Prostaglandin E₂–EP3 Signaling Induces Inflammatory Swelling by Mast Cell Activation

Four cardinal features, namely rubor (red flare), calor (heat), tumor (swelling), and dolor (pain), characterize acute inflammation. The flare and heat reactions are caused by an increase in local blood flow as a result of vasodilatation, and the swelling is elicited by an increase in vascular permeability and resultant leukocyte recruitment. These processes are triggered by tissue injury and invasion of exogenous materials and organisms (1). Such inflammatory insults are primarily detected by TLRs in immune cells followed by activation of the local cytokine network such as TNF-α and IL-1β. Because these cytokines affect vascular permeability and leukocyte recruitment, such a TLR/cytokine axis in innate immunity is one factor that governs the inflammation process (2, 3). Alternatively, by using various experimental models of acute inflammation, chemical mediators such as bradykinin, histamine, thrombin, and growth factors have been found and characterized (1). Aspirin-like drugs have been used as the first choice of drugs for acute inflammation because of their high potency to suppress the above inflammatory symptoms (4). Because these drugs exert their actions by inhibiting cyclooxygenases (COXs) and thereby inhibit the biosynthesis of PGs, endogenously synthesized PGs are thought to be involved in inflammation reactions (1, 5). Indeed, PGs have been shown to elicit vasodilatation and an increase in local blood flow, leading to red flare and local heat. It is thought that vascular permeability factors such as histamine and bradykinin are thereafter released into the inflammation site, leading to edema formation (1, 6). However, the link between the initial vasodilatation and the subsequent permeability change remains unknown.

PGE₂ is the most abundantly synthesized PG, especially at an inflammation site, and has long been regarded as an important potentiator of acute inflammation (1, 4–7). The diverse actions of PGE₂ are mediated through four receptor subtypes, EP1, EP2, EP3, and EP4 (8, 9). EP1 is coupled to intracellular Ca²⁺ mobilization via G_q, EP2 and EP4 are coupled to stimulation of adenylyl cyclase via G_s, and EP3 is mainly coupled to inhibition of adenylyl cyclase via G_i. Among the four PGE receptors, the EP3 receptor was proposed to participate in arachidonic acid (AA)–induced edema formation (10). These results suggest that PGE₂ may function as a link to an increase in vascular permeability and edema formation (11). However, it remains unsolved as to how PGE₂ increases permeability and edema formation in acute inflammation.

Mast cells (MCs) are immune cells widely distributed in various peripheral tissues and are activated in an Ag-dependent manner (12, 13). Once activated by Ag-induced cross-linking of IgE receptors, MCs release bioactive substances in their granules such as histamine and proteases, leading to acute allergic inflammation typically seen in urticaria. To date, many researchers have investigated the roles of PGE receptor signaling in the Ag-dependent activation of MCs and obtained various results because MCs show tissue-specific phenotypes. For example, PGE₂ enhances Ag-induced degranulation in mouse bone marrow–derived MCs (BMMCs) (14) and human peripheral blood–derived MCs (15), whereas it attenuates Ag-induced degranulation in MCs isolated from rat peritoneal cells (16) and human lung (17). A recent study found that MCs can be activated in an Ag-independent fashion; that is, substance P released from blood vessels or peripheral terminals of sensory neurons stimulates MC degranulation (18). Indeed, it was recently reported that MCs play a role in Ag-independent inflammation such as physical or chemical irritation (19, 20). In these models, haptens have been shown to activate TLR signaling in skin cells (21). Because TLR signal activation induces COX gene expression and PG production in many types of cells (22), PGE₂-induced MC activation is a likely mechanism underlying irritant contact dermatitis. Nevertheless, it remains unknown as to whether PGE receptor signaling is involved in such Ag-independent MC activation.

In this study, we conducted a series of experiments to clarify how PGE receptor signaling increases vascular permeability and edema formation. We show that PGE₂ by itself triggers degranulation of dermal MCs via the EP3–G_i/o–Ca²⁺ influx signaling pathway, and this functions as the underlying mechanism of PGE₂-induced vascular hyperpermeability and subsequent edema formation.

