NF-κB transcription factors regulate inflammatory responses to cytokines such as IL-1β and TNF-α. We tested whether PGE2 regulated nuclear localization of individual NF-κB subunits, p65 and p50, in synovial fibroblasts harvested from patients with rheumatoid arthritis (RA). IL-1β/TNF-α stimulated the translocation of p65 and p50 from the cytosol to the nucleus of human RA synovial fibroblasts, as well as NF-κB activation measured by luciferase reporter assay. PGE2 (10 nM, 6 h) enhanced p50, but inhibited p65 translocation and NF-κB activation. In contrast, depletion of endogenous PGE2 by ibuprofen (100 μM) and celecoxib (5 μM) enhanced p65, but inhibited p50 nuclear translocation as well as binding to NF-κB DNA binding sites. PGE2 also blocked IL-1β/TNF-α-stimulated ERK activation, and the ERK inhibitor, PD98059, mimicked PGE2 in blocking p65, but enhancing p50 nuclear translocation, suggesting that the effects of PGE2 on p65 and p50 are mediated via effects on ERK. PGE2 also enhanced the expression of IκBα in an ERK-independent manner, suggesting that PGE2 inhibits NF-κB activation by both ERK-dependent and -independent mechanisms. Our data indicate that PGE2 may act to attenuate cytokine-induced inflammatory responses in RA synovial fibroblasts via regulation of the localization of specific NF-κB family dimers.

Proinflammatory cytokines such as IL-1β and TNF-α participate in the regulation of inflammatory joint diseases, including rheumatoid arthritis (RA).4 Both IL-1 and TNF-α stimulate synovial fibroblasts (SF) to secrete mediators such as PGE2 (1, 2), matrix metalloproteinases (MMPs) (3), and vascular endothelial growth factor (4) that regulate inflammation, connective tissue degradation, and angiogenesis, respectively. Cytokines also stimulate the RA SF NF-κB signaling pathway (5), which in turn regulates the expression of a wide array of proinflammatory molecules.

NF-κB is a family of dimeric transcription factors formed by the hetero- or homodimerization of proteins from the rel family. There are five rel proteins: RelA (p65), RelB, and cRel, which contain transactivation domains, and p50 and p52, which are expressed as precursor proteins p105 (NF-κB1) and p100 (NF-κB2), respectively, and do not contain transactivation domains (6, 7). NF-κB p65/p50 heterodimers remain inactive in the cytoplasm through association with endogenous inhibitor proteins of the IκB family. Phosphorylation of IκBs in response to proinflammatory stimuli targets them for degradation by the 26S proteosome, allowing the liberated dimers to translocate to the nucleus (8, 9). Subsequently, the dimers bind to decameric κB motifs of a large set of genes and regulate their expression (10). Genes regulated by NF-κB encode defense and signaling proteins including cell surface molecules involved in immune function such as the Ig κL chain, class I and II MHC, and cytokines such as IL-1β, IL-2, IL-6, IL-8, IFN-β, and TNF-α (6). NF-κB regulates the synthesis of a number of MMPs, including MMP-1 (collagenase-1) and MMP-13 (collagenase-3) (11, 12). NF-κB dimers also bind to promoters of inducible NO synthase and cyclooxygenase-2 (COX-2), leading to inducible NO synthase and COX-2 expression and the synthesis of NO and PGs (13, 14, 15, 16). Whereas p65/p50 heterodimers stimulate these responses, binding of p50/p50 homodimers blocks the transcription of these inflammatory/immune genes.

Among the PGs synthesized by COX-2, PGE2 is perhaps the best studied. PGE2 mediates vasodilation, vascular leakiness, and pain and may regulate B cell differentiation (17). However PGE2 may also mediate anti-inflammatory effects (18). For example, PGE2 inhibits neutrophil superoxide anion generation via inhibition of the MAPK ERK (19). Studies by Zurier and colleagues (20, 21) indicate that PGEs can reduce inflammation in animal models of arthritis and nephritis. In rabbit SF, PGE2 inhibits secretion of MMP-1, again via effects on ERK (3, 22, 23). Thus, PGEs may actually be complex mediators of inflammation, with both inflammatory and anti-inflammatory effects.

In our current study, we used primary human RA SF, as well as rabbit synovial fibroblast and HeLa cell lines, to study the effects of PGE2 on NF-κB activation. Our data indicate that PGE2 dissociates nuclear trafficking of individual NF-κB subunits (p65, p50). In particular, PGE2 inhibits the sustained nuclear accumulation of p65 in cytokine-stimulated synovial fibroblasts, while promoting the nuclear accumulation of p50, indicating an anti-inflammatory effect of PGE2 on NF-κB signaling. This effect is mediated via both ERK-dependent and -independent processes. The capacity of NF-κB to regulate COX-2 expression, and of PGE2 to attenuate NF-κB activation, suggest a feedback loop in which PGE2 inhibits the activity of an enzyme required for its synthesis.

Primary cultures of human RA SF were prepared from synovial tissue obtained at the time of joint replacement surgery. Synovial tissues were minced and incubated with 1 mg/ml collagenase in serum-free DMEM for 2 h at 37°C, filtered through a nylon mesh, extensively washed, and cultured in DMEM supplemented with 10% FBS, 100 U/ml penicillin, 100 μg/ml streptomycin, and 292 μg/ml l-glutamine in a humidified atmosphere of 95% air and 5% CO2. After overnight culture, nonadherent cells were cultivated in DMEM plus 10% FBS. At confluence, cells were trypsinized, split at a 1:3 ratio, and recultured in medium. SF were used from passages seven to eight where they comprised a homogeneous population. The collection and use of human RA SF was reviewed and approved by the Institutional Board of Research Associates, New York University School of Medicine (New York, NY). For individual experiments, confluent cell monolayers were grown in serum-free medium for 40–42 h before stimulation with a cytokine mixture containing 5 ng/ml human IL-1β and 10 ng/ml TNF-α in the presence or absence of different agents.

