Crystal Structure of Manganese Lipoxygenase of the Rice Blast Fungus Magnaporthe oryzae

Abstract

Lipoxygenases (LOX) are non-heme metal enzymes, which oxidize polyunsaturated fatty acids to hydroperoxides. All LOX belong to the same gene family, and they are widely distributed. LOX of animals, plants, and prokaryotes contain iron as the catalytic metal, whereas fungi express LOX with iron or with manganese. Little is known about metal selection by LOX and the adjustment of the redox potentials of their protein-bound catalytic metals. Thirteen three-dimensional structures of animal, plant, and prokaryotic FeLOX are available, but none of MnLOX. The MnLOX of the most important plant pathogen, the rice blast fungus Magnaporthe oryzae (Mo), was expressed in Pichia pastoris. Mo-MnLOX was deglycosylated, purified to homogeneity, and subjected to crystal screening and x-ray diffraction. The structure was solved by sulfur and manganese single wavelength anomalous dispersion to a resolution of 2.0 Å. The manganese coordinating sphere is similar to iron ligands of coral 8R-LOX and soybean LOX-1 but is not overlapping. The Asn-473 is positioned on a short loop (Asn-Gln-Gly-Glu-Pro) instead of an α-helix and forms hydrogen bonds with Gln-281. Comparison with FeLOX suggests that Phe-332 and Phe-525 might contribute to the unique suprafacial hydrogen abstraction and oxygenation mechanism of Mo-MnLOX by controlling oxygen access to the pentadiene radical. Modeling suggests that Arg-525 is positioned close to Arg-182 of 8R-LOX, and both residues likely tether the carboxylate group of the substrate. An oxygen channel could not be identified. We conclude that Mo-MnLOX illustrates a partly unique variation of the structural theme of FeLOX.

Introduction

Lipoxygenases (LOX)³ are iron- or manganese-containing dioxygenases that oxidize polyunsaturated fatty acids containing one or more 1Z,4Z-pentadiene units to hydroperoxides (1, 2). These hydroperoxides are precursors of signal molecules in animals, plants, and fungi. They may take part in inflammation, asthma, cancer development, and the chemical warfare between plants, fungi, and other microorganisms (3, 4). The LOX mechanism is initiated with hydrogen abstraction from a bis-allylic carbon of the 1Z,4Z-pentadiene of fatty acids. This is followed by oxygen insertion, which usually produces cis-trans-conjugated hydroperoxy fatty acids (1, 2). Plant, mammals, and a few prokaryotes express FeLOX, whereas both MnLOX and FeLOX occur in plant pathogenic fungi (5–8).

All LOX belong to the same gene family, but plant FeLOX, mammalian FeLOX, and fungal FeLOX and MnLOX form separate subfamilies (5, 8). The prototype MnLOX is secreted by the take-all fungus of wheat, Gaeumannomyces graminis (7). The evolution of this enzyme, 13R-MnLOX, and five members of the MnLOX subfamily are illustrated in a phylogenetic tree together with pro- and eukaryotic LOX, including fungal FeLOX (Fig. 1A).

FIGURE 1.

Overview of LOX with catalytic iron or manganese. A, phylogenetic tree of MnLOX from filamentous fungi and a selection of FeLOX. The GenBank^TM numbers are for the MnLOX enzymes are as follows: G. graminis (AAK81882.2); Magnaporthe salvinii (CAD61974); M. oryzae (ALE27899) (27); F. oxysporum (EGU80482.1); Colletotrichum gloeosporioides (EQB45907.1); and Aspergillus fumigatus MnLOX (EDP47436.1). The listed FeLOX are as follows: P. aeruginosa (Q8RNT4.2); Plexaura homomalla (PDB code 4QWT); Glycine max FeLOX (P08170.2); Homo sapiens 5-LOX (P09917.2); F. oxysporum (EXK38530.1); G. graminis FeLOX (EJT77580.1), and A. fumigatus (EAL84806). The tree was generated by MEGA6 (25) as described (38). B, overview of the oxidation of linoleic acid to hydroperoxides by FeLOX and MnLOX. Both enzymes catalyze the abstraction of the pro-11S hydrogen. The formed radical is delocalized over the pentadiene, and oxygen is typically inserted in an antarafacial way at the 13S or 9R positions by FeLOX, and in a suprafacial way at the 13R, 9S and 11S positions by MnLOX. Note that if the fatty acid enters in the reverse orientation in the catalytic channel, FeLOX can abstract the pro-11R hydrogen and form hydroperoxides with 9S and 13R configuration.

The three-dimensional structures of 11 eukaryotic and two prokaryotic FeLOX are available. These are four structures of soybean LOX (sLOX-1, LOX-3, VLX-B, and VLX-D) (9–13), three human LOX (15-LOX-2, 12S-LOX, and 5-LOX) (2, 13, 14), coral 8R-LOX and 11R-LOX (15, 16), rabbit arachidonate 15-LOX-1 (17), porcine 12S-LOX (18), 15S-LOX of Pseudomonas aeruginosa (19), and linoleate 9R-LOX of Cyanotheca sp. (20). Plant and mammalian LOX consist of two domains, a relatively small eight-stranded β-barrel domain with homology to the PLAT (polycystin-1, lipoxygenase, α-toxin) domain and a larger catalytic domain of α-helices containing the substrate-binding channel and the catalytic iron (2). The PLAT domain appears to be absent in fungal MnLOX (5, 6, 21–23). The catalytic domain revealed highly conserved metal ligands and likely oxygen channels to the catalytic center in several LOX structures (2, 10, 24). Iron is usually ligated by three His residues, oxygen of an Asn residue, and a carboxyl oxygen of the C-terminal amino acid (2). A water molecule (Fe²⁺-OH₂) or a hydroxide (Fe³⁺-OH⁻) provides an additional oxygen ligand to the metal and the catalytic base for hydrogen abstraction (2, 10). Sequence alignment with FeLOX and site-directed mutagenesis suggest that MnLOX have essentially conserved metal ligands (23, 25).

