G-CSF is a hemopoietic growth factor involved in granulocytic differentiation of progenitor cells. In this study, we investigated the effects of PGE2 on G-CSF production in murine peritoneal neutrophils in vitro and in vivo. PGE2 augmented LPS-primed G-CSF release from peritoneal neutrophils. This augmentation was mimicked by a type E prostanoid receptor (EP)2-selective agonist but not by other EP-specific agonists. Indeed, the effect of PGE2 on G-CSF release was abolished in neutrophils isolated from EP2-deficient mice. PGE2 and an EP2 agonist have the ability to stimulate G-CSF gene expression even in the absence of LPS. In the casein-induced peritonitis model, the appearance of G-CSF in the casein-injected peritoneal cavity associated well with the timing of neutrophil infiltration as well as PGE2 levels in exudates, with a peak value at 6 h postinjection. Inhibition of endogenous PG synthesis by indomethacin resulted in a marked decrease in G-CSF content and neutrophil number in the peritoneal cavity. Moreover, EP2-deficient mice exhibited a strikingly reduced G-CSF content in peritoneal exudates with comparable responses in neutrophil migration and local PGE2 production at 6 h postinjection. These results suggest that the PGE2-EP2 system contributes to the local production of G-CSF during acute inflammation.
Prostaglandin E2 is one of the most important arachidonate metabolites synthesized by the action of cyclooxygenase (1). This lipid mediator is involved in a wide range of diseases, including inflammation, by exerting pleiotropic actions (2). Administration of PGE2 alone does not cause any significant responses, demonstrating that PGE2 on its own has little inflammatory capacity. In contrast, in the presence of other mediators, PGE2 can synergistically amplify the local inflammatory response. For instance, PGE2 has been shown to enhance zymosan-stimulated IL-10 production in macrophages (3). In contrast to these proinflammatory activities, PGs are also known to inhibit the production of proinflammatory cytokines by macrophages activated with LPS; PGE2 has been shown to inhibit LPS-induced IL-12 production and LPS-stimulated TNF-α production in resident macrophages (4, 5). Thus, PGE2 works as a modulator of cellular immune responses initiated by other stimulants.
G-CSF is a member of a family of hemopoietic growth factors that are required for proliferation and differentiation of hemopoietic progenitor cells (6, 7). Administration of G-CSF increases peripheral blood neutrophil counts in many species, including humans (8, 9). G-CSF is released by monocytes and endothelial cells in response to proinflammatory inputs such as LPS and TNF-α (7, 10). Such inflammation-induced G-CSF has been considered to contribute to the maintenance of a number of neutrophil lineage cells in the bone marrow (7, 11). Indeed, in the casein-induced peritonitis model, G-CSF is produced in the peritoneal cavity at the time of neutrophil migration, suggesting that neutrophils can produce G-CSF for this purpose (11). PGE2 has been shown to enhance LPS-induced G-CSF formation in human monocyte or macrophage-like THP-1 cells (12, 13). However, it remains unknown whether PGE2 modulates G-CSF formation from neutrophils, and there are no reports evaluating the contribution of PGE2 to local G-CSF production within an inflammatory area. The biological actions of PGE2, including its effects on cellular immune responses, are mediated by the G protein-coupled receptor, type E prostanoid receptor (EP; reviewed in Refs. 14 and 15). EP can be divided into four distinct pharmacological classes: EP1, EP2, EP3, and EP4. Among them, both the EP2 and EP4, which are coupled to the stimulation of adenylyl cyclase, have been shown to exist throughout a wide range of immune cells, including dendritic cells (5, 16, 17). However, the precise EP subtypes that mediate the immunoregulatory actions of PGE2 remain to be established.
In this study, we have investigated the effects of PGE2 on G-CSF production from peritoneal polymorphonuclear leukocytes (PMN 4) (3) in vitro and in vivo. We show that PGE2 in the absence of LPS stimulation is capable of inducing the production of G-CSF by peritoneal PMNs. Although both EP2 and EP4 are expressed in peritoneal PMNs, this effect of PGE2 is mediated by EP2 but not by EP4. Moreover, EP2 signaling was found to be a physiological stimulus of the local production of G-CSF within an inflammatory site.