Rabbit SF (HIG82 cell line) were obtained from American Type Culture Collection and maintained in Ham’s F-12 medium supplemented with 10% FBS, 100 U/ml penicillin, and 100 μg/ml streptomycin. HeLa cells were obtained from American Type Culture Collection and maintained in DMEM with 10% FBS containing penicillin/streptomycin.

RA SF on glass coverslips were serum-starved for 48 h and incubated with or without IL-1/TNF-α in serum-free DMEM culture medium, fixed with 4% paraformaldehyde (10 min), washed in PBS (three times for 5 min); permeabilized with acetone (1 min), washed in PBS (three times for 5 min); and blocked with 1% normal goat serum (NGS) in PBS for 1–2 h at room temperature. Polyclonal Abs to p50 and p65 (Santa Cruz Biotechnology) were diluted 1/100 in NGS and applied at room temperature (1 h), followed by washing in PBS (three times for 5 min) and incubation for 2 h at room temperature with FITC-conjugated goat anti-mouse IgG (Vector Laboratories) diluted 1/200 in NGS. Coverslips were washed (three times for 5 min), mounted on slides with Vectashield (Vector Laboratories), and examined by laser scanning confocal microscopy (×63 PlanApo 1.4 aperture objective; Zeiss).

Nuclear and cytoplasmic extracts were prepared using NE-PER Nuclear and Cytoplasmic Extraction Reagents (Pierce) according to the manufacturer’s instructions. Protein content was determined with a MicroBCA protein assay (Pierce). Equal amounts of protein (5–10 μg) were analyzed by 10% SDS-PAGE and transferred onto nitrocellulose membranes. The membranes were blocked (5% nonfat dry milk in PBS/0.1% Tween 20 (PBST), 4°C for overnight) and incubated with anti-p65, p50, IκBα, COX-2, phospho-ERK, total ERK, or β-actin Abs. Membranes were washed with PBST, incubated with peroxidase-conjugated secondary Abs (1 h, room temperature in PBST), washed, and developed using the ECL chemiluminescence detection reagent (Pierce). Immunoblot was quantitatively analyzed after scanning the image and quantitating pixels using the ImageQuant program.

Rabbit SF (HIG82), which share many phenotypic features of human RA SF (24, 25), were grown to 50% confluence in 35-mm culture dishes and transfected with 3 μg of NF-κB reporter construct pBIIX-luc or the control luciferase construct pFLUC (lacking NF-κB response elements; both constructs were a gift from S. Ghosh, Yale University, New Haven, CT), along with 0.5 μg of pSVGal plasmid using LipofectAMINE 2000 in Opti-MEM medium. Twenty-four hours after transfection, the cultures were treated with or without 10 nM PGE2 for 1 h, followed by stimulation with IL-1 (5 ng/ml) and TNF-α (10 ng/ml) for 24 h. Cells were harvested and lysed using reporter lysis buffer. Luciferase assays were performed using 20 μl of cell extract and 100 μl of luciferin substrate (Promega). Galactosidase assays were performed using a Tropix kit (Applied Biosystems) according to the manufacturer’s instructions, and luciferase and galactosidase activities were measured using a MiniLum luminometer (Bioscan). Galactosidase activity was used to confirm uniform transfection between conditions.

DNA-binding assays were performed with the NF-κB Family Transcription Factor Assay kit (Active Motif) according to the manufacturer’s instructions. In our experiments, 3 μg of nuclear extract was hybridized with the NF-κB-specific oligonucleotide bound into the ELISA plates. The bound NF-κB protein was analyzed by an ELISA system supplied with the kit.

PGE2 in culture supernatant was determined using a RIA kit (Sigma-Aldrich) according to the manufacturer’s instructions.

Constitutively active pciEGFP-MEKCA (GFP-caMek) was a gift from Michael Bell (Mayo Clinic College of Medicine, Rochester, MN). DsRed-p65 plasmid was constructed by subcloning p65 cDNA from pEGFP-p65 (gift from Johannes Schmid, University of Vienna, Vienna, Austria) into pDsRed1-N1 plasmid (Clontech) using BamH1 and HindIII restriction enzymes. Transient transfections were performed using the Superfect transfection reagent (Qiagen) according to the manufacturer’s protocol. In brief, HeLa cells were grown on 35-mm glass bottom culture dishes (MakTek) and transfected with 1.0 μg of GFP vector or GFP-caMek, and cotransfected with 0.5 μg of control plasmid or DsRed-p65 construct in 10 μl of Superfect reagent for 3 h. Cells were washed with PBS and incubated with fresh complete medium overnight. Cells were starved (0% FBS) for 24 h, then imaged by laser scanning confocal microscopy. Digital images were processed with Adobe Photoshop version 5.0 (Adobe Systems).