Specific pathogen-free 8-wk-old C57BL/6 mice and WBB6F₁-W/W^v (W/W^v) (23) and littermate control WBB6F₁^+/+ mice were obtained from Japan SLC (Hamamatsu, Japan). Kit^W-sh/W-sh mice (24) with a genetic background of C57BL/6 were obtained from The Jackson Laboratory (Bar Harbor, ME). Ptger1^−/−, Ptger2^−/−, Ptger3^−/−, and Hdc^−/− mice with a C57BL/6 background were previously generated as described (25–27), and C57BL/6 mice were used as wild-type (WT) controls. We used surviving Ptger4^−/− offspring and their WT littermates with a mixed background of 129Ola and C57BL/6 from heterozygote crosses (28–31). All mice were housed in specific pathogen-free animal facilities and mice at 8–12 wk of age (20 ± 2 g in weight) were used for each experiment. All animal experiments were performed according to the Guidelines for Animal Experiments of Kyoto University, Kumamoto University, and the Tokyo Metropolitan Institute of Medical Science. The following materials were purchased from the sources indicated: AA and La³⁺ from Nacalai Tesque (Kyoto, Japan); recombinant mouse IL-3 from R&D Systems (Minneapolis, MN); PGE₂ and butaprost (an EP2 agonist) from Cayman Chemical (Ann Arbor, MI); p-nitrophenyl-β-d-2-acetoamide-2-deoxy-glucopyranoside, LY294002, indomethacin, diphenhydramine, Hoe140, an anti-DNP mouse monoclonal IgE (SPE-7), and DNP-conjugated human serum albumin from Sigma-Aldrich (St. Louis, MO); fura 2-AM from Dojindo (Kumamoto, Japan); pertussis toxin (PTX) from Seikagaku (Tokyo, Japan); U73122, A23187, thapsigargin, and 2-aminoethoxydiphenyl borate (2-APB) from Calbiochem (Merck, Darmstadt, Germany); an anti–phospho-Akt (Ser⁴⁷³) rabbit mAb (193H12) and an anti-Akt rabbit mAb (11E7) from Cell Signaling Technology (Danvers, MA); an anti-actin mouse mAb (C4) from Santa Cruz Biotechnology (Santa Cruz, CA); and SKF-96365 from Tocris Bioscience (Bristol, U.K.). ONO-DI-004 (an EP1 agonist), ONO-AE-248 (an EP3 agonist), and ONO-AE1-437 (an EP4 agonist) were gifts from ONO Pharmaceuticals (Osaka, Japan). The basic profiles of selectivity of each EP-specific agonist were reported previously (32). ONO-AE-208 (an EP4 antagonist) was also provided from ONO Pharmaceuticals. All other chemicals were commercial products of reagent grade.

Mice were anesthetized with ether and injected i.v. with 1 mg Evans blue dye in 200 μl saline. Immediately after the dye injection, AA (0.6 mg in 20 μl acetone) was applied to the inner and outer surface of the right ear, whereas the left ear received the vehicle delivered in the same manner. Alternatively, the ears of the mice were intradermally injected with PGE₂ or each EP receptor agonist in 25 μl saline, and 0.57% ethanol in saline was used as the vehicle. Thirty minutes after AA application or the intradermal injection of PGE₂, mice were sacrificed and their ears were collected. The ear biopsies were incubated at 37°C in 1 ml of 3 N KOH. On the following day the mixtures were extracted with 1.24 M phosphoric acid and acetone. Absorbance of the resulting supernatants was measured at 620 nm, and the quantities of leaked dye were calculated by comparison with tissue samples injected with certain amounts of dye. We measured ear thickness with a micrometer for each mouse before and at the indicated times after AA application or PGE₂ injection and expressed the difference as edema formation.

The activity of a neutrophil marker enzyme, myeloperoxidase (MPO), was measured according to the procedure reported by Bradley et al. (33) with minor modifications. Briefly, 6 h after the administration of AA to mice, their ears were removed and homogenized with a Polytron homogenizer in 50 mM potassium phosphate buffer (pH 6.0). Centrifuged tissue pellets were re-extracted in potassium phosphate buffer with 0.5% hexadecanoyl-trimethyl ammonium bromide, freeze-thawed three times, and centrifuged to collect the supernatants that were then used for measurement of MPO activity. Sixty microliters of supernatant was added to 60 μl assay buffer containing 0.33 mg/ml o-dianisidine and 0.001% hydrogen peroxide, and the change of absorbance at 460 nm was measured. One unit of MPO activity was defined as that degrading one micromole of peroxide per minute at 25°C. Ear tissue lysates were centrifuged at 10,000 × g for 30 min at 4°C, and the resultant supernatants were subjected to the histamine assay. Specifically, histamine was separated on the cation exchange column WCX-1 (Shimadzu, Kyoto, Japan) by HPLC and detected by the o-phthalaldehyde method (34). For the measurement of histamine release from peritoneal MCs, 5 ml PBS was injected into the peritoneal cavity and, after a slight massage, the PBS was collected. Such peritoneal lavage fluid usually contains Kit⁺/FcεRIα⁺ MCs (2–3% of total cells). Peritoneal cells were stimulated with PGE₂ or compound 48/80 for 30 min. After centrifugation, histamine in the supernatants and cellular fractions was measured by the histamine assay.