The importance of residues in the active site of LOX has been confirmed by site-directed mutagenesis and recently with three-dimensional structures of bound substrates or inhibitors (13, 24). The regio- and stereospecificity of LOX can be a result of different head-to-tail orientations of the substrate, the depth of the active site, residues positioning the hydrogen for abstraction close to the catalytic metal, and oxygen channels (1, 2). The MnLOX and FeLOX reaction mechanisms differ in two principal ways as follows: (i) hydrogen abstraction and oxygen insertion occur in a suprafacial manner in at least five MnLOX and antarafacially in all FeLOX (Fig. 1B) (6, 21, 22, 26); (ii) MnLOX are able to oxidize and rearrange bis-allylic hydroperoxides, a reaction that FeLOX only catalyze to a very low rate (27–29). This difference is possibly related to the redox properties of protein-bound iron and manganese and to structural factors. The adjustment of the different redox potentials iron and manganese is an unresolved issue as well as the metal preference of MnLOX that occurs even though the intracellular iron concentration is higher than the manganese concentration (6, 30).

Magnaporthe oryzae (Mo) causes rice blast disease, and it is listed as the most important fungal pathogen in molecular biology (31). This fungus expresses Mo-MnLOX, which oxidizes 18:2n-6 and 18:3n-3 to 9S-, 11-, and 13R-hydroperoxides with intermediate bis-allylic 11-hydroperoxides as the main product. Mo-MnLOX catalyzes β-fragmentation of these 11-hydroperoxides to cis-trans-conjugated hydroperoxides as end products in analogy with 13R-MnLOX (Fig. 1B) (22). In addition, Mo-MnLOX catalyzes prominent sequential lipoxygenation of 18:3n-3 at C-9 and C-16 (22).

The three-dimensional structure of MnLOX may provide important information on the catalytic mechanism, metal selection, and will allow a comparison between MnLOX and FeLOX. Mo-MnLOX has recently been expressed in Pichia pastoris as a stable enzyme in high yields (22). We therefore selected Mo-MnLOX for three-dimensional structure analysis due to its suitable biochemical properties and biological importance in rice blast disease. We now report the crystallization and 2.0 Å resolution structure of Mo-MnLOX.

Experimental Procedures

Materials

Fatty acids and routine chemicals were from Larodan, Merck, and Sigma. pPICZαA, P. pastoris (strain X-33), phleomycin (Zeocin), SYPRO Orange, and yeast nitrogen base were from Invitrogen. 9S-Hydroperoxy-10E,12Z,15Z-octadecatrienoic acid (9S-HPOTrE) was prepared with potato 9S-LOX and purified by HPLC. Equipment and reagents for SDS-PAGE were from Bio-Rad. Pre-stained protein ladder (Page Ruler) and colloidal Coomassie protein staining (Page-Blue) for SDS-PAGE were from Fermentas. Crystal screens were from Hampton Research (United Kingdom), and CYMAL-7 (7-cyclohexyl-1-heptyl-β-d-maltoside) was from Molecular Dimensions (United Kingdom).

Expression and Purification

The Mo-MnLOX precursor consists of 619 amino acids, including a predicted secretion signal of 16 amino acids (GenBank^TM accession number ALE27899). Mo-MnLOX without the secretion signal was cloned in the pPICZα expression vector in-frame with the yeast α-secretion signal and was expressed in P. pastoris as described (17). Large amounts of enzyme (70 mg/liter) were obtained by expression in a bioreactor for 3–4 days. The secreted enzyme constituted of 603 amino acids with two additional amino acids (Glu and Phe) from the expression vector at the N-terminal end. Mo-MnLOX was purified essentially as described (27). The enzyme (in the expression medium with added 136 g of (NH₄)₂SO₄ per liter and pH adjusted to 6.8 with 10 m KOH) was captured by hydrophobic interaction chromatography (30 ml of butyl-Sepharose CL-4B), washed with 25 mm KHPO₄ (pH 6.8), 1 m (NH₄)₂SO₄, and eluted with 25 mm KHPO₄ (pH 6.8) using ÄKTA FPLC.

Mo-MnLOX contains seven Asn residues available for N-glycosylation as judged from Asn-Xaa-(Ser/Thr) motifs (NetNGlyc 1.0 Server) and eight Ser/Thr residues for O-glycosylation (NetOGlyc 4.0 Server).

The eluted LOX was concentrated by diafiltration, diluted with 0.1 m sodium acetate (pH 5.0), 20 mm ZnCl₂, and deglycosylated with α-mannosidase (Sigma) and endoglycosidase H (Sigma) in a protein ratio of 1:40 (w/w) at 21 °C overnight. The deglycosylated LOX (in 25 mm HEPES (pH 7.0), 0.1 m NaCl) was purified by gel filtration (Superdex-200 HiLoad 26/600). Fractions with LOX activity were pooled and concentrated to 8–14 mg/ml by diafiltration (Amicon Ultra 10K) and analyzed by SDS-PAGE.

Site-directed Mutagenesis

Site-directed mutagenesis was performed by whole plasmid PCR technology with Pfu polymerase (16 cycles) according to the QuickChange protocol (Stratagene). 10 ng of the expression vector pPICZαA with the open reading frame of Mo-MnLOX served as a template (27). The desired substitutions, R525A and F526L, were introduced with oligonucleotide primers (44 nucleotides). The PCR products were analyzed by agarose gel electrophoresis to confirm amplification of the desired product by digestion of the template DNA with DpnI (37 °C, 2 h). All mutations were confirmed by sequencing before expression (Rudbeck Laboratory, Uppsala University). Transformants were obtained after linearization with SacI, transformation of P. pastoris (strain X-33), and selection on yeast peptone dextrose agar plates with phleomycin (100 μg/ml) at 28 °C (28). Transformed cells were stored as glycerol stocks at −80 °C, and expression was performed in laboratory bench shakers as described (22). The mutated enzymes were captured by hydrophobic interaction chromatography as above, and protein expression was confirmed by SDS-PAGE.

