Human mast cells express multiple EP receptors for prostaglandin E₂ that differentially modulate activation responses

Mast cells (MCs) initiate inflammatory responses to infectious organisms and allergens. In allergic diseases, MCs are activated by cross-linkage of their high-affinity Fc receptors for immunoglobulin (Ig) E (FcϵRI), releasing preformed proteases and biogenic amines, and generating leukotriene C₄ (LTC₄), the parent molecule of the cysteinyl leukotrienes (cysLTs), and prostaglandin D₂ (PGD₂^1-3 ). Aspirin-exacerbated respiratory disease (AERD) is characterized by nasal polyposis, asthma, and MC activation in response to challenge with nonselective inhibitors of cyclooxygenase (COX).^4,5  MC activation follows allergen challenge in the lungs or nose of susceptible individuals, inducing tissue swelling, bronchconstriction, and vascular leakage (early-phase response [EPR]).⁶  In vitro, FcϵRI-induced MC activation also initiates transcription through the actions of nuclear factor κB (NF-κB), nuclear factor of activated T cells (NF-AT) and activator protein-1 (AP-1) transcription factors, resulting in sustained cytokine and chemokine generation.⁷  MC activation recruits leukocytes to allergen-challenged tissues, resulting in sustained swelling and inflammation (late-phase response [LPR]) in a significant proportion of susceptible individuals.⁸  Mediator generation by MCs stimulated through toll-like receptors (TLRs) also plays an important role in protective innate immunity.^9-11  Thus, whether activated through stimulation of FcϵRI, idiosyncratic mechanisms, or pattern recognition receptors, MCs provide eicosanoids and inducible cytokines for inflammatory responses in vivo, benefiting the host in protective immunity, but also potentially inducing exacerbations of allergic diseases.

PGE₂, a functionally versatile eicosanoid, acts at 4 divergent G protein-coupled receptors (GPCRs), called the EP₁, EP₂, EP₃, and EP₄ receptors (reviewed in Kobayashi and Narumiya¹² ). PGE₂ can amplify inflammatory gene expression and promote tissue pathology in colon cancer¹³  and in mouse models of arthritis.¹⁴  In contrast, PGE₂ strongly suppresses allergic respiratory mucosal inflammation. Inhalation of PGE₂ before allergen challenge prevents both EPR and LPR in subjects with asthma,¹⁵  and decreases the levels of PGD₂ that are detected in the bronchoalveolar lavage fluid after allergen challenge.¹⁶  In AERD, PGE₂ inhalation prevents bronchconstriction induced by challenge with nonselective COX inhibitors and abrogates the characteristic rise in urinary levels of LTE₄, the end product of the cysLTs.^5,17,18  Collectively, these observations suggest that PGE₂ inhibits MC activation in the respiratory tract. Although PGE₂ reportedly suppresses mediator release by some MC subtypes in vitro by raising intracellular levels of cyclic adenosine monophosphate (cAMP),^19,20  it also enhances mediator release from mouse MCs.^21-23  The mechanisms and EP receptor subtypes responsible for PGE₂-mediated inhibition of MC activation are incompletely understood.

We explored the mechanisms and receptors by which PGE₂ modulates activation responses of cord blood-derived human MCs (hMCs). hMCs express EP₂,EP₃, and EP₄ receptors, and respond to PGE₂ and receptor-selective analogs with anticipated biochemical signatures and signaling events. In contrast to the results of earlier studies of human lung MCs, PGE₂ does not suppress FcϵRI-dependent exocytosis by hMCs. Rather, PGE₂ causes exocytosis on its own when hMCs are primed with interleukin-4 (IL-4), a function reflecting EP₃ receptors. Moreover, EP₃ receptor-dependent stimulation of hMCs induces phosphorylaton of extracellular signal-regulated kinases (ERKs) 1 and 2 and PGD₂ generation, and also causes the production of LTC₄ when protein kinase A (PKA) is inhibited. However, PGE₂ interferes with both LTC₄ and PGD₂ generation by hMCs stimulated by FcϵRI cross-linkage, an effect mediated by EP₂ receptors, and strikingly inhibits FcϵRI-mediated production of tumor necrosis factor-α (TNF-α). The latter event reflects complementary signaling through PKA- and ERK-mediated signaling pathways, in turn reflecting combinatorial contributions from EP receptors leading to the expression of the transcriptional repressor protein inducible cAMP early repressor (ICER).^24,25  Thus while EP₃ receptor-dependent signaling causes or potentiates MC mediator release, EP₂ receptors inhibit FcϵRI-initiated mediator generation and may limit tissue pathology in the LPR and other circumstances where MC activation induces inflammation.

