Synthesis of two new lipid mediators from docosahexaenoic acid by combinatorial catalysis involving enzymatic and chemical reaction

Omega-3 polyunsaturated fatty acids (PUFAs) have been known to have beneficial effects in the prevention of various diseases. Recently, it was identified that the bioactivities of omega-3 are related to lipid mediators, called pro-resolving lipid mediators (SPMs), converted from PUFAs, so they have attracted much attention as potential pharmaceutical targets. Here, we aimed to build an efficient production system composed of enzymatic and chemical catalysis that converts docosahexaenoic acid (DHA) into lipid mediators. The cyanobacterial lipoxygenase, named Osc-LOX, was identified and characterized, and the binding poses of enzyme and substrates were predicted by ligand docking simulation. DHA was converted into three lipid mediators, a 17S-hydroxy-DHA, a 7S,17S-dihydroxy-DHA (RvD5), and a 7S,15R-dihydroxy-16S,17S-epoxy-DPA (new type), by an enzymatic reaction and deoxygenation. Also, two lipid mediators, 7S,15R,16S,17S-tetrahydroxy-DPA (new type) and 7S,16R,17S-trihydroxy-DHA (RvD2), were generated from 7S,15R-dihydroxy-16S,17S-epoxy-DPA by a chemical reaction. Our study suggests that discovering new enzymes that have not been functionally characterized would be a powerful strategy for producing various lipid mediators. Also, this combination catalysis approach including biological and chemical reactions could be an effective production system for the manufacturing lipid mediators.

Lipid mediators, such as resolvins and protectins, as bioactive signaling molecules are converted from EPA and DHA via lipoxygenase enzymatic reaction at inflamed sites 1 . These lipid mediators actively promote the termination of inflammation with general actions: blocking the conversion of pro-inflammatory mediators such as prostaglandins and leukotriene, releasing other signaling molecules with anti-inflammatory activity, inhibiting the expression of pro-inflammatory cytokines, and promoting macrophage phagocytosis 2,3 . In addition, several in vitro and in vivo studies have demonstrated the pathological effects of SPMs on cancer and vascular diseases 4,5 . Eventually, it was revealed that the health benefits of omega-3 fatty acids in humans are due to the physiological action of their downstream metabolites 6 .
In recent years, lipid mediators behind the beneficial effects of omega-3 fatty acids have attracted much attention as important pharmaceutical targets for the protection and treatment for chronic inflammatory diseases. Consequently, the resolvin E1 analog, RX-10045, is in Phase 2 clinical trials for dry eye and other retinal diseases 7 , and the lipoxin analogs BLXA4 (ZK-142 and ZK-994) are undergoing Phase 1 clinical trials for treatment of inflammatory airway diseases (e.g. asthma) 8 . In contrast to lipid mediators actively promoting the termination of the inflammatory response, traditional steroidal and non-steroidal types of anti-inflammatory drugs have focused on suppressing the initiation of the inflammation. Although conventional medication for inflammation that has been used for a long time can be effective, many of them eventually lead to immunosuppressive which may increase the risk of infection 9 . Hence, resolution therapy based on the SPMs to control various chronic inflammatory diseases may be an innovative therapeutic approach enable to replace current treatments. Gene cloning. The gene encoding the lipoxygenase, Osc-LOX, from O. nigro-viridis PCC 7112 was codonoptimized and synthesized by Bioneer Inc. The synthesized gene was used as a template ( Supplementary Fig. S1). The primers used for cloning were designed using SnapGene Viewer software and were based on the sequence of the synthesized gene (Supplementary Table S1). Polymerase chain reaction (PCR) was performed using pfu DNA polymerase premix, and the amplified PCR product was purified using a purification kit. The purified product was inserted between NdeI and XhoI sites of the pET-28a plasmid, resulting in fusion of a hexa-histidine tag to the N-terminus. The stop codon was located in front of XhoI so as to prevent expression of the C-terminal his-tag. E. coli strain DH5α was transformed with the constructed plasmids. DNA sequencing, performed by Bioneer Inc., showed that the resulting sequence was identical to the original gene sequence.
