Pharmacological profile and efficiency in vivo of diflapolin, the first dual inhibitor of 5-lipoxygenase-activating protein and soluble epoxide hydrolase

Arachidonic acid (AA) is metabolized to diverse bioactive lipid mediators. Whereas the 5-lipoxygenase-activating protein (FLAP) facilitates AA conversion by 5-lipoxygenase (5-LOX) to pro-inflammatory leukotrienes (LTs), the soluble epoxide hydrolase (sEH) degrades anti-inflammatory epoxyeicosatrienoic acids (EETs). Accordingly, dual FLAP/sEH inhibition might be advantageous drugs for intervention of inflammation. We present the in vivo pharmacological profile and efficiency of N-[4-(benzothiazol-2-ylmethoxy)-2-methylphenyl]-N′-(3,4-dichlorophenyl)urea (diflapolin) that dually targets FLAP and sEH. Diflapolin inhibited 5-LOX product formation in intact human monocytes and neutrophils with IC50 = 30 and 170 nM, respectively, and suppressed the activity of isolated sEH (IC50 = 20 nM). Characteristic for FLAP inhibitors, diflapolin (I) failed to inhibit isolated 5-LOX, (II) blocked 5-LOX product formation in HEK cells only when 5-LOX/FLAP was co-expressed, (III) lost potency in intact cells when exogenous AA was supplied, and (IV) prevented 5-LOX/FLAP complex assembly in leukocytes. Diflapolin showed target specificity, as other enzymes related to AA metabolism (i.e., COX1/2, 12/15-LOX, LTA4H, LTC4S, mPGES1, and cPLA2) were not inhibited. In the zymosan-induced mouse peritonitis model, diflapolin impaired vascular permeability, inhibited cysteinyl-LTs and LTB4 formation, and suppressed neutrophil infiltration. Diflapolin is a highly active dual FLAP/sEH inhibitor in vitro and in vivo with target specificity to treat inflammation-related diseases.


Results
Diflapolin inhibits cellular 5-LOX product formation without affecting 5-LOX in cell-free assays. Based on a pharmacophore-based virtual screening campaign, diflapolin was identified as most potent agent out of 20 hit compounds that dually inhibited FLAP and sEH in simple screening assays 15 . The structure of diflapolin is composed of a urea moiety (present in the sEH reference inhibitor AUDA) potentially binding to sEH as a mimetic of epoxides, and an aromatic heterocyclic scaffold (benzothiazole, seemingly reflecting the indole scaffold of the reference FLAP inhibitor MK886) that may primarily confer FLAP interference (Fig. 1a). We first aimed to investigate the interference of diflapolin with FLAP and thus with 5-LOX product biosynthesis in more detail. Since FLAP does apparently not possess any measurable enzyme activity that can be readily monitored in a cell-free assay, functional interference of a given compound with FLAP requires indirect analysis of 5-LOX activation and product formation in intact cells 23 . In intact monocytes and neutrophils from human peripheral blood stimulated with Ca 2+ -ionophore, diflapolin effectively inhibited the formation of LTB 4 and its isomers and of 5-H(p)ETE with IC 50 values of 30 and 170 nM, respectively (Fig. 1b). In order to exclude direct inhibition of 5-LOX, diflapolin was analyzed against 5-LOX activity in cell-free assays. Diflapolin, up to 10 µM, did not significantly inhibit the activity of isolated human recombinant 5-LOX (not shown) or of 5-LOX in homogenates of neutrophils and monocytes (Fig. 1b). The same pattern of interference with 5-LOX product formation was observed for the FLAP inhibitor MK886 (IC 50 monocytes: ~3 nM, neutrophils: 10-14 nM; IC 50 5-LOX in cell-free assays: >10 µM, not shown), which is in agreement with the literature 24 . In contrast, the direct 5-LOX inhibitor zileuton inhibited 5-LOX activity in the cell-based (monocytes, neutrophils) and cell-free assays about equally well (IC 50 = 1.5 and 0.8 µM, respectively), as reported 25 . A typical feature of FLAP inhibitors is their loss of efficiency, when cells are stimulated for 5-LOX product formation in the presence of exogenous AA, since (I) FLAP inhibitors compete with AA binding within the active site of FLAP 26 , and (II) ample AA supply may circumvent the requirement of FLAP for cellular 5-LOX product formation 27,28 . In both monocytes and neutrophils, increasing levels of AA (up to 60 µM) sequentially reduced the inhibitory potency of diflapolin and shifted the IC 50 values from 30 and 170 nM, respectively, (no AA supplementation) to >10 µM (at 60 µM AA) (Fig. 1c/d). These data suggest that diflapolin binds in the fatty acid substrate (AA) pocket of FLAP.
