohol 8-HODE. 8-HODE was originally identified as a metabolite of L. arvalis (Bowers et al., 1986) and later as a factor, which induces premature sexual sporulation of Aspergillus nidulans (Champe and El-Zayat, 1989;Mazur et al., 1990). A microsomal hydroperoxide isomerase from G. graminis transforms 8-HPODE to a diol, 7,8-DiHODE, but L. arvalis lacks this enzyme Brodowsky and Oliw, 1993;Su et al., 1995). The biological functions of 8-HPODE, 8-HODE, and 7,8-DiHODE are unknown, but it is possible that these oxylipins could affect growth and reproduction. This may be worth investigation. The ascomycete G. graminis causes the wheat root disease "take-all", which causes substantial losses for farmers all over the world (Abelson, 1995).
The main objective of this study was to purify the 8-dioxygenase from G. graminis to homogeneity and to determine whether the enzyme contains heme or non-heme iron. We also wanted to determine some properties of the purified protein.
Our results suggest that 8-dioxygenase may belong to a novel family of fatty-acid dioxygenases that are distinct from lipoxygenases and PGH synthases.
Enzyme Assays-All incubations were performed on ice for 5-10 min in a volume of 100 l unless stated otherwise. An aliquot (0.2-10 l, adjusted to 10 l with buffer) from the chromatography fractions was mixed with 85 l of 10 mM TEA, 5 mM EDTA, 1 mM GSH and incubated with 1 nmol of [ 14 C]linoleic acid (10 M), added in 5 l of 30% ethanol in the same buffer. The amount of enzyme was usually adjusted so that the conversion of the substrate was Ͻ25%; the conversion was linear with the amount of added protein and with time. The effect of BW A4C was assayed in duplicate or triplicate after 10 min of preincubation. After termination with ethanol and extractive isolation on Sep-Pak C 18 as described, the products were separated by TLC . Hydroperoxide isomerase activity was assessed by mixing aliquots of each fraction with purified 8-dioxygenase and [ 14 C]linoleic acid as described above. The formation of 14 C-labeled 7,8-DiHODE was determined by TLC. TLC was performed with toluene/dioxane/acetic acid/formic acid (8.2:1.4:0.1:0.1, v/v). The R F values for linoleic acid, 8-HPODE, 8-HODE, and 7,8-DiHODE were 0.54, 0.42, 0.32, and 0.13, respectively. For quantification, the TLC plates were scanned for radioactivity (Berthold Dü nnschichtscanner Model II). The area of the 14 C-labeled peaks on the chart was estimated by cutting and weighing the peaks. The pH optimum was determined in 50 mM potassium phosphate buffer. In kinetic experiments, an oxygen electrode was used to measure enzyme activity (oxygen monitor 5300, oxygen probe 5531, Yellow Springs Instrument Co.) in a thermostated Gilson oxygen cell (cell volume of 1.5 ml, 25°C) with 10 mM TEA, 1 mM EDTA, 1 mM GSH.
Enzyme Purification-The pH of all buffers was 7.4, and all buffers contained 1 mM GSH. The following procedures were performed at ϩ4°C. Acetone powder was prepared from 5-30 g of frozen mycelia with cold acetone (Ϫ20°C) by mechanical disruption, followed by filtration and drying with nitrogen, and stored at Ϫ80°C. Acetone powder (5 g) was homogenized in 100 ml of 0.05 M potassium phosphate, 0.75 M sorbitol, 0.15 M NaCl, 5 mM EDTA (buffer E) in a glass-Teflon homogenizer. The material was centrifuged at 10,000 ϫ g for 25 min. The supernatant was brought to 1.6 M (40%) by the addition of 3.6 M ammonium sulfate, 5 mM EDTA and centrifuged at 10,000 ϫ g for 25 min. The precipitated proteins were solubilized on ice for 30 -60 min with 30 ml of buffer E containing 0.15% Tween 20. The ammonium sulfate solution described above was then added to a final concentration of 0.8 M (20%). The solution was then centrifuged at 100,000 ϫ g for 60 min. The pellet was discarded. The supernatant was diluted with 10 mM TEA, 5 mM EDTA to 0.45 M ammonium sulfate and applied to a column containing butyl-Sepharose 4FF (1.0 ϫ 12 cm). After loading and extensive washing with 10 mM TEA, 0.45 M ammonium sulfate, 5 mM EDTA, 0.04% Tween 20, the column was eluted at 1 ml/min with a linear gradient to 10 mM TEA, 5 mM EDTA, 0.04% Tween 20 in 60 min.