Harvested E. coli cells containing Osc-LOX were resuspended in buffer A (50 mM Tris-HCl pH 7.5, 500 mM NaCl, 10 mM imidazole, 0.1 mM phenylmethylsulfonyl fluoride, 5% glycerol). The resuspended cells were disrupted by sonication on ice, and then cell debris was removed by centrifugation at 13,000×g for 30 min at 4 °C. The supernatant was loaded onto an affinity column equilibrated with buffer A. A cobalt affinity column (HiTrap Talon crude, 5 mL) was utilized in the first step of purification of Osc-LOX. The bound protein was washed with 50 mM imidazole and then eluted with a linear gradient of 10 to 700 mM imidazole at flow rate of 5 mL min −1 . Subsequently, eluted fractions containing target protein were pooled and concentrated using an Amicon Ultra-15 centrifugal concentrator (10,000 kDa molecular weight cut-off). Concentrated protein was loaded onto a HiLoad 16/600 Superdex 200 pg column (GE Healthcare) that had been previously equilibrated with gel filtration buffer (50 mM Tris-HCl pH 7.5, 150 mM NaCl) and was separated based on its molecular size and shape at a flow rate of 1 mL min −1 . And, this size-exclusion chromatography (SEC) was used to determine the molecular weight and oligomeric state. The molecular weight of the eluted protein was calculated by comparison of its retention time with that of standard materials. The gel filtration standards, thyroglobulin (670 kDa), γ-globulin (158 kDa), ovalbumin (44 kDa), myoglobulin (17 kDa) and vitamin B12 (1.35 kDa), were used for calibration. Sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) on 12% gels was used to confirm purification at each step.
Enzymatic activity assay and thermal stability. Purified Osc-LOX was used to determine the optimum reaction conditions of pH and temperature toward LA (18:2n-6), as well as the half-life of the enzyme. Unless otherwise noted, all enzymatic reactions for characterization were performed using 3 μg mL −1 enzyme and 50 μM substrate for 10 min. The optimum pH for enzyme reactions was determined over a range of pH values from 5.0 to 10.0 using 50 mM 2-(N-morpholino) ethanesulfonic acid (MES) buffer (pH 5.0), 50 mM sodium phosphate buffer (pH 6.0-7.0), 50 mM Tris-HCl buffer (pH 8.0) and 50 mM sodium tetraborate buffer www.nature.com/scientificreports/ (pH 9.0-10.0) at room temperature. The optimum temperature for enzymatic activity and thermal stability was investigated by pre-incubating the reaction mixture without substrate at different temperatures from 20 to 60 °C in 50 mM Tris-HCl buffer (pH 8.0) for 30 min, after which activity was measured following incubation with substrate for 10 min. The half-life of Osc-LOX was measured by incubating the enzyme reaction mixture excluding substrate up to 360 min at a temperature of 30 °C. The enzyme that was continuously incubated at 30 °C was taken every 30 min interval and checked for remaining activity.
Enzyme kinetics measurements. Arachidonic acid (AA), eicosapentaenoic acid (EPA), and docosahexaenoic acid (DHA) were used to determine the kinetic parameters of purified Osc-LOX. For kinetics analysis, the formation rate of the conjugated diene from the three different substrates was measured by monitoring the increase in absorbance at 234 nm. The enzymatic reaction was performed in a quartz cuvette (10 mm) using 50 mM Tris-HCl buffer (pH 8.0), 3 μg mL −1 enzyme, and varying amounts of substrate (25-175 μM) for 5 min at 30 °C. The reaction mixture was quickly mixed, and absorbance was measured every 2 s. Spectra were collected using an Amersham Biosciences Ultrospec 3100 Pro. Data represent the means (± standard deviation, SD) of three separate experiments, each carried out in triplicate. LOX activity was calculated using the extinction coefficient, ε = 23,000 M −1 cm −1 , for the conjugated diene 15  . The catalyzed products were extracted using an HP20 resin (1/20 of the reaction volume). Briefly, HP20, pre-activated with ethanol, was washed with 5 resin volumes of deionized water. The reaction mixture was loaded onto the cartridge containing HP20 and washed with 10 resin volumes of deionized water and dried under a stream of nitrogen gas. Products bound to HP20 were eluted with 3 resin volumes of ethanol. The extracted eluents were used for HPLC and LC-MS/MS analysis.
Base catalyzed hydrolysis for epoxide ring-opening. The epoxide ring-opening reaction was carried out using sodium hydroxide (NaOH) as a strong nucleophile. The aqueous solution for the reaction was prepared to contain 0.2 M NaOH in distilled water (DW), to which was added epoxidized material at a final concentration of 100 μg mL −1 . The reaction mixture was incubated overnight at 30 °C with stirring. As a reference, the hydrolysis reaction for epoxide ring opening was performed using a commercially available EH from Rhodococcus rhodochrous (Sigma-Aldrich) in accordance with the manufacturer's protocol. The hydrolyzed products were extracted using HP20, as described in the enzyme reaction section.