Unstimulated RA SF demonstrated little nuclear p65 and variable but somewhat higher levels of constitutive nuclear p50. Stimulation of RA SF for 6 h with IL-1β (5 ng/ml) or TNF-α (10 ng/ml) resulted in sustained nuclear accumulation of both p65 and p50 subunits of NF-κB, determined by subcellular fractionation followed by immunoblotting of nuclear fractions (Fig. 1,A). As we have previously reported in chondrocytes (26), NF-κB subunit translocation in RA SF was detectable by 15 min and achieved levels similar to those seen at 6 h as early as 1 h after stimulation (data not shown). Stimulation with IL-1β/TNF-α together resulted in additional increases in sustained p65 and p50 translocation relative to either cytokine alone. Because both cytokines are coexpressed in the RA joint, subsequent experiments used the combination of IL-1β/TNF-α. As shown in Fig. 1 B, sustained translocation of p65 and p50 from cytosol to nucleus at 6 h was confirmed by immunostaining IL-1/TNF-α-treated RA SF with anti-p65 or p50 Ab, as well as FITC-conjugated secondary Ab, and imaging via confocal microscopy. Parallel experiments, including nuclear counterstaining with Hoechst and examination with a Zeiss Axiovert S100 fluorescence microscope, confirmed the nuclear localization of p65/p50 translocation (data not shown).

FIGURE 1.

Cytokines stimulate nuclear translocation of NF-κB: A, RA SF were stimulated with or without IL-1β (5 ng/ml), TNF-α (10 ng/ml), or both for 6 h. NF-κB translocation was assessed in nuclear extracts by immunoblotting with Abs against p65 and p50 (representative experiment, n = 3). B, NF-κB translocation in response to stimulation with IL-1β and TNF-α together for 6 h was assessed by fixing and staining stimulated cells with FITC-labeled anti- p65 (top panels) or anti-p50 Abs (bottom panels), respectively, and imaging via confocal microscopy (representative experiment, n = 3).

FIGURE 1.

Cytokines stimulate nuclear translocation of NF-κB: A, RA SF were stimulated with or without IL-1β (5 ng/ml), TNF-α (10 ng/ml), or both for 6 h. NF-κB translocation was assessed in nuclear extracts by immunoblotting with Abs against p65 and p50 (representative experiment, n = 3). B, NF-κB translocation in response to stimulation with IL-1β and TNF-α together for 6 h was assessed by fixing and staining stimulated cells with FITC-labeled anti- p65 (top panels) or anti-p50 Abs (bottom panels), respectively, and imaging via confocal microscopy (representative experiment, n = 3).

Close modal

Because PGE2 has anti-inflammatory effects on RA SF (25), we performed detailed experiments to assess whether PGE2 regulates p65 and p50 localization in human RA SF (Fig. 2, A–C). In these experiments, IL-1β/TNF-α again stimulated persistent (as long as 6 h) p65 and p50 translocation to the nucleus. Cytoplasmic p65 and p50 levels were reduced after 1 h of stimulation (consistent with nuclear translocation of a finite cytosolic pool of these proteins), but recovered by 6 h despite persistence of p65 and p50 in the nucleus. Treatment of RA SF with cycloheximide before 6 h of stimulation, to inhibit protein synthesis, inhibited p65 and p50 reaccumulation in the cytosol, confirming that IL-1β/TNF-α stimulate de novo synthesis of p65 and p50 that is subsequently localized to the cytosol (data not shown). PGE2 (10 nM) inhibited IL-1β/TNF-α-stimulated nuclear translocation of p65 at 6 h, but not at 1 h (Fig. 2,A). In contrast, PGE2 enhanced p50 nuclear translocation both at 1 and 6 h (Fig. 2,B). Fig. 2 C shows quantitation of the nuclear accumulation of p65 and p50. These data indicate that PGE2 inhibits p65, but promotes p50 nuclear translocation. The delayed effect of PGE2 on p65 translocation, vs its more rapid effect on p50, suggests that these responses likely reflect separate effects. PGE2 also inhibited cytosolic reaccumulation of p65, but not p50, at 6 h, suggesting that PGE2 may inhibit p65 in part by specifically abrogating synthesis and/or reducing stability of this subunit.

FIGURE 2.

PGE2 uncouples p65 and p50 translocation and inhibits NF-κB-mediated transcription: RA SF incubated with or without PGE2 (10 nM) for 1 h were stimulated with or without IL-1/TNF-α for the indicated times. Nuclear and cytoplasmic p65 (A, representative experiment, n = 5) and p50 (B, representative experiment, n = 6) localization were assayed by immunoblot. β-Actin was used to control for loading of the samples. C, Quantitation (ImageQuant program) of nuclear p50 and p65 from the experiments in A and B (n = 6). ∗, p = 0.04; ∗∗, p ≤ 0.001 vs the appropriately matched IL-1/TNF-α-stimulated condition. D, HIG82 SF were transfected with a luciferase reporter construct containing two NF-κB response elements (pBIIX-luc) or a control luciferase construct with no NF-κB response elements (pFLUC). Cells were incubated with or without PGE2 for 1 h, followed by incubation with or without IL-1 (5 ng/ml)/TNF-α(10 ng/ml) for 24 h, and luminometry was used for luciferase activity (n = 4).

FIGURE 2.

PGE2 uncouples p65 and p50 translocation and inhibits NF-κB-mediated transcription: RA SF incubated with or without PGE2 (10 nM) for 1 h were stimulated with or without IL-1/TNF-α for the indicated times. Nuclear and cytoplasmic p65 (A, representative experiment, n = 5) and p50 (B, representative experiment, n = 6) localization were assayed by immunoblot. β-Actin was used to control for loading of the samples. C, Quantitation (ImageQuant program) of nuclear p50 and p65 from the experiments in A and B (n = 6). ∗, p = 0.04; ∗∗, p ≤ 0.001 vs the appropriately matched IL-1/TNF-α-stimulated condition. D, HIG82 SF were transfected with a luciferase reporter construct containing two NF-κB response elements (pBIIX-luc) or a control luciferase construct with no NF-κB response elements (pFLUC). Cells were incubated with or without PGE2 for 1 h, followed by incubation with or without IL-1 (5 ng/ml)/TNF-α(10 ng/ml) for 24 h, and luminometry was used for luciferase activity (n = 4).

