COX-2-dependent and -independent biosynthesis of dihydroxy-arachidonic acids in activated human leukocytes.

Biosynthesis of 5,15-dihydroxyeicosatetraenoic acid (5,15-diHETE) in leukocytes involves consecutive oxygenation of arachidonic acid by 5-lipoxygenase (LOX) and 15-LOX in either order. Here, we analyzed the contribution of cyclooxygenase (COX)-2 to the biosynthesis of 5,15-diHETE and 5,11-diHETE in isolated human leukocytes activated with lipopolysaccharide and calcium ionophore A23187. Transformation of arachidonic acid was initiated by 5-LOX providing 5S-HETE as a substrate for COX-2 forming 5S,15S-diHETE, 5S,15R-diHETE, and 5S,11R-diHETE as shown by LC/MS and chiral phase HPLC analyses. The levels of 5,15-diHETE were 0.45 ± 0.2 ng/10⁶ cells (mean ± SEM, n = 6), reaching about half the level of LTB₄ (1.3 ± 0.5 ng/10⁶ cells, n = 6). The COX-2 specific inhibitor NS-398 reduced the levels of 5,15-diHETE to below 0.02 ng/10⁶ cells in four of six samples. Similar reduction was achieved by MK-886, an inhibitor of 5-LOX activating protein but the above differences were not statistically significant. Aspirin treatment of the activated cells allowed formation of 5,15-diHETE (0.1 ± 0.05 ng/10⁶ cells, n = 6) but, as expected, abolished formation of 5,11-diHETE. The mixture of activated cells also produced 5S,12S-diHETE with the unusual 6E,8Z,10E double bond configuration, implicating biosynthesis by 5-LOX and 12-LOX activity rather than by hydrolysis of the leukotriene A₄-epoxide. Exogenous octadeuterated 5S-HETE and 15S-HETE were converted to 5,15-diHETE, implicating that multiple oxygenation pathways of arachidonic acid occur in activated leukocytes. The contribution of COX-2 to the biosynthesis of dihydroxylated derivatives of arachidonic acid provides evidence for functional coupling with 5-LOX in activated human leukocytes.


Leukocyte preparation and extraction
Leukocytes were isolated from peripheral blood from healthy human subjects . The study was approved by the Vanderbilt University Medical Center Institutional Review Board (protocol #091243), and informed consent was obtained from the donors. Venous blood (45 ml) was collected into a syringe containing 5 ml of citric acid and 10 ml of 6% dextran (250 kDa average molecular weight). The blood was allowed to settle for 1 h and the top layer was collected and centrifuged. The pellet was washed with PBS, and remaining red cells were lysed by treatment with a 10fold excess of deionized water for 30 s. The leukocytes were centrifuged, washed, and diluted in PBS + containing 5 mM glucose to a concentration of 1.5 × 10 7 cells/ml. 1.0 × 10 7 cells were diluted to 1 ml, LPS (10 g/ml) was added, and the cells were incubated at 37°C for 6 h. After addition of calcium ionophore A23187 (5 M), the cells were incubated for an additional 15 min before the addition of 10 ng of d 4 -LTB 4 standard, acidification to pH 4 using acetic acid, and extraction (Waters HLB cartridge). Products were eluted from the HLB cartridge with methanol, evaporated under a stream of nitrogen, and dissolved in 10 l acetonitrile and 40 l of LC-MS column solvent A.

