Manganese lipoxygenase of F. oxysporum and the structural basis for biosynthesis of distinct 11-hydroperoxy stereoisomers.

The biosynthesis of jasmonates in plants is initiated by 13S-lipoxygenase (LOX), but details of jasmonate biosynthesis by fungi, including Fusarium oxysporum, are unknown. The genome of F. oxysporum codes for linoleate 13S-LOX (FoxLOX) and for F. oxysporum manganese LOX (Fo-MnLOX), an uncharacterized homolog of 13R-MnLOX of Gaeumannomyces graminis. We expressed Fo-MnLOX and compared its properties to Cg-MnLOX from Colletotrichum gloeosporioides. Electron paramagnetic resonance and metal analysis showed that Fo-MnLOX contained catalytic Mn. Fo-MnLOX oxidized 18:2n-6 mainly to 11R-hydroperoxyoctadecadienoic acid (HPODE), 13S-HPODE, and 9(S/R)-HPODE, whereas Cg-MnLOX produced 9S-, 11S-, and 13R-HPODE with high stereoselectivity. The 11-hydroperoxides did not undergo the rapid β-fragmentation earlier observed with 13R-MnLOX. Oxidation of [11S-2H]18:2n-6 by Cg-MnLOX was accompanied by loss of deuterium and a large kinetic isotope effect (>30). The Fo-MnLOX-catalyzed oxidation occurred with retention of the 2H-label. Fo-MnLOX also oxidized 1-lineoyl-2-hydroxy-glycero-3-phosphatidylcholine. The predicted active site of all MnLOXs contains Phe except for Ser348 in this position of Fo-MnLOX. The Ser348Phe mutant of Fo-MnLOX oxidized 18:2n-6 to the same major products as Cg-MnLOX. Our results suggest that Fo-MnLOX, with support of Ser348, binds 18:2n-6 so that the proR rather than the proS hydrogen at C-11 interacts with the metal center, but retains the suprafacial oxygenation mechanism observed in other MnLOXs.

Plants form jasmonates from the 13 S -hydroperoxide of 18:3 n -3, which is sequentially transformed by plant allene oxide synthase (CYP74) and by allene oxide cyclase to 12-oxophytodienoic acid, the precursor of jasmonates ( 20 ). FoxLOX shows 13 S -LOX activities and could provide the hydroperoxide precursor for jasmonate biosynthesis ( 11 ). Fo-MnLOX might also contribute, but little is known about this enzyme other than its sequence homology to the MnLOX subfamily ( 11 ).
13 R -MnLOX was recently crystallized ( 27 ), but the 3D structure has not yet been resolved. Electron paramagnetic resonance (EPR) analysis suggests that Fe and Mn are ligated in Fe-and MnLOXs in a similar way in analogy with other Fe/Mn enzymes ( 21,28 ). The two ion pairs, Fe 2+ /Fe 3+ and Mn 2+ /Mn 3+ , differ in redox potentials, 0.77 and 1.5 V, respectively ( 29,30 ). This difference is likely reduced by the Mn coordinating amino acid residues to allow oxidation and recycling of the metal center ( 29 ).
F. oxysporum and Colletotrichum gloeosporioides belong to the top ten fungal pathogens of molecular biology, and they can devastate the harvest of a wide range of fruits and vegetables ( 31,32 ). Their MnLOXs are likely secreted and might contribute to the pathogenic process by oxidation of plant lipids and biosynthesis of oxylipins, and they are structurally related ( Fig. 1B ). For these purposes we chose to express and characterize these enzymes, which could be more suitable than 13 R -MnLOX for crystallization and diffraction.
Fusarium oxysporum was found to produce jasmonic acid over 20 years ago, but the mechanism is still unknown ( 15 ). This has sparked interest in the pattern of oxygenation of C 18 fatty acids by this fungus. The genome codes for several dioxygenases of linoleic and ␣ -linolenic acid: 9 S -dioxygenase-allene oxide synthase, 10 R -dioxygenase-epoxy alcohol synthase, a related 9 R -dioxygenase, and two LOXs designated FoxLOX and F. oxysporum manganese LOX (Fo-MnLOX) ( 18,19 ) ( Fig. 1 ).