Close modal

To confirm that the effects of PGE2 on p65 and p50 localization had consequences for transcription, we transfected HIG82 rabbit SF with a luciferase reporter (pBIIX-luc) containing two NF-κB binding sites (Fig. 2 D). Transfection with pBIIX-luc resulted in luciferase activity, as measured by luminometry. Stimulation of pBIIX-luc-transfected cells with IL-1/TNF-α resulted in increased luciferase activity. Luciferase activity in stimulated pBIIX-transfected cells was inhibited by coincubation with PGE2. Transfection with pFLUC, a luciferase reporter construct lacking NF-κB binding sites, resulted in increased luciferase activity that was unaffected by either IL-1/TNF-α or PGE2. Thus, IL-1/TNF-α stimulates and PGE2 inhibits NF-κB activity.

That exogenous PGE2 differentially regulated p50 and p65 translocation suggested that endogenous PGE2 might have a similar effect. We therefore explored the role of endogenous PGE2 produced in response to cytokine stimulation by testing the effects of COX inhibition. As shown in Fig. 3,A, IL-1β/TNF-α (6 h) induced a marked increase in PGE2 secretion by RA SF. Ibuprofen (nonselective COX inhibitor) and celecoxib (selective COX-2 inhibitor) comparably inhibited PGE2 synthesis induced by IL-1β/TNF-α, confirming that the majority of induced PGE2 synthesis was COX-2 dependent. As shown in Fig. 3, B and C, ibuprofen and celecoxib each enhanced IL-1β/TNF-α-stimulated p65 nuclear accumulation at 6 h. Simultaneously, celecoxib inhibited p50 nuclear translocation (ibuprofen was not tested in these experiments for p50 localization). In separate studies, we confirmed that IL-1β/TNF-α increased the binding of both p65 and p50 to NF-κB response elements on DNA and that celecoxib enhanced p65, but inhibited p50, binding to these DNA sites (Fig. 3 D). The selective COX-2 inhibitor rofecoxib also enhanced p65, but inhibited p50 binding to DNA (data not shown). Moreover, celecoxib also enhanced both resting and IL-1β/TNF-α-stimulated NF-κB transcriptional activity as measured by pBIIX-luc luciferase reporter activity, effects that were reversed by repletion of PGE2 (data not shown). The ability of COX inhibitors to enhance p65 and to inhibit p50 translocation and DNA binding, as well as to increase NF-κB transcriptional activity, suggests that the effects of endogenous PGE2, produced in response to inflammatory cytokines are identical to those observed following the exposure of cells to exogenous PGE2.

FIGURE 3.

COX inhibitors uncouple p65 and p50 translocation opposite to PGE2: A, RA SF were incubated with or without ibuprofen (100 μM) or celecoxib (2 μM) for 16 h, then stimulated with or without IL-1/TNF-α for 6 h. PGE2 levels in the supernatants were assessed by RIA (n = 3). B, RA SF were incubated with or without ibuprofen (100 μM) for 16 h, then stimulated with or without IL-1/TNF-α for 6 h. Nuclear extracts were prepared and levels of p65 were determined by immunoblot. C, RA SF were incubated with or without PGE2 (10 nM, 6 h) or celecoxib (2 μM, 16 h), then stimulated with or without IL-1/TNF-α for 6 h. Nuclear extracts were analyzed by immunoblot for p65 and p50 levels. D, DNA-binding capacity of p50 and p65 from nuclear extracts was determined by an ELISA-based DNA-binding assay (n = 3). ∗, p = 0.06; ∗∗, p ≤ 0.02 vs the appropriately matched IL-1/TNF-α-stimulated condition.

FIGURE 3.

COX inhibitors uncouple p65 and p50 translocation opposite to PGE2: A, RA SF were incubated with or without ibuprofen (100 μM) or celecoxib (2 μM) for 16 h, then stimulated with or without IL-1/TNF-α for 6 h. PGE2 levels in the supernatants were assessed by RIA (n = 3). B, RA SF were incubated with or without ibuprofen (100 μM) for 16 h, then stimulated with or without IL-1/TNF-α for 6 h. Nuclear extracts were prepared and levels of p65 were determined by immunoblot. C, RA SF were incubated with or without PGE2 (10 nM, 6 h) or celecoxib (2 μM, 16 h), then stimulated with or without IL-1/TNF-α for 6 h. Nuclear extracts were analyzed by immunoblot for p65 and p50 levels. D, DNA-binding capacity of p50 and p65 from nuclear extracts was determined by an ELISA-based DNA-binding assay (n = 3). ∗, p = 0.06; ∗∗, p ≤ 0.02 vs the appropriately matched IL-1/TNF-α-stimulated condition.

Close modal

We have previously reported that ERK regulates p65 translocation in rabbit SF (25). To examine ERK regulation of p65 and p50 translocation in human RA SF, we used PD98059 (2–9′-amino-3′-methoxyphenyl)-oxanaphthalen-4-one) (27), a specific inhibitor of MEK, the proximal activator of ERK. Fig. 4,A illustrates that PD98059 inhibited IL-1β/TNF-α-induced p65 nuclear translocation at both 1 and 6 h. In contrast, PD98059 enhanced IL-1β/TNF-α stimulation of p50 nuclear translocation (Fig. 4,B). Thus, ERK promotes p65, but inhibits p50, nuclear translocation. To confirm a role for ERK pathway signaling in p65 translocation, we cotransfected HeLa cells with DsRed-p65 as well as either GFP or constitutively active MEK complexed to GFP (caMek-GFP) (Fig. 4 C). Transfection with GFP resulted in a diffuse distribution of that protein. In cells cotransfected with GFP and DsRed-p65, DsRed-p65 was also evenly distributed to both cytosol and nucleus. In contrast, cotransfection with caMek-GFP and DsRed-p65 resulted in predominant localization of caMek-GFP to the cytosol, but increased localization of DsRed-p65 to the nucleus. Therefore, the presence of active MEK, and consequently of active ERK, is sufficient to drive p65 nuclear translocation and promote NF-κB activation.

