Oxidative Stress Suppresses Cysteinyl Leukotriene Generation by Mouse Bone Marrow-derived Mast Cells*

Cysteinyl leukotrienes and oxidative stress have both been implicated in bronchial asthma; however, there is no previous study that focused on the ability of oxidative stress to alter cysteinyl leukotriene generation. In this study, treatment of bone marrow-derived mast cells with prostaglandin D2 reduced their ability to generate leukotriene (LT) C4 upon calcium ionophore stimulation but had little effect on LTB4 generation. This effect could be reproduced by a selective agonist of the DP2 receptor, 15R-methyl prostaglandin D2 (15R-D2). 15R-D2 dose-dependently inhibited LTC4 generation with an IC50 of 2 μm, and the effect was not altered by a DP2/thromboxane antagonist or by a peroxisome proliferator-activated receptor-γ antagonist. 15R-D2 exerted its suppressive effect via a reduction in intracellular GSH, a mechanism that involved the conjugation of its non-enzymatic breakdown product to GSH. At 10 μm, 15R-D2 reduced LTC4 generation to 10%, intracellular GSH to 50%, and LTC4 synthase (LTC4S) activity to 33.5% of untreated cells without altering immunoreactive LTC4S protein expression or 5-lipoxygenase activity. The effects of 15R-D2 on LTC4S activity could be partially reversed by reducing reagent. The sulfhydryl-reactive oxidative agent diamide suppressed LTC4S activity and induced a reversible formation of covalent dimer LTC4S. LTC4S bearing a C56S mutation was resistant to the effect of diamide. Covalent dimer LTC4S was observed in nasal polyp biopsies, indicating that dimerization and inactivation of LTC4S can occur at the site of inflammation. These results suggest a cellular redox regulation of LTC4S function through a post-translational mechanism.

Cysteinyl leukotrienes and oxidative stress have both been implicated in bronchial asthma; however, there is no previous study that focused on the ability of oxidative stress to alter cysteinyl leukotriene generation. In this study, treatment of bone marrow-derived mast cells with prostaglandin D 2 reduced their ability to generate leukotriene (LT) C 4 upon calcium ionophore stimulation but had little effect on LTB 4 generation. This effect could be reproduced by a selective agonist of the DP 2 receptor, 15R-methyl prostaglandin D 2 (15R-D 2 ). 15R-D 2 dose-dependently inhibited LTC 4 generation with an IC 50 of 2 M, and the effect was not altered by a DP 2 /thromboxane antagonist or by a peroxisome proliferator-activated receptor-␥ antagonist. 15R-D 2 exerted its suppressive effect via a reduction in intracellular GSH, a mechanism that involved the conjugation of its non-enzymatic breakdown product to GSH. At 10 M, 15R-D 2 reduced LTC 4 generation to 10%, intracellular GSH to 50%, and LTC 4 synthase (LTC 4 S) activity to 33.5% of untreated cells without altering immunoreactive LTC 4 S protein expression or 5-lipoxygenase activity. The effects of 15R-D 2 on LTC 4 S activity could be partially reversed by reducing reagent. The sulfhydrylreactive oxidative agent diamide suppressed LTC 4

S activity and induced a reversible formation of covalent dimer LTC 4 S. LTC 4 S bearing a C56S mutation was resistant to the effect of diamide. Covalent dimer LTC 4 S was observed in nasal polyp biopsies, indicating that dimerization and inactivation of LTC 4 S can occur at the site of inflammation. These results suggest a cellular redox regulation of LTC 4 S function through a post-translational mechanism.
