Secretion of two novel enzymes, manganese 9S-lipoxygenase and epoxy alcohol synthase, by the rice pathogen Magnaporthe salvinii.

The mycelium of the rice stem pathogen, Magnaporthe salvinii, secreted linoleate 9S-lipoxygenase (9S-LOX) and epoxy alcohol synthase (EAS). The EAS rapidly transformed 9S-hydroperoxy-octadeca-10E,12Z-dienoic acid (9S-HPODE) to threo 10 (11)-epoxy-9S-hydroxy-12Z-octadecenoic acid, but other hydroperoxy FAs were poor substrates. 9S-LOX was expressed in Pichia pastoris. Recombinant 9S-LOX oxidized 18:2n-6 directly to 9S-HPODE, the end product, and also to two intermediates, 11S-hydroperoxy-9Z,12Z-octadecenoic acid (11S-HPODE; ∼5%) and 13R-hydroperoxy-9Z,11E-octadecadienoic acid (13R-HPODE; ∼1%). 11S- and 13R-HPODE were isomerized to 9S-HPODE, probably after oxidation to peroxyl radicals, β-fragmentation, and oxygen insertion at C-9. The 18:3n-3 was oxidized at C-9, C-11, and C-13, and to 9,16-dihydroxy-10E,12,14E-octadecatrienoic acid. 9S-LOX contained catalytic manganese (Mn:protein ∼0.2:1; Mn/Fe, 1:0.05), and its sequence could be aligned with 77% identity to 13R-LOX with catalytic manganese lipoxygenase (13R-MnLOX) of the Take-all fungus. The Leu350Met mutant of 9S-LOX shifted oxidation of 18:2n-6 from C-9 to C-13, and the Phe347Leu, Phe347Val, and Phe347Ala mutants of 13R-MnLOX from C-13 to C-9. In conclusion, M. salvinii secretes 9S-LOX with catalytic manganese along with a specific EAS. Alterations in the Sloane determinant of 9S-LOX and 13R-MnLOX with larger and smaller hydrophobic residues interconverted the regiospecific oxidation of 18:2n-6, presumably by altering the substrate position in relation to oxygen insertion.

Our fi rst goal was to determine the oxylipin biosynthesis by M. salvinii and whether LOXs were secreted in analogy with G. graminis . We found that M. salvinii secreted linoleate 9 S -LOX and a novel epoxy alcohol synthase (EAS) with specifi city for 9 S -hydroperoxy-10 E , 12 Z -octadecadienoic acid (9 S -HPODE). This enzyme was apparently identical to the recombinant 9-LOX of M. salvinii , which was described and patented by Sugio and Takagi at Novozymes ( 31 ) with a detailed patent description available on the Internet 2 . Our second goal was to determine the catalytic metal of this 9-LOX. To this end, we expressed 9 S -LOX of M. salvinii in Pichia pastoris . The recombinant enzyme was found to oxidize 18:2n-6 and 18:3n-3 in the same way as native 9 S -LOX and to contain Mn; it will therefore be referred to as 9 S-Mn-LOX. Our third goal was to investigate the structural basis of the regiospecifi cities of 9 S -and 13 R -MnLOX. Finally, we studied the metal selection process of 13 R -MnLOX by insertion of an amino acid (Thr/Gly) between the fourth and fi fth positions of the His-Val-Leu-Phe-His pentamer motif and by expression in the environment of augmented intracellular iron concentration.
13 R -MnLOX is unknown, but it is presumed to oxidize cellular lipids to generate harmful reactive oxygen species in the infectious process ( 18 ). The putative LOX of M. oryzae is transcribed during rice infection ( 19 ). All three pathogens are frequently found in rice fi elds world-wide, and can severely diminish the harvest ( 16 ). Oxylipin biosynthesis by G. graminis and M. oryzae has been studied ( 4,19 ), but little is known about oxylipin biosynthesis by M. salvinii . The latter infects the stem and the leaf sheats at the water line, leading to necrosis and formation of small black sclerotia with conidia ( 16,20 ).
The LOX family is characterized by sequence homology with conserved ligands to the catalytic nonheme iron ( 2,5,21 ). These octahedral ligands are three His residues, the carboxyl group of the C-terminal amino acid, usually Ile, a distant Asn or His residue, and water. The latter forms the catalytic base for hydrogen abstraction at bis -allylic carbons of FAs and subsequent oxidation. In this process of proton-coupled electron transfer, the catalytic base (Fe 3+ OH Ϫ ) is reduced (Fe 2+ OH 2 ), and oxygen is inserted in an antarafacial manner at either C-1, C-5, or C-3 of the 1,4-pentadiene radical with formation of peroxyl radicals ( 21,22 ). In the next step, hydrogen is transferred to the peroxyl radicals with regeneration of the catalytic base, Fe 3+ OH Ϫ . Iron was identifi ed as the catalytic metal of soybean LOX-1 (sLOX-1) ( 23,24 ), and later in many other LOXs. To date, the only known exception is 13 R -MnLOX.
To date, little is known about iron and manganese selection by LOXs. In pursuit of additional LOXs with catalytic manganese, we found that 9-LOX of M. salvinii (GenBank, CAD61974.1) contains the tentative manganese binding pentamer motif, His-Val-Leu-Phe-His. This sequence could be aligned to 13 R -MnLOX (GenBank, AAK81882.1) with 77% amino acid identity and contains conserved structural elements also found in mammalian LOXs ( Fig. 1A, B , and supplementary Fig. I). The tentative Mn ligands of 13 R-MnLOX are conserved in the LOX of M. salvinii except for the exchange of the C-terminal Val residue to Ile, which is the predominant C-terminal amino acid of LOXs ( Fig. 1A ). Both enzymes contain Gly in the Coffa-Brash (Gly/Ala) determinant of regiospecifi city and R/S chirality ( 29 ), but they differ at one homologous position (Phe347/Leu350) of the Sloane determinant ( 30 ) ( Fig. 1B ). An overview of hydrogen abstraction and oxygenation of 18:2n-6 by LOXs is shown in Fig. 1C . , three His, one Asn/His residue, and the C-terminal amino acid residues (Val/Ile)]. B: Sequences from the Gly/Ala (Coffa-Brash) determinant of regio-and stereospecifi city to the Sloane sequence determinant of regiospecifi city with a noticeable difference between 9 S -and 13 R -MnLOX at one position (Leu350/ Phe347) of the Sloane determinant. C: Overview of the six positions of oxygenation of 18:2n-6 by LOXs. LOXs discussed here abstract the proS hydrogen at C-11 and insert oxygen at either C-13 or C-9, and in a few cases, also at C-11.