HPLC analysis for conversion products. HPLC analyses of products converted by Osc-LOX were performed using an Agilent 1200 series equipped with a quaternary pump, solvent degasser, autosampler, thermostatted column compartment, diode array detector, and fraction collector. Normal-phase HPLC (NP-HPLC) was carried out isocratically on a SUPELCOSIL LC-DIOL column (SUPELCO, 25 cm × 3 mm, 5 μm particle size) using heptane/2-propanol/acetic acid (95:5:0.1, by volume) as the solvent. The flow rate was 0.5 mL min −1 , and the column temperature was maintained at 10 °C. Reversed-phase HPLC (RP-HPLC) was performed on a HECTOR-M C18 column (HECTOR, 25 cm × 4.6 mm, 5 μm particle size) using methanol/water/acetic acid (70:30:0.1 by volume) as a solvent. The reaction products were separated with the mobile phase at a column temperature of 30 °C and a flow rate of 1 mL min −1 . Enantiomers were separated by normal-phase chiral HPLC (NPchiral HPLC) using a CHIRALCEL OD-H column (DAICEL, 25 cm × 4.6 mm, 5 μm particle size) and reversedphase chiral HPLC (RP-chiral HPLC) using a CHIRALPAK IB column (DAICEL, 25 cm × 4.6 mm, 5 μm particle size). For NP-chiral HPLC, a mobile phase consisting of n-hexane/2-propanol/acetic acid (95:5:0.1, by volume) was used at a flow rate of 1 mL min −1 and a column temperature of 30 °C. For RP-chiral HPLC, a mobile phase consisting of acetonitrile/water/formic acid (50:50:0.1, by volume) was used at a flow rate of 0.8 mL min −1 and a column temperature of 25 °C. All samples used for NP-, RP-, NP-chiral and RP-chiral HPLC were dissolved in mobile phase after removing ethanol and were injected in a total volume of 10 μL. Substrate (PUFAs) remaining after the reaction was assessed by monitoring absorbance of samples at 210 nm, and conjugated hydroxy fatty acids were detected by monitoring absorbance at 234 nm, 237 nm, 242 nm, 270 nm, and 302 nm. The standard materials, 17S-HDHA, 17R-HDHA and resolvin D1-5, were used to identify the reaction products by comparison of retention times. The concentrations of lipid mediators were calculated by correlating peak areas with concentrations of standard materials in linear calibration curves. The calibration curves for standard materials are shown in Supplementary Fig. S2. Crude reaction products were separated by NP-HPLC using a SUPELCOSIL LC-DIOL column, and were collected with a fraction collector. Fractions were evaporated under a stream of nitrogen gas and stored in absolute ethanol at − 80 °C. Purified target compounds were used for NMR analysis. www.nature.com/scientificreports/ LC-MS/MS analysis for identification of conversion products. LC-MS analyses were performed using an Agilent TOF spectrometer (G6550A) with an electrospray ionization (ESI) interface. The converted products produced by enzymatic reactions were separated on a ZORBAX Eclipse Plus C18 Rapid Resolution High Definition column (Agilent, 10 cm × 2.1 mm, 1.8 μm particle size) using an Agilent 6200 series UHPLC system consisting of a binary pump, autosampler, multicolumn thermostat, and diode array detector. The column was eluted isocratically with a solvent of water/acetonitrile/formic acid (90:10:0.1, by volume) at a column temperature of 40 °C and a flow rate 0.3 mL min −1 . All samples used in analyses were dissolved in absolute ethanol after removing methanol and were injected in a total volume of 2 μL. LC flow was directly injected into a mass spectrometer equipped with a Dual AJS ESI source. Sample ionization was performed in negative mode using the following conditions: drying gas, nitrogen (14 L min −1 , 200 °C); nebulizer gas, nitrogen (35 psi); capillary voltage, 3.5 kV; capillary temperature, 350 °C. Spectra were recorded over an m/z range of 50-3000 with an accumulation rate of 2 spectra/s. Data were processed using Mass Hunter Workstation Acquisition software. The chemical structure was drawn and the molecular masses of its fragments were calculated using ChemDraw Professional v 15.1.

