Article Text


EP2/EP4 signalling inhibits monocyte chemoattractant protein-1 production induced by interleukin 1β in synovial fibroblasts
  1. R Largo,
  2. I Díez-Ortego,
  3. O Sanchez-Pernaute,
  4. M J López-Armada,
  5. M A Alvarez-Soria,
  6. J Egido,
  7. G Herrero-Beaumont
  1. Bone and Joint Research Unit, Fundación Jiménez Díaz, Autonoma University, Madrid, Spain
  1. Correspondence to:
    Dr G Herrero-Beaumont
    Rheumatology Department, Fundación Jiménez Díaz, Avenida Reyes Católicos 2, 28040 Madrid, Spain,


Background: Besides its proinflammatory properties, prostaglandin E2 (PGE2) acts as a regulator of the expression of inducible genes. Inhibition of PGE2 synthesis might thus result in a paradoxical deleterious effect on inflammation.

Objective: To examine the effect of PGE2 on monocyte chemoattractant protein-1 (MCP-1) expression in cultured synovial fibroblasts (SF) stimulated with interleukin (IL)1β.

Methods: MCP-1 expression was assessed in SF stimulated with IL1β in the presence of PGE2 or different NSAIDs by RT-PCR or northern blot and immunocytochemistry. Expression of cyclo-oxygenase (COX) isoforms was studied by western blot techniques. The role of PGE2 receptors (EP) in PGE2 action was assessed employing EP receptor subtype-specific agonists.

Results: PGE2 significantly inhibited IL1β induced MCP-1 expression in SF in a dose dependent manner. IL1β increased COX-2 and did not alter COX-1 synthesis in SF. 11-Deoxy-PGE1, an EP2/EP4 agonist, reproduced PGE2 action on MCP-1 expression. Butaprost, a selective EP2 agonist, was less potent than PGE2. Sulprostone, an EP1/EP3 agonist, had no effect on IL1β induced MCP-1 expression. Inhibition of endogenous PGE2 synthesis by NSAIDs further enhanced MCP-1 mRNA expression in IL1β stimulated SF, an effect prevented by addition of exogenous PGE2.

Conclusion: Activation of EP2/EP4 receptors down regulates the expression of MCP-1 in IL1β stimulated SF, while PGE2 pharmacological inhibition cuts off this signalling pathway and results in a superinduction of MCP-1 expression. The data suggest that NSAIDs may intercept a natural regulatory circuit controlling the magnitude of inflammation, which questions their continuous administration in inflammatory joint diseases.

  • AC, adenylate cyclase
  • cAMP, cyclic adenosine monophosphate
  • COX, cyclo-oxygenase
  • DCF, diclofenac
  • EMSA, electrophoretic mobility shift assay
  • GAPDH, glyceraldehyde-3′-phosphate dehydrogenase
  • IL, interleukin
  • MCP-1, monocyte chemoattractant peptide-1
  • MXC, meloxicam
  • NSAIDs, non-steroidal anti-inflammatory drugs
  • PGE2, prostaglandin E2
  • RT-PCR, reverse transcription-polymerase chain reaction
  • SF, synovial fibroblasts
  • EP2/EP4 receptors
  • monocyte chemoattractant protein-1
  • prostaglandin E2
  • synovial cells

Statistics from

Prostaglandins are lipid mediators locally increased in the synovial fluid and synovial membrane of patients with inflammatory joint diseases.1 Prostaglandin E2 (PGE2), particularly, contributes to tissue oedema, hyperalgesia, and cytokine production in inflamed tissues.2 PGE2 synthesis takes place in response to cell activation by cytokines, such as interleukin (IL)1β or tumour necrosis factor α, and its generation accounts for many of the proinflammatory actions induced by these peptides.1,3 However, different, and even opposite, effects of PGE2 have been seen depending on the experimental system tested. Thus, PGE2 induced effects on vascular smooth muscle tone,4 cellular proliferation,5 IL6 production,3 or the synthesis of adhesion molecules6 are highly variable. These diverse and cell-specific effects of PGE2 are greatly dependent on the binding of different subtypes of PGE2 receptors (EP receptors).7 Six EP receptor subtypes have been identified in humans, classified into EP1, EP2, EP3, and EP4 according to their different signalling pathways (reviewed by Narumiya et al7). Triggering of EP1 receptors increases calcium mobilisation. Both EP2 and EP4 receptors are coupled with cyclic adenosine monophosphate (cAMP) generation. In addition, three different splicing forms of EP3 receptor mediate the activation of several second messengers, leading to changes in cAMP levels, calcium mobilisation, and activation of phospholipase C.7 Different cell responses can be described for the production of inflammatory mediators, depending on the specific receptor subtype expressed at the cell surface.3