To confirm the hypothesis that diflapolin acts as FLAP inhibitor, we studied suppression of 5-LOX product formation in stably transfected HEK293 cells that either express 5-LOX plus FLAP or 5-LOX alone. Note that induction of 5-LOX product formation in HEK293 cells requires supplementation of exogenous AA (regardless of FLAP) 29 and we therefore stimulated the cells with A23187 plus 3 µM AA. Diflapolin strongly inhibited 5-LOX product formation in intact HEK293 cells expressing both 5-LOX and FLAP, whereas in HEK293 cells deficient in FLAP, 5-LOX product formation was hardly impaired by diflapolin (Fig. 1e). Together, these data show that diflapolin inhibits 5-LOX product formation only when FLAP is operative supporting FLAP as target of diflapolin.
Diflapolin inhibits epoxide hydrolase activity of sEH without affecting the phosphatase activity. sEH is a bifunctional enzyme with a C-terminal epoxide hydrolase (EH) and an N-terminal phosphatase activity that operate independent from each other 30 . In a cell-free assay, diflapolin reduced the EH activity of human recombinant sEH with an IC 50 of 20 nM (Fig. 2a), comparable to the activity of AUDA (IC 50 = 69 nM), a well-recognized reference inhibitor of sEH 31 . sEH is constitutively expressed in the human liver cancer cell line HepG2 making it suitable as cell-based test system for evaluation of diflapolin for sEH inhibition in the cellular context. HepG2 cells were pre-treated with diflapolin and control inhibitors, and incubated with the sEH substrate 14,15-EET. sEH activity was then analyzed by monitoring 14,15-DiHETrE formation using UPLC-MS/MS. Diflapolin as well as AUDA inhibited cellular sEH activity to ~50% at 1 µM (Fig. 2b). Further decrease at concentrations up to 10 µM could not be observed, probably due to EET degradation that was sEH-independent and potentially non-enzymatic, as recombinant sEH in the corresponding cell-free assay yielded comparable results (Fig. 2b). AUDA reduced the sEH activity in a comparable manner, whereas MK886 and SC57461A (LTA 4 -H inhibitor) had no impact on sEH in the cell-based sEH assay (Fig. 2c).
Next, we tested diflapolin against the phosphatase activity of sEH in a cell-free assay. Of interest, diflapolin failed to inhibit the phosphatase activity even at high concentrations (10 µM) (Fig. 2d). In order to demonstrate that diflapolin as a specific inhibitor of the hydrolase activity of sEH, its effect on LTA 4 -H was determined. LTA 4 -H hydrolyses the epoxide in LTA 4 that is produced from AA by 5-LOX in a co-incubation experiment using isolated 5-LOX and LTA 4 -H, where LTB 4 is formed. Diflapolin failed to inhibit LTA 4 -H activity (i.e. LTB 4 formation) up to 10 µM, compared to the LTA 4 -H inhibitor SC57461A (IC 50 of 0.1 µM against recombinant LTA 4 -H) 32 that blocked LTB 4 biosynthesis and shifted the conversion of LTA 4 towards the non-enzymatically formed trans-isomers of LTB 4 (Fig. 2e). AUDA (10 µM) showed the same pattern as diflapolin, whereas zileuton (3 µM) as a direct 5-LOX inhibitor reduced the formation of all LTB 4 isomers (Fig. 2e), as expected due to reduced LTA 4 formation.
