Leucine/Valine Residues Direct Oxygenation of Linoleic Acid by (10R)- and (8R)-Dioxygenases

Linoleate (10R)-dioxygenase (10R-DOX) of Aspergillus fumigatus was cloned and expressed in insect cells. Recombinant 10R-DOX oxidized 18:2n-6 to (10R)-hydroperoxy-8(E),12(Z)-octadecadienoic acid (10R-HPODE; ∼90%), (8R)-hydroperoxylinoleic acid (8R-HPODE; ∼10%), and small amounts of 12S(13R)-epoxy-(10R)-hydroxy-(8E)-octadecenoic acid. We investigated the oxygenation of 18:2n-6 at C-10 and C-8 by site-directed mutagenesis of 10R-DOX and 7,8-linoleate diol synthase (7,8-LDS), which forms ∼98% 8R-HPODE and ∼2% 10R-HPODE. The 10R-DOX and 7,8-LDS sequences differ in homologous positions of the presumed dioxygenation sites (Leu-384/Val-330 and Val-388/Leu-334, respectively) and at the distal site of the heme (Leu-306/Val-256). Leu-384/Val-330 influenced oxygenation, as L384V and L384A of 10R-DOX elevated the biosynthesis of 8-HPODE to 22 and 54%, respectively, as measured by liquid chromatography-tandem mass spectrometry analysis. The stereospecificity was also decreased, as L384A formed the R and S isomers of 10-HPODE and 8-HPODE in a 3:2 ratio. Residues in this position also influenced oxygenation by 7,8-LDS, as its V330L mutant augmented the formation of 10R-HPODE 3-fold. Replacement of Val-388 in 10R-DOX with leucine and phenylalanine increased the formation of 8R-HPODE to 16 and 36%, respectively, whereas L334V of 7,8-LDS was inactive. Mutation of Leu-306 with valine or alanine had little influence on the epoxyalcohol synthase activity. Our results suggest that Leu-384 and Val-388 of 10R-DOX control oxygenation of 18:2n-6 at C-10 and C-8, respectively. The two homologous positions of prostaglandin H synthase-1, Val-349 and Ser-353, are also critical for the position and stereospecificity of the cyclooxygenase reaction.

also be positioned with its end embedded in the interior of 7,8-LDS of Gaeumannomyces graminis (18). 7,8-LDS of G. graminis and Magnaporthe grisea and 5,8-LDS of Aspergillus nidulans have been sequenced (5,8,21). Gene targeting revealed the catalytic properties of 5,8-LDS, 8,11-LDS, and 10R-DOX in Aspergillus fumigatus and A. nidulans (3). Homologous genes can be found in other Aspergilli spp. Alignment of the two 7,8-LDS amino acid sequences with 5,8-LDS, 8,11-LDS, and 10R-DOX sequences of five Aspergilli revealed several conserved regions with single amino acid differences between the enzymes with 8R-DOX and 10R-DOX activities, as illustrated by the selected sequences in Fig. 1. Leu-306, Leu-384, and Val-388 of 10R-DOX are replaced in 5,8-and 7,8-LDS by valine, valine, and leucine residues, respectively. Whether these amino acids are important for the oxygenation mechanism is unknown, and this is one topic of the present investigation. The predicted secondary structure of 10R-DOX suggests that Leu-384 of 10R-DOX can be present in an ␣-helix with Val-388 close to its border. This ␣-helix is homologous to helix 6 of PGHS-1, which contains Val-349 and Ser-353 at the homologous positions of Leu-384 and Val-388 (Fig. 1).