In most experiments, the final purification was achieved as follows (see procedure A in Table I). The active fractions were concentrated (Biosep concentrator 10K, Filtron Technology Corp., Northborough, MA) and loaded on a gel filtration column (Sephacryl S-300 HR, 2.6 ϫ 82 cm). The column was eluted with 10 mM TEA, 0.5 mM CHAPS at 2.6 ml/min. Fractions with enzyme activity were combined and loaded on an anion-exchange column (Q-Sepharose FF, 1.0 ϫ 8.0 cm) in 10 mM TEA, 5 mM EDTA, 0.5 mM CHAPS (or 0.04% Tween 20). The proteins were eluted with a linear gradient of 0 -0.3 M NaCl in 10 mM TEA, 5 mM EDTA, 0.5 mM CHAPS in 20 min with a flow of 2.0 ml/min. Fractions with enzyme activity were combined, concentrated, and desalted (PD-10) before HPLC on Mono Q HR 5/5 in 10 mM TEA, 1 mM EDTA, 0.5 mM CHAPS (or 0.04% Tween 20). The column was eluted at 1 ml/min with a linear gradient of 0 -0.3 M NaCl in the elution buffer for 20 min. The eluted enzyme was concentrated and further studied by HPLC with a column for gel filtration, chromatofocusing, or IMAC. This purification scheme was also used with 30 g of acetone powder using larger columns of butyl-Sepharose 4FF (2.6 ϫ 10 cm) and Q-Sepharose FF (1.6 ϫ 8 cm).
For gel filtration using HPLC, an aliquot (100 l) of concentrated active fractions from anion-exchange chromatography was injected on the column (Biosep SEC-S3000), which was eluted at 1 ml/min with 0.05 M potassium phosphate buffer, 0.15 M NaCl, 1 mM EDTA, 1 mM CHAPS. An aliquot of each fraction was analyzed by SDS-PAGE and for enzyme activity.
For chromatofocusing, the Mono P HR 5/5 column was equilibrated with 25 mM piperazine (pH 6.3), 0.04% Tween 20. The purified material from the Mono Q HR 5/5 column was loaded in 3.5 ml of the piperazine buffer. After washing with 5 ml of the same buffer, a pH gradient was created by elution with 10% Polybuffer 74 (pH 4.5) with 0.04% Tween 20. Fractions (0.5 ml) were collected, pH was determined (at 25°C), and each fraction was analyzed for enzyme activity and then by SDS-PAGE.
In the later part of the investigation, the chromatographic purification of 8-dioxygenase was changed (see procedure B in Table I). Following chromatography on butyl-Sepharose 4FF (2.6 ϫ 10 cm), which was eluted without EDTA in the buffer, the enzymatically active fractions were directly purified by IMAC, followed by gel filtration (Sephacryl S-300 HR), anion-exchange chromatography (Mono Q HR 5/5), and gel filtration (Biosep SEC-S3000).
The IMAC column (chelating Sepharose FF, 1.4 ϫ 1.8 or 1.6 ϫ 12 cm) was charged with ZnCl 2 (17 mol/ml of gel) and equilibrated with 10 mM TEA, 0.15 M ammonium sulfate, 0.04% Tween 20. After loading, the column was first washed extensively with this buffer (2 ml/min) and then with buffer without ammonium sulfate. 8-Dioxygenase was eluted with 40 mM imidazole (or 25 mM EDTA), 20 mM potassium phosphate, 0.25 M NaCl, 0.04% Tween 20. The IMAC column was also charged with Cu 2ϩ and Ni 2ϩ , but the 8-dioxygenase could only be eluted from the Zn 2ϩ -loaded column with satisfactory recovery.
Spectroscopy-Spectroscopy was performed with a dual-beam spectrophotometer (Shimadzu UV-2101PC) at 22°C. The reduced spectrum of the purified protein was obtained after adding sodium dithionite to the cuvette. The cuvette was then saturated with CO for a few minutes, and the spectrum was then determined. Heme was characterized by the alkaline pyridine ferrohemochromogen at 557 nm (␣) and 525 nm (␤) (Paul et al., 1953;Schutz and Feigelson, 1972). Bovine hemin was used as a standard curve for quantification.
Other Analyses-SDS-PAGE was performed with 7.5% resolving and 3.0% stacking gels according to Laemmli (1970). Proteins were detected by silver staining (Ohsawa and Ebata, 1983;Tunón and Johansson, 1984). Protein concentration was determined, in duplicate or triplicate, as described by Bradford (1976) using bovine albumin as a standard.