Enzyme Assay

LOX activity was measured on a dual beam spectrophotometer (Shimadzu UV-2101PC). Enzyme was mixed with 50–100 μm 18:2n-6 or 18:3n-3 in 0.1 m NaBO₃ (pH 9.0) at 22 °C, and the UV absorbance was followed at 234 and 237 nm, respectively. The cis-trans-conjugated hydro(pero)xy fatty acids were assumed to have an extinction coefficient of 25,000 cm⁻¹ m⁻¹. Biosynthesis of 9,16-dihydroperoxy-10E,12Z,14E-octadecatrienoic acid (9,16-DiHOTrE) and related trienes was assayed at 270 nm. Oxidation of 20:2n-6, 20:3n-3, and 9S-HPOTrE was studied in the same way and compared with 18:2n-6 and 18:3n-3 with the same amount of Mo-MnLOX. Products were extracted on a cartridge of octadecylsilica (SepPak/C18), and hydroperoxides were reduced to alcohols with triphenylphosphine (22). The detergent CYMAL-7 (100 μm) for crystallization did not inhibit Mo-MnLOX activities.

LC-MS Analysis

RP-HPLC with MS/MS analysis was performed with a Surveyor MS pump (ThermoFisher) and an octadecyl silica column (5 μm; 2.0 × 150 mm; Phenomenex), which was usually eluted at 0.3–0.4 ml/min with methanol/water/acetic acid, 750:250:0.05. 9S,16S- and 9S,16R-DiHOTrE were resolved with methanol/water/acetic acid, 650:350:0.05 (22). The effluent was subject to electrospray ionization in a linear ion trap mass spectrometer (LTQ, ThermoFisher) as described (22).

Thermostability

The thermostability of Mo-MnLOX before and after deglycosylation was determined with SYPRO Orange (Invitrogen) and a thermocycler (CFX Connect real time PCR, Bio-Rad). Fluorescence was monitored using the FAM filter (excitation 495 nm; detection 520 nm) as the temperature was gradually increased from 20 to 90 °C (1.5 °C/min). Samples were prepared in triplicate and contained 5 μm Mo-MnLOX and SYPRO Orange (final dilution 1:700 of 5000 concentrates) in 25 mm HEPES (pH 7.0), 100 mm NaCl, in a total volume of 30 μl. Data evaluation and melting temperature determination were performed using the Bio-Rad CFX manager software.

Crystallization

Initial crystallization screens were performed with sitting-drop vapor diffusion in 0.3-μl drops in a 96-well plate with aid of a Mosquito crystallization robot (TPP Labtech, Cambridge, UK). Crystal optimization was carried out by hanging drop vapor diffusion by mixing 1 μl of 8.5 mg/ml protein with a 1-μl reservoir solution (0.1–0.2 m ammonium citrate dibasic (pH 6.5), 10–16% w/v PEG-3350) in a 15-well plate.

Data Collection and Processing

Data were collected with the focus of achieving a high sulfur and manganese signal at a wavelength of 1.77 Å at 100 K; one dataset was collected at beam line ID29 at the European Synchrotron Radiation Facility, Grenoble, France, and five additional datasets were collected for the same purpose at beam line I02 at the Diamond Light Source, Oxfordshire, UK.

The datasets were processed using XDS (32), and the integrated data were scaled using AIMLESS (33). A set of 5% of the reflections was set aside and used to calculate the quality factor R_free. None of the datasets provided sufficient anomalous signal to find the manganese and the sulfur sites. To increase the anomalous signal, all the datasets were analyzed for crystal isomorphism using BLEND (34). Four of the XDS integrated datasets were merged and scaled with R_meas and R_p.i.m. values of 0.152 and 0.017 and multiplicity of 77.4.

Structure Determination and Refinement

The positions of manganese and sulfur atoms were determined using the HKL2MAP graphical interface with SHELXC, SHELXD, and SHELXE (35, 36). SHELXC showed an anomalous signal extending to 3.5 Å resolution. Two manganese and 15 sulfur sites were identified in SHELXD. The correctness of the solution was confirmed by SHELXE. Single anomalous dispersion phasing was performed by phenix.autosol from the Phenix suite (37), using the sites obtained by HKL2MAP and the merged dataset. The phases obtained from SHELX and the protein sequence were submitted to phenix.autosol, and phenix.autobuild (37) was able to build 871 out of 1210 residues of the two molecules in the asymmetric unit, with R_work 0.37 and R_free 0.39. By using a single dataset to 2.0 Å resolution, we could build 1136 out of 1210 residues with R_work 0.17 and R_free 0.21. Model evaluation and manual model building were performed in Coot (38). Refinement was performed with phenix.refine (39). Model quality was evaluated with MOLPROBITY (40). 97% of the residues were in favored regions of the Ramachandran plot. Statistics of data collection, processing and model building are presented in Table 1.

TABLE 1.

Data collection, processing, phasing and structure refinement statistics

Statistics for the highest resolution shell are shown in parentheses.