The overall three-dimensional structures of myeloperoxidases are conserved. It is therefore conceivable that important residues for substrate binding in the cyclooxygenase channel of PGHS could be conserved in LDS and 10R-DOX. The three-dimensional structure of ovine PGHS-1 shows that Val-349 and Ser-353 are close to C-3 and C-4 of 20:4n-6, and residues in these positions can alter both position and stereospecificity of oxygenation (22)(23)(24). Replacement of Val-349 of PGHS-1 with alanine increased the biosynthesis of 11R-HETE, whereas V349L decreased the generation of 11R-H(P)ETE and increased formation of 15(R/S)-H(P)ETE (23,25). V349I formed PGG 2 with 15R configuration (22,24). Replacement of FIGURE 1. Alignments of partial amino acid sequences of five heme containing fatty acid dioxgenases and a comparison of the predicted secondary structure of 10R-DOX with ovine PGHS-1. A, top, amino acids residues at the presumed peroxidase and hydroperoxide isomerase sites. The last two residues, His and Asn, are conserved in all myeloperoxidases (1). Middle and bottom, amino acid residues of the presumed dioxygenation sites are shown. Conserved residues in all sequences are in boldface, and mutated residues of 10R-DOX and/or 7,8-LDS are marked by an asterisk. B, alignment of partial amino acid sequences of 10R-DOX with ovine PGHS-1, and a secondary structure prediction of the 10R-DOX sequence. The secondary structure of 10R-DOX was predicted by PSIPRED (43) and the secondary structure of ovine PGHS-1 from its crystal structure (Protein Data Bank code 1diy; cf. Ref 19). In short, our first strategy for site-directed mutagenesis was to switch hydrophobic residues between the enzymes with 10R-and 8R-DOX activities and to assess the effects on the DOX and hydroperoxide isomerase activities (10R-DOX/7,8-LDS: Leu-306/Val-256, Leu-384/Val-330, Val-388/ Leu-334, and Ala-426/Ile-375) and to switch one hydrophobic/charged residue (Ala-435/Glu-384). Only catalytically active pairs would provide clear information on their importance for the position of dioxygenation (e.g. L384V of 10R-DOX and V330L of 7,8-LDS, both of which were active). Unfortunately, replacements of 7,8-LDS often led to inactivation or very low activity (e.g. V330A, V330M, I375A, E384A). Our second strategy was to study replacements in two homologous positions of ovine PGHS-1 (Val-349 and Ser-353) with smaller and larger hydrophobic residues, i.e. at Leu-384 and Val-388 of 10R-DOX. Abbreviations used are as follows: oCOX-1, ovine cyclooxygenase-1; Af, A. fumigatus; Gg, G. graminis. The GenBank TM protein sequences were derived from P05979, EAL89712, AAD49559, EAL84400, and ACL14177. The amino acid sequences were aligned with the ClustalW algorithm (DNAStar).
There is little information on the hydroperoxide isomerase and peroxidase sites of LDS (18,26), but the latter could be structurally related to the peroxidase site of PGHS. PGG 2 and presumably 8R-HPODE bind to the distal side of the heme group, which can be delineated by hydrophobic amino acid residues (27). Val-291 is one of these residues, which form a dome over the distal heme side of COX-1. The V291A mutant retained cyclooxygenase and peroxidase activities (27). 5,8-and 7,8-LDS also have valine residues in the homologous position, whereas 8,11-LDS and 10R-DOX have leucine residues (Fig. 1). Whether these hydrophobic residues are important for the peroxidase activities is unknown.
In this study we decided to compare the two catalytic sites of 10R-DOX of A. fumigatus and 7,8-LDS (EC 1.13.11.44) of G. graminis (18). Our first aim was to find a robust expression system for 10R-DOX of A. fumigatus. The second objective was to determine whether C 16 and C 20 fatty acid substrates enter the oxygenation site of 10R-DOX "head" or "tail" first. Unexpectedly, we found that 10R-DOX oxygenated 20:4n-6 by hydrogen abstraction at both C-13 and C-10 with formation of two nonconjugated and four cis-trans-conjugated HPETEs. Our third objective was to investigate the structural differences between 10R-DOX and 7,8-LDS of G. graminis, which could explain that oxygenation of 18:2n-6 mainly occurred at C-10 and at C-8, respectively. The strategy for site-directed mutagenesis of 10R-DOX and 7,8-LDS is outlined in the legend to Fig. 1; an alignment of the amino acid sequences of 10R-DOX and 7,8-LDS is found in supplemental material.
Site-directed Mutagenesis-Site-directed mutagenesis of pIZ_10-DOX and pIZ/V5-His_LDS was performed with Pfu DNA polymerase and oligonucleotides of 44 -46 bases (supplemental Table 2S), as described by the QuikChange manual (Stratagene) with minor modifications (31). All constructs were confirmed by sequencing.
Enzyme Assay-Recombinant 10R-DOX, 7,8-LDS, and their mutants were incubated with 100 M of unsaturated fatty acids for 30 min on ice. The reaction (0.2-0.5 ml) was terminated with ethanol (3-4 volumes), and an internal standard was added in some experiments, and proteins were precipitated by centrifugation. The metabolites were extracted on octadecyl silica (SepPak/C 18 ), evaporated to dryness, and diluted in the mobile phase or methanol (20 -40 l), and 5-10 l was subject to LC-MS/MS analysis. TPP was used to reduce hydroperoxides to alcohols. Oxygenation rates of 100 M fatty acids by 10R-DOX and kinetic parameters of 18:2n-6 were determined in triplicate by LC-MS/MS (supplemental material).
