A Proinflammatory Role for the Cyclopentenone Prostaglandins at Low Micromolar Concentrations: Oxidative Stress-Induced Extracellular Signal-Regulated Kinase Activation Without NF-κB Inhibition¹

The cyclopentenone PGs (cyPGs)⁴ are naturally occurring derivatives of PGs E₁, E₂, and D₂ (1). The cyPGs PGA₁, PGA₂, and PGJ₂ are formed by dehydration within the cyclo-hydroxy-pentanone ring of PGE₁, PGE₂, and PGD₂, respectively. PGJ₂ is further spontaneously converted to the cyPGs Δ¹²-PGJ₂ and 15-deoxy-Δ^12,14-PGJ₂ (15d-PGJ₂).

It has been reported that cyPGs may have antitumor, antiviral, and anti-inflammatory activity (2, 3, 4, 5, 6). Although cyPGs may bind to unspecific G-protein coupled prostanoid receptors with low affinity (1, 7), reported cyPG effects are different from those mediated by specific binding to these receptors (1). The consensual opinion is that cyPGs elicit biological responses by interacting with intracellular target proteins, mainly signaling proteins and transcription factors. For example, 15d-PGJ₂ is a high affinity ligand for the peroxisome proliferator-activated receptor (PPAR)γ and is thought to exert some of its effects through binding to this nuclear receptor (1, 8, 9). cyPGs also differ from other eicosanoids in that they lack stereoselectivity (4, 5, 8, 9, 10, 11). Moreover, cyPGs require micromolar concentrations to elicit biological effects, whereas other eicosanoids, such as leukotrienes and lipoxins, usually evoke bioactions in the nanomolar or subnanomolar range (12, 13, 14).

The anti-inflammatory properties of cyPGs have generated considerable interest. Elevated endogenous 15d-PGJ₂ production has been associated with resolution of inflammation in carrageenin-induced pleurisy in rats, suggesting a physiological anti-inflammatory role for cyPGs (6). An anti-inflammatory effect of exogenous cyPGs, particularly 15d-PGJ₂, has also been demonstrated in vivo and in vitro (6, 15, 16, 17, 18, 19), raising the hypothesis that cyPGs may have therapeutic value in the treatment of many inflammatory diseases (1, 6, 15, 16, 17, 18, 19).

The anti-inflammatory activity of cyPGs appears to be mediated through inhibition of NF-κB activity (19, 20, 21, 22, 23), which plays a key role in inflammatory gene expression (24) and inflammatory and immune cell survival (25, 26, 27, 28). cyPG-induced NF-κB inhibition may proceed by three distinct mechanisms. First, cyPGs may activate PPARγ, which antagonizes NF-κB transcriptional activity (17, 18, 23). Second, cyPGs can directly inhibit the IκB kinase-β (21, 22, 23), a key activator of NF-κB (29, 30). Third, cyPGs may directly block binding of NF-κB to target DNA sequences (23, 31).

Disconcertingly, exogenous cyPGs inhibit NF-κB only at concentrations substantially higher (viz, micromolar concentrations) (20, 21, 22, 23) than those of endogenous cyPGs present at the site of inflammation (viz, nanomolar concentrations) (6), thus challenging the hypothesis that cyPGs are naturally occurring inhibitors of inflammation and suggesting that cyPGs at low concentrations might have previously unappreciated effects. Therefore, in this study, using various cell types, we examined the effects of two cyPGs, PGA₁ and 15d-PGJ₂, used at concentrations significantly lower than that required for NF-κB inhibition, on the inflammatory response to TNF-α, a potent proinflammatory cytokine.

A549, U937, and HeLa cells were obtained from the American Type Culture Collection (Manassas, VA). Human blood neutrophils were obtained from buffy coats (Transfusion Center, Liège, Belgium). Neutrophils were separated from mononuclear cells by density centrifugation (Histopaque; Sigma-Aldrich, Bornem, Belgium). Contaminating erythrocytes were removed from the neutrophil fraction by hypotonic lysis. Neutrophil purity, as determined by counting of cytospin preparations stained with Diff-Quick (Dade Behring, Dudingen, Germany), was always > 95%. Cells were cultured in either Ham F-12 (A549 cells), DMEM (HeLa cells) or RPMI 1640 (U937 cells and granulocytes), supplemented with 10% FCS, 1% glutamine, 50 μg/ml streptomycin, and 50 IU/ml penicillin.