Expression constructs
Comparison of the deduced protein from the M. salvinii LOX mRNA (GenBank, AX590415) with 13 R-MnLOX of G. graminis (GenBank, AAK81882) suggested an analogous N-terminal secretion peptide of 17 amino acids (SignalP 3.0 Server). The open reading frame without signal peptide was ordered from GenScript (Wheelock House; Hong-Kong) in pUC57, flanked by restriction sites for EcoRI and XbaI (1,815 bp). The open reading frame was ligated into the pPICZ ␣ A vector. This plasmid, pPICZ ␣ A_salvLOX, was then used to transform E. coli (NEB5 ␣ ). We replaced Phe347 of 13 R -MnLOX_580 with Leu, Val, and Ala by site-directed mutagenesis of pPICZ ␣ A_MnLOX_580 with Pfu DNA polymerase and primer pairs (forward and the reverse complement primer F347L: 5 ′ -ggttctgggac caaaac tta ggcctgcccgcctcggccgcc; F347V, 5 ′ -ggttctgggaccaaaacgtt ggcctgcccgcctcggccg; F347A, 5 ′ -cggccgaggcgggcaggccagcgttttg gtcccagaacc). In the same way, we replaced Leu350 of 9 S -MnLOX with Phe and Met (L350F, 5 ′ -gcgggttttgggaccagaactttggcctgccc gccacggcgg; L350M, 5 ′ -gcgggttttgggaccagaacatgggcctgcccgcca cggc). The PCR product was restricted with DpnI and transformed into E. coli (NEB5 ␣ ). The Thr294 insert (ACT) in the His-Val-Leu-Phe-His pentamer was also ordered as a 492 bp sequence of the cDNA of 13 R -MnLOX (GenScript) in pUC57. pUC57 with this insert was restricted with BspI and the fragment was used to replace the corresponding sequence in pPICZ ␣ A_MnLOX_580. The resulting pPICZ ␣ A_MnLOX_580_ Tins was modifi ed by replacements of Thr294 with Gly by sitedirected mutagenesis using Pfu DNA polymerase and a primer pair (5 ′ -gatgtaccacgtgctcttc ggt cacacgatccccgagatcgtg-3 ′ , and its complementary primer). All constructs were confi rmed by sequencing at Uppsala Genome Center (Uppsala University).