Chronic inflammatory diseases are characterised by proliferation of resident cells and infiltration of large numbers of mononuclear cells into the tissue. Migration of these cells is directed by the release of chemokines at the target tissue.8,9 Chemokines are a superfamily of small proinflammatory peptides that, based on structural analysis, can be divided into two subfamilies, C-X-C—mainly chemotactic for neutrophils—and C-C—primarily attracting mononuclear cells.10 A variety of chemokines are found at high concentrations in joint effusions and cells of pannus, in rheumatoid and in experimental arthritis.11–14 In particular, monocyte chemotactic peptide-1 (MCP-1) is known to attract monocytes to the joint, where these cells have a decisive contribution to the perpetuation of disease.15 The number of macrophages in the synovial membrane is one of the better markers for showing local disease activity.16,17 In a parallel way, plasma and synovial MCP-1 concentrations closely reflect the severity of joint inflammation in rheumatoid and in antigen induced arthritis.11,18,19 In antigen induced arthritis we have found that expression of synovial MCP-1 strongly correlates with leucocyte infiltration into the tissue.20 Strikingly, we also found that pharmacological inhibition of PGE2 further increases MCP-1 expression and mononuclear cell concentration, in comparison with no treatment.20 These data suggest that PGE2 may regulate MCP-1 expression, thus playing an important role in controlling mononuclear cell migration to the joint. In this study we have explored this hypothesis using synovial fibroblasts (SF) stimulated with IL1β, a well recognised proinflammatory cytokine. Our data provide new evidence for an anti-inflammatory role of PGE2 in SF acting through its EP2/EP4 receptors.


Culture of SF

SF were obtained as described previously from the knee synovial membranes of healthy New Zealand rabbits or osteoarthritic patients undergoing joint replacement surgery at the Orthopaedic Department of the Fundacion Jimenez Diaz.21 Patients were selected for the study only if they had not taken non-steroidal anti-inflammatory drugs (NSAIDs) for at least 30 days before surgery. All studies were performed strictly in accordance with current local regulations. Briefly, human or rabbit synovial explants were exposed to enzymatic digestion with 1.25 mg/ml of trypsin for 1 hour with agitation (Boehringer Mannheim). Collected cells were resuspended in RPMI 1640 medium (Gibco BRL) enriched with 10% fetal calf serum (BioWittaker) and supplemented with 2 μm glutamine (Gibco BRL), 100 U/ml penicillin, and 100 μg/ml streptomycin. Cells were cultured at 37°C in 5% CO2, and used between the sixth and twelfth passages.

Stimulation of SF

SF were seeded onto Petri dishes (Costar) for RNA or total protein extraction experiments. When cells reached confluence, the medium was changed to RPMI without fetal calf serum. After 48 hours, cells were stimulated with vehicle or human recombinant IL1β (Immugenex Corp) in the presence or absence of PGE2, 11-deoxy-prostaglandin E1, butaprost, sulprostone (all from Cayman Chemicals), or the adenylate cyclase (AC) inhibitor MDL-12,330A (Calbiochem) in the indicated concentrations. Where indicated, cells were preincubated for 1 hour with diclofenac (DCF; Sigma Chemicals) or meloxicam (MXC; Boehringer Ingelheim) before IL1β addition.