Human rTNF-α was purchased from Roche (Mannheim, Germany). PGA₁ and 15d-PGJ₂ were obtained from Cayman Chemicals (Ann Harbor, MI). PD 98059 was from Calbiochem (Darmstadt, Germany), and actinomycin D and mannitol were from Sigma-Aldrich.

Nuclear protein extracts were prepared as previously described (32). Cytoplasmic buffer contained 10 mM HEPES (pH 7.9), 10 mM KCl, 2 mM MgCl₂, 0.1 mM EDTA, 0.2% (v/v) Nonidet P-40, and 1.6 mg/ml protease inhibitors (Complete; Roche). Pelleted nuclei were resuspended in 20 mM HEPES (pH 7.9), 1.5 mM MgCl₂, 0.2 mM EDTA, 0.63 M NaCl, 25% (v/v) glycerol, and 1.6 mg/ml protease inhibitors (nuclear buffer), incubated for 20 min at 4°C and centrifuged for 30 min at 14,000 rpm. Protein amounts were quantified with the Micro BCA protein assay reagent kit (Pierce, Rockford, IL).

Binding reactions were performed for 30 min at room temperature with 5 μg of nuclear proteins in 20 mM HEPES (pH 7.9), 10 mM KCl, 0.2 mM EDTA, 20% (v/v) glycerol, 1% (w/v) acetylated BSA, 3 μg of poly(dI-dC) (Amersham Biosciences, Aylesbury, U.K.), 1 mM DTT, 1 mM PMSF, and 100,000 cpm of ³²P-labeled double-stranded oligonucleotide probes. Probes were prepared by annealing the appropriate single-stranded oligonucleotides (Eurogentec, Liege, Belgium) at 65°C for 10 min in 10 mM Tris, 1 mM EDTA, 10 mM NaCl, followed by slow cooling to room temperature. The probes were then labeled by end-filling with the Klenow fragment of Escherichia coli DNA polymerase I (Roche), with [α-³²P] dATP and [α-³²P]dCTP (NEN, Les Ulis, France). Labeled probes were purified by spin chromatography on Sephadex G-25 columns (Roche). DNA-protein complexes were separated from unbound probe on 4% native polyacrylamide gels at 150 V in 0.25 M Tris, 0.25 M sodium borate, and 0.5 mM EDTA (pH 8.0). Gels were vacuum-dried and exposed to Fuji x-ray film at −80°C for 12 h. To confirm specificity, competition assays were performed with a 50-fold excess of unlabeled wild-type probes and with mutated probes (data not shown). Binding of the noninducible transcription factor Oct-1 was always used as an internal standard (data not shown). The sequences of the wild-type probes were as follows: NF-κB, 5′-CAA CGG CAG GGG AAT TCC CCT CTC CTT AGG TT-3′; AP-1, 5′-CGC TTG ATG AGT CAG CCG GAA-3′; CREB, 5′-AGA GAT TGC CTG ACG TCA GAG AGC TAG-3′; C/EBP, 5′-CTA GGC ATA TTG CGC AAT AT-3′; NF-AT, 5′-TCG ACC AAA GAG GAA AAT TTG TTT CAT ACA GAG-3′; OCT-1, 5′-TGT CGA ATG CAA ATC ACT AGA A-3′.

The concentration of IL-8, IL-6, and GM-CSF in cell supernatants was measured using ELISA kits (BioSource International, Nivelles, Belgium).

Cell proliferation was assayed using the cell proliferation reagent WST-1 (Roche) according to the manufacturer’s instructions. Apoptosis and necrosis were assessed by staining with Annexin V-FITC and propidium iodide using the Annexin-V-FLUOS Staining kit (Roche), following the recommendations of the manufacturer. Flow cytometry analyses were performed with a FACStar^Plus (BD Biosciences, Erembodegem, Belgium).

Neutrophil chemotaxis was assessed using 48-well microchemotaxis chambers divided in two compartments by a polyvinyl propylene-free polycarbonate membrane with 5-μm pores (Nuclepore, Whatman, Maidstone, U.K.). A549 cell supernatants were added to the lower compartment of the chambers and 0.5 × 10⁵ neutrophils were added to the upper compartment. The chambers were incubated at 37°C for 90 min, and the filters were removed, fixed, and stained with Diff-Quick. A single-blind assessment of chemotaxis was conducted by coding the slides before the number of migrated neutrophils/oil-immersion field (magnification, ×400) was counted by an unbiased observer.