Materials and Methods
Female C57BL/6 mice (12 wk of age) purchased from Japan SLC were used as wild-type mice. The generation of EP2-deficient (ptger2−/−) mice has been described previously (18). EP2-deficient mice, backcrossed for 10 generations to C57BL/6 mice, were maintained on a 12-h light/dark cycle under specific pathogen-free conditions. All experimental procedures were approved by the Committee on Animal Research of Kyoto University Faculty of Pharmaceutical Sciences.
LPS from Escherichia coli O55:B5, casein, dibutyryl cAMP, and indomethacin were obtained from Sigma-Aldrich. The 125I-labeled cAMP assay system was purchased from Amersham Bioscience, and the ELISA kits for G-CSF and MIP-2 quantification were purchased from R&D Systems. PGE2 in the peritoneal fluid was quantified using an enzyme immunoassay kit (Cayman Chemical). PGE2 was purchased from Funakoshi. RPMI 1640 medium and FBS were from Invitrogen Life Technologies (LPS < 30 pg/ml). The EP-specific agonists, DI-004, AE1–259, AE-248, and AE1–329 were generous gifts from Ono Pharmaceutical. The specificities of the agonists were analyzed by measuring the binding affinity to the respective EP subtype expressed in Chinese hamster ovary (CHO) cells, as reported previously (5, 19). PGE2, EP agonists, dibutyryl cAMP (dbcAMP), other reagents, and culture media were confirmed to be endotoxin-free (<0.1 EU/ml endotoxin) using the Limulus Amoebocyte Lysate assay (Endospecy; Seikagaku Kogyo). All other chemicals were commercial products of reagent grade.
Preparation of peritoneal PMNs and immunostaining
Female mice were injected i.p. with 2 ml of 5% (w/v) casein in sterile saline and were sacrificed by cervical dislocation 6 h after injection. The lavage fluid (4 ml) was collected in a syringe, and exudated peritoneal cells were precipitated by centrifugation. PMNs were further purified from the peritoneal cells by Percoll stepwise density gradient (1.090 and 1.070 g/ml) centrifugation (600 × g for 20 min at 4°C). The purity of PMNs was >95% as determined by staining with May-Grünwald-Giemsa. PMNs were suspended in RPMI 1640 medium containing 10% heat-inactivated FBS, 150 μM 2-ME, and 100 μM sodium pyruvate. Immunofluorescence studies were performed as described previously (20). LPS-stimulated PMNs were centrifuged onto round coverglasses (φ = 18.0 mm), which were then placed in 12-well culture plates. For G-CSF staining, a goat anti-G-CSF Ab (1:50) (Santa Cruz Biotechnology) was used as the primary Ab, and a rhodamine-conjugated donkey anti-goat IgG Ab (1:200) (Chemicon International) was used as the secondary antibody. A FITC-conjugated rat anti-Gr-1 Ab (1:200) (BD Pharmingen) was used for Gr-1 staining. The stained cells were observed using a confocal laser scanning microscope, LSM5Pascal (Carl Zeiss). No significant staining was observed when cells were incubated without primary Ab (G-CSF) or incubated with FITC-conjugated isotype control (Gr-1) (Fig. 1 A).
Measurement of G-CSF production and cAMP accumulation
PMNs (5 × 105 cells/well) were incubated with medium containing FBS (10%) or BSA (1%) with or without 100 ng/ml LPS for the indicated times at 37°C in 5% CO2. After incubation, each culture was centrifuged at 300 × g for 5 min at 4°C to remove the cells. The amounts of G-CSF in the supernatant were assayed using ELISA kits, according to the manufacturer’s instructions (R&D Systems). The values obtained from the medium only were used as the value at 0 h. The cAMP levels in PMNs were determined as described previously (5). Cells (5 × 105 cells/well) were washed with Krebs-HEPES buffer (pH 7.4) containing 100 μM Ro-20-1724 and preincubated for 10 min. Reactions were started by the addition of test reagents along with 100 μM Ro-20-1724. After incubation for the indicated time at 37°C, reactions were terminated by the addition of 10% trichloracetic acid. The cAMP content of the cells was then measured by the cAMP radioimmunoassay kit (Amersham Biosciences).
