Epoxy alcohol synthase of the rice blast fungus represents a novel subfamily of dioxygenase-cytochrome P450 fusion enzymes

Abstract

The genome of the rice blast fungus Magnaporthe oryzae codes for two proteins with N-terminal dioxygenase (DOX) and C-terminal cytochrome P450 (CYP) domains, respectively. One of them, MGG_13239, was confirmed as 7,8-linoleate diol synthase by prokaryotic expression. The other recombinant protein (MGG_10859) possessed prominent 10R-DOX and epoxy alcohol synthase (EAS) activities. This enzyme, 10R-DOX-EAS, transformed 18:2n-6 sequentially to 10(R)-hydroperoxy-8(E),12(Z)-octadecadienoic acid (10R-HPODE) and to 12S(13R)-epoxy-10(R)-hydroxy-8(E)-octadecenoic acid as the end product. Oxygenation at C-10 occurred by retention of the pro-R hydrogen of C-8 of 18:2n-6, suggesting antarafacial hydrogen abstraction and oxygenation. Experiments with ¹⁸O₂ and ¹⁶O₂ gas confirmed that the epoxy alcohol was formed from 10R-HPODE, likely by heterolytic cleavage of the dioxygen bond with formation of P450 compound I, and subsequent intramolecular epoxidation of the 12(Z) double bond. Site-directed mutagenesis demonstrated that the cysteinyl heme ligand of the P450 domain was required for the EAS activity. Replacement of Asn⁹⁶⁵ with Val in the conserved AsnGlnXaaGln sequence revealed that Asn⁹⁶⁵ supported formation of the epoxy alcohol. 10R-DOX-EAS is the first member of a novel subfamily of DOX-CYP fusion proteins of devastating plant pathogens.

Oxylipins designate oxygenated metabolites of unsaturated fatty acids, and they are mainly formed by eukaryotic organisms (1, 2). Arachidonic acid is oxygenated in human cells by cyclooxygenases and cytochrome P450 (CYP) to prostaglandins, thromboxanes, prostacyclin, and related compounds and by lipoxygenases to hydroperoxides, leukotrienes, and epoxy alcohols (3, 4). These eicosanoids exert a plethora of both physiological and pathophysiological functions (e.g., in control of reproduction, cancer development, skin water permeability, fever, and inflammation). Homologous enzymes are found in plants and fungi with functions in reproduction, development, and infection, and they usually use linoleic and linolenic acids as substrates (5–7).

Plant lipoxygenases and allene oxide synthases (AOSs) catalyze the first steps during biosynthesis of jasmonic acid, which is a potent regulator of growth and defense with structural similarities to prostaglandins (7, 8). Pathogens can induce expression of lipoxygenases and fatty acid α-dioxygenase (DOX) in plants with formation of jasmonic acid and other oxylipins, which act as defense molecules. Oxylipins are also formed by fungi, which are human and plant pathogens. These fungi express lipoxygenases and DOX-CYP fusion proteins (5). The DOX domains are homologs to animal heme peroxidases, which include cyclooxygenases and α-DOX, and their CYP domains take advantage of the peroxide shunt pathway in analogy with thromboxane and prostacyclin synthases (9).

The fungal gene family of DOX-CYP fusion proteins contains five subfamilies, and an overview of their oxygenation products is shown in Fig. 1A. DOX-CYP fusion enzymes are self-sufficient, and the two domains are usually functionally linked. The first two described subfamilies were 5,8- and 7,8-linoleate diol synthase (LDS) of the genera Aspergillus and Gaeumannomyces. LDSs oxidize linoleic acid sequentially to 8(R)-hydroperoxy-9(Z),12(Z)-octadecadienoic acid (8R-HPODE) and to 5,8- or 7,8-diols, by their DOX and CYP domains, respectively (10–13). A third subfamily comprises 10R-DOX of aspergilli, and these proteins lack the functional heme thiolate ligand in the CYP domain (14, 15). Recent additions are the 9R-DOX-AOS subfamily of certain aspergilli and the 9S-DOX-AOS subfamily of the genera Fusarium and Colletotrichum (16, 17).

Fig. 1.

Overview of oxylipin biosynthesis by the genera Aspergillus, Gaeumannomyces, Magnaporthe, Colletotrichum, and Fusarium and phylogenetic tree of DOX-CYP subfamilies. A: Linoleic acid is oxidized by 5,8-LDS, 10R-DOX, and 9R-DOX-AOS of A. terreus and other aspergilli (top). Oxylipins formed from 18:2n-6 by 7,8-LDS of Gaeumannomyces and Magnaporthe (e.g., G. graminis, M. oryzae) and by 9S-DOX-AOS of F. oxysporum and C. graminicola (bottom). B: Phylogenetic tree of MGG_10859 (M. oryzae) and three homologous sequences (EJT82559, EFQ36272, and FOXB_03425, respectively) with >60% amino acid identity covering >90% of the sequence. The next sequences are from top to bottom: 10R-DOX (ACL14177, ATEG_04755, and AAT36614), 7,8-LDS (MGG_13239, Q9UUS2), 5,8-LDS (ACO55067, ATEG_03992), 9S-DOX-AOS (EFQ27323, FOXB_01332), and 9R-(DOX)-AOS [ATEG_02036, EHA25900 (Hoffman and Oliw, unpublished observation)]; the parentheses in 9R-(DOX)-AOS emphasize that recombinant ATEG_02036, in contrast to recombinant EHA25900, lacks the 9R-DOX activities.