The pNF-κB-Luc reporter construct was purchased from Stratagene (La Jolla, CA). IL-8 and IL-6 promoter-containing plasmids p1481hu.IL8P-Luc and p1168hu.IL6P-Luc were kindly provided by Dr. G. Haegeman (University of Ghent, Ghent, Belgium). Transient transfection of A549 cells was performed using Fugene 6 (Roche) according to the manufacturer’s instructions. After 24 h, cells were incubated in the presence or absence of cyPGs for 2 h, and then treated with TNF-α for another 6 h. Then cells were washed twice in PBS and lysed. Luciferase activities were determined by using the Luciferase Gene Assay Chemiluminescent kit (Roche), and were normalized for the amount of protein.

Total RNA was extracted from cells using the RNeasy Mini kit according to manufacturer’s instructions (Qiagen, Hilden, Germany). RPAs were performed as recommended by the manufacturer (BD PharMingen, San Diego, CA) using 10 μg of total RNA and [³²P]αUTP-labeled antisense RNA probes specific for human IL-8, IL-6, GM-CSF, and GAPDH.

Equal amounts of whole cell lysates were subjected to SDS-PAGE under reducing conditions, and proteins were electrotransferred to polyvinylidene difluoride membranes (Roche). The membranes were blocked for 1 h at room temperature with 5% milk in 1× TBS with 0.1% Tween 20 and incubated overnight at 4°C with 1/1000 phosphospecific anti-p38, phosphospecific anti-c-Jun N-terminal-kinase (JNK), or phosphospecific anti-extracellular signal-regulated kinase (ERK) 1/2 Abs (New England Biolabs, Beverly, MA). The blots were then incubated for 45 min with HRP-conjugated secondary Abs. Immunoreactive bands were revealed using the ECL detection method (ECL kit; Amersham Biosciences). Equal loading of proteins on the gel was always confirmed by probing the blots for α-tubulin (data not shown).

Cell lysate preparation, immunoprecipitation of active ERK1/2, and subsequent assessment of ERK1/2 activity were performed using a commercial kit (p44/42 MAP Kinase Assay kit; New England Biolabs), in which GST-Elk-1 307–428(307–428) serves as a specific substrate for ERK1/2.

Production of ROS was assessed by incubating the cells for 30 min in the presence of 100 μM 2′,7′-dichlorodihydrofluorescein diacetate (Molecular Probes, Leiden, The Netherlands). Trypsinized cells were then assessed for fluorescence emission by flow cytometry using a FACStar^Plus.

Data are presented as means ± SD. The differences between mean values were estimated using either an ANOVA with subsequent Fisher’s protected least significant difference tests or a Student t test for unpaired data. A value of p < 0.05 was considered significant. All presented results are representative of at least three similar experiments.

First, using A549 epithelial cells, we explored the effects of PGA₁ and 15d-PGJ₂, used at concentrations lower than that required for NF-κB inhibition, on the production of three cytokines critical for initiating and maintaining the inflammatory process, namely IL-8, IL-6, and GM-CSF. In TNF-α-stimulated A549 cells, both cyPGs inhibited, in a dose-dependent manner, NF-κB activation (Fig. 1,A) and cytokine production (Fig. 1,B), with peak inhibition reached at 96 μM PGA₁ and 36 μM 15d-PGJ₂. However, up to the threshold concentration of 12 μM, neither PGA₁ nor 15d-PGJ₂ inhibited NF-κB activity (Fig. 1,A) or decreased cytokine secretion (Fig. 1,B). Rather, at low micromolar concentrations ranging from 3 to 12 μM, 15d-PGJ₂, but not PGA₁, significantly promoted TNF-α-induced cytokine expression, with peak potentiation reached at 12 μM for IL-8 and 3 μM for IL-6 and GM-CSF (Fig. 1 B). When used at nanomolar concentrations (viz, 10–1000 nM), neither PGA₁ nor 15d-PGJ₂ affected TNF-α-triggered cytokine production in A549 cells (data not shown).