RT-PCR and microarray analysis
For RNA extraction, PMNs (3.5–5 × 107 cells/ml) were collected and total RNA prepared by the acid guanidinium thiocyanate-phenol chloroform method (21). For RT-PCR analysis, cDNA was amplified using primers specific for G-CSF and GAPDH. GAPDH primers were purchased from Invitrogen Life Technologies. The primer sequences for G-CSF were 5′-CTGTGGCAAAGTGCACTATGGTCAGGACG-3′ (sense primer) and 5′-GGATGTTGCCAACTTTGCCACCACCAT-CTG-3′ (anti-sense primer). For microarray analysis, total RNA was isolated by a combination of the guanidinium thiocyanate-phenol chloroform method and RNeasy column chromatography (Qiagen). The obtained RNA was amplified, labeled, and prepared for hybridization to GeneChip Murine Expression 430 oligonucleotide arrays (Affymetrix) using standard methods as described previously (22).
Mice were injected i.p. with 2 ml of a 5% (w/v) solution of casein in PBS. Indomethacin (10 mg/kg) was s.c. injected 2 h before the casein treatment. In additional experiments, peritoneal inflammatory responses were induced by an i.p. injection of 2 ml of 0.3% sodium thioglycollate (w/v in distilled water). At the indicated times after injection, mice were sacrificed by cervical dislocation. The abdominal cavity was then injected with 4 ml of PBS and massaged to ensure adequate mixing of the cell population with the harvesting fluid. Cells in the peritoneal cavity were then collected using a syringe. Cytocentrifuge preparations of peritoneal cells were made and then stained with May-Grünwald-Giemsa. The G-CSF and MIP-2 contents in harvested peritoneal washout fluids were assayed with ELISA kits, and PGE2 contents were assayed with an enzyme immunoassay kit (Cayman Chemical).
Data are shown as means ± SEM. Comparison of two groups was performed by the Student’s t test. For comparison of more than two groups with comparable variances, one-way ANOVA was performed first, and then the Dunnett’s test was used to evaluate the pairwise group difference. The presented results are representative of at least three independent experiments.
Effects of PGE2 on LPS-stimulated G-CSF release from mouse peritoneal PMNs
To uncover the potential roles of PGE2 in neutrophils, we examined the effects of PGE2 on LPS-stimulated production of G-CSF in mouse peritoneal PMNs. In PMNs purified from mouse peritoneal cells, G-CSF was immediately produced upon LPS treatment. More than 95% of the PMN preparation were Gr-1+ cells, all of which were also stained with the anti-G-CSF Ab (Fig. 1,A). No significant staining was observed in Gr-1− cell or when cells were incubated with FITC-conjugated isotype control in the absence of anti-G-CSF Ab. PGE2 at 1 μM augmented the effects of LPS on G-CSF production by ∼2- to 3-fold (Fig. 1,B). Such an enhancing effect of PGE2 on G-CSF production was observed even at 10 nM, but carbacyclin, a stable prostacyclin analog, failed to enhance G-CSF production even at 1 μM (Fig. 1 C). We further tested the possibility that the PGE2 synthesized by neutrophils themselves may stimulate LPS-induced G-CSF production. However, pretreatment of the cells with indomethacin failed to alter the LPS-induced G-CSF release from neutrophils.
PGE2 augmented LPS-induced G-CSF production via the EP2 by stimulating cAMP accumulation in mouse peritoneal PMNs
There are four subtypes of PGE2Rs (EPs), which differ in their signal transduction pathways. We previously demonstrated that peritoneal PMNs express mRNAs for the Gs-coupled EPs, EP2 and EP4, but not for the Gq-coupled EP1 nor the Gi-coupled EP3 (23). Indeed, when we performed the cAMP formation assay on peritoneal PMNs, PGE2 dose dependently increased cAMP content. In addition, AE1-329, an EP4 agonist, as well as AE1-259, an EP2 agonist, elicited cAMP accumulation, indicating that PMNs express functional EP2 and EP4 (Fig. 2,A). We next examined which EP agonist can mimic the PGE2 stimulation of G-CSF production in PMNs. However, among the four EP-selective agonists, only an EP2 agonist augmented LPS-induced G-CSF production while an EP4 agonist could not (Fig. 2,B). In the PMNs isolated from EP2-deficient mice, PGE2, as well as an EP2 agonist, failed to augment LPS-induced G-CSF production, although a membrane-permeable cAMP analog, dbcAMP, augmented the LPS-induced G-CSF response as observed in PMNs from wild-type mice (Fig. 2, C and D). These results suggested that PGE2 stimulates G-CSF production via the EP2 through an increase in cAMP levels.