PGE₂, PGD₂, LTD₄, the EP₂/EP₃ dual receptor antagonist AH6809 (EC₅₀ = 50 μM) and the EP₂ receptor selective agonist butaprost (Ki ∼10 μM) were purchased from Cayman Chemical (Ann Arbor, MI). EP₁, EP₂, EP₃, and EP₄ receptor-selective agonists (DI-004, AE1-259-01, AE-248, and AE-329, respectively) and an EP₄ receptor-selective antagonist (AE-208) were obtained from Ono Pharmaceuticals (Osaka, Japan). These agonists are potent (Ki ∼3 nM, 150 nM, 7.5 nM, and 10 nM for AE1-259-01, DI-004, AE-248, and AE-329, respectively) and reportedly show little to no cross-reactivity on cloned EP receptors even at doses 1000 times their Ki values.^26,27  In primary cells, agonist doses between 0.1 and 10 μM are reportedly sufficient to elicit maximal functional and signaling responses.^28,29  Each selective reagent was dissolved in dimethyl sulfoxide (DMSO; Sigma, St Louis, MO). Pertussis toxin (PTX) was purchased from Sigma. The selective agonist of exchange protein activated by cAMP (Epac), 8-pCPT-2′-O-Me-cAMP, and an inhibitor of glycogen synthase kinase 3 (GSK-3) (GSK-3 inhibitor II) were purchased from Calbiochem (San Diego, CA). A second GSK-3 inhibitor, SB216763, was obtained from Sigma. Inhibitors of PKA (H89; Sigma) and mitogen-activated protein kinase kinase (MEK)/ERK (UO126; Promega, Madison, WI) were used at their recommended concentrations (10 μM and 5 μg/mL, respectively).

hMCs were derived from cord-blood mononuclear cells cultured in the presence of stem-cell factor (SCF; 100 ng/mL), IL-6 (50 ng/mL), and IL-10 (10 ng/mL) (all from R&D Systems, Minneapolis, MN), as previously described,³⁰  and studied when they reached more than 95% purity based on staining with toluidine blue (6-9 weeks). For FcγRI-dependent exocytosis, cytokine generation, and eicosanoid production, hMCs were transferred to new medium containing SCF and IL-4 (10 ng/mL, R&D Systems) for 5 days before activation.^31,32  Because IL-4 did not alter the expression level of any EP receptor and did not change calcium fluxes in response to PGE₂, signaling studies were performed with unprimed hMCs. Each donor's cells were used for 1 experiment per assay.

Expression of EP receptor proteins was assessed on fixed, permeabilized hMCs as described.³³  Polyclonal antibodies (Abs) against each EP receptor (Cayman Chemical) and a monoclonal Ab against Kit (Pharmingen, San Diego, CA) were used at 2 μg/sample. The EP₃ receptor Ab was raised against a peptide sequence (amino acids 308-327) common to all isoforms of this receptor.

Total RNA was extracted from hMCs at 6 to 8 weeks of culture with TRI Reagent (Molecular Research, Cincinnati, OH). RNA samples were primed with oligo(dT) and reverse transcribed using an RT kit (Clontech, Palo Alto, CA). Primer sequences for the amplification of the EP₁, EP₂, and EP₄ receptors³⁴  are as follows, reading from the 5′ to the 3′ direction: EP₁, sense ATCTGCTGGAGGCCAATGCTGGTGT, antisense TCGTTGGGCCTCTGGTTGTGCTT; EP₂, sense CCTGGCCGTGCTGCCTGTCATCTAT, antisense CCATGGACACCCTTTCCCTCTCCT; and EP₄, sense TTTGCAGGCCATCCGAATTGCTTCT, antisense CCTGCCTCCAAGGCCATTTTCACTGG. Because the human EP₃ receptor gene gives rise to several splice variants,^35,36  we designed primers for the selective amplification of each. Splice variants I to IV were amplified using a common sense strand primer (CTGAACCAGATCTTGGATCC) with the following isoform-specific antisense primers: EP₃-I, TCACCATCATAAGCTTATAC; EP₃-II, TACGAATGGCAGACTCAACA; and EP₃-III and EP₃-IV, TCATGGAGCTTCCAGTGATG. Specific primer pairs were used to amplify EP₃-V (sense CAGAGGTTTCCCAGAGAGGAAGGCGTGG, antisense TCCTGGACCTGC CTCCATGCATGACAAA) and EP₃-VI (sense GAGATGGGGCCTGATGGAAG, antisense TCATGGAGCTTCCAGTGATG). Primers for human glyceraldehyde-3-phosphate dehydrogenase (Clontech) were run as positive controls. PCR was performed in a Perkin-Elmer Thermal Cycler with 0.4 units of Taq polymerase (Perkin-Elmer, Shelton, CT) for 35 cycles (94°C × 1 minute, 94°C × 30 seconds, 56°C × 2 minutes, and 72°C × 4 minutes). Genomic DNA and non-reverse-transcribed RNA were used as positive and negative control templates, respectively. The PCR products were resolved on ethidium bromide-stained 1.5% agarose gels. The analyses for each receptor were repeated 8 times with RNA harvested from the cells of different donors.