RNA studies

Rabbit or human SF were harvested and total RNA was obtained by the acid guanidinium thiocyanate-phenol-chloroform method.22 Messenger RNA expression in rabbit SF was analysed by reverse transcription-polymerase chain reaction (RT-PCR)11 and by northern blot22 in human SF. cDNA probes for human MCP-1 (JE*/pGEM-hJE34) and the housekeeping gene glyceraldehyde-3′-phosphate dehydrogenase (GAPDH) were obtained from the American Type Culture Collection. Probes were radiolabelled by nick translation (Boehringer Mannheim) with [32P]dCTP. For RT-PCR studies, RNA was reverse transcribed and then amplified with a commercial kit (Access RT-PCR System; Promega), employing specific primers for human MCP-1 and GAPDH as described previously.11 Autoradiographs were analysed by scanning densitometry (Biorad).

Immunofluorescence staining

Human SF were grown in eight well Titer-Tek slides (Costar). Quiescent cells were exposed to 10 U/ml IL1β with or without 10−6 M PGE2 for 4 hours. After incubation, cells were washed and fixed in methanol/acetone (50:50 vol/vol) at −20°C. Cells were incubated with 3% bovine serum albumin in phosphate buffered saline (9.1 M dibasic sodium phosphate, 1.7 M monobasic sodium phosphate, 150 M NaCl, pH 7.4) overnight at 4°C. A mouse monoclonal antibody against MCP-1 (10 μg/ml) (Immugenex Corp) was used as primary antibody, and a 1:200 dilution of FITC labelled rabbit antimouse IgG was used as a secondary antibody (Sigma). Controls were incubated with non-immune serum or with the secondary antibody alone. Slides were mounted in glycerol and examined under microscope by two “blinded” observers. Images were photographed and printed at equivalent exposures.

PGE2 assay

The amount of PGE2 in the conditioned media collected from stimulated rabbit SF was determined by a commercially available enzyme linked immunosorbent assay (ELISA; Assay Designs). Experiments were done in triplicate in two different cell cultures.

Western blot analysis

Human SF were employed for protein determinations owing to the absence of commercially available antibodies against rabbit products appropriate for these studies. For cyclo-oxygenase-1 (COX-1) and COX-2 determinations, human SF were homogenised with ice cold lysis buffer (1% Nonidet P-40 (Sigma); 0.5% sodium dodecyl sulphate, 0.1 M EDTA, 1 M dithiothreitol, and 1 M phenylmethylsulphonylfluoride in phosphate buffered saline). The lysates were transferred to Eppendorf tubes and centrifuged at 12 000 g for 10 minutes twice. Protein concentrations were determined by the BCA method (Pierce). Equal amounts of protein were run on sodium dodecyl sulphate-polyacrylamide gels and electrophoretically transferred to polyvinylidene difluoride membranes (Millipore). Membranes were blocked in 0.1 M Tris, pH 7.4, and 0.1 M NaCl containing 0.3% Tween-20 and 6% dry skimmed milk for 60 minutes at room temperature. Then, membranes were incubated overnight with antibodies to COX-1 or COX-2 (both from Santa Cruz Biotech) at 4°C. Detection was done with peroxidase conjugated antibodies, followed by development with a chemiluminescence based method (ECL; Amersham). Controls were performed by immunodetection of α-tubulin in each membrane with a specific antibody (Sigma).

Electrophoretic mobility shift assay (EMSA)

Nuclear extracts were obtained from rabbit SF as previously described.23 Briefly, cellular nuclei were isolated by precipitation and dissolved in buffer. Total nuclear protein concentration was measured by the BCA method (Pierce). Transcription factor activity was determined by EMSA employing an end labelled [32P]ATP-NF-κB consensus oligonucleotide. Nuclear extracts (5 μg) were equilibrated for 10 minutes in the appropriate binding buffer, and then the labelled probe (0.35 pmol) was added and incubated for 20 minutes at room temperature. To establish the specificity of the reaction, negative controls without cell extracts, and competition assays with a 100-fold excess of unlabelled oligonucleotide were done. DNA–protein complexes were run on a non-denaturing 4% acrylamide gel in Tris-borate. The gel was dried and exposed to an x ray film.