Ear sections (7 μm) were fixed with formalin, followed by sequential staining with Alcian blue and safranin, and then by staining with acidic toluidine blue (pH 0.5). The number of metachromatic cells was counted in five sections per mouse, and the results from five to six mice were averaged. For examination of degranulation status, tissue MCs were counted as either resting MC or MC undergoing degranulation, depending on whether its toluidine blue–positive granule contents were clustered within a cell or spread out throughout the cell, respectively.

Preparation of BMMCs was performed as described previously (35). In brief, bone marrow cells were cultured in RPMI 1640 medium containing 10% heat-inactivated FBS, 10 ng/ml mouse IL-3, 50 μM 2-ME, and 0.1 mM nonessential amino acids at 37°C in a fully humidified 5% CO₂ atmosphere. After 4–5 wk of culture, >95% of the viable cells were confirmed to be immature MCs, as assessed by staining of cytospins with acidic toluidine blue (pH 2.5), which results in MC-specific metachromatic staining. For MC reconstitution, WT or Ptger3^−/− BMMCs (1× 10⁶ cells/ear) were injected into the ear cutaneous tissues of the Kit^W-sh/W-sh mice. After 5 wk, the MC-reconstituted Kit^W-sh/W-sh mice were subjected to the PGE₂-induced inflammation assay. Reconstitution and maturation of transferred MCs were confirmed by monitoring the gene expression of Ptger3 and MC-specific genes (Fcer1a, Hdc, Mcp4, and Cpa3) using RT-PCR. Flow cytometric analysis was performed as previously described with a FACSCalibur (Becton Dickinson, Franklin Lakes, NJ) equipped with CellQuest software (36). The dead cell population was gated out by propidium iodide staining.

BMMCs were collected and washed in Tyrode’s buffer (10 mM HEPES-NaOH [pH 7.3] containing 130 mM NaCl, 5 mM KCl, 5.6 mM glucose, 1.4 mM CaCl₂, 1 mM MgCl₂, 10 μM indomethacin, and 0.1% BSA). The cells were incubated for 30 min in Tyrode’s buffer with or without PGE₂. Degranulation of MCs was evaluated by measuring the activity of a granule enzyme, β-hexosaminidase, using p-nitrophenyl-β-d-2-acetoamide-2-deoxyglucopyranoside as a substrate. For the measurement of IL-6 production, BMMCs were collected and washed in fresh culture medium. The cells were then incubated for 3 h in culture medium with PGE₂ (1 μM). The IL-6 in the medium was measured by ELISA (BD Biosciences).

Cells were loaded with 2 μM fura 2-AM in Tyrode’s buffer for 45 min at room temperature and then washed in Tyrode’s buffer. For Ca²⁺-free conditions, the buffer was replaced with Ca²⁺-free Tyrode’s buffer. Fluorescent intensities were measured at an excitation wavelength of 340 or 380 nm and an emission wavelength of 510 nm with a fluorescence spectrometer (CAF-100; Jasco, Tokyo, Japan) as described previously (36). BMMCs were stimulated with PGE₂ or Ag for 10 min in Tyrode’s buffer without BSA. For cell lysis, the cell suspension was added directly to an equal volume of 2× Laemmli sample buffer. The proteins were separated on 10% polyacrylamide gels and transferred onto polyvinyl difluoride membranes. Membranes were probed with an anti–phospho-Akt or anti–total Akt Ab. To normalize protein loading, the membranes were stripped and probed for actin.