Changes in the concentration of cytostolic free Ca⁺⁺ were assessed by two methods. First, a fluorescence imaging plate reader (FLIPR)-based calcium-imaging assay was used to determine the optimal dosing range for PGE₂-mediated calcium flux. hMCs (8 weeks old) were washed into Hanks balanced salt solution (HBSS) containing 1% bovine serum albumin (BSA), 20 mM HEPES, and 2.5 mM probenecid at a density of 1.3 × 10⁶ cells/mL and loaded with 2 μM Fluo-4 (Invitrogen, Carlsbad, CA) according to the manufacturer's protocol. The plates were then placed into a FLIPR (Molecular Devices) to monitor cell fluorescence (_ex = 488 nm, _EM = 540 nm) before and after the addition of various agonists. Triplicate samples of cells were stimulated with PGE₂ (0.01-10 μM) or similar concentrations of PGD₂ or LTD₄ as negative and positive controls. The susceptibility of PGE₂-mediated calcium flux to interference by PTX (4- to 12-hour treatment with 100 ng/mL) and the ability of EP receptor-selective agonists to elicit this response were tested with Fura-2 am (Molecular Probes, Eugene, OR)-loaded hMCs using a fluorescence spectrophotometer (Hitachi F-4500; Hitachi, Greenville, SC) using excitation at 340 and 380 nm to monitor cytosolic free Ca⁺⁺ as previously described.³⁰ 

IL-4-primed hMCs were passively sensitized with human myeloma IgE (2 μg/mL; Chemicon, Tecaluma, CA) overnight on the fourth day of priming. The cells were washed and resuspended in fresh medium at a concentration of 1 × 10⁶ cells/mL, except for the samples used for analysis of exocytosis, which were suspended at a 10 × 10⁶/mL. For studies of exocytosis, samples of 2 × 10⁶ hMCs were challenged with anti-IgE (1 μg/mL; Calbiochem) or medium alone. In some experiments, PGE₂, AE-248, AE-329, AE1-259-01, DMSO, or butaprost were added just before activation. The cells were maintained in the presence of SCF (100 ng/mL) throughout to promote optimal viability. Activation was stopped on ice. Supernatants were separated from the pellets by centrifugation at 200g in an Eppendorf microcentrifuge at 4°C. The content of β-hexosaminidase (β-hex) was measured by spectrophotometric analysis as described previously,³⁷  and the percentage of release values were calculated. Data were expressed as mean release from duplicate readings.

For eicosanoid production, triplicate samples of 1 × 10⁵ cells were stimulated with various doses of PGE₂ or its mimics (0.01-10 μM) or DMSO (1:1000), with and without anti-IgE (1 μg/mL) in the wells of 96-well flat-bottom plates. Supernatants were collected at 30 minutes and stored at -20° C until further analysis with specific enzyme-linked immunosorbent assays (ELISAs) to detect cysLTs (LTC₄, LTD₄, and LTE₄; Cayman Chemical) and PGD₂ (Amersham, Arlington Heights, IL). In some experiments, the selective inhibitor of PKA, H89 (10 μM), was added to the cells 30 minutes before activation. The assay for PGD₂ did not detect PGE₂ at concentrations as high as 10 μM. For cytokine generation, sensitized, primed hMCs were challenged with Staphylococcus aureus peptidoglycan (PGN) (10 μg/mL; Sigma), anti-IgE (1 μg/mL), or medium alone in the presence or absence of agonists or controls as described for eicosanoids. Supernatants were collected at 6 hours and frozen at -70°C until further analysis with ELISAs for TNF-α and IL-5 (both from eBiosiences, San Diego, CA). The data for each experiment were tabulated as the mean values from triplicate samples.