	S-SAD 1 crystal	S-SAD (4 crystals merged)
Data collection and processing
Beamline	Diamond I02	ESRF ID-23, Diamond I02
Detector	Pilatus 6 m	Pilatus 6 m
Wavelength (Å)	1.77	1.77
Oscillation range	0.15	0.15
No. of images	3600	4800/2400/3600/3600
Space group	P212121	P212121
Cell parameters a, b, and c (Å)	70.72, 111.37, 171.22	70.60, 111.32, 171.25
Resolution range (Å)	29.70–2.04 (2.11–2.04)	48.93–2.53 (2.62–2.53)
No. of observed reflections	1,531,004 (49,813)	3,535,869 (338,794)
No. of unique reflections	81,415 (3,816)	45,670 (4,371)
Multiplicity	18.8 (13.1)	77.4 (77.5)
Completeness (%)	98.1 (86.0)	99.9 (99.9)
Rmeas	0.12 (0.63)	0.15 (0.34)
I/σ(I)	17.9 (3.9)	44.1 (19.2)
CC1/2 (%)	0.997 (0.919)	1.0 (0.998)
Manganese and sulfur phasing
Resolution cutoff		3.5
No. of sites		2 manganese, 15 sulphur
CC anomalous		30
Map correlation		0.66
Connectivity		0.75
Contrast		0.54
Refinement statistics
Resolution used in refinement	2.04
Reflections in working/test set	84190/4403
R/Rfree factor (%)	16.8/21.3
Molecules in asymmetric unit	2
No. of atoms
Protein atoms	9741
Mn	2
N-Glycosylation (NAG) atoms	112
Water molecules	683
Protein residues	1136
Wilson B-factor	35.9
Average atomic B-factors (Å2)
Overall	43.5
Protein	42.94
Water	47.27
Mn	33.06
r.m.s. deviation
Bond lengths from ideal (Å)	0.007
Bond angles from ideal (°)	0.93
Ramachandran outliers (%)	0

Miscellaneous Methods

SDS-PAGE was performed as described (27). Sequences of FeLOX and MnLOX were aligned with the ClustalW program, and a phylogenetic tree was constructed by MEGA6 with bootstrap tests of the resulting nodes (41). All figures were generated with PyMOL (The PyMOL Molecular Graphics System, Version 1.7.4 Schrödinger, LLC).

Results

Deglycosylation

Mo-MnLOX retained more than 50% of the enzyme activity after deglycosylation, and the three-dimensional structure discussed below showed that N-acetylglycosamine residues remained at Asn-72, Asn-150, and Asn-535. The deglycosylation process decreases the melting temperature of Mo-MnLOX from 60 to 56 °C. The protein unfolding in response to temperature (assayed with SYPRO Orange (22)) indicated that the enzyme solution was not fully homogeneous (supplemental Fig. S1, A and B).

Crystallization of Mo-MnLOX

Needle-like crystals of Mo-MnLOX were formed after 1 week in polyethylene/ion screen HT (Hampton Research) by 1:1 mixing of 14 mg/ml protein and reservoir solution (4% v/v tacsimate (pH 7.0), 12% (w/v) PEG-3350, or 0.2 m ammonium citrate dibasic (pH 5.1), 12% w/v PEG-3350). Single large crystals grew after 1 month at 8 °C in 0.2 m ammonium citrate dibasic (pH 4.8), 12% (w/v) PEG-3350. Parallel optimization was performed with additive screen MemAdvantage (Molecular Dimensions) in 96-well sitting drop plates at 21 °C and in 15-well hanging drop plates at 8 °C. Crystals were formed after 3 days at 21 °C by mixing 0.1 μl of protein (8.5 mg/ml), 0.1 μl of reservoir solution (0.2 m ammonium citrate dibasic (pH 4.8), 10–14% (w/v) PEG-3350), and 0.02 μl of 1.9 mm CYMAL-7. Crystals also appeared after 30 days at 8 °C by mixing 1 μl of the same reservoir solution with 1 μl of protein and 0.1 μl of 1.9 mm CYMAL-7 in hanging drop 15-well plates (supplemental Fig. S1C). Crystals were cryo-protected in reservoir solution with 15% (v/v) glycerol and vitrified in liquid N₂ prior to data collection.

X-ray Diffraction Analysis

The crystals of Mo-MnLOX were relatively insensitive to radiation, and 4800 images were collected with 0.15° oscillation. The best dataset had a completeness of 99.6% and an average multiplicity of 25.3. The crystals of Mo-MnLOX diffracted to 2.0 Å. Mo-MnLOX contains one manganese atom in the active site and 15 sulfur atoms from 3 Cys and 12 Met residues. This made it possible to solve the structure by single anomalous dispersion phasing using both manganese and sulfur atoms as anomalous scatterers. We therefore collected data at 1.77 Å to mitigate the absorption effect at longer wavelengths while still being able to collect a useful sulfur signal. Because the K_α absorption edge of a manganese is at λ = 1.88 Å, the anomalous signal for manganese (f″ = 3.45 electrons) at λ = 1.77 Å becomes an additional source of anomalous signal, which facilitated the phase determination. The anomalous signal is weak for sulfur (f″ = 0.7 electrons) at λ = 1.77 Å. It was therefore crucial to have high redundancy data to enhance the signal to noise ratio. Four datasets were merged that resulted in a multiplicity of 77.4 and an anomalous signal that enabled us to determine the position of the manganese and sulfur atoms.

The crystals belong to space group P2₁2₁2₁ with unit cell dimensions as follows: a = 70.7 Å, b = 111.4 Å, and c = 171.2 Å. The solvent content was 47% with two molecules in the asymmetric unit with the average C_α root-mean-square (r.m.s.) deviation of 0.147 Å. The final model was refined to R_work of 0.17 and R_free of 0.21. We were able to build all the residues except for the 37 N-terminal residues of the expressed protein (EFV … PEL), possibly due to its flexibility, as well as N-acetylglucosamine groups at Asn-72, Asn-150, and Asn-535.