Real-time PCR analysis
M. salvinii sporulates in growth media containing rice leaves sterilized by chemicals or by heat ( 20 ). We therefore compared the expression of 9 S -MnLOX mRNA by analysis of cDNA prepared from M. salvinii (CBS 288.54) grown in complete medium with or without autoclaved rice leaves (7 days, 24°C, slow agitation). Total RNA was prepared from nitrogen powder using the TRI reagent according to the manufacturer's protocol with the addition of a sonication step (30s ×10, maximal intensity) and treated with DNa-seI (Promega). cDNA was synthesized with Superscript III (Invitrogen), and 250 ng was used as template for real-time PCR with SYB-RGreen. cDNA was prepared from three different experiments and analyzed in triplicate. The two primers for 9 S -MnLOX (forward, 5 ′ -acagcgtatcgtgaagccaa; reverse, 5 ′ -ttccaatggccgtcgtagag) were designed to span exon/exon boundaries to distinguish amplifi cation of genomic DNA and cDNA by agarose gel electrophoresis. Relative

Growth of M. salvinii and enzyme assays
M. salvinii was fi rst grown on oatmeal or potato dextrose agar (1-2 weeks, 22°C) ( 20 ), and then in liquid complete medium (per liter: 10 g glucose, 3 g NaNO 3 , 0.26 g KCl, 0.13 g MgSO 4 , 0.76 g KH 2 PO4, 2 g peptone, 1 g yeast extract, 1 g casamino acids, trace metals, and vitamin solution) under fl uorescent light for (1-2 weeks, 22°C) with no or only slight agitation. The mycelia were separated by fi ltration, washed with saline, blotted dry, and ground to a fi ne powder with liquid nitrogen. This nitrogen powder was stored at Ϫ 80°C and used for isolation of RNA and DNA and assay of oxylipin biosynthesis.
The EAS activity was determined by UV analysis from the decline of the UV absorbance at 234 nm with 10-100 µM 9 S -HPODE as a substrate in 0.1 M NaBO 3 buffer (pH 9.0) or by LC-MS analysis of the formed epoxy alcohol. K m was estimated from triplicate analysis by the Michaelis-Menten equation (GraFit).

Atomic emission spectroscopy of Fe and Mn and expression of 13 R -MnLOX with iron excess
The Fe and Mn concentrations of partly purifi ed 9 S -MnLOX and yeast cells transformed with 13 R -MnLOX were measured with inductively coupled plasma atomic emission spectroscopy (AES) (Spectrofl ame P) as previously described ( 17 ). For AES, the proteins were washed three times by diafi ltration with 25 mM KHPO 4 (pH 7.0)/2 mM NaN 3 /150 mM NaCl, which had passed through a resin with iminodiacetic groups (Chelex-100) to chelate divalent cations. The fi nal diafi ltration fl ow-through was collected and used as blank in AES measurements. Fe was analyzed at 259.9 nm and Mn at 257.6 nm with a standard curve of Fe and Mn from 0 to 2 ppm.
For yeast cell metal analysis, the cells were harvested after 5 days by centrifugation (5,000 g ), dried on Petri dishes (60°C, 5 days), washed, and subjected to acid digestion in Erlenmeyer fl asks. To 0.1 g of dry cells, 1.5 ml H 2 SO 4 and 2.5 ml HNO 3 were added, and the mixture was boiled. Additionally, aliquots of 1.5 ml HNO 3 were added to the mixture until white/yellow fumes appeared ( 37 ). The volume was reduced by boiling and diluted in Milli-Q water to 6 ml. A blank sample (acids without cells) was subjected to the same treatment.
The effect of an increased intracellular Fe/Mn ratio during expression of recombinant 13 R -MnLOX was analyzed as follows. Glassware was washed with 15% HNO 3 . BMM and BMGY with reduced Mn and Fe content were prepared by fi ltering yeast nitrogen base through a resin with iminodiacetic groups (10 × 30 mm, Chelex-100). CaCl 2 and MgSO 4 were added to the media, which were stored in plastic bottles. P. pastoris cells, transformed with pPICZ ␣ A_MnLOX_580, were fi rst grown in BMGY with low iron content to upregulate iron transporters ( 38 ) or with normal BMGY as control. Cells were collected (in duplicate), washed three times with Milli-Q water, and resuspended in normal BMM or BMM with low Mn 2+ and excess of Fe 2+ (100 µM (NH 4 ) 2 Fe(SO 4 ) 2 ). Protein expression was induced with methanol, and expressed LOX was purifi ed as described ( 26 ). LOX activity was assayed with 18:2n-6 and with 11 R -HPOTrE as substrates, because the latter is a poor substrate for FeLOX.