IL-8 is a potent neutrophil chemoattractant (33), and IL-6 and GM-CSF delay neutrophil apoptosis, thus contributing to the persistence of the inflammatory response (34, 35, 36). To determine whether elevated IL-8, IL-6, and GM-CSF concentrations, which are detected in the supernatants of A549 cells cotreated with TNF-α and 15d-PGJ₂ at low micromolar concentrations, have biological activity, we tested the ability of these supernatants to recruit neutrophils and increase their viability. As illustrated in Fig. 1 C, supernatants from cells cotreated with TNF-α and 15d-PGJ₂ at low micromolar concentrations displayed increased neutrophil chemotactic and prosurvival activities, when compared with supernatants from cells treated with TNF-α alone or in combination with low micromolar concentrations of PGA₁. CyPGs alone and cyPG solvents, namely ethanol (PGA₁) and methyl acetate (15d-PGJ₂), did not affect NF-κB activity and cytokine production (data not shown). CyPGs also had no toxic effect on A549 cells at the concentrations used, as determined by proliferation and apoptosis/necrosis assays (data not shown).

To elucidate the molecular mechanisms by which low micromolar concentrations of 15d-PGJ₂ potentiate cytokine expression in TNF-α-activated A549 cells, transfection experiments were undertaken using IL-8 and IL-6 promoter-luciferase reporter constructs. These experiments indicated that 15d-PGJ₂, particularly at 3 μM, can enhance induction of the IL-6, but not the IL-8, promoter in response to TNF-α (Fig. 2,A). To further explore the mechanisms by which 15d-PGJ₂ exerts proinflammatory effects, RPAs were performed. IL-8 (Fig. 2,B, upper panel), IL-6 (Fig. 2,B, lower panel), and GM-CSF (data not shown) transcripts were significantly elevated in cells cotreated with TNF-α and low micromolar concentrations of 15d-PGJ₂, as compared with cells treated with TNF-α alone or TNF-α in combination with low micromolar concentrations of PGA₁. Stability of cytokine mRNA was also assessed in similar experiments using actinomycin D, which blocks de novo RNA synthesis. The levels of IL-6 and GM-CSF mRNA decreased promptly in TNF-α-treated cells after RNA synthesis was inhibited, even in the presence of cyPGs (Fig. 2,B, lower panel, and data not shown). In contrast, 15d-PGJ₂, unlike PGA₁, stabilized IL-8 mRNA in TNF-α-stimulated A549 cells (Fig. 2 B, upper panel). Taken together, these results indicate that 15d-PGJ₂ at low micromolar concentrations potentiates IL-6 and GM-CSF expression at the transcriptional level, whereas it enhances IL-8 production at a posttranscriptional level.

To further investigate the mechanisms responsible for the proinflammatory effects of 15d-PGJ₂ at low micromolar concentrations, we next examined whether this compound might increase the activity of transcription factors, namely NF-κB, AP-1, CREB, C/EBP, and NF-AT, and mitogen-activated protein kinases (MAPK), namely p38, JNK, and ERK1/2, which all may up-regulate inflammatory gene expression at either transcriptional or posttranscriptional levels. 15d-PGJ₂ at low micromolar concentrations did not increase TNF-α-induced NF-κB activation in A549 cells, as demonstrated by EMSAs (Fig. 1,A) and assays performed with the pNF-κB-Luc reporter construct (data not shown). Moreover, 15d-PGJ₂ did not modulate AP-1, CREB, C/EBP, and NF-AT activity, as assessed by EMSAs (data not shown). Neither TNF-α nor cyPGs could activate p38 and JNK in A549 cells, even when TNF-α and cyPGs were combined (data not shown). Conversely, both PGA₁ and 15d-PGJ₂ synergized with TNF-α to activate ERK1/2 (Fig. 3,A). Indeed, although TNF-α and cyPGs alone could induce ERK1/2 activity (Fig. 3,A and data not shown), activation was significantly higher when TNF-α and cyPGs were combined. However, PGA₁ only significantly increased TNF-α-induced ERK1/2 activity at concentrations equal or higher than those required for NF-κB inhibition, whereas 15d-PGJ₂ enhanced TNF-α-mediated ERK1/2 activation in a dose-dependent manner from a low concentration of 3 μM to a high concentration of 36 μM (Fig. 3,A). To examine whether low micromolar concentrations of 15d-PGJ₂ promote cytokine expression via ERK1/2 activation, experiments were performed with PD 98059, a specific inhibitor of the kinase upstream of ERK1/2, the mitogen-activated/extracellular signal-regulated kinase 1/2. PD 98059 moderately but significantly reduced TNF-α-induced cytokine production and potently decreased 15d-PGJ₂-mediated potentiation of IL-8, IL-6, and GM-CSF expression in TNF-α-treated A549 cells (Fig. 3 B, and data not shown). Absence of PD 98059 toxicity was verified by proliferation and apoptosis/necrosis assays (data not shown). ELISAs also showed that DMSO, the PD 98059 solvent, has no effect on cytokine production at the concentrations used (data not shown).