Enzyme assay. Fatty acids were incubated with the Fo-and Cg-MnLOX in 0.1 M NaBO 3 buffer (pH 9.0). The products were identifi ed as described below. LOX activity was measured on a dual beam spectrophotometer (Shimadzu UV-2101PC). The enzyme was mixed with 100 M 18:2 n-6 or 18:3 n -3 in 0.1 M NaBO 3 (pH 9.0), and the UV absorbance was followed at 235 and 237 nm, respectively. Kinetic isotope deuterium effects were determined by incubation of the enzymes with 100 M [11 S-2 H]18:2 n -6 (>99% 2 H) and 100 M 18:2 n -6, and by comparing the increase in UV absorbance at 235 nm. The reaction rate was estimated from the linear part of the curve.
The apparent K m of Fo-MnLOX was fi rst estimated by plotting the reaction rate with 18:2 n -6 (5-100 M) from the linear rate of biosynthesis of cis-trans conjugated 9-and 13-HPODE (UV analysis). We next examined the biosynthesis of 11-HPODE, which was estimated from the amounts of 9-and 13-HPODE by RP-HPLC of samples obtained during the linear rate and zoom scan MS analysis with integration of the signal intensities of the carboxylate anions of 11-HPODE and 9-and 13-HPODE ( m/z 311). These ratios, 11-HPODE/(9-HPODE+13-HPODE), appeared to decrease with increasing linoleate concentrations. The apparent K m was recalculated with the adjustments for 11-HPODE biosynthesis.
Site-directed mutagenesis of Fo-MnLOX. We investigated two positions of Fo-MnLOX by replacement, Ser348Phe and Leu530Arg, using site-directed mutagenesis essentially, as described ( 36 ). The nucleotide changes were confi rmed by sequencing (Uppsala Genome Center, Uppsala University). These mutants were designated Fo-MnLOX·Ser348Phe and Fo-MnLOX·Leu530Arg and were expressed and partially purifi ed by hydrophobic interaction chromatography as above.

MnLOX expression constructs
The proposed open reading frames of the MnLOXs from F. oxysporum 2 and C. gloeosporioides 3 were based on one intron (64 nt and 59 nt, respectively), predicted between the codons of Ile The open reading frames of Fo-and Cg-MnLOXs without their native secretion signals were ordered from GenScript in the Pichia pastoris expression vector, pPICZ ␣ A, cloned in frame with the ␣ -factor secretion signal of yeast. The expression vectors were amplifi ed in E. coli (NEB5 ␣ ), purifi ed by Nucleobond AX (Xtra Midi kit), and linearized with Pme I and Sac I, respectively. P. pastoris (strain X-33) was transformed with these constructs, as described ( 25,35 ). Transformed colonies were selected on yeast peptone dextrose agar plates with phleomycin (100 g/ml) at 28°C ( 35 ). They were stored as glycerol stocks at Ϫ 80°C or, for short time, on agar plates.

Site-directed mutagenesis of MnLOX
pPICZ ␣ A_FoMnLOX was modifi ed by replacements of Ser 348 with a Phe residue and Leu 530 with an Arg residue, with the aid of site-directed mutagenesis using Pfu polymerase. The PCR products were restricted with Dpn I, analyzed by agarose gel electrophoresis, and used to transform E. coli (NEB5 ␣ ). P. pastoris was transformed as above, and PCR and DNA sequencing confi rmed insertion of the vector in genomic DNA.

Protein expression and purifi cation
Phleomycin-resistant P. pastoris colonies were grown in buffered glycerol yeast medium in baffl ed fl asks at 28°C (200 rpm; 48 h). Buffered minimal methanol medium (0.3-0.5 l) was inoculated with washed cells to obtain OD 600 ‫ف‬ 1 . Cells were grown at 21°C at 200 rpm for about 3 days in 3-5 l baffl ed fl asks. Protein expression was induced by addition of 0.5% methanol daily for 3-5 days. The recombinant protein was secreted in the medium and assayed daily. The enzyme activity sometimes declined at the third day, possibly due to release of yeast proteases, and the medium was then harvested. The yeast cells were precipitated by centrifugation.
