Cyclooxygenase 1 mediates IL-33-induced extracellular signal regulated kinase activation in mast cells; Implications for aspirin sensitivity

Abstract

Background.

Classical FcεRI-induced mast cell (MC) activation causes synthesis of arachidonic acid (AA)-derived eicosanoids (leukotriene (LT)C₄, prostaglandin D₂ (PGD₂), and thromboxane A₂ (TXA₂)), that mediate vascular leak, bronchoconstriction, and effector cell chemotaxis. Little is known about the significance and regulation of eicosanoid generation in response to non-classical MC activation mechanisms.

Objectives.

To determine the regulation and significance of MC-derived eicosanoids synthesized in response to interleukin-33 (IL-33), a cytokine critical to innate type 2 immunity.

Methods.

We employed an ex vivo model of mouse bone marrow-derived MCs (BMMCs) and an IL-33-dependent in vivo model of aspirin-exacerbated respiratory disease.

Results.

IL-33 potently liberates AA and elicits LTC₄, PGD₂, and TXA₂ production by BMMCs. Unexpectedly, the constitutive function of cyclooxygenase (COX)-1 is required for IL-33 to activate group IVA cytosolic phospholipase A₂ (cPLA₂) with consequent AA release for synthesis of all eicosanoids, including cysLTs. In contrast, COX-1 was dispensable for FcεRI-driven cysLT production. Inhibition of COX-1 prevented IL-33 induced phosphorylation of extracellular signal related kinase (ERK), an upstream effector of cPLA₂, which was restored by exogenous PGH₂, implying that the effects of COX-1 required its catalytic function. The administration of a COX-1-selective antagonist to mice completely prevented the generations of both PGD₂ and LTC₄ in a model of AERD in which MC activation is IL-33-driven.

Conclusions.

MC-intrinsic COX-1 amplifies IL-33-induced activation in the setting of innate type 2 immunity, and may help explain the phenomenon of therapeutic desensitization to aspirin by nonselective COX inhibitors in AERD.

Capsule Summary

IL-33 activates cytosolic phospholipase A₂ in mast cells by an unexpected mechanism requiring COX-1-dependent extracellular signal regulated kinase activation.

Introduction

Mast cells (MCs) are tissue-resident effector cells that initiate and propagate inflammation in both innate and adaptive immunity. Classical allergen-induced MC activation results from cross-linkage of the tetrameric high-affinity Fc receptor for IgE (FcεRI) occupied by antigen-specific IgE. MCs activated via this mechanism degranulate and rapidly release pre-formed mediators such as histamine and proteases in anaphylaxis, rhinitis, and allergic asthma (1). Simultaneously, MCs release arachidonic acid (AA) from cell membrane phospholipids by a calcium-dependent group IVa phospholipase A₂ (also known as cytosolic (c)PLA₂)(2) that requires serine phosphorylation by extracellular signal related kinase (ERK) and/or p38 mitogen activated protein kinase (MAPK)(3). Both calcium flux and MAPK-dependent phosphorylation are necessary for cPLA₂ to translocate from the cytosol to intracellular membranes. Once liberated by cPLA₂, AA is converted via cyclooxygenases (COX)-1 and/or -2 to prostaglandin H₂ (PGH₂) which in turn is converted by hematopoietic PGD₂ synthase to PGD₂ and by thromboxane synthase to thromboxane A₂ (TXA₂) (4, 5). MCs also metabolize AA to leukotriene (LTA)₄ by 5-lipoxygenase (5-LO)-dependent oxidation (6). LTA₄ is converted by LTA₄ hydrolase to LTB₄ (7) and by LTC₄ synthase to LTC₄ (8), the parent of the cysteinyl leukotrienes (cysLTs) LTD₄ and LTE₄. PGD₂, TXA₂, cysLTs and LTB₄ cooperate to elicit bronchoconstriction (9), changes in vascular tone and permeability (10), priming of endothelial adhesion pathways (11), and recruitment and activation of blood-born hematopoietic cells to the tissues (12). Consequently, AA-derived eicosanoids are strongly implicated in the clinical and physiological manifestations of type 1 hypersensitivity, and likely also contribute to IgE-dependent adaptive immunity to helminths (13). The mechanisms regulating their production are of substantial importance.