Nuclear magnetic resonance (NMR) spectroscopy. NMR experiments were conducted using a
Bruker Avance III HD spectrometer (800 MHz) with a 5 mm triple resonance inverse (TCI) Cryo Probe at the Research Institute of Pharmaceutical Sciences (Seoul National University, Seoul, Republic of Korea). Purified lipid mediators, prepared at a concentration of 5 to 7 mg mL −1 were dissolved in methanol-d4 (CD 3 OD) and measured at 298 K. In addition to 1D NMR ( 1 H, 13 C), the following 2D NMR experiments were performed: correlation spectroscopy (COSY), total correlation spectroscopy (TOCSY), nuclear Overhauser effect spectroscopy (NOESY), hetero-nuclear single-quantum correlation spectroscopy (Edited-HSQC), and hetero-nuclear multiple-bond correlation spectroscopy (HMBC) 17 . NMRPipe 18 and NMRView software 19 were employed for NMR data processing and peak assignment. All samples were confirmed by several characteristic proton and carbon resonances. Protons were assigned using integrated 1 H-COSY (1D 1 H and 2D 1 H) spectra, which measure the correlation between adjacent protons. Then, protons were assigned to carbons using 2D-Edited-HSQC ( 1 H and 13 C) spectra, which record single-through-bond relationships. TOCSY and HMBC spectra were used to identify relationships through either two or three bonds, and to distinguish overlapping peaks. Stereochemistry were confirmed by NOESY spectra based on the identified stereo information of the precursor.
Polarimetric analysis of conversion products. Specific rotations were measured using a 10 mL cell with 100 mm path length on a JASCO P-2000 Digital Polarimeter. All conversion products were prepared in concentrations ranging from 0.05 to 0.1 mg mL −1 in ethyl alcohol, and their specific rotations were detected in absorbance at 589 nm at room temperature. Molecular modeling and substrate docking simulation. The 3D structure of Osc-LOX was acquired using known crystal structures, soybean LOX-3 (PDB ID: 1NO3 and 1HU9), with high sequence homology (21% identity) as templates 20,21 . Initial three-dimensional (3D) structures were generated using the I-TASSER method 22,23 . The generated structure of Osc-LOX was then refined by performing loop modeling using a modeler loop-building algorithm implemented in UCSF Chimera 1.11.2 (https ://www.cgl.ucsf.edu/chime ra/) 24,25 . Finally, energy minimization of modeled structure was performed after clean up and addition of hydrogen using YASARA 19.5.5 (https ://www.yasar a.org/) [26][27][28] . Molecular surface analyses of protein structures and calculation of electrostatic surface potentials (ESP) were investigated using UCSF Chimera and YASARA. The structures of lipid mediators were drawn using MarvinSketch 17.28 (2018, ChemAxon, https ://www.chema xon.com), and 3D structures of lipid mediators were generated and minimized using UFF/GAFF force field in Avogadro 29 and Spartan 18 30 . Molecular docking of Osc-LOX with lipid mediators was performed using AutoDock Vina 31,32 , and analyses of ligand-protein interactions were performed using PLIP (Protein-Ligand Interaction Profiler) 33

Results and discussion
Conserved sequence analysis. The sequence of osc-lox (accession number K9VMV7, 1713 bp) was found in the UniProt database for gene synthesis; however, its enzymatic activity studies had not been reported. The osc-lox gene encodes a protein consisting of 571 amino acids (calculated molecular mass, 64.9 kDa). A sequencealignment analysis was performed with conserved sequences of previously reported LOX enzymes from animals, plants, fungi, proteobacteria, and cyanobacteria ( Supplementary Fig. S3). The metal-binding amino acids in all LOXs were perfectly conserved as the motif, HHHN(H/S)I; however, Osc-LOX incorporates a methionine in the last position instead of isoleucine. Alanine and glycine in the catalytic site of LOX are preserved and known as stereo-controlled determinants called the Coffa site 34 . In the Coffa site hypothesis, alanine is conserved in S-type LOX and glycine is conserved in R-type LOX; Osc-LOX contains alanine 296, and thus is an S-type LOX.