18-fold increase in specific activity from 1.3 to 23.7 nmol·mg Ϫ 1 ·min  Fig. 2 . The theoretical molecular mass of Fo-MnLOX was 66.7 kDa and Cg-MnLOX was slightly smaller, 66.2 kDa . The predicted isoelectric points were also similar, 6.9 and 6.4, respectively. Secreted proteins by P. pastoris can be subject to O-and Nlinked glycosylation, mainly with long chains of mannose residues ( 38 ). SDS-PAGE analysis showed a broad band between 80 and 90 kDa that could be reduced to about 65 kDa by deglycosylation with ␣ -mannosidase and endoglycosidase H .

Metal analysis
EPR measurements were performed on a Bruker ELEXYS E500 spectrometer using an ER049X SuperX microwave bridge, in a Bruker SHQ0601 cavity equipped with an Oxford Instruments continuous fl ow cryostat. The measurement temperature was 7 K, using an ITC 503 temperature controller (Oxford Instruments) and liquid helium as coolant. Signal processing was performed using the Xepr software package (Bruker). EPR analysis of Fo-MnLOX [0.15 mM in 0.2 ml 25 mM HEPES (pH 7.0)/ 100 mM NaCl/10% glycerol] was performed before and after denaturation of the protein with H 2 SO 4 .
Metal content in the protein sample was analyzed by inductively coupled plasma atomic emission spectroscopy (ICP-AES) (Spectrofl ame P). The protein sample was diluted in buffer with 5 mM EDTA and then washed three times by diafi ltration with 25 mM HEPES and 100 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 the ICP-AES measurements. The protein was diluted to 5 ml with above-mentioned low metal buffer and the protein concentration was estimated to be 0.49 mg/ml by UV absorbance at 280 nm.
The Mn and Cu effect on catalytic activity was analyzed by preincubation of the purifi ed enzyme with 1 M and 10 M CuSO 4 and with 1 M and 10 M MnCl 2 . The reaction rate was followed with UV absorbance as described above. The samples were prepared by mixing 5 l of the preincubated sample with 125 l of 100 M 18:2 n -6 in borate buffer.
Bioinformatics. ClustalW (DNASTAR) was used for sequence alignments. SWISS-MODEL was used for modeling of Fo-Mn-LOX with 8 R -LOX as a template ( 37 ). PyMOL Molecular Graphics system version 1.7.4 (Schrödinger, LLC) was used for visualizing the model.

Expression and purifi cation of Fo-and Cg-MnLOXs
Fo-and Cg-MnLOXs were expressed in baffl ed fl asks and secreted to the growth medium. Attempts to precipitate the enzymes with (NH 4 ) 2 SO 4 at 60% saturation were unsuccessful. The enzymes could be purifi ed by hydrophobic interaction chromatography (Butyl-Sepharose 4F) as described in the Materials and Methods. This resulted in capture of 98% of the applied enzyme activity and an that Fo-MnLOX forms the R enantiomer of 11-HPODE. This is in contrast to previously characterized MnLOXs that mainly formed the S enantiomer.
We confirmed that Cg-MnLOX formed 11-HPOTrE, which was transformed by 13 R -MnLOX to 13-HPOTrE (and not to 9-HPOTrE). It was thus identifi ed as 11 R -HPOTrE, which is the same 11-HPOTrE stereoisomer formed by 13 R -MnLOX and transformed to 13-HPOTrE. It is noteworthy that the R or S confi guration at C-11 of 18:2 n -6 and 18:3 n -3, by Cahn-Ingold-Prelog nomenclature rules, is infl uenced by the 15 Z double bond, which results in different R and S chirality at the 11 position of HPOTrE and HPODE, respectively, even though the hydrogen is facing in the same side. CP-HPLC-MS/MS analysis and partial separation of a mix of 11 S -and 11 R -HODE, which are formed by Cg-and Fo-LOX, respectively, is shown in Fig. 4B .