ERK1/2 may be activated by ROS (37). We thus examined whether ROS production might account for cyPG-induced ERK1/2 activation. PGA₁ and 15d-PGJ₂ substantially increased ROS production in A549 cells, whereas TNF-α did not generate intracellular oxidative stress (Fig. 4,A, and data not shown). Moreover, the level of cyPG-induced ROS production paralleled the level of cyPG-mediated ERK1/2 activation in TNF-α-stimulated cells (compare Fig. 3,A and Fig. 4,A). Indeed, PGA₁ significantly generated oxidative stress only at concentrations equal or higher than those required for NF-κB inhibition, whereas 15d-PGJ₂ significantly increased ROS production, even at the low concentration of 3 μM (Fig. 4,A). To definitely establish a role for ROS production in cyPG-mediated ERK1/2 activation, it was necessary to inhibit cyPG-induced oxidative stress, which raised methodological problems. Commonly used antioxidants, such as N-acetyl cysteine and pyrrolidine dithiocarbamate, possess nucleophilic groups, such as free sulfhydryls that make these compounds susceptible to Michael addition reactions with the α,β-unsaturated carbonyl group of cyPGs. Because conjugation of cyPGs with molecules bearing nucleophilic groups may totally eliminate cyPG bioactivity (1, 38), use of classical antioxidants could lead to misinterpretation of results. Therefore, we used mannitol, a ROS scavenger unable to form Michael adducts with cyPGs. Mannitol was used at nontoxic concentrations, as determined by proliferation and apoptosis/necrosis assays (data not shown). Mannitol completely inhibited ROS production following treatment with 15d-PGJ₂ at low micromolar concentrations and significantly reduced ROS production following treatment with high concentrations of both cyPGs (Fig. 4,A). Moreover, mannitol prevented enhancement of ERK1/2 activation by low micromolar concentrations of 15d-PGJ₂ in TNF-α-stimulated cells and significantly, but not completely, reduced enhancement of ERK1/2 activation by high concentrations of PGA₁ and 15d-PGJ₂ (compare Fig. 4,B and Fig. 3,A). Finally, mannitol greatly decreased 15d-PGJ₂-mediated potentiation of IL-8, IL-6, and GM-CSF production in TNF-α-stimulated cells, whereas it had no effect on cytokine expression in cells stimulated with TNF-α alone or in combination with PGA₁ at low micromolar concentrations (Fig. 4 C).

CyPGs may activate PPARγ (1, 8, 9, 10). To ascertain that the proinflammatory effects of cyPGs were independent on PPARγ ligation and not restricted to A549 cells, the experiments performed with A549 cells were repeated using HeLa epithelial cells, which do not express PPARγ (17), and U937 monocytic cells. HeLa cells expressed IL-8 and IL-6, but not GM-CSF, in response to TNF-α, whereas stimulated U937 cells produced only IL-8. Both 15d-PGJ₂ and PGA₁, when used at micromolar concentrations lower than that required for NF-κB inhibition (data not shown), considerably potentiated cytokine production in TNF-α-stimulated HeLa and U937 cells, particularly in the latter, where an ∼7-fold increase in IL-8 expression was observed following treatment with 12 μM 15d-PGJ₂ (Fig. 5 and data not shown). Optimal potentiation was obtained with 12 and 24 μM cyPGs in U937 and HeLa cells, respectively. When used at nanomolar concentrations, neither PGA₁ nor 15-dPGJ₂ modified TNF-α-induced cytokine production in these cells (data not shown). Inhibition of cyPG induced oxidative stress using mannitol and inhibition of ERK1/2 activation using PD 98059 drastically reduced cyPG-mediated potentiation of IL-8 production in TNF-α-stimulated U937 cells (Fig. 5,A), and totally blocked enhancement of IL-8 and IL-6 expression by cyPGs in activated HeLa cells (Fig. 5 B, and data not shown).