PGE2 itself stimulates G-CSF gene expression in peritoneal PMNs
We next examined whether PGE2 has the ability to stimulate G-CSF release from PMNs in the absence of LPS. When the PMNs were cultured in medium supplemented with 1% BSA in the absence of LPS and FBS, G-CSF was undetectable in the culture medium. Even under such conditions, PGE2, as well as an EP2 agonist, stimulated G-CSF production of PMNs (Fig. 3,A). dbcAMP also stimulated G-CSF release in these cells (data not shown). To test the possibility that PGE2 via cAMP accumulation stimulates G-CSF gene expression, we examined the effects of EP agonists and dbcAMP on G-CSF mRNA levels in PMNs by using semiquantitative PCR. Although we did not detect G-CSF in the serum-free medium, the vehicle-treated cells showed a low level of basal expression. The cells exposed to 1 μM PGE2 or 1 μM of an EP2 agonist for 6 h showed an ∼5-fold increase in expression of G-CSF mRNA (Fig. 3 B). The cells treated with dbcAMP showed an 8.5-fold increase, whereas treatment with an EP4 agonist failed to significantly increase G-CSF mRNA levels. These results suggested that PGE2 stimulates G-CSF production via the EP2 at least in part by increasing G-CSF mRNA levels in PMNs.
To estimate the contribution of EP2 or cAMP signaling on the effect of PGE2 on PMNs, we examined the changes in gene expression profiles of PMNs treated with PGE2, an EP2 agonist, an EP4 agonist or dbcAMP. Of the ∼24,000 genes represented on the oligonucleotide array, genes with altered expression values upon PGE2 treatment (log ratio of PGE2 vs control was >1.3 or less than −1.3) were selected and regarded as PGE2-regulated genes (193 genes) (Fig. 3,C). Approximately 80% of the PGE2-regulated genes (151 genes) were regulated by dbcAMP in the same manner. An EP2 agonist also mimicked the effect of PGE2 in 68% of the PGE2-regulated genes (131 genes), whereas an EP4 agonist did so only in 12% of the PGE2-regulated genes (24 genes). Most of the EP2-regulated genes (122 genes, 93%) were not altered by an EP4 agonist, and 84% of such “EP2 only” genes (102 genes) showed PGE2-like responses also upon dbcAMP treatment. This group of genes, with their expression levels commonly altered by PGE2, an EP2 agonist and dbcAMP but not by an EP4 agonist, includes G-CSF (Csf3) and other genes known to be up-regulated by cAMP signals such as transferrin (Trf) and phosphodiesterase 7 (Pde7b) (Table I) (24, 25). Thus, expression changes upon treatment with PGE2, an EP2 agonist, and dbcAMP are closely associated, and the molecular events induced by PGE2 are likely to be caused by the activation of cAMP signaling. To further explore the possible difference in cAMP signaling between EP2 and EP4, we examined the time course of EP agonist-induced cAMP increase (Fig. 3 D). PGE2 increased cAMP levels over the basal until 30 min and sustained this level until 60 min. An EP2 agonist mimicked this profile even though its maximal activity was lower than PGE2. On the other hand, an EP4 agonist increased cAMP levels in a manner similar to an EP2 agonist until 30 min but failed to maintain this level until 60 min. Such chronic activation in cAMP signaling may be required for G-CSF gene expression. Based on these results, we conclude that PGE2 stimulation of G-CSF production is mediated by the persistent activation of cAMP signaling.