Data are represented as means ± SEM of at least three independent experiments. Two sets of data were compared by Student t test. Multiple treatment groups were compared by ANOVA. When significant differences were found, additional comparisons were made with the Dunnett multiple comparison test for comparison with control groups or with the Tukey–Kramer multiple comparison test for comparison of all pairs of columns. A p value < 0.05 was considered statistically significant.

It was previously reported that lipid mediators synthesized by COX pathways are involved in AA-induced inflammation, and that PGE₂ was abundantly synthesized within ear tissues upon AA application (37). Indeed, also in our experimental settings, aspirin significantly suppressed AA-induced hyperpermeability (Fig. 1A). We first identified PGE receptors involved in AA-induced hyperpermeability and edema formation. Mice deficient in each of the four PGE receptors were subjected to the AA-induced and PGE₂-induced inflammation, and vascular permeability was monitored (Fig. 1B). In WT mice, vascular permeability was drastically increased 30 min after topical application of AA. In EP1-, EP2-, and EP4-deficient mice, permeability increased similarly to that in WT mice (Fig. 1B). Moreover, the pretreatment with an EP4 antagonist (10 mg/kg/d orally for 24 h) failed to affect AA-induced responses (Fig. 1C), although this condition was previously shown to block EP4 signaling systemically (29, 31). In contrast, EP3-deficient mice showed significantly attenuated responses (Fig. 1B). When we investigated edema formation, topical application of AA induced an increase in ear thickness in WT mice but failed to do so in EP3-deficient mice (Fig. 1D). These results suggest that PGE₂–EP3 is involved in the initiation of acute inflammation. We then examined neutrophil infiltration by measuring MPO activity. Indeed, in WT mice, MPO activity was augmented 6 h after the application of AA. However, EP3-deficient mice showed only a slight induction of MPO activity (Fig. 1E). Application of AA has been shown to induce the production of a broad range of eicosanoids such as PGE₂ (37). We therefore examined the direct effect of exogenous PGE₂ on vascular permeability. Mice were injected i.v. with Evans blue dye, and immediately thereafter PGE₂ was injected intradermally into the ear. Thirty minutes later, amounts of dye leaked into the ear were measured. PGE₂ exerted transient dye leakage in WT mice, which peaked at 30 min (Fig. 1F). Furthermore, PGE₂ increased vascular permeability in a dose-dependent manner (Fig. 1G). These results suggest that PGE₂ has the potential to induce hyperpermeability. Interestingly, an EP3-specific agonist also induced vascular permeability with a dose-response similar to PGE₂ (Fig. 1H). Because both PGE₂ and an EP3 agonist at a dose of 1.4 nmol were enough to stimulate dye leakage, we compared the effects of the agonists specific for each of the four EPs at this dose. Only an EP3 agonist mimicked the effect of PGE₂ (Fig. 1I), suggesting that the induction of vascular permeability by PGE₂ is mediated by the EP3 receptor. Indeed, the effect of PGE₂ was abolished in Ptger3^−/− mice (Fig. 1J). These results indicate that the PGE₂–EP3 signal induces hyperpermeability.

To uncover the mechanisms involved in PGE₂-elicited hyperpermeability, we tested the effect of various chemicals known to affect inflammation. Although PGE₂ is known to sensitize bradykinin-induced responses, a bradykinin B₂ receptor antagonist failed to alter the PGE₂-induced response (Supplemental Fig. 1A), indicating that there is little contribution of bradykinin B₂ signaling in PGE₂-elicited hyperpermeability. Alternatively, we found that a histamine H₁ blocker, diphenhydramine, inhibited the action of PGE₂ in WT mice (Supplemental Fig. 1B), suggesting that histamine mediates PGE₂-induced hyperpermeability.