Effects of diflapolin on other eicosanoid biosynthetic enzymes. We next investigated the impact of diflapolin on other enzymes within the AA cascade that are involved in the biosynthesis of various eicosanoids in addition to FLAP and sEH. Besides FLAP, the LTC 4 S and mPGES-1 belong to the membrane-associated proteins in eicosanoid and glutathione metabolism (MAPEG) family sharing high sequence and structure homology to FLAP 33,34 . Potent FLAP inhibitors (like MK886 or BRP-187) inhibit the MAPEG family members LTC 4 S and mPGES-1 as well 27,35,36 . In contrast, diflapolin failed to inhibit LTC 4 S and mPGES-1 activity in cell-free assays up to 10 µM (Fig. 3a,b), which indicates a high target specificity of diflapolin among the MAPEGs. Also, diflapolin did not significantly inhibit the activities of COX-1 and -2 in cell-free assays (Fig. 3c), whereas the reference drug indomethacin blocked COX activities, as expected. The activities of other LOXs (i.e. 12-LOX and 15-LOX) in neutrophils incubated with A23187 and 20 µM AA were not inhibited by diflapolin. In contrast, 15-HETE formation and sEH (AUDA) inhibitors. (b) Inhibition of 5-LOX product formation in human monocytes and neutrophils and in corresponding cell homogenates. Cells were pre-incubated with diflapolin (or 0.1% DMSO as vehicle) for 15 min and stimulated with 2.5 µM Ca 2+ -ionophore A23187 for 10 min. Cell homogenates were preincubated with diflapolin (or 0.1% DMSO) for 10 min at 4 °C, pre-warmed at 37 °C for 30 sec, and 20 µM AA plus 1 mM CaCl 2 was added for another 10 min at 37 °C. 5-LOX products were analyzed by HPLC. (c,d) Effects of exogenous AA on the potency of diflapolin for inhibition of 5-LOX product formation in neutrophils (c) and monocytes (d). Cells were pre-treated by diflapolin (or 0.1% DMSO as vehicle) at 37 °C for 15 min, and subsequently activated by 2.5 µM Ca 2+ -ionophore A23187 plus the indicated concentrations of exogenous AA for another 10 min. (e) HEK293 cells expressing 5-LOX or 5-LOX and FLAP were pre-incubated with diflapolin and stimulated with 5 µM Ca 2+ -ionophore A23187 plus 3 µM AA for 10 min at 37 °C. 5-LOX products were analyzed by HPLC. Data, expressed as percentage of vehicle control (=100%), are given as means ± S.E.M, n = 3 *p < 0.05; **p < 0.005; ***p < 0.001 vs. vehicle control (ANOVA + Bonferroni with logarithmized values).
was concentration-dependently elevated (up to 200% of the vehicle control) (Fig. 3d). Since diflapolin was most potent in leukocytes to suppress 5-LOX product formation from endogenous AA but ineffective when high concentrations of exogenous AA were supplied, the compound could act at the level of AA release. However, in [ 3 H] AA-pre-labelled neutrophils, diflapolin even at high concentrations (1 µM) did not inhibit the release of AA upon A23187-stimulation, as compared to the cPLA 2 inhibitor RSC-3388 that suppressed AA liberation (Fig. 3e). Finally, detrimental effects on cellular viability could be excluded, as diflapolin (10 µM) did not affect the viability of monocytes in a MTT assay at 24 or 48 hrs, while staurosporine (3 µM, positive control) strongly impaired cell viability under these conditions (Fig. 3f). Taken together, diflapolin dually and strongly inhibits FLAP and sEH with target specificity as it did not interfere with other AA pathway-related enzymes (LTA 4 -H, COX-1/2, cPLA 2 , 12/15-LOXs) and FLAP-related MAPEG enzymes such as mPGES 1

and LTC 4 S.
Effects of diflapolin on 5-LOX subcellular redistribution and 5-LOX/FLAP complex assembly. 5-LOX, a soluble cytosolic or intranuclear enzyme in resting leukocytes, translocates to the nuclear envelope upon cell activation and co-localizes with FLAP at the nuclear membrane to form a tight LT biosynthetic complex 37,38 . Immunofluorescence microscopy studies using human neutrophils or monocytes showed that neither diflapolin nor the FLAP inhibitor MK886 or the 5-LOX inhibitor zileuton prevented co-localization of 5-LOX with FLAP (Fig. 4). However, diflapolin efficiently prevented the tight 5-LOX/FLAP complex assembly, visualized by proximity-ligation assay (PLA) (Fig. 4), a common feature of FLAP inhibitors 37 . MK886 gave comparable effects, whereas the 5-LOX inhibitor zileuton failed in this respect.

Diflapolin exhibits potent anti-inflammatory properties in in-vivo experiments.