HPLC-MS/MS Analysis-A multiple-stage linear ion trap mass spectrometer (LTQ, ThermoFisher) with the Surveyor MS pump and Surveyor autosampler was used with electrospray ionization and monitoring of negative ions. The electrospray needle was set at 4.5 kV; the temperature of the heated capillary was 315°C, and PGF 1␣ was used for tuning. Data were analyzed by the Xcalibur software. The RP-HPLC column contained octadecyl silica (5-m, 150 ϫ 2 mm) and was eluted with methanol/water/acetic acid, 80:20:0.01 or 75:25:0.01, at 0.3 ml/min (3,28). For quantification we used the ratio of signal intensities of the carboxylate anions of the products and the internal standard ([ 2 H 4 ]13-HODE or [ 2 H 8 ]15-HETE) and measured the area under the peaks from experiments done in triplicate.

RESULTS
Expression of 10R-DOX-The predicted sequence of 10R-DOX of A. fumigatus from genome analysis is available at GenBank TM (accession number XM_749316). We cloned 10R-DOX by RT-PCR and sequenced 3366 bp (GenBank TM accession number ACL14177). A comparison with the predicted gene sequence confirmed 13 exons and 12 short introns (supplemental material).
Recombinant 10R-DOX was successfully expressed in insect cells (Sf21) using the plasmid-driven expression method. Two different constructs were utilized, one with the native stop codon (pIZ_10-DOX) and one without but with 34 additional C-terminal amino acids (V5 epitope and His 6 tag; pIZ/ V5His_10-DOX; see supplemental material). Both recombinant enzymes appeared to oxygenate 18:2n-6 to the same products, as described below. Expression of 10R-DOX with the His tag was confirmed by Western blot analysis (Fig. 2). Low speed supernatant of untransformed Sf21 cells did not convert 18:2n-6 to any of these metabolites, and we could therefore use the 10R-DOX activity as evidence of enzyme induction.
The oxidation of 100 M 18:2n-6 was linear for at least 30 min on ice. The apparent K m was 0.05 mM, as judged from triplicate LC-MS analysis with [ 2 H 8 ]15S-HETE as an internal standard with monitoring of the area under the carboxylate anions of the reconstructed ion chromatograms (supplemental material). By this method, 100 M 18:3n-3 (n ϭ 3) was found to be oxidized as rapidly as 100 M 18:2n-6 (n ϭ 3). 18:1n-9 (n ϭ 2) was oxidized at a rate of ϳ40% of 18:2n-6. Metabolites of 18:3n-6 could not be detected.
The products were separated by NP-HPLC into six HETEs (Fig. 3B). Steric analysis showed that the S stereoisomers of 8-, 10-, 11-, and 13-HETE were formed in access over the R stereoisomers (Fig. 4). This was also confirmed by CP-HPLC analysis using Chiralcel OD-H (with known elution order of the enantiomers of 10-and 13-HETE (33)), Reprosil Chiral-AM, and by preparation of 13S-HPETE (supplemental material). The corresponding HPETEs were also detected in some experiments, but significant amounts 5-H(P)ETE and 9-H(P)ETE could not be detected. The rate of total oxidation of 100 M 20:4n-6 was estimated after reduction of hydroperoxides to alcohols to be ϳ50% of the rate of oxidation of 100 M 18:2n-6. We conclude that hydrogen abstraction occurred at C-10 and C-13 of 20:4n-6, and this could account for formation of all six metabolites (cf. Ref. 33).
The L384A mutant changed the stereochemistry at C-8 and C-10. CP-HPLC-MS/MS analysis showed that L384A formed the R and S enantiomers of 8-HODE and 10-HODE in a ratio of ϳ3:2 (Fig. 6), whereas native and recombinant 10R-DOX form both products with Ͼ95% R configuration (28). Replacement of Leu-384 with methionine increased the relative biosynthesis of 8-HODE significantly (n ϭ 6, p Ͻ 0.05; sign test), although only with 3-4% units. Finally, we prepared the L384F mutant. The relative biosynthesis of 8-HODE was 48 Ϯ 0.06% (n ϭ 3). This mutant did apparently not shield C-8 for oxygenation. NP-and CP-HPLC separation and MS/MS showed that with this bulky replacement significant amounts of racemic 9-and 13-HODE also appeared to be formed (about 20% of 10-HODE); 10-and 8-HODE retained their R chirality (Ͼ90% R).