Genome sequences of pathogenic fungi are being published at an increasing rate (18, 19). The catalytic activities of emerging DOX-CYP fusion proteins have not yet been fully determined. Putative new DOX-CYP subfamilies can be described by amino acid identity and by the linked and specific dual catalytic activities. In contrast, mammalian P450 families and subfamilies are defined by >40% and 55% amino acid identity, respectively, and not by their catalytic activities due to the functional redundancy within subfamilies. DOX-CYP fusion enzymes of different subfamilies can generally be aligned with 35% to 50% sequence identity, whereas enzymes of subfamilies with identical catalytic activities usually align with 60% or higher sequence identity. Based on the latter, a tentative new DOX-CYP subfamily could be identified in several genera of the top 10 plant pathogens (e.g., Magnaporthe, Colletotrichum, and Fusarium) (Fig. 1B) (20).

Wheat, rice, maize, and banana are staple crops, and rice feeds half of the world’s population. Fungi constitute a constant threat to this food supply. Magnaporthe oryza³ causes rice blast disease, which annually reduces the rice harvest by 10% to 30% (21). Colletotrichum graminicola infects maize at annual costs in Northern America on the order of 1 billion US$. Both fungi share a common infection process. Their conidia develop into an injection apparatus, the appressorium, required for blast cuticle penetration (21, 22). M. oryzae is the prototype organism for studying this process. Its appressorium accumulates lipid bodies, trehalose, and glycerol from the conidium, and the latter generates the turgor pressure for blast penetration. The catalytic subunit of cAMP-dependent protein kinase A (PKA), mitogen-activated protein kinase PMK-1 (PMK1), and triacylglycerol lipases take part in mobilization of fatty acids of lipid bodies in support of the infection process (23–27).

The genome of M. oryzae is sequenced (28). It is estimated to contain about 11,000 genes, and 2 of them code for DOX-CYP fusion proteins. One of them, MGG_13239,⁴ was previously identified as 7,8-LDS by gene deletion (29). The catalytic function of the second gene, MGG_10859,⁵ is unknown, but MGG_10859 is upregulated during appressorium formation and downregulated by gene deletion of PMK1⁶ (30). Circumstantial evidence suggests that MGG_10859 may also be regulated by PKA. A mutant strain of M. oryzae (DA-99) with constitutive PKA activity (Δmac1 sum1-99) oxidized linoleic acid to a new profile of metabolites, likely due to expression of linoleate 10R-DOX with MGG_10859 as a possible but unproven candidate gene (29). The PMK1-controlled expression of MGG_10859 and the distribution of MGG_10859 homologs in devastating plant pathogens [Fig. 1B; Gaeumannomyces graminis (Take-all of wheat), C. graminicola (Anthracnose of maize), and Fusarium oxysporum (Panama disease of banana)] indicate a pathophysiological function, which is noteworthy.

The first objective of the present report was to express the two putative DOX-CYP fusion proteins of M. oryzae, MGG_10859 and MGG_13239, because the former could be the first member of a new subfamily (Fig. 1B). The second objective was to reexamine the products formed by the mutant strain DA-99 with constitutive PKA activity (Δmac1 sum1-99) of M. oryzae with regard to the catalytic activities of MGG_10859 and MGG_13239. Finally, we also expressed the MGG_10859 homolog of F. oxysporum, FOXB_03425,⁷ to determine whether MGG_10859 and this protein possess the same enzyme activities.

MATERIALS AND METHODS

Materials

Fatty acids were dissolved in ethanol and stored in stock solutions (50–100 mM) at −20°C. The 18:2n-6 (99%), 18:3n-6 (99%), and 18:3n-3 (99%) were from VWR. The 18:1n-9, 20:2n-6, 20:3n-3 (98%–99%), and [¹³C₁₈]18:2n-6 (98%) were from Larodan (Malmö, Sweden). The [8R-²H]18:2n-6 was prepared as described (13, 31). M. oryzae strain DA-99 (Δmac1 sum1-99) (25) was provided by Dr. A. Sesma, Universidad Politécnica de Madrid, Spain. Phusion DNA polymerase and chemically competent Escherichia coli (NEB5α) were from New England BioLabs. Restriction enzymes were from New England BioLabs and Fermentas. Champion pET Directional TOPO Kit was from Invitrogen. Gel extraction kit and Pfu DNA polymerase were from Fermentas. RNaseA, ampicillin, and methylene blue were from Sigma. Sequencing was performed at Uppsala Genome Center (Biomedical Center, Uppsala University). The open reading frames of MGG_13239 (3,527 bp), MGG_10859 (3,462 bp), and FOXB_03425 (3,306 bp) were ordered in pUC57 vectors from GenScript (Piscatawy, NJ). PCR primers were ordered from TIB Molbiol (Berlin, Germany). The (+)- and (−)-vernolic acids were from Lipidox, and photooxidation of vernolic acids was performed with methylene blue (32). ¹⁸O₂ gas (97%) was obtained from Isotec (Sigma-Aldrich). SepPak columns (silicic acid or C₁₈) were from Waters.

Expression of recombinant proteins

The open reading frames of MGG_13239, MGG_10859, and FOXB_03425 in pUC57 vectors were transferred to pET101D-TOPO vectors by PCR technology according to Invitrogen’s instructions (all primers are listed in supplementary Table I). Competent E. coli (BL21) cells were transformed with the expression constructs by heat-shocking. Cells were grown until A₆₀₀ of 0.6–0.8 in 2xYT medium prior to addition of 0.1 mM isopropyl β-d-1-thiogalactopyranoside to induce protein expression. After 5 h under moderate shaking (∼100 rpm) at room temperature, the cells were harvested by centrifugation (13,000 rpm, 4°C, 25 min) and sonicated (Bioruptor Next Gen, 10 × 30 s, 4°C). Cell debris was removed by centrifugation, and the supernatants were used immediately or frozen at −80°C until needed. Each protein was expressed in at least three independent expression experiments.