We next investigated the anti-inflammatory effectiveness of diflapolin in the zymosan-induced peritonitis mouse model 39 that is strongly related to the pathophysiological activities of LTs. Diflapolin pre-treatment (1, 3 and 10 mg/kg, i.p. 30 min before zymosan injection) induced a significant reduction of LTC 4 and LTB 4 peritoneal levels, starting from the dose of 1 mg/kg ( Fig. 5a and b) and comparable to the effect of MK886 (1 mg/kg, i.p. 30 min before zymosan). Since LTB 4 is a major chemoattractant for leukocytes, diflapolin and MK886 caused concomitant block of leukocyte recruitment, which was dose-dependent for diflapolin (Fig. 5c). Accordingly, at the dose of 10 mg/kg, diflapolin inhibited the activity of MPO, a typical marker protein for neutrophils to 52.8 ± 12.2% (mean ± SEM) vs. vehicle control and reduced vascular permeability to 55.7 ± 14.4% (mean ± SEM) vs. vehicle control (Table 1), compared to inhibitory effects of MK886 (1 and 3 mg/kg) to 58.5 ± 10.6% and 48.6 ± 3.2%, respectively.

Discussion
Our recent pharmacophore-based virtual screening campaign for dual FLAP/sEH inhibitors proposed diflapolin as most promising hit and novel chemotype targeting both FLAP and sEH 15 . Here, we disclose diflapolin as a potent, dual inhibitor of FLAP and sEH with marked anti-inflammatory efficacy in vivo and high target selectivity. Side-by-side studies of diflapolin with the "FLAP benchmark inhibitor" MK886 40 revealed comparable potencies for inhibition of 5-LOX product biosynthesis in human leukocytes in vitro, and about equal effectiveness in suppression of LT formation and inflammatory properties in vivo using murine zymosan-induced peritonitis models.
The identification of novel chemotypes as FLAP inhibitors is hampered due to the lack of distinct assays that unequivocally proof direct and functional interference of a given compound with FLAP 23 . Thus far, no enzymatic activity has been assigned to FLAP that can be exploited as read-out in FLAP inhibitor discovery approaches. Moreover, FLAP does not support 5-LOX activity in cell-free assays (e.g. homogenates) 41 . Nevertheless, FLAP is essential for LT biosynthesis in intact cells and in vivo, as reflected by results from various pharmacological approaches 40 and from gene intervention using FLAP knock-out mice 42 . Experimental evidence suggests that FLAP operates as a 5-LOX helper protein for transforming AA to 5-HPETE and for dehydration of 5-HPETE to LTA 4 29, 43 . FLAP is able to bind AA 26 and to stimulate conversion of AA and 5-HPETE by 5-LOX 44 , and along these lines AA or 5-HPETE were required for in situ 5-LOX/FLAP complex assembly at the nuclear membrane in activated cells 37,38 . Together, it appears that FLAP binds released AA and/or de novo-formed 5-HPETE and transfers them to 5-LOX, thus permitting optimal access of 5-LOX towards its substrates.
In light of these facts, assignment of a small molecule as FLAP inhibitor requires certain characteristics, which are all fulfilled by diflapolin. First of all, diflapolin did not randomly emerge as 5-LOX product biosynthesis inhibitor but was identified in a target-directed screening campaign for dual FLAP/sEH inhibitors applying two independently created ligand-based pharmacophore models 15 . Second, diflapolin potently inhibited LT biosynthesis only in intact leukocytes (without being cytotoxic or suppressing AA substrate release) but did not directly affect the activity of isolated human recombinant 5-LOX or 5-LOX in leukocyte homogenates. Third, ample supply of exogenous AA strongly reduced the potency of diflapolin in stimulated neutrophils and monocytes, compatible with the proposed competition between AA and FLAP inhibitors for binding to FLAP 43 . Fourth, diflapolin prevented the agonist-induced 5-LOX/FLAP complex assembly at the nuclear membrane in monocytes, visualized by PLA 27, 37 , without blocking 5-LOX translocation to the nucleus. Finally, the most striking proof for diflapolin inhibiting 5-LOX product formation by acting on FLAP but not on 5-LOX, is deduced from the fact that diflapolin suppresses cellular 5-LOX activity in 5-LOX-transfected HEK293 cells only when FLAP was co-expressed, while in cells devoid of FLAP, diflapolin failed in this respect. Note that these features of diflapolin were shared also with the FLAP inhibitor MK886 but not with the direct 5-LOX inhibitor zileuton (refs 27, 29, 37, 41 and 45 and this study). In conclusion, diflapolin is a potent LT biosynthesis inhibitor that confers its activity via inhibition of FLAP.