Preparation of Osc-LOX for enzyme assay and reaction. Osc-LOX containing a His 6 -tag had a molecular weight (theoretically calculated) of approximately 67.2 kDa ( Supplementary Fig. S4a, left). Cobalt affinity chromatography was utilized as a first step of purification. Osc-LOX was eluted over an imidazole con-  Fig. S4a, middle). To obtain highly purified enzyme, we performed size-exclusion chromatography (SEC). Osc-LOX was retained on the SEC column for 82 min, and the relative molecular weight was confirmed to be ~ 67.5 kDa (Supplementary Fig. S4b). Approximately 15 mg of Osc-LOX purified with high purity of 96% was obtained from a 1-L culture ( Supplementary Fig. S4a, right). Since the enzyme was concentrated, the specific activity (unit mg −1 ) of the enzyme increased at each purification stage, but the total activity decreased owing to loss of protein during the purification process. Total activity (unit, U) of unpurified Osc-LOX (crude enzyme) was calculated to be 146 kilounits. After the first purification step, 71.23% (104 kilounits) of the activity remained, and after the second purification step, 27.4% (40 kilounits) of the activity remained (Supplementary Table S2). Total activity refers to the amount of enzyme that can convert monohydroperoxy fatty acids from DHA. The highly purified Osc-LOX was used for characterization and kinetics assays. However, because a high concentration of sample is required for NMR analysis, the unpurified supernatant (crude enzyme) after cell disruption was used for efficient production of lipid mediators.
Optimal conditions for catalysis and thermal stability. The optimal conditions for catalysis were determined by monitoring the catalytic activity of Osc-LOX at different pH values and temperatures. The effects of pH and temperature on the production of hydroxyoctadecadienoic acid (HODE) from LA (C18:2 n-6 Δ9,12 ) were determined using spectroscopic methods. The maximum activity of Osc-LOX was defined as 100%, and the relative activity, as a percentage, was plotted against various pH values and temperature. The relative activity of purified Osc-LOX at various pH values was determined in difference buffer systems at room temperature. Osc-LOX showed optimal activity at pH 8.0 (Fig. 1a). Although it has been reported that some LOXs exhibit optimal activity at an acidic pH, most LOXs have optimal activity at a neutral or alkaline pH 35 . The optimum temperature and thermal stability of Osc-LOX was measured after incubation for 30 min in 50 mM Tris-HCl buffer (pH 8.0) www.nature.com/scientificreports/ at temperatures between 20 and 60 °C at 5 °C intervals. The optimum temperature for the enzymatic reaction of Osc-LOX was 30 °C, and the enzyme appeared to be stable from 20 to 35 °C with no aggregation (Fig. 1b). However, Osc-LOX aggregated and its activity drastically decreased at temperatures above 40 °C; at temperatures above 55 °C, its activity was completely lost. Osc-LOX might exhibit relative thermal stability at temperatures below 35 °C, but heat denaturation occurs when the temperature rises above 40 °C. The optimum temperature for Osc-LOX was similar to that of soybean LOX, which is maximally active at 30 °C 36 . However, the optimal temperature for Osc-LOX was higher than that for other LOXs, such as human 12-LOX 37 and Agrocybe aegerita LOX4 38 , which show optimal activity at 25 °C. The half-life of the enzyme activity of Osc-LOX was determined by measuring the residual activity by taking the enzyme every 30 min while incubating at 30 °C for 360 min. As a result, Osc-LOX had a half-life of 240 min (Fig. 1c). The time it took for the enzyme activity of Osc-LOX to be decreased in half at the temperature of optimum activity was 1.7 times slower than that of sesame LOX 39 .
Specific activities and kinetics against PUFAs. The specific activity and kinetics of fatty acid oxyfunctionalization by Osc-LOX were analyzed using highly purified enzyme and the substrates, AA, EPA and DHA, under optimal enzymatic conditions. The Michaelis-Meten plots of each substrates are shown in Supplementary  Fig. S5. Michaelis-Menten constants (K m ), turnover numbers (k cat ), catalytic efficiencies (k cat /K m ), and specific activities of Osc-LOX for different substrates are shown in Three products generated from DHA by Osc-LOX. In the present study, we focused on the analysis of metabolites converted from DHA (Fig. 2a) by Osc-LOX because the reaction generated previously unknown products. The catalyzed products were identified by NP-, and CP-HPLC analysis. Three different products were obtained depending on the concentration of Osc-LOX used (50-400 Unit mL −1 ). In NP-HPLC, the first product converted at an Osc-LOX concentration of 50 Unit mL −1 eluted at 5.57 min, which matched the retention time of the 17S-HDHA standard (17S-hydroxy-4Z,7Z,10Z,13Z,15E,19Z-docosahexaenoic acid) at a λ max of 237 nm (Fig. 2b,d). At 50 Unit mL −1 , Osc-LOX completely consumed DHA, but the concentration of the converted product was 45 μM (conversion rate from DHA, 95%).