These results, taken together, suggest that fatty acids likely bind tail fi rst to the active site of Fo-and Cg-MnLOX.

Metal analysis
Mn, Fe, and Cu contents of purifi ed recombinant Fo-MnLOX enzyme (0.49 mg/ml) were determined by ICP-AES to be 254.4, 4.6, and 154.6 ng/ml, respectively ( Fig. 5B ). The Mn/protein ratio was 0.6/1. We can therefore deduce that the catalytic metal of Fo-MnLOX is Mn and not Fe. The Cu content in the protein sample compared with the blank was high, suggesting binding of Cu to the enzyme in a ratio of 0.3/1.
EPR analysis of Fo-MnLOX [ ‫ف‬ 0.15 mM in 0.2 ml 25 mM HEPES (pH 7.0)/100 mM NaCl/10% glycerol] before denaturation showed a strong Cu 2+ signal, but denaturation with H 2 SO 4 led to the appearance of a prominent and characteristic sextet of Mn 2+ (between 3,100 and 3,700 G) ( Fig. 5 ). This is in analogy with the release of proteinbound Mn 2+ by denaturation of 13 R -MnLOX [compare ( 22 )]. From these results, we conclude that the catalytic metal of Fo-MnLOX is likely Mn. The detection of Cu 2+ was unexpected, as the sample was prepared to reduce contamination by divalent ions by diafi ltration as described in the Materials and Methods.
Preincubation of the purifi ed enzyme with 1 or 10 M MnCl 2 did not increase the catalytic activity. Preincubation with 1 M or 10 M CuSO 4 abolished the oxidation of 18:2 n -6. It is possible that Mn 2+ and Cu 2+ might reduce the hydroperoxides in the sample and thereby prevent the catalytic metal from being oxidized to its active form.

Replacements of Fo-MnLOX and a 3D model based on 8R-LOX
Fo-MnLOX·Leu530Arg had little infl uence on the product profi le formed from 18:2 n -6 in comparison to native label and a small ( ‫ف‬ 2) isotope effect ( Fig. 4C ). This low number is characteristic of secondary deuterium kinetic isotope effects ( 39 ). The MS analysis of the anions of the hydroperoxides formed by Fo-MnLOX ( Fig. 4C , insert) details the retention of the deuterium label, which was further supported by the MS/MS spectrum of 11-HODE ( Fig. 4D ). The latter shows retention of the label, except for two odd numbered ions [ m/z 169 and 151 (169- 18)]. We therefore conclude that 11 R -HPODE, produced by Fo-MnLOX, is formed by suprafacial hydrogen abstraction and oxygen insertion. Other MnLOXs synthesize 11 S -HPODE by suprafacial hydrogen abstraction and oxygen insertion ( 24,25 ).
Analysis of the rate of products formed from 18:2 n -6 by UV analysis (235 nm) at different substrate concentrations is shown in Fig. 4E . These data indicated an apparent K m value of 29 ± 10 M. The ratio of 11-HPODE/(9-HPODE+13-HPODE) appeared to decrease with increasing concentrations of 18:2 n -6, e.g., 3.07 (10 M), 2.39 (40 M), and 1.91 (80 M). Without these adjustments for the rate of total biosynthesis of HPODE, the apparent K m was ‫ف‬ 40 ± 12 M.
The sequences of Fo-and Cg-MnLOXs can be aligned with 65% amino acid identities, yet Fo-and Cg-MnLOXs apparently oxidize 18:2 n -6 to hydroperoxides with different chirality. Fo-and Cg-MnLOXs also oxidize L ␣ -lysophosphatidylcholine in analogy with 13 R -MnLOX ( 41 ). This observation could be in agreement with a common substrate channel, which binds the omega end of fatty acids fi rst and with the carboxylate group (head) at the orifi ce .