Triplicate samples of 2 × 10⁵ hMCs were stimulated with various agonists or antagonists for 10 minutes as described previously.²⁵  cAMP was measured with a commercial Biotrack cAMP ELISA kit (Amersham). The mean cAMP values for triplicate samples of cells stimulated with each agonist were compared with values from unstimulated cells, and the data expressed as absolute values.

Samples of 2 × 10⁵ hMCs were stimulated for various intervals with each agonist. Reactions were stopped on ice and the cells were lysed in a buffer containing 1% sodium dodecyl sulfate (SDS), 1% Triton X-100, 1 mM EDTA, 1 mM EGTA, 0.2 mM PMSF, 5 μg/mL leupeptin, and 1 μg/mL pepstatin in 10 mM Tris (pH 8.0). Western blotting was performed as previously described.²⁵  Dilutions (1:2000) of primary antibodies specific for the active, phosphorylated forms of ERK-1/ERK-2, c-Jun NH₂-terminal kinase (JNK), and p38 mitogen-activated protein kinase (MAPK) were used. The same blots were stripped and probed again with rabbit polyclonal antibodies that detect total ERK-1/ERK-2, p38, and JNK at dilutions of 1:2,000. All anti-MAPK antibodies were purchased from Cell Signaling Technologies (Beverly, MA). ICER was detected with a rabbit polyclonal antiserum that recognizes all forms of cAMP response element modulator (CREM), including ICER (provided by Dr Carlos Molina, University of New Jersey Medical School) at a concentration of 1:5000.³⁸ 

Samples of 5 to 10 × 10⁶ primed, sensitized hMCs were stimulated for 2 hours with anti-IgE or medium alone, all in the presence of SCF at a constant concentration of 100 ng/mL, with or without PGE₂. The expression of TNF-α mRNA was determined with real-time PCR performed on ABI PRISM 7700 Sequence Detection System (Applied Biosystems, Foster City, CA) as previously described.²⁵  Primers and probes for the amplification of TNF-α and human β₂ microglobulin and the corresponding VIC dye were purchased from Applied Biosystems.

Data are expressed as means ± SEM from at least 3 experiments except where otherwise indicated. Because quantities of eicosanoids and cytokines generated by hMCs from different donors varied widely, the data for some analyses were converted to percentages of the control (the samples not treated with PGE₂ or EP receptor-selective agonists). Differences between treatment groups were determined with the Student t test, with P values less than .05 considered significant. A Q test was used to eliminate outlying experiments.

We used RT-PCR to determine the profile of EP receptor mRNAs expressed by hMCs. Bands corresponding to the predicted sizes of the EP₂ and EP₄ receptors were detected in mRNA samples from all 8 donors tested (as shown for 1 donor, Figure 1A). All 8 donors showed both the EP₃-II splice variant and at least one Gi-linked EP₃ variant (5 showed EP₃-I, 4 showed the EP₃-IV, 4 showed EP₃-VI, and none showed EP₃-V transcript). EP₁ receptor transcripts were detected in the RNA from the cells of only 1 donor (not shown). All of the primer sets yielded bands of the expected sizes from the positive control DNA template, and no products were detected from the negative control template (Figure 1). The EP₂, EP₃, and EP₄ receptor proteins were all detected by flow cytometry (n = 3 as shown for 1 experiment; Figure 1B), while the EP₁ receptor protein was not detected in the cells from any of these donors. IL-4 priming of the hMCs did not alter the profile or the level of EP receptor mRNA or proteins expressed (not shown). Virtually all of the cells showed high-level cytofluorographic expression of Kit (Figure 1B), confirming the identity of the EP receptor-positive cells as hMCs.