Besides FLAP, diflapolin was identified as potential hit using pharmacophore models for sEH inhibitors 15 , and it potently inhibited the epoxide hydrolase activity of sEH in a cell-free assay (IC 50 20 nM), while the phosphatase activity was not affected. In contrast, the epoxide hydrolase activity of LTA 4 -H was not affected by diflapolin, indicating specificity for sEH. Interference of diflapolin with sEH is not surprising since the urea moiety is a typical structural feature of sEH inhibitors that prevent the degradation of EETs 12,46 . EETs formed from AA by CYP enzymes display anti-inflammatory and antihypertensive properties, maintain vascular homeostasis, and act generally cardio-protective 20,47 . Among several EET degrading pathways, sEH metabolizes EETs to the corresponding dihydroxyeicosatrienoic acids (DiHETrE) with accompanied loss of health-promoting benefits 20,48 . Accordingly, inhibition of sEH elevates EET levels leading to various beneficial effects. Little is known about the physiological role of the phosphatase activity, but an influence on the regulation of the endothelial nitric oxide synthase (eNOS) and NO-mediated effect on endothelial cells was suggested 49 .
Because FLAP is a member of the MAPEG family, other structurally-related MAPEG members such as mPGES-1 and LTC 4 S might be targeted by diflapolin as well, which is the case for the FLAP inhibitors MK866 36,50 and BRP-187 27 that interfere with all three of these proteins. Diflapolin had no impact on mPGES-1 and LTC 4 S activity up to 10 µM, suggesting target specificity within the MAPEG family. Moreover, other enzymes within the  AA cascade including COX-1/2, epoxide hydrolase activity of LTA 4 -H, 12-LOX and 15-LOXs were not inhibited by diflapolin. Interestingly, formation of 15-HETE in neutrophil incubations was concentration-dependently increased by diflapolin, which might be explained by AA shunting towards the 15-LOX pathway. This may promote the formation of anti-inflammatory and pro-resolving LMs such as lipoxins, resolvins and protectins 51 .
A shift from biosynthesis of pro-inflammatory eicosanoids and other oxylipins towards anti-inflammatory and pro-resolving LMs would certainly strengthen the power of dual FLAP/sEH inhibitors. Additionally, even though comprehensive data are yet not available, it is reasonable to assume that epoxy-fatty acids derived from other polyunsaturated fatty acids confer anti-inflammatory properties. The current development of LT biosynthesis inhibitors as therapeutics focusses on FLAP inhibitors 23 . Early representatives such as the indole MK886 18 and the quinoline BAY X-1005 19 are highly effective in vitro but probably due to their high lipophilicity they suffered from strong plasma protein binding, competition with fatty acids and, as a consequence reduced activity in vivo 17,40 . However, more advanced compounds including the MK886 follow-up GSK2190915, the tetrahydrofuran derivative AZD6642 and the oxadiazole-based BI665915 are less prone to plasma protein binding with advantageous pharmacokinetics 23 . These compounds are under active development (some entered clinical trials), and they appear to have lower risks of side effects as compared to 5-LOX inhibitors 40 . Diflapolin is structurally unrelated to these above-mentioned chemotypes and represents the first FLAP inhibitor with a polar urea moiety (seemingly the pharmacophore for sEH interference) and high efficiency in vivo.
In order to evaluate the in vivo efficacy and anti-inflammatory potential of diflapolin, we utilized the zymosan-induced peritonitis mouse model that is well established as test system for studying LT biosynthesis in vivo 39 . Our data show that diflapolin is about equally effective as MK886 in reducing LTB 4 and LTC 4 levels in the peritoneal exudates with consequent biological functions. LTB 4 is a potent chemotactic agent for neutrophils 52 and in fact, neutrophil infiltration into the peritoneal cavity was strongly reduced by diflapolin. Cys-LTs mediate plasma extravasation 53 and diflapolin significantly impaired vascular permeability during peritonitis. Conclusively, diflapolin potently inhibits LT formation in vivo connected with anti-inflammatory activity.