Statistical analysis

Densitometric data are shown as arbitrary units (AU) as a percentage of control values. Results are expressed as the mean (SEM). Comparisons between groups were carried out with the Mann Whitney U test. Values of p<0.05 were considered significant.


Effect of PGE2 on IL1β induced MCP-1 expression and synthesis

MCP-1 mRNA was studied in rabbit SF by RT-PCR techniques; in human SF northern blot assays were performed, in order to obtain quantitative data. In time curve experiments, the gene expression of MCP-1 was up regulated by 10 U/ml IL1β in both cell populations, reaching a peak at 4 hours of incubation. For rabbit SF, mRNA levels increased by 910 (230)% at 2 hours*; 1620 (330)% at 4 hours*; 190 (30)% at 9 hours, and by 120 (20)% at 24 hours (n = 4, *p<0.05 v unstimulated cells). Four hours were then established as the optimal period of incubation for the next experiments, in which human or rabbit SF were exposed to 10 U/ml IL1β in the presence of increasing concentrations of PGE2. Figure 1 shows that similar results for MCP-1 expression were found in human and rabbit cells after incubation with these molecules. Exposure to 10−6 M PGE2 alone did not modify MCP-1 expression in SF. However, addition of PGE2 to 10 U/ml IL1β-incubated SF resulted in a dose dependent inhibition of MCP-1 expression in comparison with IL1β stimulated human (fig 1A and B) or rabbit (fig 1C) SF.

Figure 1

 Effect of PGE2 on MCP-1 expression induced by IL1β in human and rabbit SF. Cells were incubated for 4 hours with 10 U/ml IL1β and increasing concentrations of PGE2. (A) A representative northern blot autoradiograph of four different experiments carried out in human SF. Panels (B) and (C) show the average increase of MCP-1 mRNA levels in human and rabbit SF, respectively, compared with non-stimulated cells (n = 4). Data are expressed in arbitrary densitometric units (AU) after correction by mRNA levels of GAPDH, which was employed as housekeeping gene. 10−6 M PGE2 did not alter MCP-1 gene expression v vehicle. *p<0.05 v vehicle; †p<0.05 v IL1β alone.

To study whether the inhibition of MCP-1 expression was associated with a decrease in the protein production, MCP-1 synthesis was determined by indirect immunofluorescence in human SF using a monoclonal antibody to MCP-1. Starved SF showed a minimal cytoplasmic immunoreactivity to anti-MCP-1 (fig 2A). After 4 hours of stimulation with 10 U/ml IL1β, enhanced cytoplasmic staining was seen, indicating an increase in MCP-1 synthesis (fig 2B). When SF were co-incubated with 10 U/ml IL1β and 10−6 M PGE2, a marked diminution in MCP-1 staining was seen as compared with IL1β stimulated cells (fig 2C). Figure 2D shows that incubation with 10−6 M PGE2 alone did not significantly alter MCP-1 staining in comparison with unstimulated cells. In negative control experiments using unrelated IgG, no staining was detected (data not shown).

Figure 2

 MCP-1 protein detection by immunofluorescence. In untreated human SF, a slight cytoplasmic immunoreactivity to anti-MCP-1 was found (A). Cells treated with 10 U/ml IL1β for 4 hours showed an intense cytoplasmic fluorescence with the antibody (B). Co-incubation with 10 U/ml IL1β and 10−6 M PGE2 markedly diminished the MCP-1 signal, in comparison with cells stimulated with IL1β alone (C). Exposure of SF to 10−6 M PGE2 did not alter the MCP-1 immunostaining pattern in comparison with vehicle (D).