While the biochemistry of FcεRI-dependent eicosanoid generation by MCs is well-studied, comparatively little is known about how this process is regulated during non-classical, antigen-independent MC activation. MCs can be activated by proteases (14), complement fragments (15), lipids (16), lipopolysaccharide (17), and endogenous cytokines (18). MCs express the IL-1 receptor family member ST2, which binds the potent type 2 cytokine IL-33 (19). Like other IL-1 receptor family members, ST2 activates ERK, p38, and c-Jun terminal kinase, and induces NF-kB-dependent transcriptional activity through the canonical MyD88 pathway (20). The release of IL-33 from injured or activated epithelial and endothelial cells initiates type 2 immune responses to allergens (21, 22), helminths (23, 24), and viruses (25) by activating cells of the innate immune system (including MCs)(26). Tissue levels of IL-33 (and expression of ST2) are strongly upregulated in diseases characterized by robust type 2 immunopathology and increased numbers of activated tissue MCs (27, 28), such as severe asthma (29, 30) and chronic rhinosinusitis (CRS)(31). Although MCs generate abundant quantities of IL-5 and IL-13 in response to IL-33 ex vivo (19), little is known about whether IL-33-driven, ST2-dependent MC activation elicits eicosanoid generation, how this process is regulated, and what the physiologic consequences are.

Aspirin exacerbated respiratory disease (AERD) is a distinctive clinical syndrome characterized by severe type 2 respiratory tract immunopathology, asthma, CRS, and idiosyncratic reactions to COX-1-active drugs (but generally not COX-2-selective drugs) (32) characterized by MC activation and high level synthesis of both COX (PGD₂) and 5-LO (LTC₄) products (33). Paradoxically, COX-1 inhibitors also elicit “desensitization”, permitting subsequent administration of COX-1 inhibitors without eliciting symptoms or MC activation, and daily administration of aspirin is therapeutic (34). IL-33 is strongly expressed in the sinonasal tissues of patients with AERD (35). We recently demonstrated that MC activation (with attendant release of LTs and PGD₂) in a mouse model of AERD depends on IL-33 and ST2 (35). We therefore undertook this study to understand the mechanisms by which IL-33 induces lipid mediator generation by MCs. We found that IL-33 induces AA release from mouse bone marrow derived MCs (BMMCs), resulting in the concomitant generations of PGD₂, TXA₂, and LTC₄. Unexpectedly, and in contrast to classical MC activation, IL-33-induced AA release and synthesis of all eicosanoids hinges critically on the constitutive activity of COX-1. COX-1 activity is necessary for IL-33-dependent phosphorylation of ERK, but not of p38, and is essential for downstream ERK-mediated phosphorylation of cPLA₂. Treatment of AERD-like mice with SC560, a selective COX-1 inhibitor, elicits desensitization to aspirin and prevents IL-33-driven MC activation and productions of both PGD₂ and LTC₄ in vivo. Our findings reveal a previously unrecognized role for COX-1 in regulating signaling events in MCs downstream of IL-33 and ST2. This mechanism may bear both on the therapeutic benefits of COX inhibitors, including desensitization to COX-1 inhibitors and treatment with aspirin in AERD, which prevents increases in MC-dependent eicosanoids with aspirin challenge and improve outcomes (34, 36).

Methods

Reagents:

Chemicals were purchased from Sigma-Aldrich unless otherwise stated. Murine IL-33, IL-3 and SCF were from Peprotech. Selective COX-1 inhibitors (SC560, FR122047), PGH₂, PGD₂-MOX, TXB₂ and cysLT EIA kits were purchased from Cayman Chemical. the MAP Kinase Kinase inhibitor U0126 and the p38 inhibitor SKF-86002 were obtained from Cell Signaling Technologies (Danvers, MA). Df was obtained from Greer Laboratories (XPB81D3A25; Lenoir, NC).

Cell culture and transfection:

BMMCs were derived from wild type mouse bone marrow cells cultured at 37°C in a humidified incubator with 5% CO₂ in RPMI1640 supplemented with 10% FBS, glutamine (4 mM), sodium pyruvate (1 mM), penicillin (100 units/ml), streptomycin (100 μg/ml), non-essential amino acids (10 ml/L), β-mercaptoethanol (50 μM), in the presence of mouse recombinant IL-3 (10 ng/ml) and SCF (10 ng/ml) (37). BMMCs were studied at 4 – 6 weeks when virtually all cells were stained with toluidine blue and used for experiments within 6 weeks after achieving purity. For downregulation of endogenous COX-1, BMMCs cells were transfected with SMART pool ON-TARGETplus Ptgs1 or a pool of four non-targeting siRNAs as control (300 nM, L-040883-01-0005, D-001810-10-05 Dharmacon, USA). Electroporation was performed using an MAXA Nucleofector II (Lonza) with mouse macrophage cell buffer (Lonza, VPA1009, USA) and program T001. Cells were suspended in mast cell culture medium immediately after electroporation for 24 h before further experiments. Downregulation of endogenous COX-1 were confirmed by qPCR of COX-1 transcripts. Human MCs were derived from umbilical cord blood progenitor cells using a combination of recombinant SCF (100 ng/ml), IL-6 (50 ng/ml), and IL-10 (10 ng/ml) (R & D) as described previously (38), and were primed for 3 d with IL-4 (10 ng/ml) before activation (39).