Purification of Linoleate 8-Dioxygenase
An outline of the purification of 8-dioxygenase and the results are summarized in Table I. A modification of this scheme gave similar results (see procedure B in Table I). The enzyme was purified Ͼ1000-fold up to a specific activity of 1.8 mol/ min/mg. The purified enzyme was associated with a pale yellow color. SDS-PAGE showed a major protein band of 130 kDa and only traces of other proteins on silver staining.
Although the 8-dioxygenase was mainly found in the cytosolic fraction, it appeared to share membrane protein property and tended to form aggregates during chromatography. It was therefore necessary to use detergents. Several detergents were tested in order to solubilize the enzyme from the ammonium sulfate precipitate without inhibiting the enzyme. CHAPS (0.5 mM), sodium deoxycholate (0.5 mM), and Tween 20 (0.15%) were most effective. Tween 20 was compatible with the hydrophobic interaction chromatography, while CHAPS improved the separation by gel filtration. Another problem was enzyme instability, particularly at the early stages of purification. It was essential to have EDTA (1-5 mM) and GSH (1 mM) present and to include sorbitol (0.75 M) during extraction.
Three media for hydrophobic interaction chromatography were evaluated (octyl-, phenyl-, and butyl-Sepharose). Chromatography on butyl-Sepharose 4FF resulted in a 10-fold purification with good recovery (Fig. 1A and Table I). The subsequent gel filtration (Sephacryl S-300 HR) was essential to remove large aggregates of proteins (Fig. 1B). Anion-exchange chromatography on Q-Sepharose FF purified the enzyme 5-fold (Fig.  1C).
Most of the remaining contaminating proteins were removed by HPLC on Mono Q HR 5/5 ( Fig. 2A), followed by gel filtration on Biosep SEC-S3000 (Fig. 2B). However, our enzyme assay showed that hydroperoxide isomerase could be present even after HPLC. The linoleate 8-dioxygenase could be separated from the hydroperoxide isomerase by IMAC with Zn 2ϩ as the ligand (Fig. 2C) and eluted with 40 mM imidazole or with 25 mM EDTA. We confirmed that 40 mM imidazole, 25 mM EDTA did not inhibit the hydroperoxide isomerase.

Prosthetic Groups
The spectrum of the purified enzyme is shown in Fig. 3A. The native protein showed absorption peaks at 280 and 408 nm (␥, Soret) and weaker and broader absorption maxima at 504 -535 nm (␤), 565 nm (␣), and 631 nm. The ratio of 408 / 280 was 0.54 Ϯ 0.03 nm (S.D.) in six different enzyme preparations. Incubation of 0.3 mM linoleic acid with 0.56 M purified enzyme appeared to shift the Soret absorption only slightly to 407 nm, and the 408 / 280 ratio appeared to decrease from 0.53 to 0.44. Treatment with sodium dithionite shifted the Soret band to 429 nm (Fig. 3B). After saturating the reduced enzyme with CO, there were a Soret band at 421 nm and a transient shoulder at 440 nm, but no distinct peak at 450 nm (Fig. 3B).
The pyridine ferrohemochromogen of the purified enzyme showed absorption maxima at 556.5 nm (␣) and 525 nm (␤) as shown in Fig. 3C. These data were almost identical to the hemin standard and are indicative of a hemoprotein of the cytochrome group b (Webb, 1992). An enzyme preparation, which showed virtually no other proteins on SDS-PAGE than the 130-kDa protein, had a specific activity of 0.7 mol/min/mg and contained 0.49 mg of protein/ml, corresponding to ϳ0.72 M enzyme. The sample contained 2.0 M heme as judged from absorption of the pyridine ferrohemochromogen at 557 nm. These figures indicated 2.8 mol of heme/mol of enzyme (520 kDa). Heme might be partly lost from the enzyme during purification. We conclude that the enzyme contains at least 2 mol of heme/mol of enzyme. Fig. 4A. After Mono Q HR 5/5 column chromatography, the band at 130 kDa was intense on SDS-PAGE, while only traces of other proteins were found. Gel filtration (Biosep SEC-S3000) yielded enzyme activity associated with a protein peak around 520 kDa (Fig. 2B). Collecting small fractions over this peak showed that enzyme activity and the intensity of the 130-kDa protein band on SDS-PAGE appeared to vary in parallel, as discussed below (Fig. 4B). The size of the active enzyme on gel filtration HPLC was thus estimated to be 4-fold larger than on SDS-PAGE, ϳ520 and 130 kDa, respectively. SDS-PAGE with or without treatment of the sample with a strong reducing agent (␤-mercaptoethanol) did not affect the results, indicating that the 130-kDa subunits were not covalently bound to each other by disulfide bridges.