Effect of IL1β on PGE2 production in SF

The negative regulation exerted by PGE2 on IL1β induction of MCP-1 was particularly remarkable, because PGE2 is quickly synthesised in response to IL1β, and acknowledged as a key mediator of its actions in different conditions. We therefore studied whether at this period of incubation 10 U/ml IL1β could increase PGE2 levels in SF culture supernatants and if the synthesis of new COX might participate in this reaction. At 4 hours of incubation, IL1β caused a 6.5-fold increase in PGE2 release by rabbit SF (fig 3A). When cells were incubated with 10−6 MXC or 10−6 DCF for 1 hour before the addition of IL1β, PGE2 fell under basal levels, while PGE2 was also almost undetectable in supernatants of cells exposed to either of the NSAIDs alone (fig 3A). In the same way, COX-1 and COX-2 synthesis were studied by western blot analysis in human SF stimulated with IL1β for 4 hours. At this time, COX-1 protein levels remained unmodified by IL1β or by the combination of IL1β plus PGE2, in comparison with vehicle (fig 3B). In contrast, COX-2 concentration significantly rose in IL1β treated SF (fig 3B). Unlike MCP-1 gene regulation, co-incubation of cells with 10 U/ml IL1β and 10−6 M PGE2 resulted in a significantly higher increase of COX-2 synthesis, in comparison with IL1β alone (fig 3B).

Figure 3

 PGE2 release (A) and COX isoforms (B) in SF incubated with IL1β. (A) PGE2 release was measured in rabbit SF culture supernatants in triplicate at 4 hours of incubation with 10 U/ml IL1β and/or 10−6 M MXC or 10−6 M DCF (pg/ml; n = 2). Untreated cells: 248 (21); 10 U/ml IL1β: 1643 (453); 10−6 M MXC: 110 (25); 10−6 M DCF: 39 (0); MXC+IL1β: 104.5 (18); DCF+IL1β: 57.5 (32). (B) Human SF were stimulated with 10 U/ml IL1β and/or 10−6 M PGE2 for 4 hours. Upper panel: representative western blots of COX-2, COX-1, and α-tubulin are shown. Lower panel: a densitometric analysis of COX-2 levels expressed in arbitrary units (AU) after correction by α-tubulin (n = 4). *p<0.05 v vehicle; †p<0.05 v IL1β alone.

Effect of PGE2 depletion in IL1β induced MCP-1 expression

According to our experiments, although only COX-2 synthesis was stimulated by IL1β, both COX isoforms were available at the cell cytoplasm at 4 hours of IL1β exposure. To evaluate the effect of endogenous PGE2 depletion on the IL1β induced MCP-1 expression, we used two non-selective COX-1/COX-2 inhibitors, DCF and MXC, with different COX-2 inhibitor activity.24 Rabbit SF were incubated for 1 hour with MXC or DCF and then stimulated with 10 U/ml IL1β. Figure 4 shows that incubation of cells with 10−6 M DCF or 10−6 MXC alone did not modify MCP-1 gene expression in rabbit SF. However, NSAID pretreated IL1β stimulated rabbit SF showed a more pronounced increase in MCP-1 mRNA expression than cells exposed to IL1β alone. Furthermore, addition of exogenous 10−6 M PGE2 to NSAID pretreated, IL1β stimulated SF prevented the up regulation of MCP-1 mRNA levels shown by the combination of NSAIDs and IL1β (fig 4).

Figure 4

 Effect of PGE2 synthesis inhibition mediated by NSAIDs on MCP-1 expression. Rabbit SF were pre-incubated for 1 hour with 10−6 M MXC or 10−6 M DCF, and then 10 U/ml IL1β and/or 10−6 M PGE2 were added for 4 hours. Upper panel shows a representative RT-PCR autoradiograph of MCP-1 and GAPDH. Lower panel shows the analysis of MCP-1 gene expression expressed in densitometric arbitrary units (AU) after normalisation by GAPDH levels (n = 4). *p<0.05 v vehicle; †p<0.05 v IL1β alone; ‡p<0.05 v IL1β+PGE2; **p<0.05 v the NSAID+IL1β.