Additional conversion products of DHA catalyzed by Osc-LOX at concentrations of 100 to 200 Unit mL −1 were detected at 12.3 min at a λ max of 242 nm identical to the resolvin D5 (RvD5; 7S,17S-dihydroxy-4Z,8E,10Z,13Z,15E,19Z-docosahexaenoic acid) standard (Fig. 2c,e). The first product was converted into the second product at a concentration of 100 Unit mL −1 , but full conversion was achieved using an Osc-LOX concentration of 200 Unit mL −1 , which yielded 38 μM RvD5, a conversion rate of 84.5% from the first product (Fig. 2f). The concentrations of 17S-HDHA and RvD5 were calculated using commercially available standards.
Surprisingly, at an Osc-LOX concentration of 400 Unit mL −1 , we obtained a remarkable peak at 21.56 min at a λ max of 237 nm; the final product was obtained at 32 μM, corresponding to a conversion rate of 82.2% from the second product (Fig. 2g). The concentration of the final product was determined by reference to a calibration curve generated using purified products prepared by preparative HPLC.
Our assumption is that this latter peak is a new type of lipid mediator, as supported by several lines of evidence. First, this peak did not overlap with the retention time of any of the resolvin D series standards (RvD1-RvD5 and protectin DX; standards spectra not shown). Second, each product was further converted in a manner that depended on the enzyme concentration, indicating that the final product was generated from the second product (RvD5). Because LOX is known to be an irreversible enzyme 42 , higher concentrations are needed to catalyze monohydroperoxy fatty acids to dihydroperoxy fatty acids. Thus, LOX might lose its activity before being able to catalyze the next oxygenation. If the final product was converted from the second product (RvD5), one double bond must be eliminated because the final product had singlet peak at a λ max of 237 nm; its spectrum indicated that it has one conjugated diene. Lipid mediators have unique absorption spectra reflecting their conjugated double bonds.
In addition, the stereo-configuration of products generated from DHA by Osc-LOX were analyzed by CP-HPLC (OD-H column) and compared with 17S-HDHA, 17R-HDHA and RvD5 standards. The retention times of the first product was 9.22 min, which is confirmed as 17S-HDHA. And the retention times of the second product was 38 min, which is identical to RvD5 standard. Although the second product did not be compared with the stereoisomer, 7R,17S-diHDHA, through CP-HPLC analysis, it has been reported that RvD5 and its stereoisomer  43 . The final product stayed on the OD-H column for 82 min and its peak did not overlap those for any of the standards, as expected ( Supplementary Fig. S6). These results indicate that the first and second products of the catalysis of DHA by Osc-LOX also matched the stereo-configuration as well as the regio-configuration of 17S-HDHA and 7S,17S-diHDHA (RvD5), further supporting the possibility that the final product may be a new type of lipid mediator.

Structure identification of conversion products of DHA by Osc-LOX.
To identify conversion products, we first confirmed its molecular mass by performing LC-MS analysis in negative mode. The first product  Fig. S7c), which was identical to tri-hydroxy lipid mediators with six double bonds, such as RvD1 and RvD2, but HPLC analysis showed that it did not match previously known resolvin D series compounds. The final product was further analyzed by LC-MS/MS. Fragment peaks were detected at m/z values of 113 (235.1, 141) and 265.1 by cleavage between C6-C7, C14-C15, and C15-C16 positions ( Supplementary Fig. S8). This MS/MS analysis of the final product suggests that, because the molecular weight of the fragment containing the methyl group formed by cleavage between C14 and C15 was detected as an m/z of 141, the final product must reflect removal of a double bond with formation of an epoxide group. Thus, the final product should have two hydroxyl groups and one epoxide group, based on the total molecular weight or molecular weight of fragments. The structural configurations of the three products generated from DHA by Osc-LOX were more accurately determined by 1D and 2D NMR spectroscopy. The first and second products were assigned using 1 H, 13 C, COSY, TOCSY, Edited-HSQC, and HMBC spectra. The first product was identified as 17S-hydroxy-4Z,7Z,10Z,13Z,15E,19Z-docosahexaenoic acid (17S-HDHA). Chemical shifts of the C17-OH hydroxyl group in the first product were recorded at 4.12 ppm ( 1 H) and 73.99 ppm ( 13 C) (Supplementary Figs. S9 and S10, and Table S3a). The second product was identified as 