MC-derived inflammatory mediators such as histamine and proteases have been shown to increase vascular permeability. Indeed, MCs were reported to be essential for Ag-independent inflammation, such as AA-induced inflammation and irritant contact dermatitis (19, 37). We therefore examined the effects of genetic depletion of MCs (W/W^v) on AA-induced hyperpermeability. As expected, AA-induced responses were significantly attenuated in W/W^v mice, although the proinflammatory actions of AA still remained (Fig. 2A). Indeed, when we examined the MC granule statuses in AA-added ear tissues, AA application remarkably induced the MC degranulation (Fig. 2B). These results suggest that AA-induced responses are at least partly due to MC degranulation. Interestingly, aspirin significantly attenuated AA-induced MC degranulation (Fig. 2B), suggesting that some COX products may elicit MC degranulation. Indeed, when we investigated the PGE₂-induced response in W/W^v mice, PGE₂ completely failed to stimulate vascular permeability (Fig. 2C). These results indicate that PGE₂–EP3-mediated vascular hyperpermeability depends on MCs. Indeed, we found that the EP3 receptor is abundantly expressed in BMMCs, as reported previously (38). To confirm that the action of PGE₂ is mediated by EP3 expressed in MCs, we employed Kit^W-sh/W-sh mice, another strain of MC-deficient mice, and reconstituted them with WT BMMCs or BMMCs deficient in the EP3 receptor (Ptger3^−/−). After 5 wk, reconstitution and maturation of the transferred MCs in ear tissues were confirmed by monitoring gene expression of MC-specific genes (Supplemental Fig. 2), and these mice were subjected to the PGE₂-induced inflammation assay (Fig. 2D, 2E). As observed in W/W^v mice, PGE₂-induced permeability changes were abolished in Kit^W-sh/W-sh mutants but were significantly recovered by reconstitution with WT BMMCs, but such a recovery was not seen in mice reconstituted with Ptger3^−/− BMMCs (Fig. 2D). Moreover, similar results were obtained when we measured ear thickness as an index of edema formation (Fig. 2E). If PGE₂ increases vascular permeability by acting on the EP3 receptor on MCs, the inability of PGE₂ to induce hyperpermeability in Ptger3^−/− mice might be due to impaired migration, differentiation, and/or maturation of Ptger3^−/− MCs. When we compared the number and maturation indexes of MCs between WT and Ptger3^−/− mice, we failed to detect any difference in the number and histological appearance of MCs, as well as in protease activity and histamine content (Fig. 3). We thus concluded that PGE₂–EP3 signaling directly triggers MC activation.

Histamine is one of the bioactive mediators released from MCs and is responsible for the extravasation reaction (25). To investigate whether PGE₂-induced vascular hyperpermeability is mediated by MC-derived histamine, we examined the effects of PGE₂ on mice deficient in histidine decarboxylase (Hdc^−/−), an enzyme responsible for histamine biosynthesis in vivo. In Hdc^−/− mice, PGE₂-induced vascular hyperpermeability was almost abolished (Fig. 2F). Once histamine is released into the extracellular space, it is immediately converted into an inactive form by the action of diamine oxidase (39) and is rapidly cleared from tissue. Indeed, the histamine content in pinna tissue was significantly decreased to 50% upon PGE₂ injection in WT mice (Fig. 2G), suggesting that PGE₂ initiates histamine release from MCs. In contrast, histamine content in Ptger3^−/− mice was constant even after PGE₂ application. These results suggest that PGE₂-induced hyperpermeability depends on the EP3 receptor, MCs, and histamine. These results also suggest that PGE₂ by itself triggers MC degranulation by acting on the EP3 receptor. To confirm this, we prepared peritoneal cells from WT mice, which include MCs (∼2–3%), and examined whether PGE₂ induces histamine release from peritoneal MCs. As expected, PGE₂ alone induced histamine release from peritoneal MCs (Fig. 2H). No apparent histamine release was detected in Ptger3^−/− cells. There were no apparent differences in histamine contents (Fig. 3D), percentage of MCs in total peritoneal cells (WT, 2.43 ± 0.38% versus Ptger3^−/−, 2.44 ± 0.52%), and compound 48/80–induced histamine release between WT and Ptger3^−/− peritoneal cells (Fig. 2H). These results demonstrate that PGE₂–EP3 signaling triggers histamine release from MCs.