LOX and EAS activity assays
The LOX activity was measured on a dual-beam spectrophotometer (Shimadzu UV-2101PC). The enzyme was mixed with 50-100 µM 18:2n-6 or 18:3n-3 in 0.1 M NaBO 3 (pH 9.0), and the UV absorbance was followed at 237 and 234 nm, respectively, at 22°C. The reaction rate was estimated from the linear part of the curve. The cis-trans -conjugated hydro(pero)xy FAs were assumed to have an extinction coeffi cient of 25,000 cm . For steric analysis of hydroxy FAs, hydroperoxides were reduced by treatment with triphenylphosphine (10 µg) before LC-MS/MS analysis. K m of 9 S -Mn-LOX was determined by triplicate analysis in the range 2.5-70 µM 18:2n-6, and calculated from the Michaelis-Menten equation (GraFit, Erithacus software). The k cat was determined with ‫ف‬ 12 nM 9 S -MnLOX and 100 µM 18:2n-6. Hydroperoxides (25-50 µM) and most of the EAS activity ( ‫ف‬ 90%) in a single step of diafi ltration. The EAS activity was reduced by several steps of diafi ltration (30 K) with sequential addition of buffer (0.1 M NaBO 3 buffer, pH 9.0).
Gel fi ltration showed that the main 9 S -LOX activity eluted directly after albumin (in the range 60-65 kDa), whereas the EAS activity eluted with approximately the same retention time as carbonic anhydrase (29 kDa; Sigma) was also used for group separation of HOTrE. The silica column (250 × 2 mm, 5 µm; Kromasil) was eluted with hexane-isopropyl alcohol-acetic acid, 97:3:0.01 at 0.3 ml/min. GC-MS analysis was performed as described ( 17 ). Methyl ester derivatives were prepared with diazomethane and silylation with N,O-bis (trimethylsilyl)trifl uoroacetamide in pyridine (70°C, 30 min). C values were determined from the retention times of FA methyl esters ( 17 ).

Miscellaneous
SDS-PAGE was performed on a 7.5% resolving and 4% stacking polyacrylamide gel, at 100 V for 10 min followed by 140 V for 30 min. The gels were washed with water, dyed for 3-24 h (colloidal Coomassie), and then destained with water. FA composition of M. salvinii was determined in nitrogen powder of mycelia after alkali treatment (0.5 M KOH in 90% methanol, 70°C, 1 h) and extractive isolation (CHCl 3 -methanol) according to Bligh and Dyer ( 40 ). The carboxylate anions were analyzed by direct injection (LTQ). Secretion of LOX and EAS. The concentrated growth medium oxidized 18:3n-3 with little position specifi city at C-13, C-11, and C-9 ( Fig. 3 ), whereas 18:2n-6 was oxidized with position specifi city to 9-HPODE, 9-HODE, and only a small amount or traces of 11-H(P)ODE. In addition, an epoxy alcohol was also formed, suggesting secretion of both linoleate 9 S -LOX and EAS ( Fig. 2B ). Steric analysis showed that the 9 S -HODE stereoisomer was formed exclusively ( Fig. 2C ). The EAS activity was obvious from the oxidation of 18:2n-6 to 9 S -HPODE by the secreted 9 S -LOX and the subsequent decline in UV absorbance at 234 nm due to transformation of 9-HPODE to products without this absorbance (see supplementary Fig. IIIA).