Resolution of 5,15-diHETE diastereomers
Resolution of the 5 S ,15 S -and 5 S ,15 R -diHETE diastereomers was achieved after derivatization to the pentafl uorobenzyl (PFB) esters using a Chiralpak AD-RH column (4.6 × 150 mm; 5 m) eluted with a solvent of acetonitrile/ethanol (90/10, by vol.) at a fl ow rate of 1 ml/min. The ‫ف‬ 77:23 mixture of 5 S ,15 Sand 5 S ,15 R -diHETE formed by reaction of recombinant COX-2 with 5 S -HETE ( 16 ) was used to determine the elution order of the diastereomers. For analysis of 5,15-diHETE formed by leukocytes the cells stimulated with A23187 were extracted using an HLB cartridge, evaporated, and dissolved in 40 l of pentafl uorobenzyl bromide (10% in acetonitrile) and 20 l of diisopropylethylamine. After 1 h incubation at room temperature, the samples were evaporated and reconstituted in 50 l of acetonitrile for LC/MS analysis. The samples were analyzed using LC/MS with atmospheric pressure chemical ionization (APCI) ( 23 ). The same ion transition as in ESI analysis ( m/z 335 to 201) was used. Instrument parameters were optimized by direct liquid infusion of a standard of 15-HETE-PFB ester dissolved in column solvent. lipoxin biosynthesis, resulting in the formation of aspirintriggered lipoxins ( 15 ).
Our interest in the biosynthesis of 5,15-diHETE stemmed from its formation as a byproduct of the oxygenation of 5 S -HETE by COX-2 in vitro ( 16 ). This fi nding implicated crossover of the 5-LOX and COX-2 pathways as an alternative biosynthetic route of 5,15-diHETE in vivo with 5-LOX forming 5 S -HETE as the fi rst step, followed by COX-2 catalysis as the second step. The main product of the reaction is a diendoperoxide that is structurally similar to the arachidonic acid-derived endoperoxide of prostaglandin biosynthesis ( 17 ). The catalytic byproducts include 5,15-diHETE (as a ‫ف‬ 4:1 mixture of the 5 S ,15 S -and 5 S ,15 R -diastereomers) and 5 S ,11 R -diHETE ( 16 ). The diHETEs are the 5 S -hydroxy analogs of the 15-and 11-HETE byproducts of the COX-1 and COX-2 reactions with arachidonic acid ( 18,19 ).
Thus, 5,15-diHETE may be biosynthesized through a multitude of pathways that involve reactions of 5-LOX, 15-LOX, COX-1, and COX-2, or acetylated COX-2 as illustrated in Fig. 1 . Here, we analyzed the biosynthesis of 5,15-diHETE and 5,11-diHETE in activated human leukocytes. The enzymatic reactions involved were elucidated by the use of inhibitors of 5-LOX and COX-2 activities, deuterated HETE substrates, and stereochemical analysis of the diHETE products.

LC/MS and NMR analysis
Samples were analyzed using a ThermoFinnigan Quantum Access triple quadrupole mass spectrometer equipped with an electrospray interface and operated in the negative ion mode. Instrument specifi c parameters (sheath and auxiliary gas pressures, temperature, and interface voltage) were optimized using direct infusion of a solution of PGE 2 . A Waters Symmetry Shield C18 column (2.1 × 150 mm; 3 m) was eluted with a linear gradient of acetonitrile/water, 10 mM NH 4 NMR spectra were recorded on a Bruker AV-II 600 MHz spectrometer equipped with a cryoprobe. CDCl 3 was used as solvent ( ␦ 7.25 ppm).

Formation of diHETEs in activated leukocytes
Reversed-phase-HPLC analysis of human leukocytes activated with LPS for 6 h followed by A23187 for 15 min showed formation of fi ve products ( I -V ) with retention times and UV spectra that were characteristic of diHETE derivatives of arachidonic acid ( Fig. 2A ). Products I and IV were identifi ed as 5,15-diHETE and 5,11-diHETE, respectively, by comparison of HPLC retention times and UV spectra with authentic standards ( Fig. 2B ) prepared by reaction of recombinant COX-2 with 5 S -HETE ( 16 ). Product III was identifi ed as LTB 4 by coelution with an authentic standard and UV analysis ( Fig. 2C, D ). Product II was tentatively identifi ed as 6 E ,8 E ,10 E ,14 Z -5 S ,12 R , S -diHETE (i.e., a mixture of 6-trans -LTB 4 and 6-trans -12-epi -LTB 4 ) based on retention time relative to LTB 4 and UV spectrum ( Fig. 2A, D ) ( 24 ). The diastereomers of II are the major products formed by nonenzymatic hydrolysis of the LTA 4 epoxide.