Crystal Structure of Mo-MnLOX

Mo-MnLOX lacks the PLAT domain found in many FeLOX and phospholipases (2). An illustration of the overall Mo-MnLOX structure is presented in Fig. 2A. The structure is composed of 21 α-helices and 7 small β-sheets. The helices α9 and α10 combined are designated the broken arched helix, and it is sheltering the active site (light green, Fig. 2A). The most striking difference to FeLOX is the orientation of helix α2 with 11 turns between residues 79 and 117 (blue helix, Fig. 2B). This long helix is slightly arch-shaped and runs over the whole length of the protein. Its orientation in animal and plant LOX varies, and in plants it has been reported to be mobile to allow access to the active site, as illustrated by a comparison of Mo-MnLOX with human 5-LOX, sLOX-1, and 15S-LOX of P. aeruginosa (Fig. 2B) (2). This helix is found in all known LOX structures and harbors several invariant hydrophobic residues (2). A structure-based sequence alignment between 8R-LOX (4QWT; the 8R-LOX domain of the allene oxide synthase-LOX fusion protein) and Mo-MnLOX shows conservation of most α-helices (supplemental Fig. S2).

FIGURE 2.

Crystal structure of Mo-MnLOX. A, schematic illustrations in two directions of the overall structure, colored in rainbow spectrum with the N terminus in blue, the C terminus in red, and the catalytic manganese in purple. The broken helix covering the active site is colored in light green. B, variation of the α2 helix in different LOX structures as follows: Mo-MnLOX (blue, PDB code 5FNO); 5-LOX (red, PDB code 3O8Y); sLOX-1 (green, PDB code 1YGE); and P. aeruginosa 15S-LOX (yellow, PDB code 4G32). The α2 helix of Mo-MnLOX is an 11-turn-long helix (blue), which runs along the whole length of the protein and leaves an open access to the substrate channel and the catalytic metal (light orange).

Metal Coordination

His-294, His-289, His-469, Asn-473, Val-605, and a water molecule coordinate the catalytic Mn²⁺ (Fig. 3A). The coordinating sphere is similar to coral 8R-LOX, but the metal ligands do not superimpose as neatly as those of coral 8R-LOX, sLOX-1, and 15S-LOX of P. aeruginosa (Fig. 3B) (2, 16, 19, 20). The distances between the coordinating residues, manganese, and water are indicated in Table 2 along with a comparison with three FeLOX.

FIGURE 3.

Overview of the metal ligands of Mo-MnLOX and FeLOX. A, catalytic manganese (orange sphere) of Mo-MnLOX (gray) is coordinated by the His-284, His-289, His-469, Asn-473, and Val-605 in a distorted octahedral configuration. The coral 8R-LOX (PDB code 4QWT, pink) is superimposed for comparison with an r.m.s. deviation of 0.57 Å. The largest differences are between the Asn residues and the C-terminal Ile/Val residues. B, metal ligands to the catalytic iron of coral 8R-LOX (PDB code 4QWT, pink), sLOX-1 (PDB code 1YGE, green), and 15S-LOX of P. aeruginosa (PDB code 4G32, yellow) are superimposed with an r.m.s. deviation between 0.23 and 0.29 Å.

TABLE 2.

Distances (Å) between the catalytic metal, the coordinating ligands, and the catalytic base

	Mo-MnLOX	8R-LOXa	sLOX-1	Pa-LOXb
His7284(NE2)	2.6	2.4	2.2	2.3
His7289(NE2)	2.2	2.3	2.3	2.2
His7469(NE2)	2.1	2.3	2.2	2.2
Asn7473(OD1)	2.9	3.1	3.1	2.2
Val7605(OXT)	2.1	2.3	2.4	2.2
H2O	2.3	2.4	2.6	2.2
Va7l605(O)-H2O	2.4	2.4c	2.5c	2.5c

^a The 8R-LOX domain of the allene oxide synthase-LOX fusion protein (UniProtKB/Swiss-Prot: O16025.1) is shown.

^b Pa-LOX is 15S-LOX of P. aeruginosa.

^c The distances between the catalytic water and the C-terminal Ile residue are shown.

A small loop of 5 residues connects helices α17 and α18 and harbors the manganese ligand, Asn-473 (Fig. 4, A and B). This loop likely brings Asn-473 to a flexible position near the catalytic metal. The side chain oxygen of Gln-281 forms hydrogen bonds to the amino group of Asn-473 with a distance of 2.91 Å (Fig. 4C). This Gln residue is conserved in all FeLOX and MnLOX. There is also a hydrogen bond between the C-terminal carboxyl (Val-605) and the catalytic water (Fig. 4C).

FIGURE 4.

Factors influencing the metal coordination of Mo-MnLOX. A, unbiased 2F_o − F_c electron density map is shown at contour level of one σ. The metal-coordinating Asn-473 residue is situated on a loop and might provide the increased flexibility necessary for the use of manganese as catalytic metal. Gln-474 and Ser-604 are in close positions. B, comparison of the structure of the loop with Asn-473 (light blue) with the corresponding part of 8R-LOX (gray). C, hydrogen bond network close to the active site of MnLOX and Asn-473, which forms three hydrogen bonds as follows: with the conserved Gln-281, the main chain oxygen of the metal coordinating His-469_, and the main chain of Glu-476. Gln-474 is forming a hydrogen bond network with Ser-604 proximate to the C-terminal Val-605; it also forms a weak interaction with a coordinated water molecule that also interacts with the main chain of Arg-528 and the side chain of Asn-527.