SDS-PAGE from the different stages of enzyme purification is shown in
The 130-kDa protein band was also present in the large a Purification of aliquots from Mono Q HR 5/5 on this gel filtration column for HPLC resulted in a loss of ϳ30% of the applied enzyme activity but the specific activity increased. Enzyme activity was measured on ice.
protein peak without enzyme activity, which eluted after the enzyme. Analysis of the Soret band at 408 nm during gel filtration gave important information. As shown in Fig. 4C, two major peaks of protein with absorption at 408 nm were separated by gel filtration (Biosep SEC-S3000). The first eluting peak (peak I) contained active enzyme, while the second peak (peak II) was inactive. SDS-PAGE and spectroscopic analysis of both peaks gave the same results, a protein band at 130 kDa and a Soret band at 408 nm, which shifted to 429 and 421 nm, as discussed above. When the material in peak I was concentrated, frozen, thawed, and reanalyzed by gel filtration under identical conditions, peaks I and II were present again (Fig.  4D). It appears as if peak I contains the active enzyme as a tetramer of ϳ520 kDa, which is followed by peak II-containing subunits in oligomers of lower orders. The tetramer did not appear to be completely dissociated by 4 M urea. The enzyme could be incubated in 4 M urea for 1 h, followed by desalting (PD-10), with only partial loss of enzyme activity. Gel filtration in 4 M urea showed the same elution pattern, with two major protein peaks with enzyme activity only in the first (after desalting).
Purified 8-dioxygenase thus showed a major band at 130 kDa and a faint diffuse protein band at ϳ100 kDa as judged by

FIG. 2. Purification of linoleate 8-dioxygenase by HPLC and by IMAC.
A, anion-exchange chromatography (Mono Q HR 5/5). Enzyme activity was found in the shaded part of the chromatogram. The dashed line shows the increase in conductivity as NaCl increases in the elution buffer from 0 to 0.3 M. B, HPLC with a gel filtration column (Biosep SEC-S3000, 300 ϫ 7.8 mm) of the active fractions from A. The highest enzyme activity coeluted with the protein peak, which is shaded. The apparent molecular mass of the enzyme was estimated to be ϳ520 kDa as judged from the elution enzyme and from four protein standards (arrow 1, 670 kDa; arrow 2, 158 kDa; arrow 3, 44 kDa; arrow 4, 17 kDa). C, immobilized metal ion chromatography of 8-dioxygenase purified by chromatography on butyl-Sepharose 4FF (see procedure B in Table I). The Zn 2ϩ -loaded column (chelating Sepharose FF, 1.6 ϫ 12 cm) was washed with 10 mM TEA, 0.15 M ammonium sulfate, 0.04% Tween 20. Arrow 1 marks change to buffer without ammonium sulfate, and arrow 2 marks elution of the column with imidazole. 8-Dioxygenase without hydroperoxide isomerase was found in the shaded peak. silver staining (Fig. 4A). The intensity of the latter band varied in different batches, and it proved to be difficult to remove completely. The intensity of this band appeared to increase with storage of the partly purified enzyme since SDS-PAGE of several batches of 8-dioxygenase, which were purified almost without interruption, showed the highest purity.
Several observations indicated that the 130-kDa band, but not the 100-kDa band, was associated with the 8-dioxygenase activity. First, purification only moderately enriched the 100-kDa protein or proteins (Fig. 4A). Second, we occasionally obtained fractions during chromatography in which these two bands appeared to be of almost equal intensity, but these fractions did not have a very high specific enzyme activity. The 130-kDa protein must be cloned and expressed to obtain certainty, but it was reassuring to find that the amino acid sequences of four oligopeptides (with 7, 8, 11, and 14 sequenced amino acid residues), which were obtained by enzymatic digestion of the 130-kDa band, did not match any previously sequenced proteins.  Table I). Lane 1, extract of acetone powder; lane 2, ammonium sulfate precipitate (40 -20%); lane 3, after butyl-Sepharose 4FF; lane 4, after Sephacryl S-300 HR; lane 5, after Q-Sepharose FF; lane 6, after Mono Q HR 5/5; lanes 7-12, proteins from small consecutive fractions from Biosep SEC-S3000 collected over the protein peak with enzyme activity (cf. Fig. 2B); lane 13, molecular mass markers (Pharmacia; 212, 170, 116, 76, and 53 kDa and an artifact at ϳ68 kDa). B, the enzyme activity in the fractions of lanes 7-12 in A is indicated by the histogram, which shows the percent conversion of the radiolabeled substrate to 8-H(P)ODE. C, shown is the separation by gel filtration (Biosep SEC-S3000) of 8-dioxygenase (from Mono Q HR 5/5) with analysis of absorption at 408 nm (␥, Soret). Enzyme activity was found only in the first of the two peaks (peak I). D, an aliquot of the material of peak I in C was injected on the gel filtration column after concentration, freezing, and thawing. The eluting material was analyzed at 408 nm (OO) and 280 nm (---).