Previous studies of the biological activities of cyPGs indicated that these compounds may act as inhibitors of inflammatory gene expression in various cell types (17, 18, 39, 40, 41). cyPG-mediated suppression of inflammatory gene induction appears to be the result of NF-κB inhibition (19, 20, 21, 22, 23). In the present study, we have explored the effects of two cyPGs, PGA₁ and 15d-PGJ₂, used at concentrations lower than that required for NF-κB inhibition, on TNF-α-induced proinflammatory cytokine expression. We showed that both cyPGs, when used at low micromolar concentrations, may significantly promote TNF-α-induced cytokine production in various cell types. Moreover, supernatants from A549 cells cotreated with TNF-α and 15d-PGJ₂ at low micromolar concentrations displayed increased neutrophil chemotactic and prosurvival activities when compared with supernatants from cells treated with TNF-α alone, demonstrating that cyPG-mediated potentiation of cytokine expression may be biologically relevant. Our study is the first to establish a proinflammatory role for cyPGs when used at concentrations lower than that required for NF-κB inhibition.

Reporter gene assays and RPAs performed in A549 cells indicated that 15d-PGJ₂ at low micromolar concentrations potentiates IL-6 and GM-CSF expression at the transcriptional level, whereas it enhances IL-8 production at a posttranscriptional level. Induction of IL-6 and GM-CSF gene transcription critically depends on NF-κB activation (24, 42, 43). However, the maximum response requires additional transcription factors, including AP-1, CREB, C/EBP, and NF-AT (42, 43). In A549 cells, 15d-PGJ₂ at low micromolar concentrations failed to enhance NF-κB, AP-1, CREB, C/EBP, and NF-AT activity, indicating that this compound does not potentiate IL-6 and GM-CSF gene expression through activation of the most important transcription factors involved in induction of these genes. Another mechanism by which inflammatory genes are up-regulated is through activation of members of the MAPK family, such as p38, JNK, and ERK1/2 (42, 44, 45, 46, 47, 48). For example, TNF-α-induced IL-6 gene expression requires IL-6 enhanceosome-activating signals delivered by MAPKs, which increase NF-κB-dependent IL-6 gene transcription (42). Moreover, MAPKs may also stabilize IL-8 mRNA, and subsequently potentiate IL-8 production, through a mechanism targeting the AU-rich elements responsible for IL-8 mRNA instability (46, 47, 48). These observations prompted us to examine whether MAPK activation could account for the proinflammatory effects of low micromolar concentrations of cyPGs. cyPGs, although unable to modulate p38 and JNK activity, synergized with TNF-α to activate ERK1/2. Moreover, ERK1/2 inhibition potently decreased PGA₁- and 15d-PGJ₂-mediated enhancement of proinflammatory cytokine expression in TNF-α-treated A549, HeLa, and U937 cells. These findings unambiguously show that the proinflammatory function of cyPGs is, at least partly, attributable to their ability to synergize with TNF-α to activate ERK1/2. Previous studies of the effects of cyPGs on MAPK activity have generated contradictory results, even when the same cell type was used, making these results difficult to interpret (16, 21, 23, 49, 50). However, our data are consistent with a previous report that 15d-PGJ₂ increases ERK1/2 activity in a dose-dependent manner in human mesangial cells, but has no effect on p38 or JNK (50).

ROS are potent activators of ERK1/2 (37). Accordingly, we postulated that ROS production might be involved in cyPG-induced ERK1/2 activation and subsequent potentiation of proinflammatory cytokine expression. Both cyPGs generated intracellular oxidative stress, the inhibition of which substantially reduced cyPG-mediated ERK1/2 activation and enhancement of cytokine production, confirming our hypothesis. A recent study also identified cyPGs as inducers of intracellular oxidative stress, 15d-PGJ₂ being more active than PGA₂, a compound structurally related to PGA₁ (11). These findings are consistent with our results that PGA₁ was less potent than 15d-PGJ₂ in inducing ROS production, and might explain the weaker ability of PGA₁ to enhance ERK1/2 activation and cytokine expression when compared with 15d-PGJ₂.