Oxidation and transformation of FAs by
To account for biological variability, we investigated two strains of M. salvinii (CBS 288.52, CBS 254.34). Both strains secreted 9 S -LOX and EAS, and appeared to form the same profi le of oxylipins. We partly sequenced over 80% of their genomic DNA of 9 S -MnLOX, but could not detect any discrepancies that would result in amino acid replacement (data not shown).
The EAS converted 9 S -HPODE to the epoxy alcohol with loss of UV absorbance at 234 nm ( Fig. 4A ) . The EAS showed high activity over a relative broad pH range (pH 6.5-9.0). The apparent K m was estimated with 9 S -HPODE (range 10 to 100 µM; triplicate analysis) to be 32 ± 5 µM (inset in Fig. 4A ).
LC-and GC-MS analysis of 10 R (11 R )-epoxy-9 S -hydroxy-12 Z -18:1. EAS transformed 9 S -HPODE to an epoxy alcohol. It is well known that isomers of 9(10)-epoxy-11-hydroxy-12 -Z -18:1 can be formed from 9 S -HPODE by LOXs ( 41,42 ). During NP-HPLC, however, the EAS metabolite eluted between the erythro and threo isomers of 9(10)-epoxy-11hydroxy-12 Z -18:1, which were obtained by hematin-catalyzed transformation of 9 S -HPODE ( Fig. 2D ) ( 33 ). All three compounds had identical MS/MS spectra. This paradox was likely due to epoxide migration (Payne rearrangement) in the mass spectrometer. 9(10)-Epoxy-11-hydroxy-12 Z -18:1 and 10(11)-epoxy-9-hydroxy-12 Z -18:1 thus have virtually identical MS/MS spectra with strong signals at m/z 201 and 171 (cf. Fig. 5A ) ( 33 ). This suggested that EAS formed the allylic epoxy alcohol, 10 (11)-epoxy-9 S -hydroxy-12 Z -18:1. This epoxy alcohol was hydrolyzed within a few min at pH 2.2, which is a typical feature of allylic epoxides ( 43,44 ). Finally, hydrolysis products of the epoxy alcohol were also detected in the growth medium and identifi ed as two triols ( Fig. 5B, C ; supplementary Fig. III). This suggested that the epoxy alcohol contained an epoxide at  analyzed after diafi ltration or purifi ed after three days of expression (to limit lysis of cells and release of proteolytic enzymes) by hydrophobic interaction chromatography and gel fi ltration. Purifi ed 9 S -MnLOX could be stored for several weeks at 4°C with retained activity. SDS-PAGE revealed an expressed protein in a broad band at 80-90 kDa on SDS-PAGE ( Fig. 6A ). The recombinant enzyme thus differed in size from the native one. This was probably due to glycosylation of the expressed protein of 66 kDa, as deglycosylation with ␣ -mannosidase reduced the size to ‫ف‬ 65 kDa.
Catalytic metal of recombinant 9 S -MnLOX. Mn and Fe contents of the recombinant LOX were determined by AES. The Mn:protein ratio of partly purifi ed enzyme was 0.2:1, which thus represents the lower stoichiometric limit. In this sample, the Mn:Fe ratio was 1:0.05 (0.1124 ± 0.0005 ppm Mn; 0.0059 ± 0.0012 ppm Fe). We therefore conclude that the catalytic metal of the 9 S -LOX is manganese and not iron.

Oxidation of 18:2n-6 and 18:3n-3 by recombinant 9 S -MnLOX.
The 18:2n-6 was oxidized to 9-HPODE ( Fig. 6B ), the end product, and during the linear phase of oxidation also to small amounts of 11-HPODE ( ‫ف‬ 5%) and to traces of 13-HPODE ( ‫ف‬ 1%). The latter did not accumulate, and were apparently transformed to 9-HPODE (see below). The MS 3 spectrum of 9-HPODE is shown in Fig. 6C , and the inset shows that the 9 S stereoisomer of 9-HPODE was formed (Reprosil Chiral NR). The modest formation of 11-HODE during the linear phase of oxidation is shown in Fig. 6D . The apparent K m of 18:2n-6 was estimated to be 3.7 ± 0.7 µM (see supplementary ( K m /k cat ‫ف‬ 0.45). This value is lower than k cat of 13 R -MnLOX (40 s Ϫ 1 ). After correction for the lower Mn content of the 9 S -MnLOX preparation, the difference in turnover numbers appears to be small (42 vs. 40 s Ϫ 1 ).

Real-time PCR analysis
The cDNA of ␤ -tubulin and 9 S -MnLOX were readily detected (with C T values of 23-25 and 27-30 (n = 3), respectively). Grown in complete medium in the presence of rice leaves, we estimated the 9 S -MnLOX mRNA levels in the mycelium, corrected for ␤ -tubulin mRNA, to be only slightly increased (1.6-to 2.6-fold increase ( ⌬ ⌬ C T : 0.65-1.4)) compared with the control grown in parallel without rice leaves.

Recombinant expression of 9 S -MnLOX
Recombinant 9 S -MnLOX was expressed in P. pastoris , but 9 S -MnLOX was unstable in the expression medium even in the presence of protease inhibitors. It was either  Fig. VII). Repeated UV spectra during the decrease in UV absorbance ( Fig. 8C ) showed the appearance of the typical UV spectrum of conjugated trienes with max at ‫ف‬ 270 nm with shoulders at ‫ف‬ 260 and ‫ف‬ 280 nm, and an isobestic point at 253 nm.
Oxidation of bis-allylic hydroperoxides by 9 S -MnLOX and 13 R -MnLOX. 9 S -and 13 R-MnLOX transformed 11 R--HPOTrE to cis-trans -conjugated products at approximately the same rate, about 10-15% of the rate of oxidation of 18:3n-3 by these enzymes (cf. insert in Fig. 8B ). The oxidation of 11 R-HPOTrE by 13 R-MnLOX (70 nM) is shown in supplementary Fig. X, and compared with the nonenzymatic oxidation by 100 µM Mn 3+ in methanol. The enzymatic oxidation to the 11-peroxyl radical with subsequent ␤ -fragmentation and rearrangement to 13 R-HPOTrE occurs at a rate approximately three orders of magnitude higher than the nonenzymatic oxidation by Mn 3+ in methanol; Fe 3+ in methanol does not catalyze this transformation at a signifi cant rate, and sLOX-1 and other FeLOXs are poor catalysts compared with 13 R-MnLOX ( 27,28 ).