The position of Val-388 was also of catalytic importance. V388L increased the relative biosynthesis of 8-HODE from 10.4 Ϯ 2 to 16.2 Ϯ 1.1% (n ϭ 5; Fig. 5B and supplemental material) and V388F increased it to 34 Ϯ 1.5% (n ϭ 3). NP-HPLC-MS/MS analysis showed that the latter also increased the relative formation of 9-and 13-HODE (about 80% of 10-HODE), suggesting that this bulky residue may have profoundly changed the oxygenation site. CP-HPLC showed that 9-and
We also prepared the 10R-DOX double mutant, L384A/ V388L. This mutant formed almost the same relative amounts of 8R-HODE (49 Ϯ 7%; n ϭ 4) as the L384A mutant. Steric analysis showed that 8-and 10-HODE were both formed with  reduced stereospecificity (supplemental material). We conclude that amino acids in the positions 384 and 388 can be crucial for the position of dioxygenation and can affect the absolute configuration of 8-and 10-HPODE.
Leu-334 is homologue to Val-388 of 10R-DOX (Fig. 1). Mutation of Leu-334 to valine in 7,8-LDS reduced the dioxygenase activity below the detection limit (n ϭ 6). Alignment of 7,8-LDS and 10R-DOX sequences around the proximal heme ligand showed several conserved differences. Ile-375 and Glu-384 were conserved in the 7,8-LDS sequences but replaced with Ala in 10R-DOX sequences (Fig. 1). These residues could conceivably be important for the position of oxygenation. I375A of 7,8-LDS (n ϭ 3) had virtually no dioxygenase activity, whereas E384A (n ϭ 5) formed only traces of 8-HODE and 10-HODE and in the same ratio as the native enzyme (LC-MS/MS analysis). These positions were therefore not further investigated. Satisfactory expression of recombinant 7,8-LDS was confirmed by Western blot analysis of all mutants described above (supplemental material).
Detection of EAS Activity of 10R-DOX-Recombinant 10R-DOX metabolized 18:2n-6 to ϳ90% 10R-HPODE and ϳ10% of 8R-HPODE, as estimated by LC-MS/MS analysis (Fig. 7A). In addition, more polar products were also formed, albeit in small amounts (ϳ1%). This may nevertheless be of interest. Recombinant 7,8-LDS has weak hydroperoxide isomerase activity, and the hydroperoxide isomerase activity of purified native 7,8-LDS varies considerably (2). This could also apply to 10R-DOX. The minor metabolites were therefore identified. The reconstructed ion chromatogram ( Fig. 7B; MS/MS analysis with monitoring of m/z 311 3 full scan) revealed elution of metabolites after 4 min (peak I) and a smaller amounts of metabolites on its right shoulder (peak II). They were further separated by NP-HPLC into one major and several minor metabolites (Fig. 8).

DISCUSSION
We have expressed 10R-DOX and studied its 10R-DOX and 8R-DOX activities with different fatty acid substrates and by site-directed mutagenesis. Our study revealed two important amino acids for the 10R-DOX and 8R-DOX activities, Leu-384 and Val-388. Our results showed similarities with replacement of the two homologous amino acids of active site of PGHS-1 (Val-349 and Ser-353). Recombinant 10R-DOX lacked hydroperoxide isomerase activity but possessed low EAS activity with biosynthesis of a novel oxylipin, 12S(13R)-epoxy-10Rhydroxy-18:1. In analogy with PGHS and LDS, 10R-DOX thus possesses dual enzyme activities.
We first examined 10R-DOX with different substrates to determine the orientation of the substrates at the active site. Oxygenation of 18:2n-6 and 18:3n-3 mainly occurred at C-10 (position n-9; ϳ90%) but also at C-8 (position n-11; ϳ10%). The oxygenation of C 16 -C 20 fatty acids suggests that they enter the site of oxygenation tail first, as hydrogen abstraction apparently occurred at the n-11 position of C 18 fatty acids and the n-9 position of C16:1n-7 ( Table 1). Insertion of molecular oxygen then occurred at positions n-11 and n-9.
Why does 10R-DOX favor C-10 of 18:2n-6 for oxygenation after hydrogen abstraction at C-8? It seems reasonable that C-8 is in proximity of Tyr-427 (cf. Fig. 1 and Fig. 9), which abstracts the pro-S hydrogen and forms the carbon-centered radical with electron density over C-8 to C-10. Dioxygen apparently then has only limited access to C-8 and will react with C-10. Conversely, the oxygen channel of LDS may lead to C-8 with little access to C-10. Can this difference be explained by substitution of single amino acids during evolution?