15d-PGJ₂ and PGA₁ are potent PPARγ activators (1, 8, 9, 10), and 15d-PGJ₂ has been demonstrated to inhibit NF-κB transcriptional activity and inflammatory gene expression in part through binding to this receptor (17, 18, 23). Interestingly, a recent study demonstrated that activation of PPARγ by 15d-PGJ₂ may specifically increase IL-8 expression in monocytes (41), thus raising the possibility that in some circumstances PPARγ may be involved in pathways leading to cytokine up-regulation. In the present study, cyPGs drastically promoted cytokine expression in HeLa cells, which do not express PPARγ (17), indicating that the proinflammatory effects of cyPGs are independent of this receptor. However, inhibition of oxidative stress-mediated ERK1/2 activation totally blocked cyPG-induced potentiation of cytokine expression in PPARγ-deficient HeLa cells, whereas it substantially, but not completely, prevented enhancement of cytokine production by cyPGs in A549 and U937 cells, which express PPARγ (40, 51). Although these observations demonstrate that ROS-induced ERK1/2 activation accounts to a large extent for the proinflammatory effects of cyPGs, they also indicate that a role for PPARγ in these effects cannot be excluded.

Taken together, our results suggest a molecular model that accounts for the proinflammatory effects of low micromolar concentrations of cyPGs (Fig. 6). In this model, cyPGs induce ROS production, thereby synergizing with TNF-α to activate ERK1/2. Once activated, ERK1/2 potentiates cytokine expression, the initiation of which essentially depends on TNF-α-induced NF-κB activity. ERK1/2 increases IL-6 and GM-CSF expression at the transcriptional level, whereas it stabilizes IL-8 mRNA. It cannot be ruled out that PPARγ activation also contributes to the proinflammatory activity of cyPGs in cell types that express this receptor, but the underlying mechanisms are presently unidentified. When used at higher concentrations, cyPGs significantly increase ROS production and subsequent ERK1/2 activation but concomitantly inhibit NF-κB, thus blocking cytokine expression.

The hypothesis that cyPGs are naturally occurring inhibitors of inflammation has been put forward based on the observation that the levels of 15d-PGJ₂ in inflammatory fluids increase during the resolution phase of inflammation (6). However, several lines of evidence challenge this concept. First, increased levels of 15d-PGJ₂ in inflammatory exudates have also been measured during the acute phase of inflammation (6). Second, although increased, the concentrations of 15d-PGJ₂ do not exceed the low nanomolar range at the site of inflammation during the resolution phase (6). Such concentrations are likely to be insufficient to generate anti-inflammatory effects. Indeed, our findings, in addition to those previously reported (19, 20, 21, 22, 23), clearly indicate that cyPGs exert their anti-inflammatory activity through inhibition of NF-κB only at micromolar concentrations. Moreover, our study shows that cyPGs at nanomolar concentrations do not affect proinflammatory cytokine production upon cell stimulation. This last observation brings to light the contrast that exists between cyPGs and other lipid mediators that are physiologically relevant. Indeed, leukotrienes and lipoxins, unlike cyPGs, mediate their effects at the nanomolar or subnanomolar concentrations (12, 13, 14). For example, lipoxins, which are established anti-inflammatory mediators, are known to exert their effects in the nanomolar range (13, 14). Taken together, all these observations suggest that cyPGs are not biologically active within the nanomolar range and are not likely to play a physiologically relevant role in inflammation compared with other eicosanoids, such as leukotrienes and lipoxins.

Two observations led to the hypothesis that cyPGs could have therapeutic value in the treatment of inflammatory diseases. First, cyPGs, when used at micromolar concentrations, may inhibit NF-κB which is a critical activator of inflammatory gene expression (19, 20, 21, 22, 23). Second, exogenous 15d-PGJ₂ at high concentrations may attenuate inflammation in vivo (6, 15, 19). However, our results that cyPGs have proinflammatory properties when used at micromolar concentrations lower than required for NF-κB inhibition raise the possibility that the pharmacological use of cyPGs could be detrimental to health, at least in some circumstances, and therefore question the therapeutic potential of these compounds.

We thank Dr. G. Haegeman (University of Gent, Gent, Belgium) for providing the p1481hu.IL8P-Luc and p1168hu.IL6P-Luc plasmids, Drs. V. Bours, P. Chatelain, B. Fuks, P. Gosset, M.-P. Merville, and F. Trottein for advice, and B. Detry, S. Gaspar, M. Leblond, and I. Sbaï for excellent technical and secretarial assistance.