Substrate Channel

There is a solvent-accessible channel leading into the catalytic center of Mo-MnLOX (Fig. 5A). Arg-525 at the entrance is positioned at helix α19. This Arg is conserved in five out of the six confirmed MnLOX in Fig. 1A (except in MnLOX of Fusarium oxysporum). Arg-525 is positioned close to Arg-182 of 8R-LOX when the two structures are superimposed. Arg-525 likely forms a salt bridge with the carboxyl end of the substrate fatty acid (Fig. 5B) in analogy with Arg-182 (2). The entrance is also defined by residues from helix α2 (Trp-93, Val-98, Ser-101, and Phe-105), helix α9 (Leu-326), and helix α11 (Val-350).

FIGURE 5.

Substrate channel entrance of Mo-MnLOX and a comparison with 8R-LOX. A, proposed substrate channel entrance of Mo-MnLOX (PDB code 5FNO) is illustrated in surface rendering (gray), superimposed with the structure of 8R-LOX (PDB code 4QWT chain C), showing arachidonic acid as substrate. The Arg-525 is positioned in the opening to the channel in suitable distance for ionic interaction between the Arg-525 side chain and the carboxyl of the fatty acid substrate. B, Mo-MnLOX (blue) and 8R-LOX (pink) are superimposed. The Arg-182 of 8R-LOX has been found to tether the carboxylate of the substrate. The Arg-525 of MnLOX is provided by a helix closer to the C terminus, but these two Arg residues seem nevertheless to play similar roles in the tethering of the carboxyl group.

The assumed substrate channel in Mo-MnLOX is composed of several hydrophobic residues (Table 3). The closest residues surrounding the coordinated catalytic water are Phe-526 (4.2 Å), Val-323 (4.2 Å), Phe-332 (5.6 Å), Leu-331 (6.6 Å) and Gln-281 (5.9 Å) as discussed above. These residues likely constitute the direct environment of the pentadiene of the substrate and define the hydrophobic channel.

TABLE 3.

Comparison of amino acid residues surrounding the active site in MnLOX and FeLOX

Comment	Consensus	MnLOX, Mo-MnLOX	FeLOX
			Coral 8R-LOXa	Human 5-LOX	Soybean LOX-1
Channel entrance FeLOX			Tyr-178	Phe-177	Ala-254
Channel entrance FeLOX			Arg-182	Tyr-181	Gly-258
Interaction with Asn-473	Gln	281	380	363	495
Metal coordination	His	284	384	367	499
Oxygen channel		Val-285	Leu-385	Leu-368	Trp-500
Metal coordination	His	289	389	372	504
		Val-323	Ile-423	Ile-406	Ile-538
Stereocontrol (Coffa-Brash)	Gly/Ala	Gly-327	Gly-427	Ala-410	Ala-542
Clamp substrate	Leu	331	431	414	546
Supra/antarafacial	Phe/Ile	Phe-332	Ile-432	Ile-415	Ile-547
		Leu-337	Ile-437	Leu-420	Ile-552
		Phe-338	Val-438	Phe-421	Ile-553
Pocket depth (Sloane)		Phe-342	Leu-442	Asn-425	Phe-557
Pocket depth		Thr-489	Ala-589	Pro-569	Thr-709
Pocket depth		Gln-519	Ala-620	His-600	Ser-747
		Leu-522	Thr-623	Ala-603	Val-750
Channel entrance MnLOX		Arg-525b	Ile-626	Ala-606	Ile-753
Steric shielding	Phe/Leu	Phe-526	Leu-627	Leu-607	Leu-754
C terminus	Val/Ile	Val-605	Ile-693	Ile-674	Ile-839

^a The 8R-LOX domain of the allene oxide synthase-LOX fusion protein is shown.

^b Arg-525 is positioned close to Arg-182 of 8R-LOX (Fig. 5B).

Leu-331 is situated at the bottom of the arched helix where it shelters the active site, and the corresponding residue of 8R-LOX, Leu-431, has been shown to clamp the substrate in the active site (2, 15, 24). It appears to play the same role in Mo-MnLOX (Fig. 6). The side chain of the next residue, Phe-332, points into the hydrophobic channel and might shield one side of the pentadiene from oxygenation. This Phe residue is conserved in all MnLOX, but not in FeLOX, which have either Ile or Val at this position (Table 3).

FIGURE 6.

Possible oxygen access routes in the U-shaped substrate channel of Mo-MnLOX. Leu-331 from the arched helix is defining the upper wall of the channel at the bottom of the U-shaped substrate channel in analogy with Leu-431 of 8R-LOX. Phe-332 may shield the pentadiene for oxygen insertion in an antarafacial way so that oxygen may reach the pentadiene radical from the other side as indicated in by the arrows in the two side pockets. Phe-526 is likely to bend the substrate to allow oxygen access from the same side as the catalytic metal. Arachidonic acid, bound to coral 8R-LOX (PDB code 4QWT, chain C), is included for clarity; the natural substrates of Mo-MnLOX are linoleic and α-linolenic acids, but 20:2n-6, 20:3n-3, and 22:5n-6 are also oxidized by the enzyme (supplemental Fig. S4).

Phe-526 is also conserved in MnLOX, whereas the corresponding residue in FeLOX is a conserved Leu residue (Table 3). The distance between the side chains of Leu-331 and Phe-526 is only 3.9 Å. The substrate could be clamped by these residues and bent to allow oxygen to access the 11S position after the hydrogen abstraction. The distance between the two corresponding Leu residues in 8R-LOX is similar, but with the substrate in the active site Leu-627 of 8R-LOX is bent backwards, and the distance is increased to 5.2 Å (24).