Effect of PGE2 receptor subtype-specific agonists on IL1β induced MCP-1 expression

To establish which of the PGE2 receptors played a part in the inhibition of MCP-1 expression shown by PGE2, rabbit SF were incubated with 10 U/ml IL1β together with EP receptor subtype-specific agonists. 11-Deoxy-PGE1, 10−6 M, a specific EP2/EP4 agonist, suppressed MCP-1 expression evoked by IL1β similarly to PGE2 (fig 5). Butaprost, 10−6 M, a specific EP2 agonist, was less potent than PGE2 and 11-deoxy-PGE1 in inhibiting IL1β induced MCP-1 expression. When incubated alone at the same concentrations, 11-deoxy-PGE1 and butaprost had no effect on MCP-1 expression. In contrast, co-incubation with IL1β and 10−6 M sulprostone, a selective EP1/EP3 agonist, did not have any significant effect on IL1β induced MCP-1 expression (fig 5).

Figure 5

 Effect of specific EP receptor-subtype agonists on MCP-1 expression induced by IL1β. The role of EP receptors in MCP-1 regulation was studied in rabbit SF by RT-PCR, at 4 hours of incubation with 10 U/ml IL1β and/or the following molecules: the EP2/EP4 agonist 11-deoxy-PGE1 (11-D; 10−6 M), the EP2 agonist butaprost (BT; 10−6 M), the EP1/EP3 agonist sulprostone (SP; 10−6 M), PGE2 (10−6 M), and the AC inhibitor MDL-12,330A (MDL, 10−5 M, 30 minutes preincubation). Upper panel shows a representative autoradiograph of MCP-1 and the housekeeping gene GAPDH. Lower panel shows the analysis for MCP-1 expression in densitometric arbitrary units (AU) relative to that of GAPDH (n = 4). *p<0.05 v vehicle; †p<0.05 v IL1β alone.

EP2/EP4 are G-protein coupled receptors which activate AC and subsequently increase cAMP levels.7 An inhibitor of AC, MDL-12,330A, was used to study the role of cAMP as mediator of PGE2 in the regulation of MCP-1 expression. MDL-12,330A was added to rabbit SF cultures at 10−5 M, 30 minutes before incubation with 10 U/ml IL1β and 10−6 M PGE2. This compound completely abolished the down regulation exerted by PGE2 over IL1β induced MCP-1 gene expression, confirming that an AC dependent pathway was involved in this PGE2 action (fig 5). However, incubation of cells with 10−5 M MDL-12,330A and 10 U/ml IL1β did not result in a superinduction of MCP-1 expression in comparison with IL1β alone.

Role of nuclear factor κB (NF-κB) activation on MCP-1 gene control by PGE2

The MCP-1 promoter region contains binding sites for several transcription factors including NF-κB.25 Because IL1β is a potent activator of NF-κB translocation, we tested whether the inhibition of MCP-1 expression exerted by PGE2 was associated with interception of the NF-κB pathway. Figure 6 shows the effect of addition of PGE2 to IL1β on NF-κB activation in cultured rabbit SF. At 1 hour of incubation, 10 U/ml IL1β provoked a strong activation of NF-κB in comparison with vehicle, as shown by EMSA (lanes 1 and 2). The addition of 10−6 M PGE2 to 10 U/ml IL1β stimulated SF significantly reduced the NF-κB binding (lane 3). Furthermore, simultaneous incubation of SF with 10 U/ml IL1β and the EP2/EP4 agonist 11-deoxy-PGE1 at 10−6 M, reproduced PGE2 inhibition of IL1β induced NF-κB translocation (lanes 3 and 4). Neither the incubation of SF with PGE2 nor with 11-deoxy-PGE1 induced a significant modification of NF-κB activity in comparison with unstimulated SF (lanes 5 and 6).

Figure 6

 Effect of PGE2 or the EP2/EP4 specific agonist 11-deoxy-PGE1 (11-D) on NF-κB DNA binding induced by IL1β. Human SF were incubated with 10 U/ml IL1β and/or 10−6 M PGE2 or 10−6 M 11-deoxy-PGE1 for 60 minutes, and NF-κB activity was measured by EMSA. A representative autoradiograph of four different experiments with similar results is shown.