LDS and 10R-DOX can be aligned with 42-48% amino acid identity. A conserved motif, Tyr-Val-(Xaa) 3 -Leu, which occurs in two 7,8-LDS and five 5,8-LDS sequences, corresponds to the conserved motif Tyr-Leu-(Xaa) 3 -Val in 10R-DOX sequences of five Aspergilli (and to Tyr-Val-(Xaa) 3 -Ser in PGHS). We found that these leucine and valine residues were critical for the oxygenation at C-10 and C-8 of 18:2n-6. The L384V and L384A mutants of 10R-DOX increased the relative oxygenation at C-8 from 10% (native recombinant 10R-DOX) to 22 and 53%, respectively, and led to biosynthesis of 8-and 10-HODE with less R stereoselectivity. Conversely, replacement of the homo-  Hydrogen is likely abstracted by Tyr-427, and O 2 reacts with the carbon-centered radical at C-10 (ϳ90%) and at C-8 (ϳ10%). The radical at C-8 is partly shielded from O 2 by Leu-384, but smaller hydrophobic residues (valine and alanine) in this position increase the oxygenation at C-8. Conversely, replacement of Val-388 with larger hydrophobic residues (leucine and phenylalanine) increase oxygenation at C-8, possibly by shielding C-10 from O 2. The bulky phenylalanine residue may also induce other steric effects, as both L384F and V388F increased the relative biosynthesis of 8-, 9-, and 13-HODE to 10-HODE. The corresponding conserved motif of PGHS is Tyr-Val-(Xaa) 3 -Ser, and the three amino acids are important for the cyclooxygenase reaction. The LDS motif is Tyr-Val-(Xaa) 3 -Leu. Replacement of Tyr-329 with phenylalanine reduces the catalytic activity of 7,8-LDS (18), whereas V330L increases the relative 10R-DOX activity, albeit modestly. Unfortunately, V330M, V330A, and L334V of 7,8-LDS were inactive. These mutants emphasize the steric importance of this motif for catalytic activity.
Val-388 of 10R-DOX also affected oxygenation. Replacement of Val-388 with leucine (the homologue in 7,8-LDS) and with phenylalanine increased the relative oxygenation at C-8 from 10 -16 and 36%, respectively. The bulky phenylalanine residue as position 388 may have additional steric effects as judged from prominent formation of other products (9-and 13-HODE), and this was also the case with L384F.
The results of our replacement experiments with 10R-DOX can be visualized in a model of the conserved motif Tyr-Leu-(Xaa) 3 -Val and a tentative oxygen channel with access to C-10 (90%) and partially to C-8 (10%) of 18:2n-6 ( Fig. 9). Leu-384 could be partly shielding C-8 for oxygen, and Val-388 could delineate the channel near C-10. Replacement of Leu-384 with smaller hydrophobic residues (valine, alanine) increased oxygen access to C-8, whereas replacement of Val-388 with larger hydrophobic residues (leucine and phenylalanine) shielded C-10 and thus augmented the relative oxygenation at C-8. L384F was expected to yield mainly 10R-DOX activity, but this bulky residue may have induced other structural changes. Three-dimensional structural analysis will be needed to accommodate all information into a comprehensive model. It is therefore interesting to compare these results with replacements of the corresponding residues of PGHS, valine and serine, where three-dimensional information is available.
The peroxidase site of PGHS forms a hydrophobic dome over the distal site of the heme with the sequence Val-Phe-Gly-Leu-(Leu/Val) (27). Replacement of Val-291 with alanine had only little effect on PGHS, as judged from its cyclooxygenase and peroxides activity (27). The Val-291 homologue of 10R-DOX is Leu-306, as judged from the alignment illustrated in Fig. 1. We found that L306A and L306V retained the EAS activity of recombinant 10R-DOX. It seems possible that some of the other hydrophobic residues could be important for the peroxidase and EAS activities.
In summary, we have expressed 10R-DOX with EAS activity and found that its 10R-and 8R-dioxygenase activities can be modified in a logical way by replacements of two hydrophobic residues in a common conserved motif of 10R-DOX, 7,8-LDS, and PGHS, viz. Tyr-(Leu/Val)-(Xaa) 3 -(Val/Leu/Ser). Replacement of hydrophobic resides in this motif also changed or abolished the 8R-DOX activity of 7,8-LDS.