|Gene Symbol .||Gene Name .||Log2 (Fold) (vs Vehicle) .||.||.||.||Unigene ID .|
|.||.||.||PGE2 .||AE1–259 .||AE1–329 .||dbcAMP .|
|Crem||cAMP-responsive element modulator||3.6||2.2||NC||5.9||1238|
|Gene Symbol .||Gene Name .||Log2 (Fold) (vs Vehicle) .||.||.||.||Unigene ID .|
|.||.||.||PGE2 .||AE1–259 .||AE1–329 .||dbcAMP .|
|Crem||cAMP-responsive element modulator||3.6||2.2||NC||5.9||1238|
The effect of PGE2, AE1–259 (EP2 agonist), AE1–329 (EP4 agonist), and dbcAMP is indicated as a logarithm of the fold change vs the expression level of the vehicle. NC, not changed.
Inhibition of PGE2 synthesis and EP2 gene deficiency lead to decreased G-CSF levels in peritoneal exudates in casein-induced peritonitis
To explore the possibility that PGE2 synthesized locally within an inflammation site enhances G-CSF production in vivo, we used the casein-induced peritonitis model, in which neutrophils accumulate within the peritoneal cavity, and examined the relationship between the number of infiltrated neutrophils and G-CSF and PGE2 content in the peritoneal fluid (Fig. 4). In this model, casein injection into the peritoneal cavity was followed by immediate PMN infiltration preceding macrophage accumulation; infiltrated PMNs appeared at 2 h postinjection and by 6 h were increased by >10-fold. PMN number was maintained at high levels until 16 h postinjection. Macrophages were constant in number until 6 h postinjection and then transiently increased up to 5-fold at 16 h (data not shown). The PMN ratio in total peritoneal cells was 0% before the injection and increased after the injection to ∼50 and 90% for 2 and 6 h, respectively. In this peritonitis model, the casein injection initially activates peritoneal macrophages; activated macrophages are then thought to produce PGE2. Indeed, when we measured the PGE2 contents in the peritoneal exudates, rapid PGE2 production was observed; a significant amount of PGE2 could already be detected at 2 h postinjection and increased by 2-fold reaching a peak level at 6 h and was decreased by 16 h. On the other hand, G-CSF was undetectable at 2 h postinjection, but a large amount was detected at 6 h and decreased by 16 h. Thus, PGE2 production precedes the appearance of G-CSF in the peritoneal cavity.
To assess whether PGE2 triggers G-CSF production, we tested the effects of indomethacin pretreatment on the appearance of G-CSF release, as well as PMN infiltration. Subcutaneous administration of indomethacin 2 h before casein injection decreased PGE2 levels both at 2 and 6 h after the injection (Fig. 5). Under such conditions, the increase in PMN infiltration was markedly diminished at 6 h postinjection while the number of macrophages was not affected significantly. In parallel with the decrease in the number of infiltrated PMNs, the levels of G-CSF in the peritoneal exudates also decreased. No significant changes were observed in the levels of MIP-2, one of the critical chemokines that attracts neutrophils to the site of infection. Thus, the inhibition of endogenous PG synthesis was followed by a decrease in PMN infiltration, resulting in a reduction of G-CSF levels.
To evaluate the positive correlation between EP2 signaling and local G-CSF production, we examined the effects of EP2 gene deficiency on G-CSF contents in peritoneal exudates. In EP2-deficient mice, the casein injection was followed by an increase in PGE2 levels as seen in wild-type animals (Fig. 6). Although the injection-primed PMN infiltration and the numbers of peritoneal macrophages in EP2-deficient mice were indistinguishable from those of wild-type mice, the G-CSF content at 6 h postinjection was diminished to 30% in EP2-deficient mice compared with wild-type mice. Thus, EP2 deficiency results in low levels of G-CSF in the peritoneal exudates. Because the infiltrated PMNs would be exposed to PGE2, a decrease in local G-CSF production may at least be partly due to the lack of EP2 in PMNs.