We speculate that dual inhibition of FLAP and sEH inhibitor may have synergistic anti-inflammatory and cardio-protective actions. Indeed, increased levels of EETs seem to be cardio-protective 47 and FLAP was reported to be linked to certain cardiovascular diseases 54 . DMLs that block sEH and COX or sEH and 5-LOX are proposed to have improved anti-inflammatory activities over compounds that interfere with only one target enzyme 12,13,21,55 . Such DMLs dually targeting sEH and FLAP are thus far unknown. Future studies addressing the pharmacological relevance of suppression of LTs with accompanied elevation of EETs may reveal potential benefit in the therapy of inflammatory and cardiovascular diseases.
Taken together, here we provide substantial evidence that diflapolin acts as potent dual FLAP/sEH inhibitor with high target specificity. The compound lacks acute cytotoxicity and efficiently suppresses LT biosynthesis in vivo connected with potent anti-inflammatory activity in a mouse model. Based on these features, diflapolin might be a valuable chemical tool for studying the biology of FLAP and sEH, particularly as synergizing targets, and may represent a useful lead for evaluation of the therapeutic potential of dual FLAP and sEH inhibition in inflammatory and cardiovascular disorders.

Determination of 5-LOX products in intact cells and corresponding homogenates.
In order to examine 5-LOX product formation in intact human neutrophils and monocytes, freshly isolated cells were resuspended in PBS buffer containing 0.1% glucose and 1 mM CaCl 2 (PGC buffer) to a final cell density of 5 × 10 6 or 2 × 10 6 , respectively. Cells were pre-incubated with the test compounds or vehicle (0.1% DMSO) at 37 °C for 15 min prior to stimulation with 2.5 µM Ca 2+ -ionophore A23187 for 10 min (37 °C) with or without supplementation of the indicated concentrations of AA. 5-LOX product formation was stopped by addition of one volume of ice-cold methanol, samples were subjected to solid phase extraction after addition of 200 ng PGB 1 as internal standard and 5-LOX products were analyzed by RP-HPLC as described above.
Determination of 5-LOX products in corresponding homogenates was performed by resuspending neutrophils (final density of 5 × 10 6 cells/mL) or monocytes (2 × 10 6 cells/mL) in PBS containing 1 mM EDTA and sonicated on ice (3 × 15 s). Aliquots of homogenates were pre-incubated with the test compounds or vehicle (0.1% DMSO) on ice for 15 min and stimulated with 20 µM AA and CaCl 2 (2 mM) at 37 °C for 10 min. 5-LOX product formation was assayed as described for intact cells above.
For analysis of 5-LOX product formation in HEK293 cells stably expressing 5-LOX with or without FLAP, cells were harvested by trypsinization, pelleted (1,200 rpm, 5 min, 4 °C) and resuspended in PGC buffer to a final concentration of 1 × 10 6 cells/mL. Aliquots were pre-incubated with test compounds or 0.1% DMSO for 15 min, respectively, and stimulated with 2.5 µM A23187 and 3 µM AA. After termination of the incubations by addition of one volume of ice-cold methanol, samples were subjected to solid phase extraction and analysis of 5-LOX products as described above.

Determination of LTA 4 hydrolase activity.
In order to determine the activity of LTA 4 -H, aliquots of human recombinant 5-LOX (0.5 µg) and 10 µg of human recombinant LTA 4 -H (kindly provided by Dr. E. Proschak, Goethe University, Frankfurt, Germany) were suspended in 1 mL PBS containing 1 mM EDTA and pre-incubated with test compounds or vehicle (0.1% DMSO) for 10 min on ice. The specific LTA 4 -H inhibitor SC57461A (0.3 µM) was used as reference drug. Subsequently, incubations were stimulated with 20 µM AA and 2 mM CaCl 2 for additional 10 min at 37 °C. The reaction was stopped by 1 volume ice-cold methanol, 530 µL acidified PBS and 200 ng of PGB 1 as internal standard were added and subjected to solid phase extraction. All LTB 4 isomers were analyzed by HPLC as described above.
In order to analyze the phosphatase activity of sEH, a recently published assay 61 was performed. In brief, purified sEH-phosphatase-domain was pre-incubated with test compounds or vehicle (0.1% DMSO), in acetate buffer (50 mM, pH 5.8) containing MgCl 2 (10 mM) and Triton-X-100 (0.01%) for 30 min at RT prior to addition of 6,8-difluoro-4-methylumbelliferyl phosphate (DiFMUP, 300 µM). Phosphatase activity was assayed by measurement of fluorescence (λ ex 360 nm, λ em 450 nm) of the dephosphorylated DiFMU for 45 min at 37 °C. . Ion spray voltage was set to 4000 V, the heater temperature to 500 °C, the declustering potential to 50-80 eV, the entrance potential to 10 eV, the collision cell exit potential to 10-13 eV, collision energies of 10 eV, the spray gas pressure to 40 psi, medium collision gas and the curtain gas pressure to 35 psi.