The cysteinyl leukotrienes (LT) 2 LTC 4 , LTD 4 , and LTE 4 are potent inflammatory lipid mediators. They are implicated in asthma (1)(2)(3) and allergic inflammation (4,5). The biological actions of the cysteinyl leukotrienes as mediators of human asthma include contraction of bronchi (2), stimulation of mucous secretion from human airways in vitro (6), and induction of eosinophil infiltration into the airway of asthmatic individuals in vivo (7,8). Biosynthesis of LTC 4 occurs in limited numbers of effector cells, such as mast cell, basophils, eosinophils, and macrophages that express the 5-lipoxygenase (5-LO) pathway of enzymes as well as LTC 4 synthase (LTC 4 S). In mast cells, LTC 4 biosynthesis is initiated upon antigen cross-linking of the high affinity IgE receptors, which leads to influx of Ca 2ϩ ion. An increase in intracellular Ca 2ϩ activates calcium-dependent cytosolic phospholipase A 2 , which subsequently translocates to the perinuclear membrane and hydrolyzes the release of arachidonic acid from the membrane phospholipids (9). The released arachidonic acid interacts with the 5-LO-activating protein FLAP, an integral nuclear membrane protein (10), during presentation to 5-LO, which also translocates to the perinuclear membrane from both the cytosol and the nucleoplasm (11,12). 5-LO converts arachidonic acid to 5-hydroperoxyeicosatetraenoic acid and subsequently to LTA 4 (13). LTA 4 either is hydrolyzed by cytosolic LTA 4 hydrolase to form LTB 4 (14) or is conjugated to GSH to form LTC 4 , the parent compound of cysteinyl leukotrienes (15,16). This conjugation is the function of LTC 4 S (17). LTC 4 S is an 18-kDa integral membrane protein located at the nuclear envelope of LTC 4 -generating cells. Cloning of the human and mouse LTC 4 S cDNAs revealed the highest amino acid identity to FLAP (17) and to bifunctional microsomal GST2 and GST3 (18). These proteins are all members of a superfamily of membrane-associated proteins involved in eicosanoid and glutathione metabolism (MAPEG) (18), which also includes microsomal PGE 2 synthase-1 and microsomal GST1. LTC 4 conjugates GSH to LTA 4 to form LTC 4 , and in contrast to microsomal GST2 and GST3, LTC 4 S does not conjugate GSH to xenobiotics. Recently, it was demonstrated that LTC 4 S can also conjugate GSH to 5-oxohydroxyeicosatetraenoic acid to form FOG-7 (19). X-ray crystallography of human LTC 4 S with GSH shows that the LTC 4 S monomer has four transmembrane ␣-helices and forms a 3-fold symmetric trimer as a unit with functional domains across each interface with GSH resides in a U-shaped conformation within an interface between adjacent monomers (20, 21). This binding is stabilized by a loop structure at the top of the interface. X-ray crystallographic data suggest that LTA 4 would fit into the interface so that Arg-104 of one monomer activates GSH to provide the thiolate anion that attacks C6 of LTA 4 to form a thioether bond, and Arg-31 of the neighboring monomer would donate a proton to form a hydroxyl group at C5, resulting in the formation of 5Shydroxy-6R,S-glutathionyl-7,9-trans-11,14-cis-eicosatetra-enoic acid (LTC 4 ). In addition, X-crystallography shows that Arg-51, Asn-55, Glu-58, Tyr-93, Tyr-97, and Arg-104 of LTC 4 S are involved in GSH binding.
Recently, studies on prostaglandin biosynthesis have demonstrated that nitric oxide S-nitrosylates cytosolic phospholipase A 2 (22) and COX2 (23), resulting in the activation of these two enzymes and an increase in prostaglandin synthesis. These results suggest that oxidative stress is able to induce post-translational modification of enzymes associated with eicosanoid biosynthesis. Although a decrease in intracellular GSH upon 1-chloro-2,4-dinitrobenzene (CDNB) treatment of human polymorphonuclear leukocytes has been shown to increase 5-LO activity (24) and to modify 5-LO activity by ERK2 and PKA phosphorylation (25,26) and 5-LO translocation by MAPKAP2 phosphorylation (27), there is no information on the effect of oxidative stress on the post-translational modification of 5-LO or its pathway of enzymes. In this study, we examined the regulation of LTC 4 biosynthesis by oxidative stress. On the basis of preliminary experiments indicating that prostaglandin (PG) D 2 , the major cyclooxygenase product of mast cells, could suppress LTC 4 generation when applied exogenously to mouse bone marrow-derived mast cells (BMMCs), we unexpectedly recognized that PGD 2 and the DP 2 receptor agonist 15R-methyl-PGD 2 (15R-D 2 ) suppressed LTC 4 generation by BMMCs through a reduction in intracellular GSH and suppression of LTC 4 S activity. In addition, diamide, a sulfhydrylreactive oxidant, induced oxidative stress, suppressed LTC 4 S activity, and induced the formation of covalent dimer LTC 4 S, normally a noncovalent trimer. Importantly, the effect of oxidative stress was abolished in C56S mutant LTC 4 S, and the covalent dimer was present in nasal polyps of asthmatics, implying that LTC 4 S activity levels may be modified in vivo by oxidation-related dimerization.