Cell stimulation:

BMMCs (1.5 × 10⁶ cells/ml) were stimulated with IL-33 (10 ng/ml), or were sensitized overnight with mouse IgE SPE-7 (0.2 μg/ml). Excess IgE was washed off the next day prior to DNP-HSA (1 μg/ml). In some conditions, COX-1 specific antagonists (SC560, 10 nM or FR122047 300 nM) were added. Both supernatant and cell pellets were collected for the measurement of eicosanoid production and COX-1 and COX-2 transcripts over time. In some experiments, supernatants were analyzed for eicosanoids by mass spectrometry at the University of California at San Diego Lipid Maps Core Facility (www.ucsd-lipidmaps.org/).

Quantitative PCR:

RNA was extracted from BMMC pellets using the RNeasy Plus Mini Kit (Qiagen, Germantown, MD). Levels of COX-1 and COX-2 transcription over time were measured by RT² SYBR Green qPCR Master mixes using primers designed specifically to COX-1 or COX-2 mRNA. Relative quantities of mRNA were determined by comparison with GAPDH.

Western blotting:

Total BMMC lysates were prepared by centrifuging cells into ice-cold cell lysis buffer (RIPA, 5 mM EDTA and 1× complete protease inhibitor cocktail^™ [Roche]), incubation on ice for 10 min with repeated vortexing followed by denaturing in boiling Laemmli sample buffer. Proteins were resolved by SDS-PAGE and transferred onto PVDF by wet transfer. Membranes were blocked in TBS, 0.05% Tween-20 containing 5% blotting-grade blocker. The following primary antibodies were used (Cell Signaling Technology) according to manufacturer’s recommended concentration: Phospho-p44/42 MAPK (Erk1/2)197G2, p44/42 MAPK (Erk1/2), Phospho-p38 MAPK (Thr180/Tyr182), p38 MAPK Antibody, Phospho-cPLA₂ and cPLA₂. Secondary antibodies used were HRP-conjugated goat anti-mouse IgG or HRP-conjugated goat anti-rabbit IgG (BioRad). Detection was done with Amersham ECL and Kodak X-ray film. Where required, membranes were stripped in Restore Western Blot Stripping Buffer (Thermo Scientific) and re-probed. For densitometric analysis, films were scanned and band intensities quantified using ImageJ.

ELISA analysis of lipids:

Cultured BMMCs supernatant stimulated with IL-33 or IgE were analyzed for PGD₂, TXB₂ and CysLTs with Cayman ELISA assays per the manufacturer’s protocol.

Mice:

All animal protocols were approved by Animal Care and Use Committee from Brigham and Women’s Hospital. Wild type C57BL/6 mice were purchased from Charles River. C57BL/6 mice lacking mPGES-1 (Ptges^−/− mice) were from Dr. Shizuo Akira (Osaka University, Japan) (40). Animals were housed for at least 1 week at the Smith Building before experiments were done. Ptges^−/− mice were primed with 6 intranasal doses of Df (3 μg) before lysine-aspirin (Lys-ASA) challenge (41). Animals receive two doses of SC560 (or vehicle control) intraperitoneally (10 mg/kg) at 48 h and 24 h prior to challenge with aerosolized Lys-ASA. Measurement of airway resistance (R_L), collection of BAL fluid, and measurements of PGD₂ and cysLT levels were carried out as before (41).

Statistics:

Statistical analysis was unpaired student t-test for 2 samples, one-way ANOVA for 3 or more samples, followed by appropriate multiple comparison tests, as appropriate. Data with irregular distribution or unequal variance were assessed by unpaired student t-test with Welch’s correction or by non-parametric tests, either Mann-Whitney for 2-sample analysis or Kruskal-Wallis for 3 or more samples, as appropriate. P values <0.05 were considered statistically significant.