Determination of 14,15-DiHETrE-formation in HepG2 cells and in a cell
In order to confirm the identity of detected metabolites, human recombinant sEH was diluted in 25 mM Tris buffer pH 7.0 to a final concentration of 0.3 µg/mL and pre-incubated with test compounds or 0.1% DMSO on ice for 15 min and stimulated with 14,15-EET (1.5 µM, 30 min, 37 °C). Extraction and detection of metabolites was performed as described for the cell-based assay.

Determination of LTC 4 synthase (LTC 4 S) activity. LTC 4 S activity was assayed by using microsomes
of HEK293 cells stably expressing LTC 4 S, as previously published 63 . In brief, HEK293 expressing LTC 4 S were cultivated as described above and selected using geneticin (400 µg/mL). Isolation of microsomes was performed as described for mPGES-1 above and microsomes were diluted in potassium phosphate buffer (0.1 M, pH 7.4) with glutathione (5 mM) to a final concentration of 2.5 µg protein per mL. After pre-incubation with the test compounds or vehicle (2% DMSO) for 10 min at 4 °C, reactions were started by addition of 1 µM LTA 4 -methyl ester and stopped by addition of one volume of ice-cold methanol after 10 min of incubation at 4 °C. To determine enzyme activity, acidified PBS and d 5 -LTC 4 -methyl ester (5 ng) as internal standard were added prior solid phase extraction and LTC 4 -methyl ester formation was analyzed by UPLC-MS/MS as described 63 .
Determination of COX activity. COX activity was assayed by using purified ovine COX-1 and recombinant human COX-2, respectively. Enzymes were diluted in Tris buffer (100 mM, pH 8) supplemented with glutathione (5 mM), EDTA (100 µM) and hemoglobin (5 µM) to a final concentration of 50 U/mL (COX-1) or 20 U/mL (COX-2) and pre-incubated with test compounds or vehicle (0.1% DMSO) for 5 min at RT. After 30 sec at 37 °C, reactions were started with 5 µM AA (COX-1) or 2 µM AA (COX-2) and stopped after 5 min at 37 °C by addition of one volume of ice-cold methanol. Solid phase extraction was performed as described above after addition of 200 ng of internal PGB 1 standard and COX product formation was determined by analysis of 12-HHT formation as reported before 64 .

Determination of [ 3 H]-labeled arachidonic acid release.
Freshly isolated human neutrophils were resuspended in RPMI 1640 medium to a final cell density of 10 7 cells/mL and incubated with 0.5 µCi/mL of [ 3 H]-labeled arachidonic acid ([ 3 H]-AA), corresponding to a concentration of 5 nM of the fatty acid, for 2 hrs (37 °C, 5% CO 2 ). Cells were washed twice to remove unincorporated [ 3 H]-AA and resuspended in PBS containing glucose (0.1%), fatty acid-free BSA (2 mg/mL) and CaCl 2 (1 mM). Aliquots of 10 7 cells were pre-incubated with the test compounds or vehicle (0.1% DMSO) at 37 °C for 10 min and stimulated with 2.5 µM A23187 for another 10 min. The reaction was stopped on ice and cells were centrifuged at 1,200 rpm (10 min, 4 °C). The collected supernatants were combined with 2 mL of liquid scintillation counting solution (Rotiszint eco plus, Carl Roth, Karlsruhe, Germany) and assayed for radioactivity by scintillation counting (Micro Beta Trilux, Perkin Elmer, Waltham, MA).
In situ protein interaction of 5-LOX and FLAP was analyzed by proximity ligation assay as described before 37 and referring to the manufacturer's protocol 65 . In brief, freshly isolated monocytes were treated as described for IF above. Overnight incubations with primary antibodies were then treated for 1 h (37 °C) with oligonucleotide-labeled specific secondary antibodies (PLA probes anti-mouse MINUS and anti-rabbit PLUS). Formation of circled DNA sequences was induced by addition of ligase and oligonucleotide mixture (30 min at 37 °C). Rolling-circle-amplification of newly generated DNA template was performed (90 min, 37 °C) including hybridization of fluorescently-labeled oligonucleotides within the formed DNA strands, resulting in visualization of protein-protein interactions recognized as magenta-stained dots. Nuclear DNA staining with DAPI and image acquisition was performed as described above. Overview images were obtained using a Plan Neofluar 40/1.30 Oil (DIC III) objective (Carl Zeiss).