One of the remarkable findings in the current study is that PGE2 appears to induce G-CSF release from peritoneal PMNs. PGE2 has been shown to positively or negatively regulate the production of cytokines and chemokines such as TNF-α, IL-1β, IL-8, IL-12, MCP-1, and MIP-1 from monocytes and neutrophils (2, 3, 23, 24, 25, 26, 27, 28, 29). PGE2 in most cases promotes or inhibits the production of cytokines and chemokines induced by stimuli such as LPS or TNF-α and thus has been considered to be a modulator of immune responses (30). Indeed, it has been shown that endogenous PGE2 contributes to LPS-induced G-CSF gene expression in monocytes (12, 13). Hareng et al. (13) recently demonstrated that dbcAMP enhances LPS-primed promoter activity via a cAMP-responsive element located at −240 bp of the G-CSF gene in THP-1 cells. We also found that indomethacin inhibits LPS-induced G-CSF production and that PGE2 reverses this inhibition in the murine resident peritoneal macrophages (D. Mori, S. Tsuchiya, and Y. Sugimoto, unpublished observation). The enhancing effects of PGE2 on G-CSF production in monocytes/macrophages require an input of the LPS signal; PGE2 itself does not induce basal G-CSF production. In contrast, in peritoneal PMNs, PGE2, or dbcAMP alone appears to induce G-CSF production as shown in the current study. However, although G-CSF levels were undetectable in the serum-free culture medium, we detected a faint but significant amount of mRNA expression for the G-CSF gene in the PMNs in the absence of PGE2 (Fig. 3, A and B). Because the peritoneal PMNs have already received activation signals from multiple inputs such as adhesion molecules and chemoattractants during migration, the infiltrated cells may still have active signals in particular intracellular signaling cascades such as the MAPK- or ρ-cascades (31, 32, 33). We tested whether PGE2 alone can really induce G-CSF production in bone marrow-derived PMNs (20), but marrow neutrophils failed to stimulate production and gene expression of G-CSF upon PGE2 or dbcAMP treatment (data not shown). The inability of marrow neutrophils to respond to PGE2 suggested that a preactivated signal associated with migration is necessary for the basal expression of the G-CSF gene. Thus, PGE2 is likely to augment such basal promoter activity of the G-CSF gene. However, this PGE2 potentiation is considered to play a key role in local G-CSF production in the peritoneal cavity as discussed below.
The peritoneal PMNs used in the current study express functional prostanoid receptors EP2 and EP4, but only EP2 can sustain intracellular cAMP increase for >30 min (Fig. 3,D). We previously found in the peritoneal macrophages that both gene expression and cAMP-producing activity of EP4 is down-regulated by LPS treatment (5). In the peritoneal PMNs, EP4 gene expression was constant during PGE2 and/or LPS treatment (data not shown), indicating that the inability of EP4 to increase cAMP for a longer period is not due to down-regulation of EP4 gene expression. In our initial analysis, the two Gs-coupled receptors EP2 and EP4 were different in their sensitivity to agonist-induced desensitization when expressed in CHO cells (34, 35); EP4 undergoes agonist-induced desensitization, but no desensitization occurs in EP2. Such agonist-induced desensitization of the EP4 may cause its inability to sustain increases in cAMP levels for longer periods. When the PMNs were preincubated with an EP2 agonist for 30 min followed by incubation in the absence of this agonist for 6 h, the cells were no longer able to produce a significant amount of G-CSF (data not shown). Therefore, persistent increase in cAMP levels is likely to be required for G-CSF production in the peritoneal PMNs, although its precise mechanism is unknown. Moreover, PGE2-elicted changes in gene expression profiles of PMNs were mimicked by an EP2 agonist and dbcAMP but not by an EP4 agonist (Fig. 3,C). Because membrane permeable dbcAMP is considered to constantly activate cAMP-dependent signaling, these results support the notion that PGE2 alters gene expression profiles through chronic activation of cAMP signaling by the EP2. It is interesting in this respect that genes functionally antagonizing cAMP signaling such as cAMP-specific phosphodiesterase Pde7b and cAMP-responsive element modulator Crem were up-regulated by an EP2 agonist and dbcAMP (Table I).