Results

IL-33 induces the generations of both COX and 5-LO products by BMMCs

To determine whether MCs generated lipid mediators in response to stimulation with IL-33, we activated BMMCs with IL-33 over a range of concentrations (5 ng/ml - 40 ng/ml) for 30 min. Supernatants were analyzed using respective commercial ELISAs specific for the detections of PGD₂, TXB₂ (a stable metabolite of TXA₂), and cysLTs. IL-33-stimulated BMMCs generated all three mediators at 30 min, with the optimal effect at 10 ng/ml (Fig. 1A). IL-33-induced PGD₂, TXA₂, and cysLT generations peaked at 0.5~ 3h (Fig. 1B). The concentration of PGD₂ in the media declined after 3 h, likely reflecting its conversion to stable metabolites. TXB₂ and cysLT levels remained stable after 3 h.

Figure 1.

IL-33 induced eicosanoid generation by BMMCs. A. Cells were stimulated with IL-33 at various concentrations. Media were collected at 3 hours and the indicated lipid mediators were analyzed by ELISA. B. Time course of IL-33-induced eicosanoid production at 10 ng/ml. Results are mean ± range pooled from 2 independent experiments for A and mean ± SEM from 10 independent experiments for B. ****p < 0.0001 across all time points using two-way ANOVA.

COX-1 function is necessary for IL-33-induced expression of COX-2 by BMMCs

We next sought to determine which COX isoforms were necessary for IL-33 to induce PGD₂ and TXA₂ production by BMMCs. BMMCs were stimulated with IL-33 in the absence or presence of SC560, a selective inhibitor of COX-1. SC560 markedly suppressed the IL-33-induced productions of PGD₂ and TXA₂ (Fig. 2A, left and center panels), approaching complete inhibition throughout the time course (Fig. 2B). Unexpectedly, SC560 also prevented the generation of cysLTs in response to IL-33 (Fig. 2A, right panel), with complete inhibition at the 30 min time point and 50-75% thereafter (Fig. 2B). In contrast, treatment of the cells with the selective COX-2 inhibitors valdecoxib and SC236 attenuated the generations of PGD₂ and TXA₂, but not of the cysLTs (Supplemental Fig. 1). Stimulation of the BMMCs with IL-33 did not change COX-1 expression levels, but induced a sharp, transient increase in the expression of COX-2 mRNA at 0.5 h, decreasing to baseline levels by 6 h (Supplemental Fig. 2A). IL-33-induced upregulation of COX-2 mRNA was completely blocked by SC560 (Supplemental Fig. 2A), but not by SC236 (Supplemental Fig. 2D). To verify that the effects of SC560 were not off-target, we performed similar studies with a second COX-1 selective inhibitor, FR122047, and studied the effects of small interfering RNA knockdown (KD) of COX-1 on both LTC₄ production and inducible COX-2 expression in response to IL-33. Both FR122047 and COX-1 knockdown (~70% efficiency, Supplemental Fig. 3B) resulted in diminished IL-33-mediated productions of PGD₂, TXA₂, and LTC₄ (Supplemental Fig. 3A, 3C), and expression of COX-2 mRNA (Supplemental Fig. 2B, 2C).

Figure 2.

Effect of COX-1 blockade on IL-33-induced eicosanoid production. BMMCs were stimulated with IL-33 (10 ng/ml) in the presence of the COX-1 antagonist SC560 (10 nM) or vehicle control. A. The levels of PGD₂, TXB₂ and cysLTs were measured in the culture supernatant harvested at the indicated time points. B. The net percent inhibition (subtracting the quantities present in the mock control) for each experiment was calculated at each time point. Data are mean ± SEM from 5 independent experiments. ****p < 0.0001, **p < 0.01 comparing SC560-treated samples to controls across all time points using two-way ANOVA.

To determine whether the effects of COX-1 blockade were stimulus-specific, we treated passively sensitized BMMCs with SC560 before challenge with specific allergen. SC560 suppressed the FcεRI-mediated productions of both PGD₂ and TXA₂ by BMMCs (Fig. 3, left and middle panels). However, in sharp contrast to its effects on IL-33-induced activation, it did not alter the FcεRI-mediated production of LTC₄ (Fig. 3, right panel) and did not alter the induced expression of COX-2 in response to this stimulus (Supplemental Fig. 4).

Figure 3.

Effect of COX-1 inhibition on IgE-induced eicosanoid generation. BMMCs were sensitized with IgE (SPE-7 0.2 μg/ml) overnight prior to antigen stimulation (1 μg/ml), in the presence or absence of SC560. Eicosanoids were measured in the culture supernatant harvested at the indicated time points. Results are mean ± SEM from 3 independent experiments. ****p < 0.0001 comparing SC560-treated samples to controls across all time points using two-way ANOVA.