In a FLIPR assay, PGE₂ (0.1-10 μM), but not PGD₂, induced a calcium flux (an expected signature of both EP₁ and EP₃ receptors; Figure 2A). Subsequently, fluorescence spectrophotometry was used to measure calcium flux in hMCs following treatment with or without PTX. The cells from 8 donors were tested in this assay, and 3 exhibited calcium flux to PGE₂ (1-10 μM); this flux was completely blocked by treatment of the cells with PTX (Figure 2B). LTD₄-mediated calcium flux occurred in every donor and was unaffected by PTX. Of the receptor-selective reagents, only the EP₃ receptor-selective agonist AE-248 (10 μM) stimulated calcium flux, and was much weaker than PGE₂ at the same concentration (n = 2, data not shown). Priming of the cells with IL-4 did not change the strength of PGE₂-induced calcium flux.

Stimulation of hMCs with PGE₂ (1 μM) rapidly (5 minutes) induced phosphorylation of ERK-2 and ERK-1, with ERK-2 being expressed more strongly (Figure 2C). ERK was also phosphorylated in response to stimulation of the hMCs with the EP₃ receptor-selective agonist AE-248 at 1 μM, but not with the EP₁, EP₂, and EP₄ receptor-selective agonists at the same concentration (n = 4 as shown for a single experiment; Figure 2C). ERK phosphorylation in response to stimulation with both PGE₂ (Figure 2C) and AE-248 (not shown) was completely blocked by pretreatment of the cells with PTX or with the MEK inhibitor UO126 (5 μg/mL), but was unaffected by treatment with H89 (10 μM, n = 3, not shown). p38 and JNK were constitutively phosphorylated and not affected by PGE₂ or its analogs.

Stimulation of the cells with PGE₂ dose-dependently induced the accumulation of cAMP for every donor tested (n = 3 as shown for 1 representative experiment; Figure 3A). The EP₂ receptor-selective agonist AE1-259-01 modestly stimulated cAMP accumulation at concentrations of 0.1, 1, and 10 μM (n = 5, Figure 3B). A second EP₂ receptor agonist, butaprost, weakly stimulated cAMP accumulation, reaching significance at 10 μM (210 ± 6 vs 158 ± 10 fmol for buffer treated cells; P = .03, n = 3, data not shown). The EP₄ receptor-selective agonist AE-329 also induced cAMP accumulation at 1 and 10 μM (Figure 3B). cAMP values in cell samples stimulated with AE-248 at 10 μM tended to be higher than controls, but did not reach significance (P = .08 relative to unstimulated controls, n = 5). The EP₁ receptor-selective agonist D1-004 was inactive (n = 3, not shown). Even at the highest doses tested, none of the selective agonists stimulated as much cAMP accumulation as PGE₂, and various combinations of the selective agonists did not additively stimulate cAMP over a wide concentration range (0.01-1 μM, n = 3, not shown). The EP₄ receptor-selective antagonist AE-208 (10 μM) abrogated cAMP accumulation in response to 10 μM AE-329, but its effect on cAMP accumulation occurring in response to PGE₂ (1 μM) was not significant (P = .054, n = 4, Figure 3C). The EP₂/EP₃ receptor-selective antagonist AH6809 (50 μM) blocked cAMP accumulation in response to 1 and 10 μM PGE₂ by 69% ± 6% and 42% ± 16%, respectively, but failed to block cAMP accumulation in response to AE-329 (mean ± ½ range, n = 2, data not shown). PTX pretreatment failed to alter cAMP accumulation in response to any agonist (not shown).

PGE₂ at doses as high as 10 μM failed to attenuate FcϵRI-dependent or PGN-dependent exocytosis of β-hex by IL-4-primed hMCs and tended to induce some exocytosis by itself under these conditions (Figure 4A). To determine whether this unanticipated response depended on IL-4, we tested PGE₂-mediated exocytosis in both IL-4-primed and unprimed cells. PGE₂ induced exocytosis only in the IL-4-primed cells, and this response was blocked by pretreatment of the cells with PTX (Figure 4B). Stimulation of the primed hMCs with the EP₃ receptor-selective agonist AE-248 caused a small amount of exocytosis (4.3% ± 1% net release, n = 3, not shown), whereas none of the other agonists induced this response.