In the in vivo studies, we found a positive association between PGE2 levels and G-CSF contents in peritoneal exudates in the casein-induced peritonitis model; both G-CSF and PGE2 contents increased at 6 h and decreased at 16 h postinjection (Fig. 4), and attenuation of PGE2 production by indomethacin decreased G-CSF levels at 6 h postinjection (Fig. 5). The absence of an effect of indomethacin in the in vitro system suggested that PMNs themselves produce only a small amount of PGE2. Therefore, resident macrophages are likely to be a source of PGE2 critical for G-CSF production in the peritoneal cavity. In a previous report, Metcalf et al. found in the casein-induced peritonitis that the appearance of G-CSF in the peritoneal cavity correlates with the appearance of PMNs and that the concentration of G-CSF in the peritoneal fluid is higher than in the serum (11). Based on these findings, they concluded that G-CSF locally produced in the peritoneal cavity should be responsible for the increased G-CSF levels in the circulation. Then, which cell type is responsible for the G-CSF production, resident macrophages, or the newly elicited neutrophils? We found that in the murine resident peritoneal macrophages, PGE2 augmented LPS-induced G-CSF production as Hareng et al. (13) reported for THP-1 cells. However, in the case of the macrophages, EP4 is responsible for the PGE2 potentiation of G-CSF release because an EP4 agonist mimicked the effect of PGE2 more effectively than an EP2 agonist (D. Mori, S. Tsuchiya, and Y. Sugimoto, unpublished observation). Therefore, EP2 is unlikely to be involved in G-CSF release from macrophages, and a striking reduction in G-CSF levels in the EP2-deficient mice can be ascribed to loss of G-CSF production from the PMNs (Fig. 6). Thus, we conclude that EP2 signaling is essential for local G-CSF production and also that the PMNs are the physiologically critical source of G-CSF production at least in this peritonitis model.
Casein-induced PMN infiltration was markedly attenuated by the inhibition of endogenous PG synthesis but not by EP2 gene disruption. Such a difference in cell profiling suggests that some prostanoid receptor other than EP2 participates in this process. We previously demonstrated that PGI2 is critical for an inflammation-associated increase in vascular permeability; PGI2R-deficient mice showed a striking reduction in exudate volume in carrageenan-induced pleurisy (36), suggesting that PGI2 is the major prostanoid affecting vascular permeability in this model. Alternatively, other prostanoid receptor systems may be suppressing some chemotactic factors essential for neutrophil migration such as MIP-2 (37). Indeed, some prostanoids, including PGE2, have been suspected to suppress MIP-2 production (38). However, we failed to detect a significant decrease in peritoneal MIP-2 levels in the indomethacin-treated mice (Fig. 5). Recently, mice null for the IL-17R were reported to be impaired in neutrophil recruitment and in the local production of G-CSF in a Klebsiella pneumoniae lung infection model (39). Involvement of the endogenous PG system in IL-17 signaling is a likely possibility that should be examined.
In conclusion, we found that PGE2-EP2 signaling plays a key role in inflammation-related G-CSF production. Recombinant G-CSF has been clinically used as a therapeutic drug for neutropenia (9). Interestingly, it has been reported that the occurrence of neutropenia is associated with the use of aspirin-like drugs (40) and that chronic aspirin treatment in vivo but not in vitro results in functionally immature blood neutrophils (41). Whether PGE2 contributes to physiological G-CSF production in marrow cells or granulocytes is an interesting issue to be addressed.
We thank Dr. Sachiko Oh-ishi (Kitasato Institute, Tokyo, Japan) for her invaluable discussions and helpful advice on this work. We are grateful to Helena A. Popiel and Sachiko Terai-Yamaguchi for careful reading and secretarial assistance.
The authors have no financial conflict of interest.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This work was supported in part by a grant from the Sankyo Foundation of Life Science, Grant-in-Aid for Scientific Research on Priority Areas (14013036, 14021053, 15019050, 15012234, and 16012238) from the Ministry of Education, Culture, Sports Science, and Technology of Japan, and from the Ministry of Health and Labor of Japan.
Abbreviations used in this paper: PMN, polymorphonuclear leukocyte; CHO, Chinese hamster ovary; dbcAMP, dibutyryl cAMP.