Properties of Linoleate 8-Dioxygenase
pH Optimum-Enzyme activity was assayed at pH 6.4, 6.8, 7.2, 7.4, 7.6, and 8.0. The highest activity of the purified enzyme was noted at pH 7.2 and 7.4 (31% conversion of substrate at both pH values).
Isoelectric Point-Enzyme activity eluted as one peak on chromatofocusing (Mono P HR 5/5) at pH 5.2 (25°C). SDS-PAGE showed that this fraction contained a protein band at 130 kDa.
Stability-The enzyme was quite unstable at the early steps of purification, but active fractions from anion-exchange chromatography could be kept for 1 week on ice with only a small loss of enzyme activity. We routinely stored the enzyme at Ϫ80°C, but repeated freezing and thawing, particularly of dilute enzyme solutions, appeared to inactivate the enzyme, as discussed above (Fig. 4D).
Kinetic Constants-The enzyme was purified to a specific activity of 1.8 mol/min/mg at 0°C (Table I). The turnover number on ice with 10 M linoleic acid as a substrate was calculated to be 15/s. The K m and V max values for oxygen were calculated from incubations of enzyme (after Mono Q HR 5/5 purification; see Table I) with 0.2 mM linoleic acid at 25°C. Initial oxygen consumption was measured with an oxygen electrode. The reaction was started by adding the enzyme. The oxygen concentration was varied from 0.247 to 0.016 mM (five points in duplicate, r ϭ 0.99). The V max and K m values were estimated to be ϳ2.2 mol/min/mg of protein and 30 M, respectively. The K m and V max values for linoleic acid were determined in the same way, but with oxygen-saturated buffer using 7-200 M linoleic acid (five points in duplicate, r ϭ 0.99), and were found to be 8 M and 4 mol/min/mg of protein, respectively.
CO, KCN, and H 2 O 2 -The enzyme was not inhibited by 1 mM KCN. The enzyme was mixed with 4 volumes of CO-or N 2saturated buffer and then incubated with substrate. CO did not inhibit the enzyme. Finally, H 2 O 2 (up to 9 mM) did not support enzyme activity under anaerobic conditions. DISCUSSION We have purified linoleate 8-dioxygenase Ͼ1000-fold to one major protein band on SDS-PAGE ( Fig. 4A and Table I). Linoleate 8-dioxygenase appeared to consist of four noncovalently bound subunits of 130 kDa. The enzyme contained heme. This finding was unexpected. Only a few dioxygenases contain heme (Webb, 1992). Linoleate 8-dioxygenase might be related to other heme-containing oxygenases, e.g. PGH synthases, tryptophan 2,3-dioxygenase, or cytochrome P450.
Linoleate 8-dioxygenase was inhibited by micromolar concentrations of BW A4C, which is a hydroxamic acid derivative and a potent lipoxygenase inhibitor (McMillan and Walker, 1992). Hydroxamic acid derivatives can chelate iron, but they may also have redox properties (Nelson et al., 1991). Some other lipoxygenase inhibitors and reducing agents can inhibit linoleate 8-dioxygenase Su et al., 1995). In spite of pharmacological similarities, linoleate 8-dioxygenase and lipoxygenases are clearly different. Lipoxygenases do not contain heme and do not catalyze hydrogen abstraction and oxygen insertion at the same carbon. Lipoxygenases and linoleate 8-dioxygenase differ in substrate requirements. The 8-dioxygenase can metabolize oleic acid , while lipoxygenases require fatty acids with methyleneinterrupted double bonds (Gardner, 1991;Siedow, 1991;Yamamoto, 1992). We conclude that lipoxygenases and 8-dioxy-genase belong to different families of enzymes.