Product V was identifi ed as 6 E ,8 Z ,10 E ,14 Z -5 S ,12 S -diHETE by comparison of retention time and UV spectrum with an authentic standard ( Fig. 2C, D ). Peak V coeluted with the 5 S ,12 S -diastereomer while the 5 R ,12 Sdiastereomer was resolved by about 1 min ( Fig. 2C ). The confi guration of the double bonds of the authentic standard was confi rmed by NMR analysis ( J 6,7 = 15.2 Hz ( trans ), J 8,9 = 10.6 Hz ( cis ), J 10,11 = 15.3 Hz ( trans ), J 14,15 = 9.4 Hz ( cis )). The double bond confi guration of the 6 E ,8 Z ,10 E triene implied that V was formed by consecutive oxygenation of arachidonic acid by 5-LOX and 12-LOX in either order, rather than by hydrolysis of the LTA 4 epoxide. The 12-LOX activity was likely due to platelet 12-LOX. It has been shown that in vitro mixtures of platelets and neutrophils form 6 E ,8 Z ,10 E , 14 Z -5 S ,12 S -diHETE in response to stimulation with A23187 ( 25 ). by eosinophils in the leukocyte preparations. Unexpectedly, NS-398 also reduced the levels of 5-HETE and LTB 4 in some of the samples (not shown). This could have been due to the ability of NS-398 to inhibit leukocyte lipid body formation independent of COX-2 inhibition leading to the suppression of LOX-derived eicosanoids ( 27 ). Aspirin did not completely inhibit formation of 5,15-diHETE (0.10 ± 0.05 ng/10 6 cells, n = 6), compatible with the expected formation 5 S ,15 R -diHETE from 5 S -HETE ( 16 ). None of the differences between the means of the treatment groups were statistically signifi cant using a two-tailed unpaired t -test.
Elution with acetonitrile/ethanol (90/10, by vol.) achieved baseline resolution of the 5 S ,15 R -diHETE-PFB and 5 S ,15 S -diHETE-PFB diastereomers. Using recombinant COX-2, the peak areas showed the expected ratio of 20:80 for 5 S ,15 R to 5 S ,15 S ( Fig. 5A ). Analysis of 5,15-di-HETE formed by LPS/A23187-activated leukocytes showed a ratio of ‫ف‬ 15:85 of the 5 S ,15 R -to the 5 S ,15 S -diastereomer ( Fig. 5B ). Thus, compared with the ratio formed by recombinant COX-2 the proportion of the 5 S ,15 R diastereomer in the leukocytes was largely retained implying a major contribution of COX-2 to the synthesis of 5,15-di-HETE. The slight enrichment ( ‫ف‬ 10-20%) of the 5 S ,15 Sdiastereomer was likely due to synthesis by a 15-LOX enzyme. In a second sample, <5% of 5,15-diHETE was the 5 S ,15 R -diastereomer, implicating a lesser contribution of COX-2 and higher contribution of 15-LOX. Aspirin treatment of activated leukocytes led to a smaller than expected increase in the proportion of 5 S ,15 R -diHETE, possibly due to incomplete acetylation of COX-2. A representative sample ( Fig. 5C ) showed that about 25% of total 5,15-diHETE was the 5 S ,15 R -diastereomer.

Transformation of exogenous 5-HETE and 15-HETE to 5,15-diHETE
We analyzed the ability of activated leukocytes to utilize either 5-HETE or 15-HETE in the biosynthesis of 5,15-diHETE by measuring formation of octadeuterated 5, 15-HETE from exogenously added d 8 -5-HETE and d 8 -15 S -HETE, respectively. d 8 -5-HETE was used as a racemic mixture. The enantiomers did not resolve on SP or RP Chiralpak AD chiral phase HPLC columns although both give exceptional resolution of unlabeled HETEs ( 21,28 ).