FIGURE 4.

ERK regulates IL-1/TNF-α-stimulated translocation of p65 and p50: RA SF were incubated with or without PD98059 (MEK inhibitor, 100 μM) for 1 h, then stimulated with or without IL-1/TNF for 1 or 6 h. Nuclear fractions were assayed by immunoblot for p65 (A), p50 (B), or actin (representative experiment, n = 3). C, HeLa cells were cotransfected with DsRed and GFP (top) or GFP-caMek (bottom) and analyzed by immunofluorescence for protein localization (representative experiment, n = 6).

FIGURE 4.

ERK regulates IL-1/TNF-α-stimulated translocation of p65 and p50: RA SF were incubated with or without PD98059 (MEK inhibitor, 100 μM) for 1 h, then stimulated with or without IL-1/TNF for 1 or 6 h. Nuclear fractions were assayed by immunoblot for p65 (A), p50 (B), or actin (representative experiment, n = 3). C, HeLa cells were cotransfected with DsRed and GFP (top) or GFP-caMek (bottom) and analyzed by immunofluorescence for protein localization (representative experiment, n = 6).

Close modal

Since PGE2 and PD98059 identically inhibited p65 and enhanced p50 nuclear translocation, we hypothesized that PGE2 might regulate NF-κB by inhibiting ERK. Accordingly, we tested the effect of PGE2 on ERK activation. IL-1β/TNF-α-stimulated ERK activation rapidly (1 h after stimulation) and persistently (sustained 6 h after stimulation) (Fig. 5 A). PGE2 inhibited IL-1β/TNF-α-stimulated ERK activation at both 1 and 6 h. Consistent with the observation that PGE2 requires longer exposure times to inhibit p65 translocation, PGE2 inhibition of ERK was greater at 6 h than at 1 h. Thus, PGE2 regulates p65 and p50 localization, at least in part, via its effects on ERK.

FIGURE 5.

PGE2 regulates NF-κB via ERK-dependent and -independent processes: A, PGE2 inhibits ERK. RA SF were incubated with or without PGE2 (10 nM) for 1 h, followed by IL-1/TNF stimulation for 1 or 6 h. Cell lysates were analyzed by immunoblot for levels of phospho- (pERK, top) and total (tERK, bottom) ERK (representative experiment, n = 3). B, PGE2 but not PD98059 inhibits IκBα levels. RA SF were incubated with or without PGE2 (10 nM) for 1 h, followed by IL-1/TNF stimulation for 1 or 6 h. Cell lysates were analyzed by immunoblot for levels of IκBα (top) or β-actin (bottom) (representative experiment, n = 2).

FIGURE 5.

PGE2 regulates NF-κB via ERK-dependent and -independent processes: A, PGE2 inhibits ERK. RA SF were incubated with or without PGE2 (10 nM) for 1 h, followed by IL-1/TNF stimulation for 1 or 6 h. Cell lysates were analyzed by immunoblot for levels of phospho- (pERK, top) and total (tERK, bottom) ERK (representative experiment, n = 3). B, PGE2 but not PD98059 inhibits IκBα levels. RA SF were incubated with or without PGE2 (10 nM) for 1 h, followed by IL-1/TNF stimulation for 1 or 6 h. Cell lysates were analyzed by immunoblot for levels of IκBα (top) or β-actin (bottom) (representative experiment, n = 2).

Close modal

Phosphorylation and degradation of IκBα is the initiating event in NF-κB activation/p65 translocation, and up-regulation of IκBα expression constrains p65/p50 to the cytosol and inhibits inflammatory NF-κB activity (8). Accordingly, we also tested the hypothesis that PGE2 and ERK act on p65/p50 localization via effects on IκBα protein expression (Fig. 5 B). Consistent with a previous report (28), IL-1β/TNF-α stimulation of RA SF resulted in persistent IκBα levels at later time points (1–6 h), consistent with IκBα synthesis after early degradation. PGE2 significantly enhanced IκBα expression at 6 h, indicating that PGE2 may inhibit p65 translocation, in part, by increasing the presence of its regulator. However, in contrast to PGE2, PD98059 did not enhance IκBα expression, indicating that IκBα levels are not regulated by ERK. Therefore, the ability of PGE2 to regulate IκBα levels must be ERK independent, and PGE2 regulates NF-κB through both ERK-dependent and -independent mechanisms.

Since NF-κB has been reported to regulate COX-2 expression (13), we also asked whether PGE2 and ERK regulate COX-2 expression in RA SF. As shown in Fig. 6, both PGE2 and PD98059 inhibited IL-1β/TNF-α-stimulated COX-2 expression. Since PGE2 also inhibits ERK, these data suggest a model in which PGE2 acts on the ERK pathway to inhibit NF-κB-dependent expression of COX-2, providing a mechanism for feedback inhibition of PGE2 synthesis

FIGURE 6.

PGE2 and PD98059 inhibit IL-1/TNF-α-induced COX-2 expression: RA SF were incubated with or without PGE2 (10 nM) or PD98059 (100 μM) for 1 h, then stimulated with or without IL-1/TNF-α for 1 or 6 h. Cell lysates were assayed by immunoblot for COX-2 protein expression (top) or β-actin (bottom) (representative experiment, n = 2).

FIGURE 6.