Culture of BMMCs-Bone marrow cells were collected from femurs and tibiae of BALB/c mice and cultured for 6 -10 weeks in RPMI 1640 medium containing 10% fetal bovine serum, 2 mM L-glutamine, 0.1 mM nonessential amino acids, 100 units/ml penicillin, 100 g/ml streptomycin, and 1% culture supernatant from CHO cells expressing mouse interleukin-3. The culture medium for the BMMCs was changed every week, and the cell density was adjusted to 5 ϫ 10 5 /ml at every passage. After 4 weeks, Ͼ97% of the cells were BMMCs as assessed by staining with toluidine blue.
Ionophore-stimulated 5-LO Product Generation-BMMCs were first treated with PGD 2 , 15R-D 2 , or vehicle (ethanol) for up to 48 h. Cells were then harvested, washed, and suspended in Hanks' buffer containing calcium, magnesium, and 2 mg/ml bovine serum albumin (buffer A). Stimulation of BMMCs were initiated by the addition of ionophore (2 M), followed by incubation at 37°C for 15 min. Reactions were terminated by the addition of 3 volumes of methanol containing 400 ng/ml PGB 2 . Samples were analyzed for 5-LO product formation by reversephase (RP) HPLC with a Beckman System Gold Model 126 pump and Model 168 diode array UV detector using a programmed gradient and solvent system as described previously (28) and are expressed as picomoles/10 6 cells.
Enzyme Assay-To assay for LTC 4 S activity, BMMCs were first pelleted by centrifugation and then resuspended in 200 l of assay buffer (50 mM HEPES and 20 mM MgCl 2 , pH 7.6) and lysed by microtip sonication. The cell lysates were incubated with 20 M LTA 4 methyl ester and 10 mM GSH in assay buffer at room temperature for 10 min. Samples were analyzed by RP-HPLC (17). To assay for 5-LO activity, BMMCs were lysed in 5-LO assay buffer (0.1 M Tris, 30 mM KH 2 PO 4 , and 1.5 mM EDTA), and the reactions were initiated with the addition of 50 M arachidonic acid, 2 mM ATP, and 1 mM CaCl 2 and incubated at 37°C for 15 min. Reaction products were analyzed by RP-HPLC.
Western Blot Analysis-Cells were resuspended in lysis buffer (10 mM Tris-HCl, pH 7.6, 100 mM NaCl, 50 mM NaF, 100 M Na 3 VO 4 , 50 g/ml PMSF, 10 g/ml aprotinin, and 10 g/ml leupeptin), boiled in SDS-PAGE sample buffer under reducing and nonreducing conditions, and separated on Tris/ glycine-14% SDS-polyacrylamide gel (Novex). After transfer to a PVDF membrane, the blot was blocked with 4% nonfat dry milk and incubated with affinity-purified anti-LTC 4 S IgG (1 g/ml) for 2 h. After washing, the blot was incubated with horseradish peroxidase-linked secondary antibody for 1 h, washed, and developed with chemiluminescence substrate. For nasal polyps, samples were minced in lysis buffer, homogenized, and boiled in SDS-PAGE sample buffer under reducing and nonreducing conditions.
Real-time PCR-To determine the effect of 15R-D 2 on LTC 4 S mRNA, we examine the time-dependent changes in LTC 4 S mRNA by real-time PCR. Cells were treated with 10 M 15R-D 2 for 0, 3, 6, and 24 h and harvested, and total RNA were then isolated with TRIzol reagent (Invitrogen). The relative LTC 4 S mRNA level was determined by real-time PCR (GeneDirect, Princeton, NJ) and by comparing the values of the ratio of the LTC 4 S gene product to the 18 S housekeeping gene product.