Site-directed mutagenesis of recombinant proteins

Site-directed mutagenesis was performed according to the Quick Change protocol (Stratagene) with 10 ng of the pUC57 constructs as templates and Pfu DNA polymerase (16 cycles). PCR products were incubated with DpnI (37°C, 2 h) to digest maternal DNA. Gel electrophoresis confirmed amplification of one distinct PCR product, which was then used for transformation of E. coli (NEB5α) cells by heat-shocking. All mutations were confirmed by sequencing before subcloning to pET101D-TOPO vectors described previously. All primers are listed in supplementary Table II.

Enzyme assays

Recombinant proteins were incubated with 100 µM 18:2n-6 or other fatty acids for 30 min on ice. The reaction (0.3–0.5 ml) was terminated with methanol (2–4 vol), and proteins were removed by centrifugation. The metabolites were extracted on octadecyl silica (SepPak/C₁₈), evaporated to dryness, and diluted in ethanol (40–60 µl), and 7–10 µl was subject to LC/MS² analysis.

Nitrogen powder of ground fungus was homogenized (glass-Teflon, 10 passes; 4°C) in 10 vol (w/v) of 0.1 mM KHPO₄ buffer [(pH 7.3)/2 mM EDTA/0.04% Tween-20], centrifuged at 13,000g (10 min, 4°C), and immediately used for enzyme assay. An aliquot (0.5–1 ml) of the supernatant was incubated with 100 µM of fatty acids for 30–40 min on ice. The products were extracted with ethyl acetate or on a cartridge with octadecyl silica (SepPak/C₁₈). Triphenylphosphine or NaBH₄ was used to reduce hydroperoxides to alcohols.

Photooxidation of vernolic acids

Ten milligrams of 12S(13R)-epoxy-9(Z)-octadecenoic acid [(+)-vernolic acid] was photooxidized in 4 ml methanol under oxygen gas at room temperature with 0.01 molar equivalent methylene blue for 15 h (32). The reaction was followed by TLC until almost all substrate was oxidized. The reaction was evaporated to 1 ml and cooled on ice, and hydroperoxides were reduced to alcohols by addition of NaBH₄. Nine milliliters of 0.2 M sodium acetate buffer (pH 4.0) was added, and the material was extracted on a cartridge of octadecyl silica (SepPak/C₁₈). The products were then separated on a silicic acid cartridge (SepPak). Unreacted (+)-vernolic acid was eluted with 50 ml diethyl ether-hexane-acetic acid, 5:95:0.1, and the epoxy alcohols were eluted with diethyl ether-hexane-acetic acid, 50:50:0.1. The fractions with 12S(13R)-epoxy-10(S/R)-hydroxy-8(E)-octadecenoic acids were combined, and the syn and anti stereoisomers were separated by normal phase (NP)-HPLC (250 × 4.6 mm; Nucleosil 50-5) with hexane-isopropyl alcohol-acetic acid, 97:3:0.1, as eluent.

The 12R(13S)-epoxy-9(Z)-octadecenoic acid [(−)-vernolic acid] was oxidized by the same procedure, and the syn and anti isomers were separated in the same way (NP-HPLC).

Growth of M. oryzae (Guy11) and strain DA-99

M. oryzae Guy11 was grown in liquid culture for 5–16 days in complete medium at 22°C with a cycle of 16 h of fluorescent light (True-light) followed by 8 h of darkness (29). The genetically modified strain DA-99 of M. oryzae [Δmac1 sum1-99 (25)] was grown in the same way. Mycelia (0.5–20 g wet weight) were ground to a powder in liquid nitrogen and stored at −80°C. Aliquots were thawed and homogenized, and the supernatants were assayed for enzyme activity with 18:2n-6 as substrate as described previously. Nitrogen powders of both strains were assayed in parallel at different time points of growth. Each experiment was performed at least in duplicate. Conidia were harvested from mycelia, which had grown for 8–12 days on complete medium agar plates with a glass spreader and by rinsing with distilled water.

LC/MS analysis

Reversed phase (RP)-HPLC with MS² analysis was performed with a Surveyor MS pump (ThermoFisher) and an octadecyl silica column (5 µm; 2.0 × 150 mm; Phenomenex), which was eluted at 0.3 ml/min with methanol-water-acetic acid, 800:200:0.05, or 750:250:0.05. The effluent was subject to ESI in a linear ion trap mass spectrometer (LTQ, ThermoFisher). The heated transfer capillary was set at 315°C, the ion isolation width at 1.5 amu, the collision energy at 35 (arbitrary scale), and the tube lens varied between 90 and 120 V. Prostaglandin F_1α was infused for tuning. Samples were injected manually (Rheodyne 7510) or by an autosampler (Surveyor Autosampler Plus, ThermoFisher).

NP-HPLC with MS² analysis was performed with a silicic acid column (5 µm; Kromasil 100SI, 250 × 2 mm, Dalco Chromtech, or 250 × 4.6 mm; Nucleosil 50-5) using 3% isopropyl alcohol in hexane for separation of hydroxy fatty acids and epoxy alcohols (0.3–0.5 ml/min; Constrametric 3200 pump, LDC/MiltonRoy). The effluent was combined with isopropyl alcohol-water (3:2; 0.2–0.3 ml/min) from a second pump (Surveyor MS pump). The combined effluents were introduced by ESI into the ion trap mass spectrometer (LTQ, ThermoFisher). Steric analysis of 8-hydroperoxyoctadecamonoenoic acid (HPOME) and 10-hydroperoxyoctadecadienoic acid (HPODE) was performed by chiral phase (CP)-HPLC/MS² analysis on Reprosil Chiral NR as described (33).

Hydrogenation was performed with Pd/C in methanol and a small stream of hydrogen for 90 s. The catalyst was removed by filtration through Na₂SO₄. Chlorohydrin adducts of epoxides were prepared by reaction with 40 μl of methoxamine HCl in pyridine (20 mg/ml; 60°C, 1 h) (34). Water was added, and the products were purified on a C₁₈ cartridge (SepPack/C₁₈) column before analysis by HPLC/MS².