The 3D structure of 8 R -LOX with bound substrate, arachidonic acid, was recently published ( 5 ). The active site is U-shaped and mainly surrounded by hydrophobic amino acid residues. Table 1 illustrates the sequence homology recombinant Fo-MnLOX. Fo-MnLOX·Ser348Phe altered the product profi le ( Fig. 6A ). 13 R -and 13 S -HODE were formed in a ratio of 74/26, and 9 R -and 9 S -HODE in a ratio of 6/94. 11-HODE consisted mainly of the 11 S stereoisomer ( Fig. 6B ). The relative amounts of 9-and 13-HPODE were unchanged. The rate (UV analysis) of oxidation of 18:2 n -6 by this mutant after capture on the Butyl-Sepharose column was 0.025 AU/min with 0.5 mg/ml protein.
Fo-MnLOX·Ser348Phe oxidized both 18:2 n -6 and 1-linoleoyl-2-hydroxy-glycero-3-phosphatidylcholine of L ␣lysoglycero-3-phosphocholine, as judged from UV analysis, and the rate of oxidation of the latter appeared to be reduced by ‫ف‬ 15%. This suggests that substrates enter the active site of the mutant with the end ("tail") fi rst in, the same as it was in the native Fo-MnLOX .
We prepared a model of Fo-MnLOX with 8 R -LOX (pdb: 3fg1) as template ( Fig. 7 ). The fi gure illustrates the position of Ser 348 ( Fig. 7A ) and the Phe residue after replacement ( Fig. 7B ), as well as Leu337 and Phe338, previously shown to be important for binding of the substrate in the active site. Leu 530 is not shown.

DISCUSSION
We report as our main fi nding that the MnLOX homolog of F. oxysporum , Fo-MnLOX, retained the same oxidation mechanism as 9 S -and 13 R -MnLOX, but formed the unexpected 11 R stereoisomer of 11-HPODE as a major metabolite ( Fig. 8A ). In contrast, the Cg-MnLOX homolog formed the expected 11 S stereoisomer of 11-HPODE in analogy with 13 R -MnLOX.
Fo-MnLOX forms 11 R -HPODE by suprafacial hydrogen abstraction and oxygen insertion, whereas Cg-, 9 S -, and 13 R -MnLOX form 11 S -HPODE by the same mechanism. Fo-MnLOX apparently binds 18:2 n -6 so that the pro R hydrogen at C-11 is presented to the metal center, whereas the pro S hydrogen is presented in Cg-, 9 S -, and 13 R-Mn-LOXs. 11 R -HPODE is also formed by the mini iron LOX  profi le ( 11,43,44 ). Ser 348 of Fo-MnLOX aligns with Leu 442 of 8 R -LOX and this residue is positioned near the end of the arachidonic acid binding site of 8 R -LOX. This position, which is well-known as the Sloane determinant, has previously been shown to infl uence the position of the substrate in the active site of 15 S -LOX and the regiospecifi city ( 25,36,45 ). Our report shows that replacement at this position, Fo-MnLOX·Ser348Phe, can alter the hydrogen abstraction at C-11 from R to S , which is a characteristic feature of an opposite head-to-tail orientation ( 46 ). The fact that Fo-MnLOX and Fo-MnLOX·Ser348Phe metabolized 1-linoleoyl-2-hydroxy-glycero-3-phosphatidylcholine does not support an altered head-to-tail orientation. Alternatively, the 12 Z double bond of 18:2 n -6 could be repositioned by the Phe residue so that 18:2 fi nds a new orientation with the pro S hydrogen at C-11 presented to the catalytic metal (compare Fig. 8B and Fig. 7B ). Our results describe a unique way by which the Sloane determinant can alter the chirality of a specifi c carbon. A remarkable and unexpected feature of Fo-MnLOX· Ser348Phe was the marked increase in chirality of 9-HPODE, from racemic to 94% S confi guration. Enzymes are believed to be optimized for their catalysis, and replacements in their active sites are not, to the best of our knowledge, known to improve stereoselectivity in this way. It may therefore be argued that the genomic sequence of Fo-MnLOX could be incorrect at the Ser 348 position. This appears to be unlikely. It was reassuring to fi nd that there are hypothetical proteins 2 from three F. oxysporum formae speciales , and these proteins are identical to Fo-MnLOX .