Measurement of Intracellular GSH-Intracellular GSH levels were measured by enzymatic conjugation of CDNB to GSH present in the mast cell lysate to form glutathionyl-S-dinitrophenyl (GS-DNP). Mast cells were lysed with 200 l of 100 mM Tris-HCl, pH 8.0, and then incubated with 200 M CDNB at room temperature for 30 min in the presence of 1 unit of GST from human placenta. Reactions were terminated by the addition of 10 l of acetic acid (17.4 M) and 3 volumes of methanol containing 400 ng/ml PGB 2 and then analyzed by RP-HPLC for GS-DNP levels, with the ultraviolet absorbance monitored at 280 and 340 nm (for GS-DNP). Intracellular GSH was calculated by comparing the peak area of GS-DNP with that of a standard curve generated by incubating known amounts of GSH (125-4000 pmol) with 200 M CDNB and 1 unit of GST under identical conditions and is expressed as picomoles/10 6 cells.
Effect of BSO and Diamide-To examine the effect of the oxidizing condition on the ability of BMMCs to generate 5-LO products, BMMCs were pretreated with either 200 M BSO, a GSH synthesis inhibitor, for 24 h or with 300 M diamide, a sulfhydryl-reactive oxidative agent, for 30 min to deplete their intracellular GSH. Cells were then harvested and separated into three aliquots. One aliquot of cells was suspended in buffer A and stimulated with 2 M ionophore to determine the ability to generate 5-LO products, and the other two aliquots were lysed and assayed for LTC 4 S activity and GSH levels, respectively.
Metabolism of PGD 2 and 15R-D 2 -BMMCs were incubated with 10 M PGD 2 or 10 M 15R-D 2 for 24 h and then analyzed by RP-HPLC for product profiles. For non-enzymatic conversion of these prostanoids, we incubated PGD 2 and 15R-D 2 with culture medium alone for 24 h and then analyzed the product profiles.

Effect of C56S and C82V Mutations on Diamide-induced Suppression of Enzyme Activity and Covalent Dimer Formation-
To determine whether Cys-56 of LTC 4 S is involved in covalent dimer formation, we transfected wild-type, C56S, and C82V plasmid cDNAs into CHO cells by lipofection using FuGENE 6 reagent. Two days after transfection, cells were harvested, washed, and then incubated with 300 M diamide. After 30 min, cells were lysed and assayed for LTC 4 S activity and molecular size. 4 Generation in BMMCs-Like cysteinyl leukotrienes, PGD 2 is abundant in allergic inflammation and a product of activated mast cells. No previous study had addressed whether PGD 2 could modify the generation of LTC 4 . In addition, as LTC 4 S catalyzes the conjugation of LTA 4 to GSH, we were also interested in determining whether changing intracellular GSH through oxidative stress will affect LTC 4 generation. Because PGD 2 has previously been shown to increase LTC 4 generation by mouse eosinophils (29), and its cyclopentenone metabolite has been proposed as a potential endogenous regulator of redox-sensitive transcription factors (30) and has been shown to deplete intracellular GSH (31), we thus examined the effect of PGD 2 on LTC 4 generation by BMMCs. BMMCs were incubated with 10 M PGD 2 for 48 h, and for specificity purposes, we also included 10 M PGE 2 and 10 M PGF 2␣ for comparison. After harvesting, cells were stimulated with 2 M calcium ionophore A23187 for 10 min. As shown in Fig. 1, whereas both PGD 2 and PGE 2 inhibited LTC 4 generation by Ͼ70 and 65%, respectively, PGF 2␣ had no effect. We concentrated on the effect of PGD 2 , as it is also a mast cell-derived product. None of these prostaglandins affected the biosynthesis of 12-hydroxyheptadecatrienoic acid (data not shown).

PGD 2 and PGE 2 Inhibit LTC
Inhibitory Effect of PGD 2 Does Not Act through the DP 1 or DP 2 Receptor or Peroxisome Proliferator-activated Receptor-␥ (PPAR␥)-To investigate the possible receptor involved in the suppression of LTC 4 generation by PGD 2 , BMMCs were incubated with the DP 1 agonist BW 245C or the DP 2 agonists 15R-D 2 and 15-deoxy-⌬ 12,14 -PGD 2 (15d-D 2 ; also a PPAR␥ agonist), and we then examined the ionophore-stimulated generation of 5-LO products. As shown in Fig. 2A, the effect of PGD 2 could be mimicked by the DP 2 agonists 15R-D 2 and 15d-D 2 but not by the DP 1 agonist BW 245C. To further investigate whether the suppressive effect is mediated through the DP 2 receptor or PPAR␥, we examined the effect of the DP 2 antagonist Ramatroban and the PPAR␥ antagonist GW 9662 using 15R-D 2 as an agonist. As shown in Fig. 2B, the effect of 15R-D 2 on LTC 4 generation was not inhibited by the DP 2 antagonist Ramatroban or by the PPAR␥ antagonist GW 9662, suggesting that the inhibitory effect of PGD 2 or 15R-D 2 is not mediated through DP 1 , DP 2 , or PPAR␥. Because the effect of PGD 2 could be reproduced by 15R-D 2 , we performed subsequent experiments using 15R-D 2 as an agonist.