Bioinformatics

The ClustalW algorithm was used for sequence alignments (Lasergene, DNASTAR Inc.), and the MEGA6 software for construction of phylogenetic trees with bootstrap tests of the resulting nodes (35). The distance between the branches is indicative of the expected number of substitutions per amino acid position.

RESULTS

Expression and oxidation of 18:2n-6 and 18:3n-3 by MGG_10859 (10R-DOX-EAS)

Recombinant MGG_10859 was obtained by expression in E. coli BL21 cells. This protein readily oxidized 18:2n-6 and 18:3n-3 as shown in Fig. 2. The18:2n-6 was oxidized to an epoxy alcohol as main metabolite and to smaller amounts of 10(R)-hydroperoxy-8(E),12(Z)-octadecadienoic acid (10R-HPODE), 8-HPODE, and their corresponding alcohols (Fig. 2A). The 18:3n-3 was oxygenated in an analogous way as 18:2n-6 with oxidation at C-8, C-10, and at C-10 with formation of an epoxy alcohol (Fig. 2B).

Fig. 2.

LC/MS analysis of oxygenation products formed from 18:2n-6 and 18:3n-3 by recombinant MGG_10859 (10R-DOX-EAS) and the mechanism of biosynthesis of hydroperoxides and epoxy alcohols. A: Products formed from 18:2n-6. Peak I contains 12S(13R)-epoxy-10(R)-hydroxy-8(E)-octadecenoic acid (see text). TIC, total ion current. B: Oxygenation products from 18:3n-3. Peak II contains 12(13)-epoxy-10(R)-hydroxy-8(E),15(Z)-octadecadienoic acid (see text). C: CP-HPLC of 10-HPODE. The R enantiomer of 10-HPODE is formed with high stereoselectivity. D: Molecular anions of the epoxy alcohol formed from 18:2n-6 under an atmosphere of both ¹⁶O₂ and ¹⁸O₂ were noted at m/z 311 {[10,12(13)-¹⁶O₂]labeled epoxy alcohol anion} and 315 {[10,12(13)-¹⁸O₂]labeled epoxy alcohol anion (A⁻)}. E: MS² spectrum (m/z 295–296 → full scan) of HODEs, which were obtained after incubation of recombinant 10R-DOX-EAS with [8R-²H]18:2n-6 (64% ²H) and reduction to alcohols. The 8- and 10-HODE are formed with retention of the deuterium label as indicated by the values m/z 158 [⁻OOC-(CH₂)₆-C²HO] and 184 [⁻OOC-(CH₂)₆-CH = CH-C²HO], respectively.

Steric analysis showed that 10R-HPODE was formed with high stereoselectivity (Fig. 2C). This metabolite was apparently transformed to an epoxy alcohol, which was identified as 12S(13R)-epoxy-10R-hydroxy-8(E)-octadecenoic acid as described subsequently. The epoxy alcohol was formed from 10R-HPODE by an intramolecular rearrangement mechanism as judged by experiments performed under a mixture of ¹⁶O₂ and ¹⁸O₂ gas. The anions of 12(13)-epoxy-10-hydroxy-8(E)-octadecenoic acid, which were obtained under this condition, had m/z values of either 311 or 315, but not 313 (Fig. 2D). This demonstrates that the oxygen atom in the epoxy group of one molecule must have evolved from the hydroperoxy group of the same molecule. MGG_10859 will therefore be referred to as 10R-DOX-EAS.

MGG_10859 oxidized [8R-²H]18:2n-6 at C-10 and at C-8, and this occurred with retention of the pro-R deuterium label as shown in Fig. 2E. The retention of the deuterium label suggested that the hydroperoxides were formed by antarafacial hydrogen abstraction and oxygen insertion in analogy with hydroperoxides formed by 7,8-LDS of G. graminis and by 10R-DOX and 5,8-LDS of Aspergillus fumigatus (11, 13).

LC/MS analysis of the epoxy alcohols

The MS² spectra of the epoxy alcohols formed from 18:2n-6 and 18:3n-3 are shown in Fig. 3. Both spectra contain unexpected rearrangement ions.

Fig. 3.

MS² spectra of the epoxy alcohols formed from 18:2n-6 and 18:3n-3 by recombinant 10R-DOX-EAS, analysis of chlorohydrin derivatives, and separation of isotopes and stereoisomers. A: MS² spectrum of the epoxy alcohol [12S(13R)-epoxy-10(R)-hydroxy-8(E)-octadecenoic acid] formed from 18:2n-6. B, C: LC/MS analysis (m/z 347–349 full scan) of chlorohydrin derivatives, which were obtained by treatment of products formed from 18:2n-6 by recombinant MGG_10589 with methoxime HCl. The two main peaks, I and II, in B yielded the same MS² and MS³ spectra. The molecular anions at m/z 347 and 349 with incorporation of one chlorine atom (³⁵Cl or ³⁷Cl). D: Products formed during oxidation of 12R(13S)-epoxy-(9Z)-octadecenoic acid by MGG_10859 and separation of the syn and anti isomers of 12R(13S)-epoxy-10-hydroxy-8(E)-octadecenoic acid by NP-HPLC (peaks marked I and II). The partial chromatogram of the inset was obtained after addition of the epoxy alcohol formed from 18:2n-6 by MGG_10859 to the products formed from 12R(13S)-epoxy-(9Z)-octadecenoic acid, illustrating that biological product formed from 18:n-6 coelutes with the first eluting stereoisomer in peak I [syn; 12S(13R)-epoxy-10(R)-hydroxy-8(E)-octadecenoic acid; see text for details]. E: MS² spectrum of 12(13)-epoxy-10(R)-hydroxy-8(E),15(Z)-octadecenoic acid formed from 18:3n-3 by 10R-DOX-EAS.

12(13)-Epoxy-10-hydroxy-8(E)-octadecenoic acid.