Linoleate 8-dioxygenase could be related to cytochrome P450, but the reaction mechanisms of 8-dioxygenase and P450 have little in common. This work shows that purified linoleate 8-dioxygenase does not appear to require any cofactors or other enzymes for full activity. P450 can hydroxylate fatty acids in the presence of NADPH and cytochrome P450 reductase. In addition, P450 cannot form hydroperoxy fatty acids enzymatically (Capdevila et al., 1981;Oliw et al., 1993a;Oliw, 1994). Cysteine is the proximal heme iron ligand of P450, and hemethiolate proteins show a characteristic absorption at 450 nm after reduction and treatment with CO (Nelson et al., 1993;Oliw, 1994). It cannot be excluded that 8-dioxygenase also might be a heme-thiolate protein, but it seems unlikely. We were unable to obtain the characteristic spectrum with a distinct peak at 450 nm.
It seems relevant to compare linoleate 8-lipoxygenase with heme-containing dioxygenases. Tryptophan 2,3-dioxygenase has been well investigated (Schutz and Feigelson, 1972;Leeds et al., 1993), and its reaction mechanism is interesting. Tryptophan 2,3-dioxygenase cleaves the indole ring of tryptophan and forms N-formylkynurenine. The initial reaction involves hydrogen abstraction from the indole nitrogen (N-1), migration of the C-2-C-3 double bond to N-1-C-2, insertion of molecular oxygen, and formation of a peroxyl radical at C-3 (Leeds et al., 1993). Tryptophan 2,3-dioxygenase from mammalian liver is a tetrameric hemoprotein with four identical 48-kDa subunits (Schutz and Feigelson, 1972;Maezono et al., 1990). The purified enzyme contains 2 mol of heme, and no other cofactor is associated with the enzyme. The ligand of heme iron is a nitrogen atom of a histidine residue. Exogenous heme stimulates the activity, which suggests a stoichiometry of one heme/ subunit (Schutz and Feigelson, 1972;Leeds et al., 1993). The ferrous form of the enzyme is active, and the function of heme is likely to bind and activate molecular oxygen rather than the substrate (Leeds et al., 1993).
There is a specific reason to compare the reaction mechanism of linoleate 8-dioxygenase with that of PGH synthases. With 5,8,11-eicosatrienoic acid as a substrate, PGH synthase-1 and -2 2 can perform an abortive cyclooxygenase reaction by abstracting hydrogen and inserting molecular oxygen at C-13 (Elliott et al., 1986;Oliw et al., 1993b). This reaction appears to be similar to that of linoleate 8-dioxygenase. PGH synthase-1 contains two essential elements for its cyclooxygenase activity, heme and Tyr 385 . This tyrosine residue is located at the end of the substrate channel within 10 Å of the heme iron (Picot et al., 1994). Tyrosyl radicals can be detected by electron resonance spectroscopy during enzyme catalysis, but how heme and Tyr interact is not known (Tsai et al., 1994). Whether the active site of linoleate 8-dioxygenase contains heme and tyrosine in close relation and whether tyrosyl radicals are formed during catalysis are unknown. Linoleate 8-dioxygenase can now be purified in milligram amounts, and it should therefore be possible to address this question.
Linoleate 8-dioxygenase is a tetrameric hemoprotein with no other cofactors, and it contains at least 2 mol of heme. The low heme number indicates that it could be composed of two types of subunits of equal size. There are at least two possible oxygenation mechanisms for the enzyme, oxygen activation or substrate activation, as illustrated by tryptophan 2,3-dioxygenase and PGH synthase, respectively. In both models, the pro-S-hydrogen at C-8 is abstracted, a carbon-centered radical is formed at C-8, and molecular oxygen is inserted with inversion of configuration at C-8 . The (8R)peroxy radical is then reduced to 8-HPODE. Additional work is needed to elucidate the reaction mechanism and the precise interaction between substrate and enzyme, oxygen, and heme.
In summary, we have purified linoleate 8-dioxygenase from the ascomycete G. graminis, which is a devastating pathogen of wheat all over the world. The biological function of this enzyme is unknown, but it may affect sexual sporulation and fungal growth. Our results suggest that linoleate 8-dioxygenase is a hemoprotein that is distinct from previously described fattyacid dioxygenases. The reaction mechanism of the enzyme may have properties in common with tryptophan 2,3-dioxygenase or PGH synthases.