Exogenous PGH₂ reverses COX-1-mediated inhibition of IL-33-induced mast cell activation responses and induces eicosanoid formation

COX-1 and COX-2 catalyze the conversion of AA to PGH₂, the immediate precursor of all bioactive prostaglandins. To determine whether the inhibition of IL-33-driven mast cell activation by SC560 was due to a loss of COX-1 catalytic activity and not to an off-target effect of the drug, we added PGH₂ (the product of COX-mediated AA metabolism) to some samples of BMMCs stimulated with IL-33 in the absence or presence of SC560. We also treated some samples of cells with PGH₂ in the absence of other stimuli or inhibitors. Exogenous PGH₂ restored the productions of both PGD₂ (Fig. 4A) and TXA₂ (Fig. 4B) by SC560-treated BMMCs stimulated with IL-33. PGH₂ also restored the IL-33-mediated generation of cysLTs in a dose-dependent manner (Fig. 4C), and also elicited the production of all three lipids in the absence of IL-33 (Fig. 4A-C). Exogenous PGH₂ restored the IL-33-induced upregulation of COX-2 mRNA expression that was inhibited by SC560, and induced some COX-2 expression in the absence of IL-33 (Supplemental Fig. 5). In contrast to PGH₂, neither PGD₂ nor the TP receptor selective agonist U46619 restored IL-33-induced COX-2 expression nor production of cysLTs in cells treated with SC560 (not shown).

Figure 4.

Effect of exogenous PGH₂ on eicosanoid production alone or in the presence of IL-33 with COX-1 blockade. BMMCs were stimulated with IL-33 (10 ng/ml) with or without SC560 (10 nM). PGH₂ was added to some samples at the indicated concentrations. A. PGD₂, B. TXA₂ and C. cysLTs were measured in the culture supernatant harvested at the indicated time points. Results are mean of duplicates from a single representative experiment, repeated twice with the same trends.

COX-1 inhibition selectively blocks IL-33-mediated AA release and formation of all AA-derived eicosanoids, but not release of ω-3 fatty acids

To identify the control points at which COX-1-derived PGH₂ controls IL-33-induced eicosanoid formation by MCs, and to verify the surprising ELISA results, we analyzed a broad spectrum of lipid mediators generated by BMMCs stimulated with IL-33 in the absence or presence of SC560 using a mass spectrometry-driven lipidomic approach. Consistent with the ELISA data, supernatants from the IL-33-actived BMMCs contained substantial quantities of PGD₂ and TXB₂ (Fig. 5A), total 5-LO products, LTB₄, and LTC₄ (reflected by the stable conversion product LTE₄, Fig. 5B) at 30 min. At 24 h, the levels of PGD₂ had declined, while the corresponding levels of the PGD₂ metabolite PGJ₂ had increased (not shown), and the levels of TXB₂ and LTE₄ remained constant. The total quantities of both COX and 5-LO pathway products were markedly reduced in the samples from BMMCs that were stimulated with IL-33 in the presence of SC560 compared with controls (Fig. 5A, 5B). Stimulation with IL-33 induced the rapid (30 min) release of AA (Fig. 5C, left panel), as well as the ω-3 fatty acids EPA (Fig. 5C, center) and (to a lesser extent) DHA (Fig. 5C, right). Treatment of the cells with SC560 markedly suppressed the release of AA, but not those of EPA or DHA (Fig. 5C).

Figure 5.

Lipidomic profiling of IL-33-stimulated BMMCs. BMMCs (1.5 × 10⁶ cells/ml) were stimulated with IL-33 (10 ng/ml) or medium control in the absence or presence of SC560 (10 nM). Culture supernatants obtained at 0.5 and 24 h were analysed by mass spectrometry. A. Effects of COX-1 blockade with SC560 on IL-33-induced PGD₂ and TXA₂ (reflected by the stable metabolite TXB₂). B. Effects of COX-1 blockade on total 5-LO activation-derived products (consisting of LTB₄, LTE₄, 14,15-LTD₄, 5-HETE, 5-HEPE, and 5-HETrE), LTB₄, and cysLTs (reflected by the stable metabolite LTE₄). C. Effect of COX-1 blockade on IL-33-driven arachidonic acid, EPA and DHA release by BMMCs. Results are mean ± range from biological replicate samples.