Oxygen Access to the Catalytic Center

MnLOX utilize suprafacial hydrogen abstraction and oxygenation in contrast to the antarafacial oxidation mechanism of FeLOX (6, 21, 22, 26). This implies that O₂ can access the pentadienyl radical from the same side as the catalytic complex, Mn³⁺OH⁻ (cf. Fig. 1B). A possible oxygen channel has been identified in several three-dimensional structures of FeLOX (2, 10, 24). The Coffa-Brash determinant, Gly-427 of coral 8R-LOX, appears to be in a critical position in its oxygen channel (2). No equivalent channel could be found in Mo-MnLOX, and the corresponding Gly-327 residue may have little influence on the position of oxygenation (42). There are two pockets in the Mo-MnLOX substrate channel that could harbor oxygen if it enters via the substrate channel (Fig. 6). It is tempting to speculate that these pockets could explain the stereospecific oxygenation of all three positions of the pentadiene radical.

Site-directed Mutagenesis

The structure discussed above suggested that Arg-525 and Phe-526 might be of structural importance for tethering of the carboxyl group and for oxygenation, respectively. We examined the following two mutations: R525A and F526L. Protein expression was confirmed by SDS-PAGE after protein isolation by hydrophobic interaction chromatography.

The mutant R525A transformed 16, 36, and 100 μm 18:3n-3 to small amounts of 11-HPOTrE without apparent substrate inhibition (24). 11-HPOTrE was detected by RP-HPLC-MS/MS analysis (supplemental Fig. S3). The marked reduced catalytic activities could be in agreement with the proposed function of Arg-525 in tethering the carboxyl group of 18:2n-6 and 18:3n-3.

We also examined 9S-HPOTrE as a substrate of R525A. A substantial fraction of 9S-HPOTrE was transformed to 9,16-DiHOTrE as shown in Fig. 7. RP-HPLC analysis showed that it consisted mainly of the expected 9S,16S diastereoisomer (22). We conclude that Arg-525 is not essential for the lipoxygenation of 9S-HPOTrE.

FIGURE 7.

RP-HPLC-MS/MS analysis of the biosynthesis of 9S,16S-DiHPOTrE from 9S-HPOTrE by the R525A mutant and an overview of the sequential biosynthesis of 9,16-DiHPOTrE. A, RP-HPLC-MS/MS analysis of the lipoxygenation of 9S-HPOTrE by the R525A mutant of Mo-MnLOX after reduction of hydroperoxides to alcohols with triphenylphosphine. B, overview of the biosynthesis of 9S,16S-DiHPOTrE by Mo-MnLOX and the R525A mutant. NL, normalized to 100%. TIC, total ion current.

To assess the importance of the chain length for the interaction with the Arg-525 residue, we compared the oxidation of 20:2n-6, 20:3n-3, and 22:5n-6 with 18:2n-6 and 18:3n-3. The two C₂₀ fatty acids were both oxidized at C-13. 20:2n-6 was also oxidized at C-11 and C-15 in a ratio of ∼10:1 (supplemental Fig. S4A).20:3n-3 was oxidized at both C-11 and C-15. The latter also formed 11,18-dihydroperoxy-12E,14Z,16E-eicosatrienoic acid (supplemental Fig. S4B). UV analysis (235 nm) and LC-MS analysis to estimate the relative amounts of bis-allylic hydroperoxides suggested that 20:2n-6 as oxidized at a rate of 70% of 18:2n-6, whereas UV analysis (270 nm) indicated that 20:3n-3 was oxidized to trienes twice as rapidly as 18:3n-3. 22:5n-6 was oxidized at C-13 and C-17 (supplemental Fig. S4C) at about 25% of the rate of 18:2n-6. We conclude that the substrate chain length is not critical for catalysis.

The F526L mutant did not oxidize 18:3n-3, but it transformed 9S-HPOTrE to 9S,16S-DiHOTrE (supplemental Fig. S5), which suggests that the catalytic center was intact.

Discussion

We report as our main finding the first three-dimensional structure of MnLOX. This structure relates to three fundamental differences between MnLOX and FeLOX as follows: (i) the coordinating spheres of Mn²⁺ and Fe²⁺ and the metal preferences; (ii) adjustment of the redox potentials of protein-bound Mn²⁺/Mn³⁺ and Fe²⁺/Fe³⁺ and the catalytic base by hydrogen bonds, and (iii) the active sites and the supra- and antarafacial oxygenation mechanisms of MnLOX and FeLOX, respectively. An overview of the active site is shown in Fig. 8, and a comparison of important residues with FeLOX is presented in Table 3. The overall amino acid sequence identity of Mo-MnLOX and 8R-LOX is about 23% with an overall r.m.s. deviation of 3.48 Å.

FIGURE 8.

Overview of the active site of Mo-MnLOX. Arachidonic acid, bound in the substrate channel of coral 8R-LOX (PDB code 4QWT, chain C), is included in the U-shaped active site of Mo-MnLOX for clarity. The carboxyl group of arachidonic acid is likely tethered by Arg-525 and the ω end by Phe-342. Leu-332 clamps the substrate in position, and Phe-332 and Phe-526 may position pentadiene for suprafacial hydrogen abstraction and oxygenation. Three His residues, Asn-473, Val-605, and the catalytic water are coordinating manganese (pink). Hydrogen bonds are likely formed between Gln-281 and Asn-473 and between Val-605 and the catalytic water (red).

Metal Coordinating Sphere

Mn²⁺ is bound in a distorted octahedral configuration in analogy with Fe²⁺ in eukaryotic and prokaryotic LOX by three His residues, an Asn residue, and the carboxylate of the C-terminal residue (Fig. 8; Table 3). The metal ligand residues of MnLOX and FeLOX are thus identical except for the replacement of the C-terminal Ile residue with Val in 5 out of 6 MnLOX (Fig. 1A), but the ligands are not identical in space. This is shown by a comparison of the metal ligands of Mo-MnLOX with coral 8R-LOX (Fig. 3A), which align with an r.m.s. deviation of 0.57 Å. The largest differences are found between the Asn residues and the C-terminal Ile or Val residues (Fig. 3A). In contrast, the F-coordinating residues of coral 8R-LOX, sLOX-1, and 15S-LOX of P. aeruginosa can be aligned almost perfectly with an r.m.s. deviation of 0.23–0.29 Å (Fig. 3B) (2).