Primed hMCs generated both LTC₄ (as determined by the detection of cysLTs) and PGD₂ when stimulated by FcϵRI cross-linkage.³⁴  Treatment of the cells with PGE₂ inhibited the production of both PGD₂ and LTC₄ in a concentration-dependent manner (n = 2, as shown for 1 dose-response experiment; Figure 5A). PGE₂-induced suppression of both eicosanoids was reversed completely by treatment with AH6809, but not with AE-208 (n = 1, not shown). Of the selective agonists, only AE1-259-01 (Figure 5B) tended to block LTC₄ and PGD₂ production (n = 5, Figure 5B). The inhibition of LTC₄ generation, but not that of PGD₂ production, was reversed by the PKA inhibitor H89 (n = 2, data not shown). When used by itself, PGE₂ (0.1-10 μM) induced the production of PGD₂ (15, 18, and 22 ng/10⁶ hMCs in response to 0.1, 1 and 10 μM PGE₂ in a dose-response experiment, data not shown). Both PGE₂ and AE-248 (0.1-10 μM each) also induced LTC₄ production by the primed hMCs when the cells were pretreated with H89 (as shown for the 10 μM concentration; Figure 5C). PGE₂-mediated eicosanoid generation was completely blocked by pretreatment of the cells with PTX (n = 2, not shown). The selective agonists of the EP₁,EP₂, and EP₄ receptors failed to induce eicosanoid generation by themselves, or when used in combination with H89, even at the highest doses tested (10 μM, n = 3, data not shown).

Primed hMCs generated IL-5 and TNF-α when stimulated with either anti-IgE or PGN. PGE₂ interfered with the production of each cytokine in a dose-dependent manner (as shown in a representative experiment; Figure 6A), in response to both anti-IgE and to PGN (Figure 6B). AH6809 (50 μM) and AE-208 (10 μM) reversed the inhibitory effect of PGE₂ on TNF-α generation by 24% and 50%, respectively (n = 1, not shown). AE1-259-01 inhibited TNF-α generation by hMCs in a dose-dependent manner, being statistically significant at 10 μM (n = 3; Figure 6C). Butaprost was also active for this function at 10 μM (n = 2, not shown). Both AE-248 and AE-329 also inhibited TNF-α production by hMCs (Figure 6C). H89 failed to reverse the inhibitory effect of any agonist on cytokine generation (n = 3, not shown). PGE₂ at 10 μM strongly inhibited the FcϵRI-induced steady-state expression of TNF-α mRNA, as determined by real-time PCR (n = 2); this effect was reproduced by both AE1-259-01 and AE-329 (as shown for 1 experiment; Figure 6D). Pretreatment of the cells with the selective Epac agonist 8-pCPT-2′-O-Me-cAMP (10 μM) modestly inhibited FcϵRI-mediated TNF-α generation (43% inhibition), but had no effect on IL-5 production, and 2 inhibitors of GSK-3 (GSK-3 inhibitor II (20 μM) and SB 216763 (20 μM) had no effect on TNF-α generation (14% and 12% inhibition, respectively; not shown).

At 0.1-10 μM, PGE₂ dose-dependently induced ICER expression (Figure 7A), peaking at 2 to 3 hours after stimulation (Figure 7B). This effect was partly mimicked by stimulation of the cells with forskolin at 200 μM (Figure 7D). AE1-259-01 and AE-329 both also induced ICER expression at 1 μM (Figure 7B,D), while the EP₃ receptor-selective agonist AE-248 did not strongly induce ICER expression. Pretreatment of the cells with either UO126 or H89 (5 μg/mL and 10 μM) tended to attenuate the PGE₂-induced expression of ICER (P = .08 and .06 relative to PGE₂ alone, n = 4, as shown for 1 experiment; Figure 7C). Combined treatment with H89 and UO126 totally abrogated ICER induction, and pretreatment with either AH6809 or AE-208 modestly reduced ICER induction (n = 2, Figure 7D; as shown for 1 experiment, Figure 7C). H89, but not UO126, attenuated ICER induction by AE1-259-01 (n = 3, not shown). Treatment of the cells with 8-pCPT-2′-O-Me-cAMP did not induce ICER expression (n = 2, not shown).

Both COX-1- and COX-2-dependent synthetic pathways mediate PGE₂ production in models of experimentally induced allergic pulmonary disease in mice.^39,40  The abrogation of functional COX-1 or COX-2 activity through genetic³⁹  or pharmacologic⁴⁰  approaches amplifies mucosal inflammation, eosinophilia, and airway hyperresponsiveness in such models by enhancing cytokine generation by T-helper cells.⁴¹  Because MCs are an apparent target of the protective effects of PGE₂ against provocative challenges in humans with asthma^15,16  and AERD,¹⁷  we used well-characterized nontransformed hMCs derived in vitro to better define the EP receptors contributing to the effects of PGE₂ on mediator release, and to determine whether PGE₂ might also activate hMCs in certain contexts.