IL-33-induced activation of ERK, but not p38, depends on COX-1-derived PGH₂

cPLA₂, the major enzyme responsible for AA release in MCs, requires phosphorylation of p38 and/or ERK MAPKs for its translocation and activation (3, 42). p38 and ERK can also facilitate the transcriptional induction of COX-2 (43, 44). Since COX-1 blockade altered both the release of AA and the induction of COX-2 mRNA expression by IL-33-stimulated BMMCs, we speculated that COX-1-derived PGH₂ and/or conversion products may regulate MAPK activation (and consequently cPLA₂ activation) downstream of IL-33 stimulation. Stimulation of BMMCs with IL-33 induced modest phosphorylation of cPLA₂ at 30 min (52 ± 15% over baseline, n = 5) (Fig. 6A), corresponding to the peak of mediator generation. IL-33 also induced strong phosphorylation of ERK at 30 min that decline at 3 h and returned to baseline at 6 h (Fig. 6B). Similar kinetics were observed for phosphorylation of MEK (not shown), as well as p38 (Fig. 6C). Treatment with SC560 eliminated the transient IL-33-induced phosphorylation of cPLA₂ (Fig. 6A), MEK (not shown) and ERK (Fig. 6B) in all five experiments performed, but not p38 (Fig. 6C). The addition of PGH₂ to the IL-33-stimulated cells treated with SC560 restored ERK phosphorylation, and induced phosphorylations of ERK and cPLA₂ even in otherwise unstimulated cells (Fig. 6D. Supplemental Fig. 6). In contrast, PGH₂ tended to decrease the phosphorylation of p38 (Supplemental Fig. 6). Treatment of the cells with the MEK inhibitor U0126 prevented the productions of cysLTs, PGD₂, and TXA₂ stimulated with IL-33 (Fig. 6E). Similar results were obtained with the ERK inhibitor FR 180204 (10 μM) (Supplemental Fig. 7A). In contrast, the p38 inhibitor SKF-86002 blocked IL-33-induced productions of PGD₂ and TXA₂, but had no effect on the production of LTC₄ (Supplemental Fig. 7B). Both U0126 and SKF-86002 blocked the IL-33-induced expression of COX-2 nearly completely (not shown). IL-33 stimulated human cord blood-derived MCs also exhibited strong ERK phosphorylation that was blocked by SC560, but did not show cysLT generation (not shown).

Figure 6.

Effect of COX-1 activity on IL-33-induced MAP kinase-dependent activation of cPLA₂. Western blots generated from BMMCs stimulated with IL-33 for the indicated time points in the absence or presence of SC560 were probed for phosphorylated and total cPLA₂ (A) ERK (B), and p38 (C). Blots from representative experiments are displayed (left), and quantitative densitometry (corrected for total proteins, mean ±SEM) are shown (right) from 3–5 experiments. D. Effect of exogenous PGH₂ on ERK activation in unstimulated or IL-33-stimulated BMMCs in the absence or presence of COX-1 blockade. Results are representative of two experiments with similar results. E. Effect of ERK inhibition with U0126 (1 μM) on IL-33-driven eicosanoid production by BMMCs. Results are mean ± SEM from 3 independent experiments. ****p < 0.0001, **p < 0.01 across all time points by two-way ANOVA.

COX-1 inhibition blocks eicosanoid production in response to Lys-ASA challenge in a mouse model of AERD

To determine whether IL-33-driven MC activation and eicosanoid formation in vivo depend on COX-1, we studied the effect of pharmacologic inhibition of COX-1 in AERD-like Ptges^−/− mice after the establishment of allergic airway inflammation by repetitive intranasal priming by Df (41). We administered SC560 or vehicle control intraperitoneally daily for the 48 h before challenge, then challenged the mice with inhaled lysine aspirin (Lys-ASA). As expected, vehicle control-treated mice displayed sharp increases in airway resistance (R_L) (Fig. 7A) when challenged with Lys-ASA, accompanied by sharply increased BAL fluid levels of both PGD₂ and cysLTs (Fig. 7B). In contrast, SC560-treated mice displayed dramatically reduced changes in R_L, cysLTs, and PGD₂ when challenged with Lys-ASA. To determine whether the absence of cell-intrinsic mPGES-1 altered the response of BMMCs to IL-33, we activated WT and Ptges^−/− BMMCs with IL-33. Ptges^−/− BMMCs generated similar quantities of LTC₄, PGD₂, and TXA₂ in response to IL-33 as did WT controls (not shown). SC560 blocked the production of all three mediators by the Ptges^−/− BMMCs (Supplemental Fig. 8).