The Asn-473 ligand of Mo-MnLOX is positioned on a short loop, whereas the corresponding Asn of 8R-LOX and other FeLOX is positioned on an α-helix (Fig. 4, A and B). This appears to be one of the most striking differences between the coordination spheres. Oxidation of Mn²⁺ to Mn³⁺ may lead to Jahn-Teller distortion from the octahedral coordination of Mn²⁺ (43). This might be facilitated by the position of Asn-473 on a relatively flexible loop in comparison with position on an α-helix.

Fe²⁺ is usually present in a much larger intracellular concentration than Mn²⁺ (30). The incorporation of Mn²⁺ by apoproteins therefore likely occurs in specific cellular compartments, which are enriched in Mn²⁺ (30). Whether the three-dimensional differences between the coordinating spheres of MnLOX and FeLOX also can affect metal selection will await future studies.

Adjustment of Redox Potentials

FeLOX and MnLOX catalyze the same enzymatic reactions, and their redox properties are therefore likely similar, about 0.6 V (44). As far as is known, manganese-substituted FeLOX are catalytically inactive (9, 38). Two differences between MnLOX and FeLOX are the capacity of Mo-MnLOX to catalyze β-fragmentation of 11-hydroperoxides of 18:2n-6 and its prolonged catalytic lag phase (22, 27, 28). Hydrogen bonds to the catalytic center and steric factors likely adjust the redox potential of Mn²⁺OH₂ close to that of Fe²⁺OH₂ in analogy with manganese and iron superoxide dismutases (45). The catalytic water forms an almost identical hydrogen bond with the carboxylate of the C-terminal Val and Ile residue of Mo-MnLOX and coral 8R-LOX, respectively, but there were no additional hydrogen bonds to the catalytic water. We therefore examined the network of hydrogen bonds to the manganese ligands and to the second coordinating sphere, respectively (Figs. 4C and 8). A hydrogen bond likely occurs between the metal coordinating Asn-473 and Gln-281 (2.8 Å) of Mo-MnLOX. A hydrogen bond was also noted between Ser-604 and Gln-474 (2.8 Å), but site-directed mutagenesis of the corresponding Gln residue of 13R-MnLOX did not abolish the catalytic activity (25). The tuning of the redox potential of protein-bound Mn²⁺/Mn³⁺ will need further investigation. This will include further analysis of the hydrogen bond network.

Active Site and the Oxygenation Mechanism

The deduced substrate channel of Mo-MnLOX appears to be similar to the U-shaped channel of coral 8R-LOX and related FeLOX (2). The substrate channel of Mo-MnLOX is solvent-exposed (Fig. 5A), and its interior has spacious pockets close to the presumed position of the pentadiene for hydrogen abstraction and oxygenation (Fig. 6). Arg-525 likely tethers the carboxylate of the substrate in the same way as Arg-182 of 8R-LOX (Figs. 5 and 8). Replacement of Arg-182 of 8R-LOX with Ala led to a dramatic change in the kinetic properties due to striking substrate inhibition (24). The R525A mutant transformed 18:3n-3 to only small amounts of 11-HPOTrE, but it oxidized 9S-HPOTrE to 9S,16S-DiHOTrE more efficiently (Fig. 7). Arg-525 likely also interacts with the carboxyl group of C₁₈ and C₂₀ fatty acids. The oxidation of 20:2n-6, 20:3n-3, and 22:5n-6 suggests sufficient space in the active site to allow productive configurations. In analogy with FeLOX, the depth of the substrate channel is likely controlled by Phe-342 at the position of the Sloane determinant (Fig. 8; Table 3) and by Phe-353 (not shown in Fig. 8).

Two residues, Phe-332 and Phe-526, may directly influence the stereospecific oxygenation of fatty acids. Phe-332 is positioned in the active site above the catalytic metal and likely holds the substrate in place and might shield the opposite side for oxygen insertion (Fig. 8). Mutagenesis of the corresponding Phe-337 residue in 13R-MnLOX to Ile, which is found at this position of sLOX-1 and other FeLOX (Table 3), switched the oxygen insertion in relation to hydrogen abstraction from suprafacial to mainly antarafacial (46). Phe-526 is also positioned near the catalytic metal and might position the substrate for oxygenation. Site-directed mutagenesis of Phe-526 to Leu resulted in loss of oxidation of 18:3n-3 but retention of the oxidation at C-16 of 9S-HPOTrE. The altered LOX activity suggested that this residue could be essential for catalysis, but further steric analysis of this mutant could not be performed as 18:3n-3 was not oxidized. The three-dimensional structure of Mo-MnLOX with a substrate or a substrate mimic will be needed to exactly define the structural importance of the Phe-526 residue.

Conclusion

We report the three-dimensional crystal structure of MnLOX of the rice blast fungus M. oryzae. The results confirm that the metal ligands of MnLOX and FeLOX are essentially conserved but with geometric differences between the coordinating spheres. Arg-525 likely tethers the carboxyl group of the substrate, and a pair of conserved Phe residues near the catalytic center of MnLOX might be key contributors to the unique suprafacial reaction mechanism.

Author Contributions

A. W. purified and crystallized the protein, determined the x-ray structure, prepared the figures, and wrote the paper. E. H. O. initiated the study, wrote the paper, and prepared the figures. S. K. provided assistance in crystallization, data collection, and interpretation. Y. C. determined the x-ray structure together with A. W., prepared the figures, and wrote the paper. All authors reviewed the results and approved the final version of the manuscript.