PGE2 and PD98059 inhibit IL-1/TNF-α-induced COX-2 expression: RA SF were incubated with or without PGE2 (10 nM) or PD98059 (100 μM) for 1 h, then stimulated with or without IL-1/TNF-α for 1 or 6 h. Cell lysates were assayed by immunoblot for COX-2 protein expression (top) or β-actin (bottom) (representative experiment, n = 2).

Close modal

Inflammatory responses carry the potential for tissue destruction and must be closely regulated. Mammalian immune systems use a variety of strategies to suppress inflammation, including 1) decoy receptors for proinflammatory cytokines (e.g., IL-1R antagonist, soluble p55TNF-α receptor) (29), 2) anti-inflammatory cytokines (e.g., IL-4, IL-10) (30), and 3) anti-inflammatory lipids (e.g., lipoxin A4, B4) (31). In the present study, we show that PGE2, a COX product usually considered to be an important proinflammatory mediator, may have unexpected anti-inflammatory effects via attenuation of cytokine-induced NF-κB activation.

NF-κB is a family of both pro- and anti-inflammatory dimeric molecules. Of these, the p65/p50 heterodimer is the predominant proinflammatory molecule and p65 nuclear translocation and DNA binding is typically defined as NF-κB activation. In contrast, p50/p50 homodimers move to the nucleus and bind to NF-κB transcription sites, but fail to induce gene expression (32, 33, 34). Thus, p50/p50 homodimers may compete for binding with p65/p50, resulting in an anti-inflammatory effect (35, 36). Our data confirm that IL-1β/TNF-α stimulate the sustained nuclear accumulation of p65/p50, p65 and p50 DNA binding to the NF-κB promoter site, and NF-κB-regulated transcription. Exogenously added PGE2 inhibited p65 nuclear translocation as well as NF-κB-dependent transcription, but enhanced nuclear translocation of p50. We suggest that PGE2 inhibits p65/p50 translocation to the nucleus, while stimulating the translocation of p50/p50, resulting in a net movement of p50 into the nucleus. Our data do not formally exclude other possibilities, for instance, the possibility that PGE2 physically dissociates p65/p50 heterodimers or the possibility that PGE2 stimulates p65/p50 translocation to the nucleus while also stimulating export of p65/p65 homodimers (resulting in net import of p50, net export of p65, and a likely anti-inflammatory effect). However, Ma et al. (36) have recently reported that p202a, an IFN-inducible protein that inhibits NF-κB activity, acts by simultaneously inhibiting p65/p50 DNA binding, while enhancing p50/p50 DNA binding. Thus, our proposed mechanism of action of PGE2 on NF-κB is consistent with an already established cellular mechanism. Whether PGE2 acts on NF-κB via activation of p202a was not determined in our experiments.

To our knowledge, ours is the first demonstration of PGE2-induced p50 nuclear translocation. However, other investigators have documented dissociation between p50 and p65 translocation in IL-1β/TNF-α-stimulated cells treated with ostensible NF-κB inhibitors (37). Moreover, both D’Acquisto et al. (38) and Poligone et al. (39) have demonstrated the ability of PGE2 and/or PGE analogs to regulate NF-κB in other cell types. However, D’Acquisto’s studies indicated that both p50 and p65 translocation were inhibited by PGE2, whereas Poligone reported that PGE2 enhanced NF-κB activity. Whether these differing responses are cell (rat mesangial vs intestinal epithelial cells)- or stimulus (TNF/IFN vs LPS)-specific remains to be determined. In our own studies, the ability of PGE2 to prevent p65 nuclear translocation while promoting p50 nuclear translocation suggests a potentially important regulatory effect on RA SF, key mediators of RA inflammation and cartilage erosion.

The ability of exogenous PGEs to regulate NF-κB suggested that RA SF might employ endogenous PGE2 for a similar purpose. To test this hypothesis, we used ibuprofen and celecoxib to inhibit COX activity and to deplete endogenous PGs. In results reciprocal to the effects of exogenous PGE2, ibuprofen and celecoxib depleted endogenous PGE2 and enhanced p65 nuclear translocation, while inhibiting p50 translocation. PGE2 inhibition also enhanced p65 and inhibited p50 DNA-binding activities using a highly quantitative, ELISA-based assay, as well NF-κB regulation of transcription using a luciferase reporter. Thus, endogenous PGE2 regulates not only p65 and p50 localization, but also the ability of p65 and p50 to interact with chromosomal DNA and regulate transcription. Whether these effects of COX-2 inhibition have clinical implication for anti-inflammatory therapy remains to be determined. Since celecoxib (selective COX-2 inhibitor) produced effects similar to ibuprofen (nonselective COX-1/2 inhibitor) (40), endogenous PGE2 in this system is likely to be generated via COX-2 activity. Because the COX-2 product (PGH2) is used by terminal synthases to generate PGE2, but also a variety of other prostanoids, our data do not formally exclude the possibility that other prostanoids, such as PGJ2, may also regulate NF-κB and thereby promote resolution of inflammation (39). However, the ability of PGE2 to rescue the effect of COX-2 inhibition on NF-κB activation suggests that PGE2 plays an important role in mediating the effects of COX-2 inhibition on NF-κB.