LC/MS analysis revealed a complex fragmentation of the epoxy alcohol formed from 10R-HPODE. The MS² spectrum (m/z 311 → full scan) is shown in Fig. 3A. The spectrum shows two signals, m/z 183 [⁻OOC-(CH₂)₆-CH = CH-CHO] and m/z 155 [183–28; ⁻OOC-(CH₂)₆-CH = CH₂], which are formed by α-cleavage at both sides of C-10. Signals were also noted at m/z 211 (A⁻-100), 193, and 181, likely due to rearrangements. This spectrum suggested that the epoxide group could possibly be positioned at C-12 and C-13, but not at C-8 and C-9 (cf. Ref. 36).

We next examined the fragmentation of the epoxy alcohol with regard to the signals at m/z 211, 193, and 181 by recording the MS² spectra of three derivatives: i) the hydrogenated epoxy alcohol, ii) the uniformly ¹³C-labeled epoxy alcohol, and iii) and the epoxy alcohol formed under ¹⁸O₂. These mass spectra are shown in the supplementary Figs. I–III.

The MS² spectrum of the hydrogenated epoxy alcohol showed the rearrangement ions at m/z 183 and 211 (supplementary Fig. I). The MS² spectrum of the uniformly labeled epoxy alcohol showed that the rearrangement ions at m/z 211 and 181 contained 12 and 11 carbon atoms, respectively, as they were increased to m/z 223 and 192, respectively (supplementary Fig. II). The MS² spectrum of the epoxy alcohol with ¹⁸O-labeling of the hydroxy and the epoxide groups suggested that the ion at m/z 211 retained both oxygen atoms, as this ion shifted to m/z 215 (supplementary Fig. III). These results suggest that the signal at m/z 211 (A⁻-(18+82)) could be due to loss of water and C₆H₁₀ from the carboxylate anion (A⁻). This mechanism was also supported by analysis of the corresponding rearrangement ions in the MS² spectra of the epoxy alcohol formed from 18:3n-3, 20:2n-6, and 20:3n-3 described subsequently. The precise fragmentation mechanism remains to be determined.

The complex spectra of the epoxy alcohol discussed previously did not clearly show the presence of an epoxide. We therefore prepared chlorohydrin adducts, and they were partly separated by RP-HPLC (Fig. 3B). The resulting molecular anions confirmed opening of an epoxide with incorporation of ³⁵Cl and ³⁷Cl and formation of a hydroxyl group (Fig. 3C). The MS² spectrum of the chlorohydrin derivatives showed an intense signal at m/z 311 [A⁻-(36 or 38), loss of HCl], and the MS³ spectrum (m/z 347–349 → 311 → full scan) yielded the same mass spectrum as recorded for the original epoxy alcohol (Fig. 3A). We finally confirmed the structure of the epoxy alcohol by comparison with chemical standards.

Steric analysis of 12(13)-epoxy-10-hydroxy-8(E)-octadecenoic acid.

The epoxy alcohol formed from 10R-HPODE was identified as the syn stereoisomer by its retention time on NP-HPLC in comparison with synthetic and biological standards. Photooxidation of (+)- and (−)-vernolic acids, 12S(13R)-epoxy- and 12R(13S)-epoxy-9(Z)-octadecenoic acids, respectively, yielded two pairs of syn and anti stereoisomers with the same MS² spectrum as discussed previously (Fig. 3A). The syn and anti stereoisomers were separated by NP-HPLC, but the two syn and the two anti stereoisomers were mirror images, and they were not separated by NP-HPLC.

Two stereoisomers of the epoxy alcohol were also obtained by oxidation of (−)-vernolic acid [12R(13S)-epoxy-9(Z)-octadecenoic acid] with 10R-DOX-EAS (after reduction of the hydroperoxides to alcohols) at C-10 in a 1:10 ratio (Fig. 3D). Products were also formed by oxidation at C-8 [12R(13S)-epoxy-8-hydroxy-9(Z)-octadecenoic acids; data not shown]. Based on the stereoselective oxidation at C-10, we identified the major metabolite, which eluted with the longest retention time, as the anti stereoisomer [2R(13S)-epoxy-10(R)-hydroxy-9(Z)-octadecenoic acid]. The epoxy alcohol formed from 18:2n-6 eluted as a single peak, which coeluted with the first eluting stereoisomer of these two biological standards on NP-HPLC (insert in Fig. 3D). It was thus identified as the syn stereoisomer, 12S(13R)-epoxy-10(R)-hydroxy-9(Z)-octadecenoic acid.

12(13)-Epoxy-10-hydroxy-8(E),15(Z)-octadecadienoic acid.

The LC/MS spectrum of the epoxy alcohol formed from 18:3n-3 is shown in Fig. 3E. The presence of an alcohol at C-10 was evident from the strong fragments at m/z 183 and 155 formed by α-cleavage at both sides of C-10. The rearrangement ion at m/z 209 (A⁻-100) was apparently formed by the same mechanism as the m/z 211 ion discussed previously. Importantly, this spectrum shows that the rearrangement cannot be formed by cleavage at C-11, which would yield a loss of 98 (OCH-CH₂-CH = CH-CH₂-CH₃) and not a loss of 100.

Oxidation of 18:1n-9, 18:3n-6, 20:2n-6, and 20:3n-3 by MGG_10859 (10R-DOX-EAS)

18:1n-9.

NP-HPLC/MS² analysis showed that oleic acid was oxidized by the 10R-DOX activity to 8-H(P)OME and 10-H(P)OME in a ratio of >5:1. CP-HPLC analysis showed that 8-HPOME consisted of the 8R stereoisomer to >95%.

18:3n-6.

This fatty acid was a poor substrate, and it was not oxidized to specific products.

20:2n-6.