7S,17S-dihydroxy-4Z,8E,10Z,13Z,15E,19Z-docosahexaenoic acid (RvD5). The two hydroxyl groups, C7-OH and C17-OH, of the second product were detected as overlapping peaks at 4.13 ppm ( 1 H) and 73.93-73.96 ppm ( 13 C), respectively (Supplementary Figs. S11 and S12, and Table S3b). Two of the three products were confirmed to be identical to previously reported chemical structures of 17S-HDHA and RvD5 (7S,17S-diHDHA), identified by Dobson 44  The final product was assigned using COSY, TOCSY, NOESY, Edited-HSQC, and HMBC spectra. The structure of the novel lipid mediator was identified as 7S,15R-dihydroxy-16S,17S-epoxydocosa-4Z,8E,10Z,13Z,19Zpentaenoic acid, which has two hydroxyl groups introduced at C7 and C15, with an epoxide group formed across C16 and C17. Chemical shifts corresponding to C7-OH, C15-OH, and C16-C17 (epoxide ring) were recorded as follows: C7-OH, 4.15 ppm ( 1 H) and 73.71 ppm ( 13 C); C15-OH, 4.23 ppm ( 1 H) and 69.97 ppm ( 13 C); C16-epoxide, 2.79 ppm ( 1 H) and 62.79 ppm ( 13 C); C17-epoxide, 2.9 ppm ( 1 H) and 57.42 ppm ( 13 C) (Supplementary Figs. S13-S15, and Table S3c). C7-H and C17-H were identified as the S-form because the precursor (RvD5) has a 7S, 17S stereo-configuration. Epoxide ring was determined to be the trans geometry based on the coupling constants of J16 and J17, which are 2.4 Hz and 2.2 Hz, respectively. Thus, C16-H was determined to be the R-form because C17-H has an S-form configuration. In addition, C15-H was identified as R-form because a strong NOE correlation signal was monitored between C17-H confirmed as S-form and C15-H in NOESY NMR (Supplementary Fig. S16). Although C15 is closer to C16 than C17, NOE signal was detected very weakly between C15-H and C16-H. This result indicates that a hydrogen atom of C15 has an opposite direction to a hydrogen atom of C16 hydrogen having an S-form configuration, and has a similar direction as a hydrogen atom of C17. The specific rotation value of 7S,15R-dihydroxy-16S,17S-epoxy-docosapentaenoic acid was measured as [α] D + 2.1° (c 0.1, EtOH).
Ultimately, final product was confirmed as a new type of lipid mediator because it does not match any molecules registered in chemical databases. Lipoxygenases classified as ARA15LOX type are known to convert intermediate 17S-HDHA into 7S,17S-dihydroxy-docosahexaenoic acid as RvD5 and 10S,17S-dihydroxy-docosahexaenoic acid as PDX 45 . Here, our study is the first to report the production of new structure lipid mediator via RvD5 converted from 17S-HDHA. Two products converted from 7S,15R-dihydroxy-16S,17S-epoxy-docosapentaenoic acid. We performed a hydrolysis reaction for epoxide ring-opening on purified 7S,15R-dihydroxy-16S,17S-epoxy-docosapentaenoic acid (Fig. 3a) using sodium hydroxide (NaOH). The converted products were confirmed using RP-HPLC (C18) and RP-chiral HPLC. RP-HPLC analysis showed that 7S,15R-dihydroxy-16S,17S-epoxy-docosapentaenoic acid was completely consumed and converted into two products: a main product with a conversion rate of 90% or less (~ 90 μg), and a minor product accounting for 10% (10 μg). The concentration of the minor product was determined by reference to a calibration curve prepared using a commercially available resolvin D2 (RvD2, 7S,16R,17S-trihydroxy docosahexaenoic acid) standard. However, the concentration of the major peak was estimated by subtracting the amount of the quantified minor peak from the concentration of the starting material because a standard curve could not be generated for the major peak owing to difficulties in purification. The retention times of the major and minor products were 6 and 8.8 min, respectively, values that were identical to those of the standards (Fig. 3b). These standards correspond to the conversion product of the EH reaction and commercially available RvD2 (Fig. 3c,d). In addition, the chirality of a minor product was identified by a RP-chiral HPLC (IB column) comparing with RvD2 standards. The retention times of the minor product was 6.78 min, which are identical to RvD2 standard ( Supplementary Fig. S17). EHs have been known to catalyze the conversion of an epoxide ring into diol in vivo. However, the base-catalyzed epoxide ring-opening method using NaOH could be more cost-effective, and simple than EH enzyme reaction. www.nature.com/scientificreports/ cleavage between C14 and C15 ( Supplementary Fig. S18a). These results provide additional supporting evidence that the two hydroxyl groups form by epoxide ring-opening because the molecular weight of the fragment containing the methyl group formed by cleavage between C14 and C15 was detected as an m/z of 159.1. Thus, the major product contains four hydroxyl groups. The minor product obtained by hydrolysis of the epoxide ring was identified as RvD2 using MS/MS analysis. The total molecular mass of the minor product and the RvD2 standard (376.  