Time-dependent Effect of 15R-D 2 on the Ability of BMMCs to Generate LTC 4

and LTC 4 S Protein or mRNA Expression-
When 15R-D 2 was incubated with BMMCs, it caused a timedependent decrease in LTC 4 generation when stimulated with the calcium ionophore. When BMMCs were treated with 15R-D 2 for 24 h, there was a Ͼ75% reduction in ionophorestimulated LTC 4 generation (Fig. 3A), with little additional reduction when treated for 48 h (data not shown). Thus, we examined the dose-dependent effect of 15R-D 2 over 24 h of treatment.
To determine whether 15R-D 2 reduced cellular LTC 4 S activity through reduced LTC 4 S transcription and protein expression, we incubated BMMCs with 10 M 15R-D 2 for 0, 3, 6, and 24 h. LTC 4 S mRNA expression was then analyzed by real-time PCR, and protein expression was examined by immunoblot analysis. As shown in Fig. 3 (B and C), 15R-D 2 did not alter the expression of either the mRNA or immunoreactive LTC 4 S protein. This result indicates that 15R-D 2 does not reduce the cellular LTC 4 activity of BMMCs through a reduction in LTC 4 S

8) of untreated
BMMCs. An IC 50 of ϳ2 M was observed for the inhibition of LTC 4 generation by 15R-D 2 . There were no significant changes in all-trans-diastereoisomers of LTB 4 or LTB 4 or 5-hydroxyeicosatetraenoic acid generation ( Fig. 2A), suggest-ing that 15R-D 2 selectively reduces the ability of BMMCs to generate LTC 4 without affecting the 5-LO activity. To confirm that 15R-D 2 did not affect 5-LO enzyme activity, both the 5-LO and LTC 4 S activities were measured in BMMC lysates obtained  from BMMCs that were treated with various concentrations of 15R-D 2 and from untreated cells. As shown in Fig. 4 (B and C), 15R-D 2 did not affect cellular 5-LO activity but dose-dependently reduced LTC 4 S activity. The mean LTC 4 S activity of 15R-D 2 -treated BMMCs was 33.5 Ϯ 3.5% (mean Ϯ S.E., n ϭ 8) (Fig. 3B) of the control cells, thus confirming that the reduction in LTC 4 S activity upon 15R-D 2 treatment contributed to the decrease in ionophore-stimulated LTC 4 generation. Furthermore, the effect of 15R-D 2 on LTC 4 generation was not restricted to the calcium ionophore A23187, as 15R-D 2 also suppressed LTC 4 generation induced by ionomycin and by cross-linking of IgE. 15R-D 2 -reduced ionomycin (2 M) stimulated LTC 4 generation to 8.6 Ϯ 2.9% (mean Ϯ S.E., n ϭ 10) of the untreated cells. In two experiments, 10 M 15R-D 2 also reduced LTC 4 generation induced by cross-linking of IgE to 10.6 and 11.8% of the untreated cells, respectively.
15R-D 2 Reduces Intracellular GSH Levels in BMMCs-15R-D 2 inhibited LTC 4 generation without a significant effect on LTC 4 S protein expression or 5-LO activity, suggesting that 15R-D 2 may affect LTC 4 generation by altering the intracellular levels of GSH, the second substrate of the LTA 4 conjugation reaction. Thus, we investigated the level of intracellular GSH by assaying the ability of the cell lysate to enzymatically conjugate CDNB to endogenous GSH to form GS-DNP at pH 8.0 by adding excess CDNB and exogenous GST. As shown in Fig. 4D, 15R-D 2 dose-dependently decreased the intracellular GSH levels as indicated by a reduction in the amount of GS-DNP generated when cell lysates were incubated with CDNB. At 10 M 15R-D 2 , the mean GSH level of treated BMMCs was 43.9 Ϯ 5.8% (mean Ϯ S.E., n ϭ 5) of the control cells.