Figure 7.

Effect of COX-1 inhibition in vivo on Lys-ASA-induced changes in airway physiology and MC-dependent eicosanoid generation in AERD-like mice. A. Df-primed Ptges^−/− mice were challenged with Lys-ASA by inhalation. R_L was monitored continuously for 45 min and the peak change from baseline was recorded for each animal. The indicated groups received intraperitoneal doses of SC560 or vehicle control 48 and 24 h before challenge. B. Levels of PGD₂ and cysLTs detected in BAL fluid collected from the same mice 45 min after the challenge. Results are mean ± SEM from 3–5 mice per group in one of two experiments performed with similar results.

Discussion:

MCs are strategically located at tissue interfaces with the external environment and the microvasculature to serve sentinel functions in both adaptive and innate immunity. Their role in IgE-dependent type 1 hypersensitivity, a process involved an all allergic diseases, is undisputed. This role includes mobilization of AA for conversion to potent eicosanoids in response to allergen exposure. LTC₄ and its metabolites LTD₄ and LTE₄ are powerful smooth muscle constrictors that also elicit rapid increases in vascular permeability and mucous secretion through three different receptors (45-48). LTB₄ is a potent chemoattractant for neutrophils and effector memory T cells (7, 12). PGD₂ can drive chemotaxis of eosinophils, basophils and Th2 cells (49), and can also directly induce Th2 cytokine production from Th2 cells (50) and innate group 2 lymphocytes (51). TXA₂ activates platelets and endothelial cells, facilitating both thrombosis and leukocyte recruitment (11, 52). The clinical efficacy of 5-LO inhibitors (53), CysLT receptor antagonists (54), and COX inhibitors (55) all validate the importance of AA metabolites in allergic disease and inflammation in general. MCs have recently been implicated strongly in various models of innate type 2 immunity initiated by endogenous cytokines (IL-33, TSLP, IL-25)(24), and IL-33 potently induces cytokine generation by MCs (IL-5, IL-13) without eliciting exocytosis of histamine (19). We undertook this study to determine whether IL-33-dependent MC activation elicits eicosanoid synthesis, and if so, whether this process is regulated differently from the calcium- and MAP kinase-dependent cPLA₂ activation reported for MC activation induced by the classical mechanism.

IL-33 signals through ST2, an IL-1 receptor family member. Whereas IgE-dependent PGD₂ and TXA₂ production by MCs occur dominantly through COX-1 (56, 57), signaling through IL-1 receptor family members can induce COX-2 expression in MCs and other cell types, resulting in sustained PG production (44). As expected, IL-33 induced the productions of both PGD₂ and TXA₂ by BMMCs, but with rapid kinetics that was consistent with the actions of constitutive COX-1 (Fig. 1). Two unexpected observations formed the basis of our subsequent inquiries. First, both pharmacologic blockade (using two different inhibitors) and targeted knockdown of COX-1 not only decreased the generations of PGD₂ and TXA₂ (Fig. 2) but also blocked the induced expression of COX-2 (Supplemental Figs. 2,3). In contrast, selective inhibition of COX-2 decreased PGD₂ and TXA₂ productions, but did not affect COX-2 expression (Supplemental Fig. 2). Second, and surprisingly, COX-1 activity (Fig. 2) (but not COX-2 activity (Supplemental Fig. 1)) was necessary to produce LTC₄ (as reflected by the effects of both COX-1 inhibitors tested on the levels of cysLTs) in response to IL-33), differing sharply from FcεRI-induced activation (Fig. 3). The fact that these functional responses were restored by exogenous administration of PGH₂ (Fig. 4) confirmed that the effects did not reflect off-target actions of SC560, and that PGH₂ derived from COX-1, but not that derived from COX-2, permitted at least one step critical to elicit both cysLT and COX-2-driven PGD₂ generation by MCs activated by IL-33, but not FcεRI cross-linkage. Moreover, the fact that neither PGD₂ itself nor the TP receptor agonist AU46619 mimicked the effect of PGH₂ implies the involvement of a minor PG, or a distinct role for PGH₂ itself in regulating intracellular responses to an extracellular signal.