DISCUSSION
We have characterized the oxylipin biosynthesis of the rice stem rot pathogen M. salvinii and found it to secrete a novel EAS and linoleate 9 S -LOX. The mycelium mainly displayed 7,8-and 5,8-LDS activities, whereas intracellular 9 S -LOX activity could not be detected. 7,8-LDS is also found in G. graminis and M. oryzae ( 47,48 ). Native 9 S -LOX oxidized 18:2n-6 and 18:3n-3 to the same profi le of metabolites as recombinant 9 S -MnLOX, expressed in P. pastoris . With the aid of the recombinant enzyme, we could determine that the catalytic metal was manganese and not iron. The oxidation of 18:2n-6 by the two secreted enzymes is summarized in Fig. 10 .

Epoxy alcohol synthase
Gel fi ltration showed that the 9 S -LOX activity could be separated from the EAS activity, which implies that the EAS activity could be due to a distinct enzyme (see supplementary Fig. IV). The reaction mechanism of the EAS is probably related to that of peroxidases/epoxygenases with heterolytic cleavage of the O-O bond, and may differ from the homolytic cleavage and subsequent radical mechanism

Site-directed insertion in the pentamer motif and expression of 13 R -MnLOX with excess of Fe
Metal ligands. The Thr and Gly insertion mutants of 13 R -MnLOX were designed to convert the pentamer motif to a hexamer motif to mimic FeLOX, but the mutants with the His-Val-Leu-Phe-Thr-His and His-Val-Leu-Phe-Gly-His motives were inactive. The inactive proteins were analyzed for expression on SDS -PAGE, and the Thr and Gly insertion mutants showed strong bands between 75 and 90 kDa (see supplementary Fig. XI) in the same way as 13 R -MnLOX.
Excess of iron. P. pastoris was fi rst grown with low Fe 2+ concentration to induce metal transporters ( 38 ), and then with excess Fe 2+ in the expression medium. These yeast cells were able to take up ‫ف‬ 20 times more Fe than cells grown under normal conditions, giving a Fe/Mn ratio of 79/1 (see supplementary Fig. XII). The expressed proteins were analyzed for LOX activity with 18:2n-6 and 11 R -HPOTrE as substrates, but no differences were noted. The proteins expressed in cells with high Fe content thus catalyzed the conversion of both 18:2n-6 and 11 R -HPOTrE to products with UV absorbance (235-237 nm) at the same relative rates as 13 R -MnLOX, and formed the same products. 18:2n-6 to peroxyl radicals, which were converted to 9 S-HPODE. The oxidation of 18:3n-3 differed from that of 18:2n-6, inasmuch as 18:3n-3 was transformed to an equal mixture of 9 S-and 13 R-HPOTrE as end products and with signifi cant formation of 11-HPOTrE as an intermediate; the double bond at the 3 position thus shifted the oxidation toward C-11 and C-13.
The mechanism of transformation of 11 S -HPODE to 13 R -HPODE by 13 R -MnLOX was determined with 18 O-labeled 11 S -HPODE by Hamberg,Su,and Oliw ( 22 ). The results seem applicable to the transformation of 11 S -and 13 R -HPODE by 9 S -MnLOX. Incubation with 18 O-labeled 13 R -HPODE suggested that 9 S -HPODE could be formed by ␤ -fragmentation, inasmuch as the 18 O label was partly retained in 9 S -HPODE ( Fig. 7C ). The rate constants for of LOXs and ferrous hemoproteins ( 42,49 ). A schematic overview of the mechanism of biosynthesis of epoxy alcohols by metal enzymes is summarized in Fig. 11 .
The catalytic properties of EAS revealed an unexpected substrate and product specifi city. The EAS of M. salvinii appeared to be specifi c for 9 S -HPODE inasmuch as a series of other hydroperoxides were either not transformed (e.g., 9 S -HPOTrE, 9 S -HPOME, 13 R -HPODE) or were poor substrates (13 S -HPODE and 13 S -HPOTrE). It is therefore unlikely that 13 S -HPODE formed by rice LOX during fungal infection will be metabolized by this enzyme. The EAS transformed 9 S -HPODE to only one product, the threo isomer of 10(11)-epoxy-9 S -hydroxy-12 Z -18:1. EAS of red beets is also product specifi c and transforms 9 S -HPODE to 12 R (13 S )-epoxy-9 S -hydroxy-10 E -18:1 ( Fig. 11 ) ( 50 ), but the EASs of the fungus Saproglenia parasitica and potato leaves can epoxidize several double bonds of FAs ( 43,44 ).
The genome of M. salvinii has not been sequenced. Further characterization of the EAS of M. salvinii will probably require extensive enzyme purifi cation or analysis of expression libraries. It will be of interest to determine whether EAS homologs are secreted by plant pathogens with sequenced genomes. This may disclose the function of EAS in plant-pathogen interactions and whether EASs constitute a distinct gene family.