BSO Reduces LTC 4 Generation-To confirm that 15R-D 2 reduces ionophore-stimulated generation of LTC 4 by BMMCs through a reduction in intracellular GSH, we compared the effect of BSO, a glutathione synthase inhibitor, with that of 15R-D 2 on ionophore-stimulated LTC 4 generation. When BMMCs treated with 15R-D 2 were stimulated with ionophore, LTC 4 S activity was reduced to 15.2 Ϯ 5.9% (mean Ϯ S.E., n ϭ 3) of the control BMMCs. Treatment with BSO reduced LTC 4 generation to undetectable levels. Similarly, BSO reduced intracellular GSH levels to 0.57 Ϯ 0% (mean Ϯ S.E., n ϭ 3) of the control cells.
Mechanism of Reduction in Intracellular GSH by 15R-D 2 -Because cyclopentenone metabolites of PGD 2 have previously been shown to be able to reduce intracellular GSH through conjugation to GSH and because mast cells have been shown to be poor producers of reactive oxygen species (32), we examined the metabolism of 15R-D 2 and PGD 2 in BMMCs. When PGD 2 was incubated with BMMCs or with culture medium alone for 24 h and then analyzed by HPLC for its metabolites present in the culture medium, we found three major non-enzymatic metabolites that corresponded to synthetic standards of ⌬ 12 -PGJ 2 , 15d-J 2 , and 15d-D 2 , respectively, in both the BMMC culture medium (Fig. 5A) and the medium with no BMMCs (Fig.  5B). However, there was an additional polar metabolite (compound X) in the HPLC chromatograph of BMMC culture medium that was absent in the medium-alone sample. Furthermore, the relative peak heights of ⌬ 12 -PGJ 2 (also monitored at its UV maximum of 244 nm; chromatograph not shown) and 15d-J 2 culture media were nearly identically with or without BMMCs, whereas the peak height of 15d-D 2 in the medium with BMMCs was 24.9 Ϯ 1.82% (mean Ϯ S.E., n ϭ 3) lower than that with no BMMCs. Similarly, incubation of 15R-D 2 with BMMCs and culture medium alone yielded metabolite profiles identical to those of PGD 2 with three 15R-methyl derivatives of the corresponding PGD 2 metabolites (data not shown). We thus examined whether compound X is derived from 15d-D 2 or from 15d-J 2 . We first incubated 15d-D 2 and 15d-J 2 ) with and without GSH for 4 h at 37°C and then analyzed the polar metabolite formation, as they are readily available commercially. Both 15d-D 2 and 15d-J 2 were conjugated to GSH nonenzymatically to form polar metabolites with a similar retention time in HPLC in the presence of GSH (data not shown). However, only 15d-D 2 was converted to a polar metabolite with a HPLC retention time identical to that of compound X when incubated with BMMCs (Fig. 6, A and B). There was no metabolism of 15d-J 2 observed when it was incubated with BMMCs (data not shown). These results suggest that compound X is a PGD 2 metabolite derived from the conjugation of 15d-D 2 to GSH and that 15R-D 2 spontaneously degrades to form 15Rderivatives of 15d-D 2 , which are then conjugated to GSH, and lowers intracellular GSH. The chemical structure of compound X is not known and is being elucidated. This result is very similar to that reported for the conjugation of 9-deoxy-⌬ 9,12 (E)-PGD 2 to GSH (31).

Effect of 15R-D 2 on LTC 4 S Is Partially
Reversible-Because ␤-mercaptoethanol (␤-ME) counters the effect of 15R-D 2 on LTC 4 generation, we examined whether ␤-ME can rescue LTC 4 S activity 24 h after 15R-D 2 treatment. BMMCs were treated with 10 M 15R-D 2 for 24 h, harvested, washed, and incubated with and without 10 mM ␤-ME. After 20 min, cells were washed once and then assayed for LTC 4 S activity. Treatment of BMMCs with 15R-D 2 for 24 h reduced LTC 4 S activity to 29.9 Ϯ 8.5% (mean Ϯ S.E., n ϭ 4) of the control cells. LTC 4 S activity increased to 65.4 Ϯ 9.8% with ␤-ME treatment. These results suggest that the effect of 15R-D 2 on LTC 4 S is at least a partially reversible process.