To investigate the mechanism of PGE₂-induced MC activation, we used mouse BMMCs in which EP3 is abundantly expressed (38). We first compared the basic characteristics of WT and Ptger3^−/− BMMCs but failed to detect any difference in toluidine blue staining, expression level of FcεRI or c-Kit, histamine content, or degranulation upon Ag or secretagogue stimulation (Supplemental Fig. 3 and data not shown), suggesting that Ptger3^−/− BMMCs differentiate normally. We next examined whether PGE₂ alone triggers degranulation of WT BMMCs. PGE₂ rapidly induced degranulation in WT BMMCs, which was saturated in 3 min (Fig. 4A). PGE₂ induced degranulation in a concentration-dependent fashion. In contrast, in Ptger3^−/− BMMCs, such a PGE₂-induced response was hardly detected although crosslinking of FcεRI induced degranulation in a manner similar to WT cells (Fig. 4B). Because MC activation includes the release of a wide variety of cytokines (12, 13), we next examined the effect of PGE₂ on IL-6 release from BMMCs. In WT cells, PGE₂ induced IL-6 production (Fig. 4C). In contrast, PGE₂ failed to induce IL-6 release in Ptger3^−/− BMMCs, although crosslinking of FcεRI stimulated IL-6 release to levels similar to WT cells (Fig. 4D). These results indicate that PGE₂ triggers not only degranulation but also cytokine release via the EP3 receptor in BMMCs. Indeed, an EP3 agonist mimicked the effects of PGE₂ on degranulation and IL-6 release (Fig. 4E, 4F). Because the EP3 receptor was shown to be coupled to PTX-sensitive G proteins, mainly G_i (40), we investigated whether PTX affects PGE₂-induced responses in WT BMMCs. PTX inhibited PGE₂-induced degranulation and IL-6 production (Fig. 4G, 4H) but failed to suppress Ag/IgE-induced responses. These results indicate that PGE₂–EP3 signaling induces MC activation in a PTX-sensitive manner.

It has been shown that various stimuli induce MC degranulation through intracellular Ca²⁺ mobilization (41–43). To investigate whether a similar signaling pathway is stimulated by EP3 activation, we examined the effects of PGE₂ on intracellular Ca²⁺ concentration in BMMCs. PGE₂ stimulated intracellular Ca²⁺ mobilization in WT BMMCs in a dose-dependent manner. In contrast, such a PGE₂-induced Ca²⁺ response was not observed in Ptger3^−/− BMMCs, although thapsigargin-induced Ca²⁺ responses were observed at levels similar to those in WT cells (Fig. 5A, 5B). Among the agonists specific for each EP, only an EP3 agonist elicited significant Ca²⁺ mobilization in WT BMMCs (Fig. 5C). Pretreatment with PTX abolished PGE₂-induced but not Ag/IgE-induced Ca²⁺ mobilization in WT BMMCs (Fig. 5D), indicating that PGE₂–EP3 signaling induces Ca²⁺ mobilization via the PTX-sensitive G protein, G_i/o. In MCs, PTX-sensitive G proteins such as G_i1 and G_i2 are thought to induce phospholipase C (PLC) activation and Ca²⁺ release from the endoplasmic reticulum store. Indeed, when we treated WT BMMCs with the PLC inhibitor U73122, the PGE₂-induced Ca²⁺ response was abolished (Fig. 5E). Endoplasmic reticulum Ca²⁺ store depletion–induced Ca²⁺ influx via the Stim/Orai pathway has been shown to be important for MC activation (44). We hence examined the effects of inhibitors of Orai channels, La³⁺, 2-APB, and SKF-96365, on PGE₂-induced Ca²⁺ responses. These inhibitors significantly attenuated PGE₂-induced Ca²⁺ responses (Fig. 5E). We further investigated whether Ca²⁺ entry from the extracellular space is actually required for PGE₂-induced MC activation. Both PGE₂-induced degranulation and IL-6 production were abolished by the depletion of extracellular Ca²⁺ (Fig. 5F, 5G), indicating that PGE₂–EP3 signaling induces MC activation via Ca²⁺ influx from an extracellular source. Recent studies demonstrated that MC degranulation processes are divided into two events: the Ca²⁺-independent microtubule-dependent translocation of granules to the plasma membrane, and Ca²⁺-dependent membrane fusion and exocytosis (45). The former event is mediated by PI3K activation and production of phosphatidylinositol-3,4,5-trisphosphate leading to phosphorylation of the Ser⁴⁷³ residue within the C-terminal region of Akt (46, 47). We therefore examined PI3K activation by detecting Akt phosphorylated at Ser⁴⁷³ upon PGE₂ stimulation of WT and Ptger3^−/− BMMCs. As reported previously, FcεRI aggregation induced Akt phosphorylation within 10 min in WT and Ptger3^−/− cells (Fig. 6A). Interestingly, PGE₂ also rapidly stimulated Akt phosphorylation in WT but not Ptger3^−/− cells. Pretreatment of the WT cells with PTX abolished PGE₂-induced but not IgE-mediated Akt phosphorylation. These results suggest that PGE₂ induces PI3K activation via an EP3-G_i/o–dependent pathway. Because the depletion of extracellular Ca²⁺ failed to affect IgE-mediated and PGE₂-induced PI3K activation, PGE₂-induced PI3K activation appears to be Ca²⁺-independent, as reported previously (47). To examine whether PI3K activation is involved in PGE₂-induced MC activation, we finally examined the effect of a PI3K inhibitor, LY294002. Both degranulation and IL-6 release were inhibited by LY294002 (Fig. 6B, 6C). These results indicate that Ca²⁺ influx along with PI3K activation is required for PGE₂–EP3-induced MC activation.