S -MnLOX and its relation to 13 R -MnLOX
The oxidation of 18:2n-6 by recombinant 9 S-MnLOX to the end product, 9 S-HPODE, is outlined in Fig. 12 . 9 S-MnLOX also formed and oxidized 11-and 13-hydroperoxides of  The epoxy alcohol is unstable and can be partly hydrolyzed to the two triols, which were noted as minor products, isolated from incubation with the concentrated growth medium. Fig. 11. Schematic overview of formation of epoxy alcohols from 9 S -HPODE. Epoxy alcohols can be formed by ferrous and ferric hemoproteins, designated <Fe 2+ > and <Fe 3+ >, respectively, by certain LOXs and by EASs. The scheme shows within brackets to the right, heterolytic cleavage of the O-O bond, probably with formation of a hydroxyl anion and ferryl oxygen (not shown) in analogy with peroxidases ( 49 ), and to the left, homolytic cleavage with formation of a trans epoxide and a dislocated radical at C-11 to C-13 [or, in at least one case, formation of a cis epoxide and the disclocated radical ( 42 )]. The latter are formed from a peroxyl radical at C-9 (not shown). 10 R (11 R )-epoxy-9 S -hydroxy-12Z-18:1 is formed by the EAS of M. salvinii and 12 R (13 S )-epoxy-9 S -hydroxy-10 E -18:1 by EAS of beetroot ( Beta vulgaris) ( 50 ). F347L•13 R-MnLOX acquired many properties of 9 S--MnLOX, inasmuch as it formed the same end products as 9 S -MnLOX from both 18:2n-6 and 18:3n-3. F347V•13 R-MnLOX and F347A•13 R-MnLOX also increased the oxidation at C-9 of 18:2n-6. The Sloane determinant is positioned at the bottom of the substrate channel ( 30,54 ). These three mutants of 13 R -MnLOX with smaller hydrophobic residues than phenylalanine probably allow 18:2n-6 to be positioned further down the substrate channel with retained hydrogen abstraction at C-11, but with increased oxygenation at C-9 or C-11. The products thus shifted from hydroperoxides at 13 R to hydroperoxides at positions 11 S and 9 S (cf. Fig. 1C ). We also examined L350M•9 S -MnLOX and L350F•9 S -MnLOX. The former shifted oxygenation of 18:2n-6 from C-9 toward C-13, suggesting that the methionine residue reduced space for 18:2n-6 at the bottom of the substrate channel. L350F•9 S -MnLOX yielded only low enzyme activity with little effect on the regiospecifi city.
We conclude that alterations in the Sloane determinant can change the oxidation of 18:2n-6 by both enzymes, presumably by altering the substrate position, as fi rst described for mammalian 12 S -and 15 S -LOX, but with retention of hydrogen abstraction ( 30 ).

Metal selection
An intriguing issue is the mechanism of metal selection by LOXs. We examined the insertion of Gly/Thr in the characteristic pentamer motif of 9 S-and 13 R -MnLOX to mimic the hexamer motif of FeLOX, His-(Leu/Trp)-Leu-(Asn/Arg)-(Thr/Gly)-His. Mutants of 13 R -MnLOX with His-Val-Leu-Phe-(Gly/Thr)-His sequences were both inactive. We also assessed whether expression of 13 R -MnLOX in P. pastoris with augmented intracellular iron and reduced manganese concentrations affected the catalytic capacity. This recombinant protein, expressed in cells with an Fe/Mn ratio of 79/1, isomerized 11 R -HPOTrE and oxidized 18:2n-6 as rapidly as native 13 R -MnLOX. Incorporation of iron to a signifi cant extent therefore seemed too unlikely to merit further investigation. The mechanism of iron and manganese selection by 13 R -MnLOXs appears to be strictly regulated, presumably by the fi rst and second coordinating spheres.