EP receptors induce differential signaling events through specific patters of G protein utilization. Heterologously expressed EP₂ and EP₄ receptors both use Gs proteins to stimulate adenylyl cyclase and induce accumulation of cAMP,⁴²  with EP₂ receptors being more potent. We detected both of these receptors at the protein and mRNA levels (Figure 1), and showed that both were functional based on the cAMP accumulation induced by the selective agonists AE1-259-01 and AE-329 (Figure 3). The complete blockade of AE-329-induced cAMP accumulation by the EP₄ receptor antagonist AE-208 (Figure 3C) confirmed the specificity of this agonist at 10 μM, and the functionality of the receptor. However, the failure of AE-208 to significantly suppress PGE₂-induced cAMP stimulation suggests a more dominant role for EP₂ receptors in this response, also reflected by inhibition by AH6809. Despite reported nanomolar range potency on EP transfectants,^26,27  none of the selective agonists, alone or in various combinations, approached the efficacy of PGE₂ itself. We speculate that primary cells that express comparatively low levels of individual EP receptor proteins (Figure 1) may require the assembly of ligand-induced, hetero-oligomeric complexes between different EP receptors to amplify PGE_2-dependent signaling events, as reported for other coexpressed GPCRs.⁴³  Such complexes may not be efficiently induced by the selective agonists. We also cannot exclude the potential existence of previously unrecognized EP receptors among the orphan GPCRs.

Splice variants of the human EP₃ receptor mRNA produce multiple isoforms with identical ligand binding properties but different C-termini that confer the ability to either stimulate PTX-sensitive Gi proteins (all isoforms) or Gs proteins (EP₃-II and EP₃-IV isoforms),^35,36  mediating opposing effects on adenyly cyclase/cAMP.^44,45  The consistent expression of the EP₃-II message with a combination of Gi-linked variants suggests that the EP₃ receptor protein expressed by hMCs (Figure 1B) reflects more than one isoform. The modest increment in cAMP that was induced by AE-248 (Figure 3) suggests the function of the EP₃-II receptor, although this effect did not reach statistical significance and only occurred at a dose potentially capable of crossover effects.²⁷  Since cAMP elevation attenuates calcium signaling,⁴⁶  concomitant Gs-induced signaling from EP₃-II or other receptors may explain why EP₃ receptor-dependent calcium flux was observed in only 3 of the 8 donors tested (Figure 2). Nonetheless, PTX-sensitive ERK phosphorylation in response to PGE₂ and AE-248 was observed in all donors tested (Figure 2), reflecting a robust signal through the Gi-linked EP₃ receptor variants responsible for the secretory events observed (Figures 4, 5).

For mediator release, we primed hMCs with IL-4 because this cytokine augments both exocytosis and eicosanoid generation.^32,47  While earlier studies reported that PGE₂ (by raising cAMP levels) inhibited FcϵRI-dependent exocytosis and eicosanoid release from MCs,^20,21  other studies reported that PGE₂ potentiates histamine release²²  and IL-6 production²³  by mBMMCs through effects at the Gi-linked EP₃ receptor. Despite raising cAMP, PGE₂ did not inhibit FcϵRI-dependent exocytosis, instead inducing some exocytosis by itself (Figure 4A). This response required IL-4, was blocked by PTX pretreatment (Figure 4B), and was partly mimicked only by AE-248. IL-4 may thus alter the coupling of Gi-linked EP₃ receptor signaling to exocytosis, presumably distal to phospholipase C, inositol phosphate production, protein kinase C activation, and calcium flux since priming was not required for the latter event (Figure 1). IL-4, which is abundant in allergic inflammation, could promote MC activation responses to PGE₂ in tissues where MCs express EP₃ receptors.