Three different concentrations of d 8 -5-HETE (50, 100, and 500 ng/10 6 cells, equivalent to 0.75, 1.5, and 7.5 M) were added to LPS-activated leukocytes together with  cells is used in the biosynthesis of leukotrienes, lipoxins, and other lipid autacoids. Thus, analysis of mixtures of leukocytes can lead to identifi cation of eicosanoids that are absent or less prominent in monotypic cell populations. A major goal of this study was to provide evidence for a functional biosynthetic coupling of 5-LOX expressed in granulocytes and COX-2 expressed in monocytes.
The main product of the transformation of 5-HETE by COX-2 in vitro has been identifi ed as an unstable diendoperoxide ( 17 ); 5,11-and 5,15-diHETE are reaction by-products ( 16 ). The chemical instability makes direct identifi cation of the di-endoperoxide in vivo a diffi cult task. We hypothesized that, instead, formation of 5,15-di-HETE and 5,11-diHETE could provide evidence for the oxygenation of 5 S -HETE by COX-2 in vivo. While these studies were in progress, we identifi ed two hemiketal eicosanoids as the major transformation products of the unstable di-endoperoxide. The hemiketals are present in activated human leukocytes and induce tubulogenesis of endothelial cells ( 32 ).
Biosynthesis of 5,15-diHETE through crossover of the 5-LOX and 15-LOX pathways has been documented in leukocytes from normal volunteers ( 12 ) and from patients with eosinophilia ( 33 ) and asthma ( 34 ). Consecutive oxygenation of arachidonic acid by the 5-LOX and 15-LOX enzymes is illustrated as routes 3 and 6 in Fig. 1 . Alternative pathways could entail the 15-oxygenase activities of COX-1 and COX-2 (routes 1 and 4 in Fig. 1 ), and the 15 Roxygenase activity of aspirin-acetylated COX-2 (routes 2 and 5). Since testing of the alternative pathways required 5-LOX and COX-2 activities we used a crude fraction of human leukocytes containing neutrophils and eosinophils in order to provide 5-LOX activity, and monocytes/macrophages for COX-2 activity. COX-2 activity was induced by LPS ( 35 ), followed by stimulation of 5-LOX activity with calcium ionophore ( 36 ). Involvement of COX-2 in the biosynthesis of 5,15-diHETE in activated human leukocytes was evident from three independent fi ndings: 1 ) significant formation of the 5 S ,15 R -diastereomer (about 15% of total 5,15-diHETE) was indicative of COX-2 activity, resembling the ‫ف‬ 20:80 ratio formed by the recombinant enzyme ( 16 ) [and no 15 R -LOX enzyme exists in humans or any other mammal ( 37 )]; 2 ) the levels of 5,15-diHETE were markedly decreased in the presence of the COX-2 specifi c inhibitor NS-398; and 3 ) aspirin treatment of the leukocytes increased the ratio of 5 S ,15 R -diHETE versus 5 S ,15 S -diHETE (while decreasing overall synthesis of 5,15-diHETE) and abolished formation of 5,11-diHETE. In addition, biosynthesis of 5,11-diHETE was due to COX-2 reaction with 5-HETE because it was blocked by NS-398, and no 11-LOX activity has been reported in mammals ( 3 ).
An interesting facet of the double oxygenation of arachidonic acid at carbons 5 and 15 is that the reactions can occur in either order. Routes 4, 5, and 6 in Fig. 1 are essentially the inverse of routes 1, 2, and 3, respectively. What is the order of reaction in the biosynthesis of 5,15-di-HETE in the activated leukocytes? Because inhibitors are not a suitable tool to distinguish the order of events, we The levels of LTB 4 (which cannot be formed from 5-HETE or 15-HETE) were reduced about 10-fold at the highest concentration of either of the exogenous HETEs added ( Fig. 6A, C ). It is not clear why exogenous 5-HETE inhibited LTB 4 formation but inhibition by 15-HETE could have been due to outcompeting arachidonic acid as a substrate for 5-LOX ( 29 ). Inhibition of endogenous 5,15-diHETE by the highest concentration of 15-HETE could partially be due to inhibition of COX-2 by 15 S -HETE ( 30 ).