EPR and metal analysis confi rmed that Mn was present in large excess over Fe. EPR analysis before and after denaturation showed that Mn was mainly protein bound. EPR and metal analysis also revealed that Cu was present at high concentration in spite of purifi cation, washing with EDTA buffer followed by diafi ltration. The EPR signal of Cu 2+ was not augmented by protein denaturation to the same extent as the EPR signal of Mn 2+ . According to the Irving-Williams series of metal binding to proteins, Cu binds more strongly than both Fe and Mn ( 47 ). The high copper content in the protein sample could be due to either nonspecifi cally protein-bound Cu 2+ or specifi cally bound Cu 2+ at the active site. In the latter case, a catalytic role of Cu + /Cu 2+ in lipoxygenation is unprecedented and therefore seems unlikely.
It is of interest to compare Fo-MnLOX and FoxLOX of F. oxysporum . Both enzymes oxygenate 18:2 n -6, but Fox-LOX forms 13 S -HPODE almost exclusively ( 11 ). Fo-Mn-LOX oxidized all three possible positions of 18:2 n -6, and 13 S -and 11 R -HPODE were formed with stereoselectivity. The oxidation of 18:3 n -3 by Fo-MnLOXs led to 11 S -HPOTrE as the main metabolite (11 S or 11 R is defi ned by the Cahn-Ingold-Prelog nomenclature rules, which takes the presence or absence of the 15 Z double bond into account). 11 S -HPOTrE cannot be transformed to jasmonates, as far as is known . FoxLOX therefore remains the prime candidate LOX for fungal biosynthesis of jasmonates from 13 S -HPOTrE in the plant pathway.
within the tentative active sites of MnLOXs, and for comparison, the active site of 8 R -LOX ( 5,42 ). The most conspicuous differences between four MnLOXs and Fo-MnLOX are replacement in Fo-MnLOX with Ser 348 and Leu 530 instead of Phe and Arg in the corresponding position in the other four MnLOXs. Table 1 also illustrates that the geometry of the Mn metal ligands may not be uniform within the MnLOX subfamily, as judged from their primary sequences: The fi rst described enzymes, 13 R -and 9 S -MnLOX, separate two of the presumed histidine Mn ligands with three residues, whereas Mo-, Cg-, A. fumigatus (Af)-, and Fo-MnLOXs separate them with four residues (compare Table 1 ).
We examined whether replacement of Leu 530 and Ser 348 of Fo-MnLOX with the conserved Arg and Phe residues in the other four MnLOXs ( Table 1 ) could alter the product profi le. Fo-MnLOX·Leu530Arg oxidized 18:2 n -6 to products essentially in analogy with the native enzyme. Fo-MnLOX·Ser348Phe had profound effects on the chirality of the products ( Fig. 8B ). 9 R -and 9 S -HODE were now formed in a ratio of 6/94, and the chirality of 13-HODE shifted from mainly S to mainly R , and the chirality of 11-HPODE from R to S . Previous work has shown that replacement of a single amino acid in the active site of LOXs can alter the product  Identity  19%  18%  21%  20%  20%   Tyr-178  Tyr-112  Phe-119  Phe-111  Phe-109  Ser-112  Arg-182  Asn-116  Ala-123  Asn-115  Asp-113  Ala-116  Gln-380  Ser-286  Gly-294  Gly-285  Ser-284  Gly-286  His-384  290  298  289  298  290  Leu-385  Val-291  Val-299  Leu-290  Leu- Identical residues in fi ve or all six sequences are marked in bold. The two positions that were replaced are marked by an asterisk. An absent amino acid in the alignment is marked "gap" a 13 R -MnLOX of G. graminis (Gg-MnLOX) differs from 9 S -MnLOX of M. salvinii (Ms-MnLOX) in only one of the above positions (Phe347Leu).
b The two residues within brackets of 8 R -LOX (pdb: 4QWT) do not interact with arachidonic acid ( 5 ). c Phe-337 of 13 R -MnLOX is important for the suprafacial hydrogen abstraction and oxygen insertion ( 36 ).