Previous studies implicate the ERK signaling pathway in the persistent activation of NF-κB. Sustained activation of NF-κB via ERK activation has been observed in human atherosclerotic lesions and in vascular smooth muscle cells following arterial injury. Transfection with ERK antisense, but not sense, oligodeoxynucleotides, decreased the protracted activation of NF-κB induced by IL-1β in these models (41, 42). Our present data confirm that ERK regulates NF-κB in human RA SF and that specific ERK pathway inhibition with PD98059 inhibits p65, while enhancing p50 translocation to the nucleus. Similar results were obtained using U0126, another specific MEK inhibitor (data not shown). Moreover, transfection of HeLa cells with constitutively active MEK resulted in nuclear translocation of fluorescent-tagged p65, confirming that ERK pathway activation is sufficient to regulate p65 localization. We attempted to reproduce these latter experiments in SF using a variety of transfection techniques but were hindered by low efficiency of transfection (i.e., ≤3% transfection using standard techniques) and/or inconsistent dual transfection of the MEK and p65 constructs. However, we note that our data using HeLa cells is consistent with our studies using MEK inhibitors in RA SF.

We have previously observed that PGE2 inhibits ERK activation in a number of cell types (3, 43), suggesting that PGE2 might regulate NF-κB in RA SF via its effects on ERK. Our current data support such a mechanism. In particular, ERK inhibition duplicated the effects of PGE2 on p65 and p50 localization in our current experiments, and PGE2 inhibited ERK activation. However, we further observed that PGE2, but not PD98059, increased the expression of IκB, suggesting that PGE2 may also regulate NF-κB independently of ERK by spatially restricting p65/p50 to the cytosolic compartment via up-regulation of its regulatory component. In the present study, we did not assess whether PGE2 might also regulate NF-κB activity through effects on the phosphorylation state of IκBα or through effects on other IκB isoforms.

Since NF-κB regulates COX-2 enzyme expression, the ability of PGE2 to inhibit, and ERK to enhance, NF-κB activation suggested that PGE2 and ERK would differentially regulate COX-2 expression. Our data confirm that PGE2 inhibits, but ERK promotes, COX-2 expression. Thus, increased COX-2 expression in response to inflammation may be subsequently down-regulated by the resultant PGE2 produced in a feedback regulatory loop. The delay inherent in this process (NF-κB activation, followed by COX-2 up-regulation and subsequent PGE2 accumulation, resulting in ERK and NF-κB inhibition and, after turnover of preexisting COX-2, PGE2 depletion) is consistent with the hypothesis that this feedback loop may participate in the resolution phase of inflammation.

In summary, our data indicate that PGE2 exerts inhibitory effects on NF-κB via ERK-dependent and -independent mechanisms. In Fig. 7, we present a best-fit model for these effects. In human RA SF, IL-1β/TNF-α stimulates translocation of p65/p50 heterodimers and inhibits translocation of p50/p50 homodimers via distinct, ERK-dependent processes (Fig. 7,A). p65/p50 translocation induces transcription of multiple inflammatory genes, including COX-2, resulting in PGE2 production. PGE2 (whether endogenously generated or exogenously added) abrogates IL-1β/TNF-α-stimulated ERK activation, inhibiting p65/p50 but enhancing p50/p50 translocation and blocking the expression of inflammatory genes (Fig. 7 B). PGE2 also increases the expression of IκBα by an ERK-independent mechanism, thus inhibiting p65/p50 translocation. These data provide insight into mechanisms by which PGE2 may paradoxically inhibit the action of catabolic cytokines and act to prevent cartilage degradation in arthritis.

FIGURE 7.

A model for the resolution of inflammation by PGE2 in rheumatoid SF. See text for details.

FIGURE 7.

A model for the resolution of inflammation by PGE2 in rheumatoid SF. See text for details.

Close modal

We thank Mark R. Philips for providing reagents and for helpful discussions, David Michelson for technical assistance, Chuanju Liu for providing control vectors, and Madeline Rios for assistance in preparing this manuscript.

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.

1

This work was supported by a grant from the Arthritis Foundation New York Chapter (to M.H.P.).

4

Abbreviations used in this paper: RA, rheumatoid arthritis; SF, synovial fibroblast; MMP, matrix metalloproteinase; COX-2, cyclooxygenase-2; NGS, normal goat serum; PBST, PBS/0.1% Tween 20.