DISCUSSION
Eicosanoid biosynthesis in vivo occurs in an environment that enables transcellular biosynthesis ( 31 ). The exchange of arachidonic acid substrate and its oxygenated products between phagocytic, immune, and endothelial diHETEs together with 5-HETE accounted for about half of the observed 5-LOX metabolites, the other half consisting of LTB 4 and nonenzymatic hydrolysis products of the LTA 4 epoxide. We did not quantify the levels of cysteinyl LTs in the samples.
Little is known about the biological role of 5,15-diHETE. In line with its biosynthesis by leukocytes, 5,15-diHETE plays a role in the infl ammatory response and potentiates the degranulation of human neutrophils in response to platelet activating factor, but not f-Met-Leu-Phe, calcium ionophore A23187, or LTB 4 ( 41 ). 5 S ,15 S -DiHETE is also a chemoattractant for eosinophils with an ED 50 of 0.3 M ( 42 ) and a precursor to the highly potent eosinophil chemoattractant 5-oxo-15-HETE with an ED 50 of 5 nM ( 43 ).
In summary, formation of 5,11-diHETE and 5,15-di-HETE in activated human leukocytes is evidence for a functional biosynthetic cross-over of the 5-LOX and COX-2 pathways. Our studies implicate inhibition of formation of 5,15-diHETE as an additional mechanism of action of NSAIDs and COX-2 selective inhibitors with as of yet incompletely understood biological consequences. added octadeuterated 5-HETE or 15-HETE to the activated leukocytes and determined their transformation to 5,15-HETE relative to the endogenous pathway. We found that exogenous 5-HETE was transformed to 5,15-diHETE, and inhibition by NS-398 proved that the transformation was catalyzed by COX-2. This confi rmed that 5,15-diHETE in the activated leukocytes was, not to the least part, formed via route 1 in Fig. 1, i .e., by crossover of the 5-LOX and COX-2 pathways.
Similar to previous studies ( 29,38 ), we found that addition of exogenous 15-HETE led to inhibition of the formation of LTB 4 ( Fig. 6 ). This effect has initially been interpreted as an inhibitory effect of 15-HETE on 5-LOX activity ( 38 ) but was later found to be due to effi cient utilization of 15-HETE by 5-LOX at the expense of reaction with arachidonic acid ( 29 ). Two additional lines of evidence support this explanation: 1 ) studies with 5-LOX purifi ed from porcine leukocytes showed that 15-HPETE reacted at about 30% of the maximum velocity obtained with arachidonic acid ( 39 ); and 2 ) Mancini et al. ( 40 ) showed that recombinant FLAP was able to stimulate oxygenation of 15-HETE about twofold (and oxygenation of 12-HETE almost 200-fold). It appears that in activated leukocytes the availability of arachidonic acid or 15-HETE is limiting the formation of 5,15-diHETE. Taken together, the abundant formation of 5,15-diHETE from 15-HETE implies that in the leukocytes 5-LOX can react effi ciently with 15-HETE (route 6 in Fig. 1 ). This reaction is also a major pathway of biosynthesis of lipoxins ( 13,14 ).
The transformations of arachidonic acid in activated leukocytes analyzed in this study are summarized in Fig. 7 . Use of a mixed population of leukocytes revealed effi cient exchange of HETEs between LOX isozymes expressed in different cells. 5,15-DiHETE, 5,11-diHETE, and 5,12-di-HETE are produced by the cross-over of 5-LOX with 15-LOX, COX-2, and 12-/15-LOX. These three "crossover"