Murine peritonitis model. The animal studies are reported in accordance with the ARRIVE guidelines for reporting animal research 66 . Male CD-1 mice (33-39 g, 8-9 weeks, Charles River Laboratories, Calco, Italy) were housed in a controlled environment (21 ± 2 °C) and provided with standard rodent chow and water ad libitum. Mice received a standard diet containing 5.7% fat, 18.9% protein and 57.3% carbohydrate (Global Diet 2018, ENVIGO, Italy). The fatty acid composition was according to Matias et al. 67 .
Prior to experiments, all mice were allowed to acclimate for 5 days and kept at 12 h light-dark schedule, in which experiments were performed during the light phase. Animal care was in compliance with Italian regulations on protection of animals used for experimental and other scientific purpose (Ministerial Decree 116/92) and with the European Economic Community regulations (Official Journal of E.C. L 358/1 12/18/1986). Animal studies were approved by the local ethical committee of the University of Naples Federico II on 27 February 2014 (approval number 2014/18760). Mice were treated with diflapolin (1, 3 or 10 mg/kg), MK886 (1 or 3 mg/ kg), zileuton (10 mg/kg) or vehicle (0.9% saline solution containing 2% DMSO), received as intraperitoneal (i.p.) injection, 30 min prior induction of peritonitis according to well-recognized experimental design for studying LT synthesis inhibitors in acute inflammation 27 . Zymosan (Sigma, Milan, Italy) was prepared and injected i.p. as a final suspension (2 mg/mL) in 0.9% saline solution after boiling, centrifugation and sonication. Peritoneal lavage (3 mL of cold PBS) was performed after CO 2 -euthanasia at indicated time points, followed by 60 sec of gentle manual massage. Two mL of exudates were collected and infiltrated cells were determined using a Burker chamber and vital trypan blue staining. Pelleted samples (18,000 × g, 5 min, 4 °C) were frozen (−80 °C) and assayed for myeloperoxidase (MPO) activity (pellet) or LTC 4 and LTB 4 formation (supernatant), respectively.
MPO of neutrophils was examined as follows: pellets from exudates were resuspended in PBS (50 mM, pH 6) containing 0.5% hexadecyltrimethyl-ammonium bromide and sonicated, followed by 3 freeze-thawing cycles and a final sonication. Supernatants of centrifuged samples (18,000 × g, 30 min) were added to a 96-well plate and reactions were initiated by addition of PBS (50 mM, pH 6) containing o-dianisidine (0.167 mg/mL) and hydrogen peroxide (0.0005%). Absorbance was monitored in the kinetic mode (Biorad Imark microplate) and levels of MPO were determined using a calibration curve with human neutrophils as reference standard. MPO levels were expressed as units MPO per mouse. LTC 4 and LTB 4 formation within the supernatants were determined by EIA (Enzo Life Sciences International Inc., Lörrach, Germany) according to manufacturer's protocol 39 .
Vascular permeability was assessed according to a previous report 27 . Briefly, 0.3 mL of 0.9% saline solution supplemented with Evans blue dye (40 mg/kg) was injected intravenously (i.v.) into the caudal vein followed by immediate peritonitis induction (using zymosan). After 30 min, peritoneal lavage exudates of CO 2 -euthanized were collected as described above. Absorbance of the centrifuged supernatants (3,000 × g, 5 min) was measured at 650 nm (Beckman Coulter DU730).

Statistics.
Results are presented as mean ± standard error of the mean out of n independent experiments, where n represents the number of performed experiments on different days or with different donors or the number of animals for in vivo studies. IC 50 values were calculated from at least 5 different concentrations using a nonlinear regression interpolation of semi-logarithmic graphs in GraphPad Prism (GraphPad Software Inc., San Diego, CA). Statistical evaluation was performed by one-way ANOVA using GraphPad InStat (Graphpad Software Inc., San Diego, CA) followed by a Bonferroni post-hoc test for multiple or student t-test for single comparisons, respectively. P-values < 0.05 were considered as significant.