Of the multiple PLA₂ enzymes capable of releasing AA from the sn-2 position of membrane phospholipids, cPLA₂ is indispensable for the productions of both LTC₄ and PGD₂ with classical MC activation (42). cPLA₂ translocates from the cytosol of resting cells to intracellular membranes with activation and carries out its enzymatic function after it is phosphorylated by ERK and/or p38 MAPKs (3), particularly in response to stimuli that also elicit calcium flux (2). Several lines of evidence in our study suggest that COX-1 activity is necessary to elicit ERK dependent cPLA₂ activity in response to IL-33. First, IL-33-induced AA release itself (and all downstream AA-derived products, including LTE₄, LTB₄, and combined 5-LO pathway products) were sharply reduced by treatment of BMMCs with SC560 (Fig. 5). In contrast, releases of ω-3 FAs (which are not the preferred substrates for cPLA₂) were minimally to negligibly affected by COX-1 inhibition. Second, the modest cPLA₂ phosphorylation and strong ERK phosphorylation induced by IL-33 were eliminated by treatment of the cells with SC560, whereas p38 phosphorylation was unaffected (Fig. 6). SC560 also suppressed IL-33-induced ERK activation by human cord blood MCs. Third, both cPLA₂ and ERK activation were restored by the administration of exogenous PGH₂ (Supplemental Fig. 6), which was also able to induce ERK activation (Fig. 6D) and cPLA₂ phosphorylation (Supplemental Fig. 6A) on its own, consistent with its ability to induce cysLT production even without IL-33 (Fig. 4). It is likely that COX-1-derived PGH₂ provides a requisite signal that is essential for IL-33 to initiate AA release and the surge in both COX and 5-LO products characteristic of MC activation. Since most previous studies of eicosanoid production focused on FcεRI-dependent activation mechanisms, in which COX-1 is required for PGD₂ production but not needed to activate cPLA₂, this requirement for COX-1 in regulating ERK and cPLA₂ activity and LTC₄ production had not been recognized previously. Previous studies demonstrating IL-33-induced PGD₂ generation by MCs (58, 59) had neither examined cysLT production, nor recognized the proximal role of COX-1-derived PGH₂ in initiating eicosanoid synthesis detailed in our study.

Airway MC activation produces airflow obstruction that is largely attributable to the contractile effects of the cysLTs (60). Whereas allergen inhalation in sensitized hosts elicits FcεRI-dependent MC activation and accompanying releases of both cysLTs and PGD₂ (61, 62), challenges of individuals with AERD with COX-1 inhibitors (but not COX-2 inhibitors) result in MC activation by an idiosyncratic mechanism that requires depletion of COX-1-derived PGE₂ (63). Notably, COX-1 inhibitors also elicit a subsequent state of “desensitization” in which subsequent challenges with aspirin (or any other COX-1-active drug) fail to elicit characteristic MC activation and cysLT release (36). In Ptges^−/− mice, both aspirin and ketorolac (but not the COX-2-selective inhibitor celecoxib) provoke AERD-like reactions (41) that depend on release of IL-33, which is absolutely required for MC activation in this model (1). Although the time frame of the aspirin challenge protocol used in our study does not permit serial invasive assessment of lung function in response to sequentially administered agents, it seems likely that SC560 also elicits a reaction (caused by PGE₂ depletion) followed by desensitization. This desensitization may be due, at least in part, to “stabilization” of MCs through inhibition of cell-intrinsic COX-1 and impairment of their responses to endogenous IL-33 (Fig. 7). Since IL-33-driven MC activation may be central not only to AERD, but also to severe asthma, chronic sinusitis and other diseases, our findings have potential implications for the targeted use of COX inhibitors in other pathologic conditions. Our findings may also help to explain the therapeutic effects of desensitization to aspirin in AERD, which prevents incremental mediator release in response to re-challenge (36).

Key Messages.

• IL-33 activates induces mast cells to release arachidonic acid and generate eicosanoids involved in allergic disease.

• In contrast to classical, IgE-dependent mast cell activation, IL-33 requires COX-1 activity to activate cPLA₂ and induce arachidonic acid release

• COX-1 inhibition may limit IL-33-driven mast cell activation in AERD and could explain part of the mechanism of aspirin desensitization

Abbreviations

AA

arachidonic acid

AERD

aspirin exacerbated respiratory disease

BAL

bronchoalveolar lavage

COX

cyclooxygenase

cPLA₂

cytosolic phospholipase A₂

CysLT

cysteinyl leukotriene

ERK

extracellular signal regulated kinase

LT

leukotriene

MAPK

mitogen activated protein kinase

MC

mast cell

MEK

mitogen activated kinase kinase

R_L

lung resistance

TX

thromboxane

5-LO

5-lipoxygenase