What is the biological function of 9 S-MnLOX and epoxy alcohol synthase?
Our investigation is the fi rst report on oxylipin biosynthesis by M. salvinii , which infects the rice stem and causes tissue necrosis. Hydroperoxides, peroxide radicals, and other reactive oxygen species, formed by the secretome of the fungus, may oxidize intracellular lipids and proteins, which can be one of many factors contributing to infection and necrosis ( 18 ). According to current dogma, plants express LOXs and generate oxylipins in defense reactions ( 3 ). LOXs of fungi may also form specifi c oxylipins and participate in this complex chemical warfare ( 12 ). This is probably one of the major functions of 9 S -LOX and EAS of M. salvinii . Fungal oxylipins can also participate in sporulation ( 12 ). M. salvinii forms black sclerotia with conidia in the necrotic rice stem. We therefore examined whether exposure of ␤ -fragmentation of 11-and 13-HPODE are 2.4 × 10 6 and 690 s Ϫ 1 , respectively ( 51 ). These two rate constants are probably larger than the rate constants for oxidation of hydroperoxides to peroxyl radicals by the MnLOXs. This oxidation probably involves a process of facilitated proton-coupled electron transfer, which appears to differ from the oxidation of FAs ( 28 ). The redox potentials of Fe 2+/3+ (0.77 V) and Mn 2+/3+ (1.54 V) render Mn 3+ a higher potential than Fe 3+ in organic LOX mimics ( 52 ). 9 S -and 13 R-MnLOX may therefore be stronger oxidants than FeLOXs. This difference may be refl ected by the rapid oxidation of bis-allylic hydroperoxides by 13 R-MnLOX in comparison with FeLOXs and Mn 3+ in methanol (see supplementary Fig. X) ( 27,28,53 ), which is a characteristic landmark of MnLOXs.
13 R-MnLOX differs from 9 S-MnLOX in regiospecifi city, but they share many biochemical and catalytic properties. Both enzymes are secreted by rice pathogens, abstract the proS hydrogen at C-11 with a marked kinetic deuterium isotope effect, and form hydroperoxides by suprafacial hydrogen abstraction and oxygenation ( 21,22 ). 11 S-HPODE and 11 R-HPOTrE are thus formed in the same way and with the same absolute confi guration (designated S and R because of the effect of the 3 double bond on Cahn-Ingold-Prelog nomenclature rules). Both enzymes also contain glycine in the Coffa-Brash determinant ( Fig. 1B ). The relationship between 9 S -MnLOX and 13 R-MnLOX was illustrated by replacements of the single amino acids Leu350 and Phe347, respectively, in the Sloane determinant ( Fig. 1 ), and by the fact that 13 R-MnLOX binds substrates with entry of the end fi rst ( 27 ). We replaced Leu350 with larger and Phe347 with smaller hydrophobic residues. Fig. 12. Schematic overview of the oxidation of 18:2n-6 by 9 S--MnLOX to the end product, 9 S-HPODE, and to two intermediates. 11 S-HPODE and 13 R-HPODE are probably oxidized to peroxyl radicals by 9 S-MnLOX and then subjected to ␤ -fragmentation and rearrangement to 9 S -HPODE. 13 R-HPODE is mainly converted to 9 S-HPODE with 10 E ,12 Z double-bond confi guration. The 18:3n-3 was oxidized in a similar way, but 9 S -HPOTrE and 13 R -HPOTrE accumulated as end products.
M. salvinii to rice tissues, which induces sporulation ( 20 ), upregulated the expression of mRNA of 9 S -MnLOX. We found only a modest 2-fold increase. A role of 9 S -MnLOX in development of conidia thus seems unlikely.

CONCLUSION
The stem rot fungus of rice secretes a novel substrateand product-specifi c EAS and 9 S-MnLOX with sequence homology and mechanistic similarities to 13 R -MnLOX. Replacement of 9 S -MnLOX in the Sloane determinant with a larger residue (Leu350Met) augmented the oxidation at C-13 of 18:2n-6, whereas replacement of 13 R -MnLOX with smaller hydrophobic residues (Phe347Leu, Phe347Val, Phe347Ala) directed the oxygenation from C-13 toward C-9 and C-11.