www.nature.com/scientificreports/ These results show that chemical shifts of the newly formed two hydroxyl groups at C16 and C17 by ring opening were significantly changed compared with 7S,15R-dihydroxy-16S,17S-epoxy-docosapentaenoic acid (Supplementary Fig. S22). Stereochemical configuration of C7-OH and C15-OH was determined as the S-and R-form, respectively, because the precursor was identified as 7S-and 15R-form. In the NOESY spectrum, NOE correlation signal was detected between C15-H and C16-H, and C16-H and C17-H, respectively ( Supplementary Fig. S23). However, unlike the NOESY NMR result of 7S,15R-dihydroxy-16S,17S-epoxy-docosapentaenoic acid, NOE signal was not detected between C15 and C17. This result showed that the chirality of carbon 16 was reversed when the epoxide ring was opened. The specific rotation value of 7S,15R,16S,17S-tetrahydroxy-pentaenoic acid was measured as [α] D − 6.4° (c 0.1, EtOH). 7S,15R,16S,17S-tetrahydroxy-4Z,8E,10Z,13Z,19Z-pentaenoic acid was also identified as a new type of lipid mediator because it does not match any molecules registered in chemical databases. We found we could accomplish two things at once because 7S,15R-dihydroxy-16S,17S-epoxy-docosapentaenoic acid was further converted to 7S,15R,16S,17S-tetrahydroxy docosapentaenoic acid and RvD2 by base-promoted epoxide ring-opening. This finding is interesting because, although an additional chemical reaction is required, it bypasses the difficulty and high cost of preparing an additional enzyme (EH) for converting epoxide into diol; moreover, RvD2 did not be converted by EH. To date, RvD2 has been produced only by total synthesis methods, with no reports of the efficient production of RvD2 by an in vitro enzymatic reaction. However, total synthesis methods have some limitations, including low yield, toxicity of chemicals, and the time-consuming 20-step process required 12 . The combinatorial approach using both enzymatic and chemical reactions described here could be a very attractive option for overcoming difficulties in the production of RvD2.

Structure identification of two products produced by hydrolysis.
3D structure and molecular docking of the Osc-LOX. A 3D structural analysis of the catalytic site of the Osc-LOX protein also showed that multiple products could be produced from the DHA substrate. Because the 3D structure of the Osc-LOX protein has not yet been elucidated, the binding pose of the DHA substrate at the catalytic site was predicted using a 3D structure generated by homology modeling (Figs. 4 and 5). In the predicted structure of Osc-LOX, like other LOX proteins, conserved amino acid residues (His253, His258, His434, Asn438, and Met571) are known to bind a metal ion, and Ala296 is close to the catalytic site and affects the binding pose of the substrate 46 . Ala296 is known to be an important residue that regulates regioselectivity during catalysis of other LOX proteins, an interaction that is explained by the Coffa theory.
In addition, using AutoDock Vina, we were able to calculate the binding energy for each binding pose (B, 8.2020 kcal/mol; C, 7.6950 kcal/mol; and D, 6.7720 kcal/mol), which predicts that substrate-enzyme complexes could be readily formed at higher binding energies (Fig. 5). These results indicate that low concentrations of Osc-LOX produce an intermediate complex predominantly in the form of the energy-favorable Conversion pathway of DHA into lipid mediators by combinatorial catalysis. Five types of bioactive hydroxyl fatty acids have been successfully converted from DHA used as start material via collaboration including enzymatic reaction and base promoted epoxide ring-opening. Three lipid mediators were produced by Osc-LOX. The conversion rate and purity of 17S-hydroxy-DHA from DHA were 95%, respectively. RvD5 (7S,17S-dihydroxy-DHA) was generated from 17S-hydroperoxy-DHA with 84.5% conversion rate and its purity was 76%. The yield of 7S,15R-dihydroxy-16S,17S-epoxy-DPA produced from 7S,17S-dihydroperoxy-DHA was 82.2% and the purity was 64%. The yield and purity of two products generated from 7S,15R-dihydroxy-16S,17Sepoxy-DPA by NaOH were followed: 7S,15R,16S,17S-tetrahydroxy-DPA (< 90% and < 90%); 7S,16R,17S-trihydroxy-DHA (10% and 10%). The combinatorial synthetic pathway is depicted in Fig. 6, and all of the lipid mediators produced in this study are presented in Supplementary Table S4.