Migration of circulating leucocytes to the joint is a central process in the development and progression of inflammatory joint diseases.15 MCP-1 promotes recruitment of mononuclear cells and their activation in the inflamed synovium.13,26,27 It also induces cytokine production by SF, therefore displaying a major role in the interactions between resident and incoming cells which participate in the inflammatory reaction.28 Proinflammatory mediators locally increased in the rheumatoid synovium, such as IL1, interferon γ, and tumour necrosis factor α, induce MCP-1 expression in both cell populations.29,30

PGE2 is the eicosanoid achieving higher concentrations in inflamed joints.1 A complicated network of interactions between PGE2 and some cytokines has been described. Although the lipid mediator regulates production of these cytokines, the latter can, in turn, stimulate PGE2 synthesis in most cells.31

This work underlines the possibility that despite being considered a key proinflammatory mediator which contributes to joint injury, PGE2 has an anti-inflammatory role. We have demonstrated that PGE2 efficiently and dose dependently inhibits IL1β induced MCP-1 production in SF. Increasing evidence supports the statement that the diverse and specific effects of PGE2 rely on the cell expression of different EP receptor subtypes. Among these, EP2, EP3, and EP4 have been localised in human SF.32 We have shown that incubation of IL1β stimulated SF with an EP2/EP4 agonist evokes PGE2 effects on MCP-1 expression. No significant effect was noted for an EP1/EP3 agonist, while the effect of a specific EP2 agonist was less potent than that shown by the EP2/EP4 agonist. Activation of EP2/EP4 receptors has been associated with cAMP generation.7 In our experimental conditions, AC inhibition completely abolished the effect of PGE2 on MCP-1 expression. Our results are in agreement with recent data supporting the statement that EP2 and EP4 transduce anti-inflammatory actions of PGE2. In brief, it has been demonstrated that activation of these receptors inhibits expression of integrins and synthesis of IL6 in stimulated fibroblasts and in mononuclear cells3,6,33 A recent report suggests that EP4 activation would also down regulate chemokine expression in tissue infiltrated macrophages.34 In vivo, EP4 activation has been shown to improve clinical signs of inflammation in an experimental model of colitis.35 Interestingly, we have recently observed that IL1β induces EP4 gene expression in SF (data not shown), an effect also found by other authors in cervical fibroblasts.36 The hypothesis might therefore be proposed that PGE2 dependent responses drawn by EP4 activation are favoured in the inflamed joint owing to up regulation of the expression of these receptors by IL1β.

Our data provide new evidence suggesting that a complete counter regulating circuit is started in response to IL1β generation in the inflammatory environment, directed towards tempering the inflammatory reaction. Figure 7 shows this circuit. Briefly, in our experimental conditions IL1β was used to reproduce an inflammatory status in SF cultures. The cytokine activated PGE2 production by SF and also potentially provided a further source of prostanoid synthesis by increasing COX-2 protein levels. In turn, PGE2, acting through its EP2/EP4 receptors, could regulate migration of mononuclear cells to the joint, as suggested by its pronounced down regulation of MCP-1 production by SF. Finally, IL1β would increase the responsiveness of SF to this PGE2 anti-inflammatory pathway, through a selective up regulation of EP4 receptors.