Effect of Diamide on LTC 4 S-Our demonstration that a reduction in intracellular GSH suppresses LTC 4 S function suggested that oxidative stress may also exert an effect similar to that of 15R-D 2 . Thus, we utilized the sulfhydryl-reactive oxidant diamide to acutely induce oxidative stress and to examine the enzyme activity of LTC 4 S. Incubation of diamide with BMMCs for 30 min reduced LTC 4 S activity by Ͼ75% (data not shown). To further study the effect of oxidative stress on LTC 4 S, we examine the effect of diamide on CHO cells transfected with wild-type and mutant LTC 4 S cDNAs. When diamide was incubated with CHO cells transfected with wild-type LTC 4 S cDNA, it reduced LTC 4 S function to 22.8 Ϯ 3.4% (mean Ϯ S.E., n ϭ 4) of the control cells. In addition, diamide induced the formation of covalent dimer LTC 4 S in a nonreducing gel (Fig. 7A). In contrast, C56S mutant LTC 4 S retained Ͼ91% of the enzyme activity (n ϭ 2) with diamide treatment and no covalent dimer formation (Fig. 7A). Mutation of the only other cysteine residue (Cys-82) did not protect enzyme activity from diamide treat-ment and formation of a covalent dimer (Fig. 7A). The effect of diamide was also reversible, as diamide-treated CHO cells that were subsequently incubated with 10 mM ␤-ME restored LTC 4 S activity to 89.2 Ϯ 9.1% (n ϭ 4) of the control enzyme, and there was no covalent dimer formation.
To determine whether oxidative stress similarly affects purified LTC 4 S, we incubated purified LTC 4 S with diamide and then assayed its conjugating activity and analyzed the formation of the covalent dimer by Western blotting. Similar to the transfected CHO cells, diamide treatment suppressed the enzyme activities of purified LTC 4 S by Ͼ90% (data not shown) as well as the formation of a covalent dimer (Fig. 7B).
These results indicate that oxidative stress can reversibly suppress LTC 4 S function at least in part through induction of disulfide bridging of Cys-56 from neighboring monomers and the formation of an enzymatically inactive covalent dimer.
To determine whether the covalent dimer exists in vivo, we examined nasal polyps from asthmatic subjects, which have been reported to contain large numbers of infiltrated activated eosinophils (33). Western blot analyses demonstrated the presence of a 34-kDa covalent dimer LTC 4 S (Fig.  8) under nonreducing conditions, similar to wild-type LTC 4 S treated with diamide. In addition, the presence of covalent dimer LTC 4 S was also observed in nasal polyps obtained from a normal individual. These results suggest

DISCUSSION
This study has demonstrated that PGD 2 , the DP 2 agonist 15R-D 2 , and the PPAR␥ agonist 15d-J 2 suppress ionophorestimulated LTC 4 generation. The suppressive effect is a receptor-mediated event, as it was not block by either the DP 2 antagonist Ramatroban or the PPAR␥ antagonist GW 9662 (Fig. 2B), and on the contrary, these antagonists significantly enhanced the suppressive effect of 15R-D 2 on LTC 4 generation. This potentiation effect has not been investigated further. The effect of 15R-D2 is not mediated through an alteration of transcription or protein expression, as neither was affected by 15R-D2 (Fig. 3). Because reduction in intracellular GSH has previously been shown to affect various cell functions, including gene transcription and cell proliferation (34,35), we thus examined if the suppressive effect could be mediated through a reduction in intracellular GSH, a second substrate for LTC 4 S. We demonstrated that the DP 2 agonist 15R-D 2 reduces intracellular GSH in BMMCs (Fig. 4D) and reduces the ability of these cells to synthesize cysteinyl leukotrienes in response to ionophore stimulation as well as suppression of LTC 4 S activity through post-translational modification (Fig. 3, A and B, and Fig. 4). The mechanism by which 15R-D 2 reduces intracellular GSH is at least partially through conjugation of GSH to its non-enzymatic breakdown product and does not involve any known PGD 2 receptor (Figs. 5 and 6). When 15R-D 2 was incubated with BMMCs, it non-enzymatically converted to 15R-derivatives of ⌬ 12 -PGJ 2 , 15d-J 2 , and 15d-D 2 . These determinations were based on the identical product profiles when PGD 2 was utilized (Fig. 5) and