The 20:2n-6 was oxidized at C-10 and C-12 as judged from NP-HPLC/MS analysis of hydroperoxides before and after reduction to alcohols. Both hydroperoxides were also converted to epoxy alcohols. The MS² spectrum of one epoxy alcohol was consistent with 10(11)-epoxy-12-hydroxy-14(Z)-eicosenoic acid (data not shown), possibly formed by homolytic cleavage of the 10-hydroperoxy metabolite. The MS² spectrum of the other epoxy alcohol, 14(15)-epoxy-12-hydroxy-10(E)-eicosenoic acid, is shown in supplementary Fig. IVA, and it allows a comparison with the spectrum in Fig. 3A. This illustrates that many fragments differed by 28, which supports the rearrangement mechanism discussed previously with loss of C₆H₁₀ and water from the carboxylate anion.

20:3n-3.

The 20:3n-3 was oxidized in analogy with 20:2n-6 at C-10 and C-12. The hydroperoxide at C-10 was transformed to significant amounts of 10(11)-epoxy-12-hydroxy-14(Z),17(Z)-eicosadienoic acid and the hydroperoxide at C-12 to 14(15)-epoxy-12-hydroxy-14(Z),17(Z)-eicosadienoic acid. The MS² spectrum of the latter is shown in supplementary Fig. IVB for comparison with Fig. 3B.

Conversion of other hydroperoxides by the EAS activity

The capability to convert hydroperoxides other than 10R-HPODE was evaluated by incubations with 10S-, 9R-, 9S-, 13R-, and 13S-HPODE, respectively. Only 13R- and 13S-HPODE were converted to epoxy alcohols in significant amounts. Both formed the threo stereoisomer of 12(13)-epoxy-11-hydroxy-9(Z)-octadecenoic acids as the main metabolite (erythro/threo, 1:4) as judged by NP-HPLC analysis (supplementary Fig. V). Because the hydroperoxides were mostly converted to the threo isomers, it is possible that this occurred enzymatically.

Crucial amino acid residues of the EAS domain of MGG_10859

Catalytically important amino acid residues of the EAS domains are likely conserved residues of the I-helices and the heme thiolate pocket, as outlined by partial alignment of the CYP domains of 10R-DOX-EAS with three putative DOX-EAS and two 7,8-LDS sequences (Fig. 4A).

Fig. 4.

Comparison of partial sequences of four members of the 10R-DOX-EAS subfamily with 7,8-LDS of M. oryzae and G. graminis and effects of site-directed mutagenesis of Cys and Asn residues. A: The alignment covers the conserved AsnGlnXaaGln sequences, which are believed to be important for substrate positioning, the GluXaaXaaArg (“EXXR”) sequences, functioning as a salt bridge, and the motif with the heme thiolate ligand. The latter two sequences are typical of P450s. Asterisks indicate the three amino acid residues of MGG_10589, which were subject to site-directed mutagenesis. The sequence identity within the putative 10R-DOX-EAS subfamily is 60% or higher. B: RP-HPLC analysis of products formed from 18:2n-6 by 10R-DOX-EAS (MGG_10859). C: RP-HPLC/MS analysis of products formed from 18:2n-6 by 10R-DOX-EAS•C1086S. D: RP-HPLC/MS analysis of products formed from 18:2n-6 by 10R-DOX-EAS•N965V. Peak I designates 12S(13R)-epoxy-10(R)-hydroxy-8(E)-octadecenoic acid, which was identified by MS² analysis in all three chromatograms.

The Cys1086Ser replacement of MGG_10859 fully supported the oxidation of 18:2n-6 by the 10R-DOX activities, but only traces of the epoxy alcohol were noted in comparison with native 10R-DOX-EAS (Fig. 4B, C). These small amounts were presumably formed by the heme group of the 10R-DOX domain in analogy with 10R-DOX of A. fumigatus and Aspergillus nidulans, which lack the heme thiolate ligand (14, 15).

We next examined two amide residues in the conserved Asn⁹⁶⁵GlnXaaGln motif of 10R-DOX-EAS, which is present in three homologs of 10R-DOX-EAS and in two 7,8-LDS sequences (Fig. 4A). The Gln968Leu replacement of 10R-DOX-EAS did not appear to alter the biosynthesis of the epoxy alcohol from 18:2n-6, but it was strongly influenced by the Asn965Val replacement (Fig. 4D). It is difficult to exactly quantify this reduction in formation of the epoxy alcohol, but the relative amount of epoxy alcohol to 10-HPODE was altered from 92% in the wild-type enzyme to 29% in the Asn965Val mutant. Asn⁹⁶⁵ thus supports the EAS activity, but it is not absolutely required. This is also the case with the catalytically important Asn residues of CYP74 and 9R-(DOX)-AOS of A. terreus (17, 37); the parentheses indicate that the latter appears to lack 9R-DOX activity.

Reanalysis of oxylipins formed by M. oryzae strain DA-99 with constitutive PKA activities

The oxidation of 18:2n-6 by nitrogen powder of mycelia of DA-99 (Δmac1 sum1-99) yielded a different profile of metabolites than the native strain (Guy11) as reported previously (29). DA-99 consistently formed 10R-HPODE and 8R-HPODE as main metabolites along with variable but relatively small amounts of 7,8-dihydroxy-9(Z),12(Z)-octadecadienoic acid (DiHODE) and 5,8-DiHODE (Fig. 5A), whereas M. oryzae (Guy11) transformed 18:2n-6 to 7,8-DiHODE and 8-HPODE as the main products (insert in Fig. 5A). We decided to reexamine these findings in the light of the properties of recombinant 10R-DOX-EAS and 7,8-LDS (MGG_13239).

Fig. 5.