FcϵRI-dependent MC activation simultaneously induces preformed mediator release (histamine and proteases) and eicosanoid generation (LTC₄ and PGD₂) de novo. Eicosanoid generation requires ERK-dependent phosphorylation of cytosolic PLA₂,⁴⁸  providing arachidonic acid to both the COX-PGD synthase and 5-lipoxygenase-LTC₄ synthase (5-LO/LTC₄S) pathways. Because PGE₂ blocks the production of both PGD₂ and cysLTs in response to provocative challenges in vivo, it was not surprising that pretreatment of the cells with PGE₂ (or with the EP₂ receptor-selective antagonist AE1-259-01) attenuated FcϵRI-mediated generation of both PGD₂ and cysLTs (Figure 5A-B). Unexpectedly, hMCs stimulated with PGE₂ also generated PGD₂, but not LTC₄ (reflected by the cysLT-specific ELISA). cAMP-dependent PKA phosphorylates 5-LO, preventing its translocation to the nuclear envelope to synthesize the LT precursor LTA₄.⁴⁹  This likely explains why cysLT generation occurred in response to PGE₂ when PKA was inhibited by H89 (Figure 5C). This response was also elicited, albeit weakly, by AE-248. Combined with the abrogation of PGE₂-induced exocytosis and eicosanoid generation by pretreatment of the cells with PTX, these findings firmly implicate EP₃ receptors in these effector responses. The clearly opposing cellular and signaling responses to EP₂ and EP₃ receptor agonists strongly supports the selectivity of these reagents for their intended receptors at the doses used. It appears likely that EP₂ and EP₃ receptors counterbalance one another's effects on cellular activation responses when coexpressed. EP₂ receptor-selective stimulation was recently reported to induce the production of vascular endothelial growth factor by hMCs.⁵⁰  Thus EP₂ receptor signaling in allergic disease may attenuate effector eicosanoid formation while simultaneously promoting tissue remodeling to facilitate the resolution phase of mucosal inflammation.

TNF-α localizes to MCs in airway biopsy specimens from patients with severe asthma who respond therapeutically to TNF-α-selective antagonists⁵¹  and promotes granulocyte influx in MC-dependent innate immune responses in mice.¹¹  In our previous studies, adenine nucleotides suppressed TNF-α generation by hMCs,²⁵  initiating a cAMP-dependent, PKA-independent signaling cascade resulting in the expression of ICER, a transcriptional repressor protein that is the product of an alternate promoter in the gene encoding cAMP response element modulator.³⁸  ICER prevents binding of NF-AT and AP-1 transcription factors to the promoter regions of the genes encoding TNF-α and other cytokines.⁵²  The marked suppression of TNF-α protein and mRNA expression by PGE₂ (Figure 6) suggested a process involving ICER. Indeed, PGE₂ strongly induced ICER expression (Figure 7), as did the adenylyl cyclase activator forskolin. Importantly, this response likely involved input from more than 1 EP receptor, since both AE1-259-01 and AE-329 both induced ICER on their own at 1 μM (Figure 7B,D). Moreover, H89 (10 μM) only partly blocked ICER induction, but completely blocked it when used in combination with UO126 (Figure 7C,D). Thus PKA (reflecting the EP₂ and/or EP₄ receptors) and ERK (implicating EP₃ receptors; Figure 2) cooperate to induce ICER. Since the Epac agonist and GSK-3 antagonists did not block cytokine generation or induce ICER expression, the effect of PGE₂ in our study was not a consequence of the recently reported Epac/GSK-3 transcriptional repressor pathway.⁵³  Thus PGE₂ suppresses cytokine generation by cell types with different EP receptor expression profiles^34,52-55  through multiple pathways. This versatility may ensure that PGE₂, when not suppressed by COX inhibitors, tempers cytokine generation and consequent inflammation in the hours after MC activation occurs in allergic disease.

Our data reveal that PGE₂ can either induce (via EP₃ receptors) or suppress (via EP₂ receptors) mediator release from MCs that induce the EPR. EP₂ receptor-dependent signaling is an attractive candidate to explain PGE₂-dependent suppression of eicosanoid generation in allergen or aspirin challenge. Moreover, the suppressive effect of PGE₂ on allergen-induced LPRs¹⁵  could in part reflect transcriptional repression through cooperative signaling from multiple EP receptors. Finally, EP₃ receptor-dependent MC activation is reported to occur in the skin in a mouse model of cutaneous inflammation,²¹  and this may reflect both the release of histamine and PGD₂. It seems likely that the pro- and anti-MC-activating effects of PGE₂ reported for respectively for skin²¹  and lung¹⁵  reflect regional differences between MCs EP receptor expression and function.