identical HPLC retention times compared with available authentic standards of these PGD 2 metabolites. Of these three cyclopentenone metabolites, both 15d-J 2 and 15d-D 2 conjugated to GSH non-enzymatically, with 15d-J 2 being a better substrate for conjugation (data not shown). When incubated with BMMCs, however, a GSH-conjugated metabolite was observed in only incubations of 15d-D 2 with BMMCs (Fig. 6). No metabolism was observed with 15d-J 2 (data not shown), suggesting that the unknown polar metabolite X is a conjugated product of 15d-D 2 . The reason for the difference in metabolism of 15d-D 2 and 15d-J 2 by BMMCs is not known. It could be that uptake of these eicosanoids requires a membrane carrier and that there is an uptake carrier in BMMCs (36) for 15d-D 2 but not for 15d-J 2 . The ability to nonenzymatically break down to cyclopentenone metabolites may account for the observed ability of PGD 2 and PGE 2 to suppress LTC 4 generation, whereas PGF 2 ␣ did not (Fig. 1). Furthermore, the ability of PGD 2 to reduce intracellular GSH did not alter the biosynthesis of 12-hydroxyheptadecatrienoic acid, indirectly suggesting that there is no change in thromboxane synthesis, as 12-hydroxyheptadecatrienoic acid is an equal molar byproduct of thromboxane synthesis (37). We also examined if the reduction in intracellular GSH could be a result of an increase in reactive oxygen species generation, but our experiments showed that BMMCs produced negligible amounts of reactive oxygen species upon A23187 and phorbol 12-myristate 13-acetate stimulation (data not shown), an observation that is in agreement with that of Swindle et al. (32). In addition, the NADPH oxidase inhibitor diphenyleneiodonium sulfate did not overcome the effect of 15R-D 2 (data not shown), thus confirming that the effect of 15R-D 2 is not mediated through an increase in reactive oxygen species generation.
Both 15R-D 2 and oxidative stress induced by diamide caused a possible post-translational modification of LTC 4 S, as they suppressed LTC 4 S activity without affecting protein expression (Figs. 3 and 7). Diamide also induced the formation of inactive covalent dimer LTC 4 S (Fig. 7). The suppressive effect of 15R-D 2 on LTC 4 S activity can only be partially reversed with reducing agent, whereas the effect of diamide is totally reversible. Although the rationale for the difference in reversibility between 15R-D 2 and diamide is not known, one possible explanation may be that the cyclopentenone metabolite of 15R-D 2 can directly modify Cys-56 of LTC 4 S, an effect that has been reported for the p50 subunit of NFB using 15d-J 2 (38). Unlike the disulfide bridge of the covalent dimer, the metabolite-modified Cys-56 cannot be reduced by reducing agent, and therefore, that portion of LTC 4 S is irreversibly inhibited.
Importantly, our experiments suggest that cellular LTC 4 S may exist in two different forms, an enzymatically active form and an inactive form, and that the relative ratio of these two forms will depend on the oxidative state of the cells. Thus, our findings that 15R-D 2 alters cell function through a reduction in cellular GSH further suggest that the cellular redox state may control the homeostasis of inflammatory cell function and subsequently the inflammatory processes. In a preliminary experiment with one nasal polyp, we also observed a 3-fold increase in LTC 4 S activity after it was treated with the reducing agent ␤-ME (data not shown). This observation, together with the demonstration of the presence of inactive covalent dimer LTC 4 S in a nasal polyp biopsy (Fig. 8), suggests that the cellular oxidation/reduction state of the tissue can modify its ability to generate 5-LO products and that tissue immunohistochemistry alone is not sufficient in determining the true capacity of the tissue in leukotriene biosynthesis in situ. Acknowledgment-We are thankful to Dr. Joshua Boyce for helpful comments on the manuscript.  1 and 2), C56S mutant (lanes  3 and 4), and C82V mutant (lanes 5 and 6) LTC 4 S cDNAs treated with diamide (lanes 2, 4, and 6) and untreated controls (lanes 1, 3, and 5) (A) and of purified LTC 4 S treated with and without diamide (B). SDS-PAGE was performed under nonreducing conditions. CNT, control.