LC/MS analysis of the oxidation of 18:2n-6 by M. oryzae DA-99, Guy11, and recombinant 7,8-LDS (MGG_13239). A: RP-HPLC analysis of products formed from nitrogen powder of DA-99. The inset shows separation by RP-HPLC of metabolites formed by nitrogen powder of M. oryzae Guy11. B: NP-HPLC analysis of metabolites, which were formed by recombinant 7,8-LDS (MGG_13239). The chromatograms show the TIC. Products were identified by their MS² spectra as indicated.

A time curve covering a growth period of DA-99 for up to 16 days showed a variable formation of 5,8- and 7,8-DiHODE and a prominent biosynthesis of 10R-HPODE and 8R-HPODE (cf. Ref. 29). The P450 activity of 7,8-LDS thus appeared to be reduced in comparison with M. oryzae Guy11. 12(13)-Epoxy-10-hydroxy-8(E)-octadecenoic acid could be detected at some time points by its characteristic MS² spectrum (Fig. 3A). The metabolite eluted on the right shoulder of 5,8-DiHODE (supplementary Fig. VI). NP-HPLC analysis of additional experiments showed that the epoxide was a racemic mixture of syn and anti stereoisomers. This showed that 10R-DOX-EAS did not form these epoxy alcohols.

Recombinant expression of 7,8-LDS of M. oryzae

We examined recombinant 7,8-LDS for biosynthesis of major and minor metabolites, which might contribute to the oxylipin profile of DA-99. The recombinant protein transformed 18:2n-6 sequentially to 8-HPODE and 7,8-DiHODE as major products. This enzyme was therefore confirmed to possess 7,8-LDS activities, as previously indicated by gene deletion (29). NP-HPLC/MS² analysis also revealed significant biosynthesis of several minor products as illustrated in Fig. 5B. Trace amounts of 10-H(P)ODE were also formed, but 5,8-DiHODE could not be detected.

The protein sequence of 7,8-LDS contains several predicted phosphorylation sites with Ser and Thr residues (38). Single residues of five of these sites were altered by replacements with Asp or Glu residues (Ser350Asp in the 8R-DOX domain, and Ser753Glu, Ser975Asp, Thr998Glu, and Thr1098Asp in the CYP domain), but this did not seem to affect the enzymatic oxidation of 18:2n-6 (data not shown).

Site-directed mutagenesis studies revealed that Asn⁹⁴⁶, which is present in the AsnGlnXaaGln motif mentioned previously (Fig. 4A), facilitates the conversion of 8-HPODE to 7,8-DiHODE. MGG_13239•Asn946Val only formed small amounts of the 7,8-diol (supplementary Fig. VII) compared with the native enzyme (Fig. 5B).

Recombinant expression of FOXB_03425 of F. oxysporum

Recombinant FOXB_03425 converted 18:2n-6 to 10-HPODE, 8-HPODE, and an epoxy alcohol, as shown in Fig. 6A. LC/MS analysis of the epoxy alcohol yielded the same MS² spectrum as 12(13)-epoxy-10-hydroxy-8(E)-octadecenoic acid (Fig. 3A). This epoxy alcohol was also formed from 10R-HPODE, but not from 10S-HPODE (Fig. 6B). The relative amounts of HPODE and the epoxy alcohol varied. CP-HPLC/MS² analysis confirmed that 18:2n-6 was oxidized mainly to the R stereoisomer of 10-HPODE (data not shown). We conclude that FOXB_03425 is a 10R-DOX-EAS fusion enzyme. It aligns with 64% amino acid identity to MGG_10859.

Fig. 6.

RP-HPLC/MS² analysis of oxygenation products of 18:2n-6 formed by recombinant FOXB_03425 and transformation of 10-HPODE by this enzyme. A: RP-HPLC/MS² analysis revealed biosynthesis of 8- and 10-H(P)ODE, and 12(13)-epoxy-10-hydroxy-(8E)-octadecenoic acid (peak I) from 18:2n-6. B: The top chromatogram shows transformation of 10R-HPODE by FOXB_03425 to 12(13)-epoxy-10(R)-hydroxy-(8E)-octadecenoic acid (peak I), and the bottom chromatogram shows that 10S-HPODE was a poor substrate.

DISCUSSION

The discovery of the first fungal EAS with its unique self-sufficient reaction mechanism is our main finding. The dual enzyme activities are combined in a DOX-CYP fusion protein, MGG_10859, of the rice blast fungus M. oryzae. The enzyme is designated 10R-DOX-EAS because it sequentially converts linoleic and α-linolenic acids to 10R-hydroperoxides and to 12S(13R)-epoxy-10R-hydroxy metabolites as end products by intramolecular oxygen transfer (Fig. 7).

Fig. 7.

Enzymatic conversion of 18:2n-6 by recombinant 10R-DOX-EAS. Abstraction of the pro-S hydrogen of C-8 is followed by antarafacial oxygen insertion mainly at C-10 and to some extent at C-8, yielding 10R- and 8R-HPODE, respectively. 10R-HPODE is further converted by the EAS activity to 12S(13R)-epoxy-10(R)-hydroxy-8(E)-octadecenoic acid by heterolytic scission of the O-O bond with formation of P450 compound I in a process supported by Asn⁹⁶⁵ (cf. Fig. 4).

Epoxy alcohols can be formed from fatty acid hydroperoxides by two main mechanisms: homolytic or heterolytic cleavage of the hydroperoxide oxygen-oxygen bond. Hematin, hemoproteins, and many enzymes (e.g., lipoxygenases and P450) catalyze homolytic cleavage with formation of epoxy alcohols (36, 39, 40). 10R-DOX-EAS likely forms epoxy alcohols by heterolytic cleavage of the O-O bond of 10R-HPODE with formation of a hydroxyl anion and P450 compound I for subsequent epoxidation (Fig. 7). To the best of our knowledge, 10R-DOX-EAS is the first described EAS with an intramolecular and position-specific oxygenation mechanism associated with heterolytic scission of the dioxygen bond. Heterolytic scission of dioxygen bonds are also catalyzed by heme peroxidases and presumably also by the secreted EAS of Magnaporthe salvinii (41, 42). Intramolecular oxygen transfer has also been described by arachidonate 12-Lipoxygenase linked to diol synthase activities in the red alga Gracilariopsis lemaneiformis and by 15-Lipoxygenase linked to epoxygenase activities of the fungus Saprolegnia parasitica (1, 43).