Edema formation is a prominent feature of the inflammatory response and is due to the movement of fluid and plasma proteins into the extracellular space from leaky blood vessels. It is thought that an increased vascular permeability is induced by a range of injurious stimuli and by several chemical mediators generated during an inflammatory response (1). Almost 40 y ago, Williams and Morley (6) proposed the “two mediator hypothesis,” which states that the magnitude of inflammatory swelling is dependent on two factors: the degree of vasodilatation, which is induced by PGs including PGI₂ and PGE₂, and the extent of endothelial cell permeability, which is induced by a variety of permeabilizing substances, including bradykinin and histamine. Furthermore, according to this hypothesis, when both factors are present they synergize to produce a greater net swelling response. Although this hypothesis has been frequently cited and is currently embraced (1, 4–7), the link between the initial vasodilatation and the subsequent permeability change remains unclear. Recently, it was shown that EP2 and EP4 receptors are involved in edema formation mainly by eliciting vasodilatation and the subsequent increase in local blood flow in UV-induced skin inflammation (48). Therefore, PGE₂ may elicit acute inflammation via two different actions: by stimulating vasodilatation via the EP2 and EP4 receptors on smooth muscle cells, and by stimulating vascular permeability via the EP3 receptor on MCs. Furthermore, in the present study we also show that PGE₂–EP3 signaling triggers IL-6 release from MCs. IL-6 has been shown to stimulate the local production of neutrophil-attracting chemokines such as IL-8 and GLO-α (49) and to induce the endothelial expression of adhesion molecules such as VCAM-1 and E-selectin (50), both of which stimulate leukocyte rolling. Indeed, MPO activity, a hallmark of neutrophil recruitment, was significantly reduced in EP3-deficient mice with AA-induced inflammation (Fig. 1C). These results indicate that EP3-mediated MC activation plays a critical role not only in permeability change but also in neutrophil infiltration. Thus, EP3 signaling in MCs appears to function as an underlying mechanism of PGE₂-induced vasodilatation and subsequent edema formation.

Potential involvement of PGE₂-induced MC activation in various kinds of inflammatory diseases is significant from a clinical perspective. In particular, Ag-independent acute inflammatory reactions such as physical or chemical irritation may be attributed to this PGE₂ action. Indeed, MCs were recently found to play critical roles in Ag-independent irritant contact dermatitis (19, 20), and haptens have been shown to activate TLR signaling in skin cells (21). Because TLR signal activation induces COX gene expression and PG production in many types of cells (22), PGE₂-induced MC activation is a likely mechanism underlying irritant contact dermatitis. Therefore, we propose that EP3-mediated MC activation is involved not only in the previously reported PG-related acute inflammation responses but also in Ag-independent innate immune reactions. Moreover, in addition to degranulation and cytokine release, PGE₂ has been shown to stimulate chemotaxis (or chemokinesis) and adhesion activities of MCs via the EP3 receptor (39, 51, 52). If this is the case, PGE₂–EP3 signaling could function as a sensor of MCs against microenvironmental changes.

In summary, the present study demonstrates that PGE₂ directly triggers MC degranulation and cytokine release via the EP3/G_i pathway, resulting in enhancement of vascular permeability and leukocyte recruitment in an MC-dependent manner (Supplemental Fig. 4). Thus, EP3-mediated MC activation may be potentially involved in Ag-independent innate immune reactions.

We thank Prof. Kazuhisa Nakayama and Atsushi Ichikawa for precious advice and continuous support. We also thank Dr. Hirotsugu Takano and Daisuke Mori for technical support. We are grateful to Dr. H Akiko Popiel for careful reading of the manuscript.

The authors have no financial conflicts of interest.