Figure 7

 Signalling pathways controlling MCP-1 expression in SF, in which IL1β and PGE2 are involved. Mechanisms triggered by IL1β (pink arrows); PGE2 dependent signals (blue track); effects of NSAIDs (orange); action of MDL-12,330A, an AC inhibitor (green). Crossed bars: blockade of pathways; continuous lines: mechanisms explored in our study; broken lines: data from published reports. PGE2 receptors (EP1 to EP4) are shown in connection with each specific agonist employed in this study and with their second messengers. IL1β is one of the strongest activators of SF in inflammation. Many of its actions are mediated by PGE2 prompt synthesis. This is warranted by (a) activation of phospholipase A2 (PLA2) with generation of arachidonic acid (AA), which is the substrate for the action of COX and (b) increased availability of COX-2, whose synthesis is activated by the cytokine. IL1β promotes recruitment of mononuclear cells to the inflamed joint by the induction of MCP-1 gene expression in SF. IL1β is an activator of NF-κB, which controls transcription of MCP-1. At the same time, levels of PGE2 increase in response to IL1β. When targeting EP4, PGE2 could act by tempering inflammation. Addition of PGE2 or EP4 agonists leads to a down regulation of MCP-1 expression induced by IL1β in SF. Thus, IL1β could initiate lesion producing mechanism as well as restoring mechanisms in the target cell. It may even increase cell sensitivity to PGE2 by up regulating the expression of EP4 receptors. EP4 is coupled with cAMP generation, being the second messenger able to block translocation to the nucleus of NF-κB active components, where they trigger transcription of inducible genes, such as MCP-1. Thus, addition of MDL-12,330A, an inhibitor of AC, to IL1β plus PGE2, impairs PGE2 regulation of MCP-1 gene transcription. Finally, the regulating circuit is intercepted by the employment of NSAIDs owing to the dramatic reduction of PGE2 levels that these molecules provoke. Furthermore, NSAID blockade of COX may account for deviation of AA metabolism to the production of proinflammatory mediators generated by lipoxygenases (LO), some of which can activate NF-κB.

In this work we have also shown that PGE2 depletion, induced by MXC or DFC, on IL1β stimulated SF further increased MCP-1 expression in comparison with SF stimulated with IL1β alone. Although addition of exogenous PGE2 reversed this NSAID effect, other intercellular pathways should be examined to find out the precise mechanism of action of the drugs, because in this case, inhibition of AC did not result in a similar superinduction of MCP-1 mRNA expression in cells exposed to IL1β. One interesting possibility might be deviation of arachidonate metabolite production after COX blockade (fig 7); in fact, both accumulation of arachidonic acid and its oxidation by lipoxygenases might result in up regulation of proinflammatory gene products through an NF-κB dependent pathway.37 Whatever the mechanisms involved, this effect of NSAIDs on SF might explain some inflammatory flares seen during NSAID treatment in arthritic patients.

We must be cautious in the extrapolation of these in vitro results obtained with short term incubations to the clinical setting. Kinetics of both the EP receptor and cAMP mediated signalling pathways are processes tightly regulated. For instance, there seems to be a concentration dependent biphasic curve for cAMP generation yielded by PGE2, and also a deactivation phenomenon induced by prolonged exposure to the stimulating eicosanoid.38 None the less, our present work has shown similar results to those of our previous in vivo study, which found that NSAID mediated PGE2 inhibition increased synovial mononuclear cell concentration and MCP-1 expression in arthritic joints,20 resulting in the suggestion that total inhibition of PGE2 might be non-desirable in chronic inflammation.

One of the possible cellular mechanisms involved in the regulating action of PGE2 has also been studied in the present work. We have shown that EP2/EP4 activation inhibits NF-κB binding induced by IL1β. Because MCP-1 is an NF-κB dependent gene, this action might account for the PGE2 observed regulation of MCP-1 gene expression (fig 7). Previous studies have found that cAMP generation inhibits NF-κB dependent transcription of inflammatory genes in monocytic and endothelial cells.39

In summary, our results show that PGE2 mediated EP2/EP4 activation inhibits MCP-1 expression and synthesis in IL1β stimulated SF. Blockage of pathways elicited by EP2/EP4 by inhibition of PGE2 synthesis significantly superinduces MCP-1 expression in a proinflammatory setting. Our results suggest that endogenous PGE2 may control the propagation of inflammation in the inflamed synovium. Thus, chronic administration of NSAIDs might impair the beneficial effects of this autoregulatory circuit.


This work was supported by grants from FIS (01/0259; G03/152) and Mapfre Medicina Foundation.

RL is a fellow from Comunidad Autonoma de Madrid. ID-O, MJL-A, and MAA-S are fellows from Fundación Conchita Rábago.


View Abstract


  • * R Largo and I Díez-Ortego contributed equally to this work.

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.