10R-DOX-EAS contains the conserved Asn⁹⁶⁵GlnXaaGln motif, presumably located in the I-helix of the EAS domain (Fig. 4A). The Asn residue of this sequence supports catalysis of both 7,8-LDS of G. graminis and M. oryzae (cf. Fig. 6 and supplementary Fig. VII) (9). In analogy, 10R-DOX-EAS•N965V restrained the conversion of 10R-HPODE to 12S(13R)-epoxy-10(R)-hydroxy-8(E)-octadecenoic acid (Fig. 3). It is therefore possible that Asn⁹⁶⁵ is involved in positioning of 10R-HPODE for heterolytic scission of the oxygen-oxygen bond in analogy with the homologous Asn residues of 7,8-LDS. The position of the heme iron in the active sites appears to differ in relation to the hydroperoxides at C-8 and C-10. 7,8-LDS catalyzes hydroxylation at C-7 of 8R-HPODE, whereas 10R-DOX-EAS catalyzes epoxidation of the 12(Z) double bond of 10R-HPODE. Future work will reveal whether 8R- and 10R-HPODE bind the active sites with their conserved residues in opposite directions.

10R-DOX-EAS of M. oryzae is the first member of a distinct subfamily of DOX-CYP fusion proteins (Fig. 1B). The amino acid identity of this enzyme with 10R-DOX of A. fumigatus and A. nidulans with nonfunctional CYP domains is ∼47% over the entire sequence and only slightly higher (∼51%) over the 10R-DOX domains (∼680 N-terminal amino acids), whereas these two aspergilli enzymes can be aligned with 78% amino acid identity (cf. Fig. 1B). The n-6 double bond is essential for oxygenation at C-10, as linoleate 10R-DOX-EAS transforms 18:1n-9 to 8-HPOME as main metabolite in analogy with 10R-DOX of aspergilli (14, 15). Alignments of 10R-DOX and 10R-DOX-EAS sequences clearly indicate two distinct subfamilies, and 10R-DOX does not appear to be an inactive form of 10R-DOX-EAS (Fig. 1B and supplementary Fig. VIII). The fact that two 10R-HPODE-producing subfamilies have evolved in parallel strengthens previous reports that 10R-HPODE is of biological importance (44).

A second 10R-DOX-EAS (FOXB_03425⁷) was identified in F. oxysporum by recombinant expression. Sequence homologs are present in at least two other devastating fungal pathogens (Fig. 1B), and it seems likely that the latter also possess 10R-DOX-EAS activities. Virtually identical are the sequences of the homologs around the proximal and distal heme ligands of the10R-DOX domains (not shown) and critical regions (I-helices) of the EAS domains (Fig. 4A). Two sequences with five conserved amino acids are found in P450s with relatively few exceptions, namely Glu and Arg residues in the salt bridge (“EXXR”) and Cys and two Gly residues in the heme thiolate pocket (“GXXXCXG”) (19). The other amino acids in these two sequences are characteristic of fungal P450 families (19).

The PMK1 and cAMP-activated PKA constitute two important pathways for the infectious process of M. oryzae (25–27, 45). The biological function of 10R-DOX-EAS is unknown, but constitutive PKA expression alters the oxylipin profile of M. oryzae, and PMK1 affects expression of MGG_10859 mRNA. The latter is upregulated during the early and late phases of appressoria formation and downregulated by gene deletion of PMK1⁶ (30). It raises the question of whether 10R-DOX-EAS also could be regulated by cAMP-activated PKA.

The strain DA-99 of M. oryzae with the suppressor mutant Δmac1 sum1-99 (23) has constitutive PKA activities. Nitrogen powder of this fungus formed 8R-HPODE and 10R-HPODE as main metabolites, but without significant further transformation by EAS and only with modest biosynthesis of 7,8-DiHODE in comparison with M. oryzae Guy11 (Fig. 5). There are no other known gene candidates for these 8R and 10R activities than 10R-DOX-EAS and 7,8-LDS of M. oryzae. The altered oxylipin profile is therefore likely related to these enzymes, either by direct effects of PKA on these proteins or by indirect effects on their gene expression.

Phosphorylation, in particular by PKA, has previously been found to downregulate the catalytic activity as a functional switch of several isoforms of the CYP2 family (46). 7,8-LDS and 10R-DOX-EAS contain several phosphorylation sites. We hypothesized that phosphorylation of 7,8-LDS might alter the catalysis. However, replacement of Ser and Thr with Asp or Glu residues in five predicted phosphorylation sites did not change the oxylipin profile (cf. Ref 47). PKA influences gene expression by phosphorylation of transcription factors and by alternative splicing of pre-mRNA (48). A. terreus forms a splice variant of 5,8-LDS with retention of the last intron, leading to a premature stop codon with complete loss of the P450 catalysis (17). Alternative splicing of 7,8-LDS and 10R-DOX-EAS in this way could lead to accumulation of 8R- and 10R-HPODE, and this hypothesis merits further investigation.

CONCLUSION

We have discovered and characterized catalytic properties of the first 10R-DOX-EAS. This fusion enzyme catalyzes “oxygen transfer” epoxidation and belongs to a novel DOX-CYP subfamily with homologs in several of the top 10 fungal plant pathogens (20). Previous work suggests that the underlying gene, MGG_10859, could be regulated by the gene PMK1, which is a key regulator of the infectious process (30). Based on these results, the biological function of 10R-DOX-EAS and its enzymatic products can now be assessed.