1
Alsalameh, S., R. J. Amin, E. Kunisch, H. E. Jasin, R. W. Kinne.
2003
. Preferential induction of prodestructive matrix metalloproteinase-1 and proinflammatory interleukin 6 and prostaglandin E2 in rheumatoid arthritis synovial fibroblasts via tumor necrosis factor receptor-55.
J. Rheumatol.
30
:
1680
.-1690.
2
Di Battista, J. A., H. Fahmi, Y. He, M. Zhang, J. Martel-Pelletier, J. P. Pelletier.
2001
. Differential regulation of interleukin-1β-induced cyclooxygenase-2 gene expression by nimesulide in human synovial fibroblasts.
Clin. Exp. Rheumatol.
19
:
S3
.-S8.
3
Pillinger, M. H., P. B. Rosenthal, S. N. Tolani, B. Apsel, V. Dinsell, J. Greenberg, E. S. Chan, P. F. Gomez, S. B. Abramson.
2003
. Cyclooxygenase-2-derived E prostaglandins down-regulate matrix metalloproteinase-1 expression in fibroblast-like synoviocytes via inhibition of extracellular signal-regulated kinase activation.
J. Immunol.
171
:
6080
.-6089.
4
Jackson, J. R., J. A. Minton, M. L. Ho, N. Wei, J. D. Winkler.
1997
. Expression of vascular endothelial growth factor in synovial fibroblasts is induced by hypoxia and interleukin 1β.
J. Rheumatol.
24
:
1253
.-1259.
5
Barchowsky, A., D. Frleta, M. P. Vincenti.
2000
. Integration of the NF-κB and mitogen-activated protein kinase/AP-1 pathways at the collagenase-1 promoter: divergence of IL-1 and TNF-dependent signal transduction in rabbit primary synovial fibroblasts.
Cytokine
12
:
1469
.-1479.
6
Ghosh, S., M. J. May, E. B. Kopp.
1998
. NF-κB and Rel proteins: evolutionarily conserved mediators of immune responses.
Annu. Rev. Immunol.
16
:
225
.-260.
7
Pessara, U., N. Koch.
1990
. Tumor necrosis factor α regulates expression of the major histocompatibility complex class II-associated invariant chain by binding of an NF-κB-like factor to a promoter element.
Mol. Cell. Biol.
10
:
4146
.-4154.
8
Ghosh, S., M. Karin.
2002
. Missing pieces in the NF-κB puzzle.
Cell
109
:(Suppl.):
S81
.-S93.
9
DiDonato, J., F. Mercurio, C. Rosette, J. Wu-Li, H. Suyang, S. Ghosh, M. Karin.
1996
. Mapping of the inducible IκB phosphorylation sites that signal its ubiquitination and degradation.
Mol. Cell. Biol.
16
:
1295
.-1304.
10
Makarov, S. S..
2001
. NF-κB in rheumatoid arthritis: a pivotal regulator of inflammation, hyperplasia, and tissue destruction.
Arthritis Res.
3
:
200
.-206.
11
Vincenti, M. P., C. I. Coon, C. E. Brinckerhoff.
1998
. Nuclear factor κB/p50 activates an element in the distal matrix metalloproteinase 1 promoter in interleukin-1β-stimulated synovial fibroblasts.
Arthritis Rheum.
41
:
1987
.-1994.
12
Vincenti, M. P., C. E. Brinckerhoff.
2002
. Transcriptional regulation of collagenase (MMP-1, MMP-13) genes in arthritis: integration of complex signaling pathways for the recruitment of gene-specific transcription factors.
Arthritis Res.
4
:
157
.-164.
13
Catley, M. C., J. E. Chivers, L. M. Cambridge, N. Holden, D. M. Slater, K. J. Staples, M. W. Bergmann, P. Loser, P. J. Barnes, R. Newton.
2003
. IL-1β-dependent activation of NF-κB mediates PGE2 release via the expression of cyclooxygenase-2 and microsomal prostaglandin E synthase.
FEBS Lett.
547
:
75
.-79.
14
Crofford, L. J., B. Tan, C. J. McCarthy, T. Hla.
1997
. Involvement of nuclear factor κB in the regulation of cyclooxygenase-2 expression by interleukin-1 in rheumatoid synoviocytes.
Arthritis Rheum.
40
:
226
.-236.
15
Sherman, M. P., E. E. Aeberhard, V. Z. Wong, J. M. Griscavage, L. J. Ignarro.
1993
. Pyrrolidine dithiocarbamate inhibits induction of nitric oxide synthase activity in rat alveolar macrophages.
Biochem. Biophys. Res. Commun.
191
:
1301
.-1308.
16
Eberhardt, W., D. Kunz, J. Pfeilschifter.
1994
. Pyrrolidine dithiocarbamate differentially affects interleukin 1β- and cAMP-induced nitric oxide synthase expression in rat renal mesangial cells.
Biochem. Biophys. Res. Commun.
200
:
163
.-170.
17
Chace, J. H., A. L. Fleming, J. A. Gordon, C. E. Perandones, J. S. Cowdery.
1995
. Regulation of differentiation of peritoneal B-1a (CD5+) B cells: activated peritoneal macrophages release prostaglandin E2, which inhibits IgM secretion by peritoneal B-1a cells.
J. Immunol.
154
:
5630
.-5636.
18
Weissmann, G..
1993
. Prostaglandins as modulators rather than mediators of inflammation.
J. Lipid Mediat.
6
:
275
.-286.
19
Pillinger, M. H., M. R. Philips, A. Feoktistov, G. Weissmann.
1995
. Crosstalk in signal transduction via EP receptors: prostaglandin E1 inhibits chemoattractant-induced mitogen-activated protein kinase activity in human neutrophils.
Adv. Prostaglandin Thromboxane Leukotrienes Res.
23
:
311
.-316.
20
Zurier, R. B., I. Damjanov, D. M. Sayadoff, N. F. Rothfield.
1977
. Prostaglandin E1 treatment of NZB/NZW F1 hybrid mice. II. Prevention of glomerulonephritis.
Arthritis Rheum.
20
:
1449
.-1456.
21
Zurier, R. B., F. Quagliata.
1971
. Effect of prostaglandin E1 on adjuvant arthritis.
Nature
234
:
304
.-305.
22
Salvatori, R., P. T. Guidon, Jr, B. E. Rapuano, R. S. Bockman.
1992
. Prostaglandin E1 inhibits collagenase gene expression in rabbit synoviocytes and human fibroblasts.
Endocrinology
131
:
21
.-28.
23
Suzuki, K., B. E. Rapuano, R. S. Bockman.
1997
. Role of protein kinase A in collagenase-1 gene regulation by prostaglandin E1: studies in a rabbit synoviocyte cell line, HIG-82.
J. Bone Miner. Res.
12
:
561
.-567.
24
Georgescu, H. I., D. Mendelow, C. H. Evans.
1988
. HIG-82: an established cell line from rabbit periarticular soft tissue, which retains the “activatable” phenotype.
In Vitro Cell Dev. Biol.
24
:
1015
.-1022.
25
Pillinger, M. H., V. Dinsell, B. Apsel, S. N. Tolani, N. Marjanovic, E. S. Chan, P. Gomez, R. Clancy, L. F. Chang, S. B. Abramson.
2004
. Regulation of metalloproteinases and NF-κB activation in rabbit synovial fibroblasts via E prostaglandins and Erk: contrasting effects of nabumetone and 6MNA.
Br. J. Pharmacol.
142
:
973
.-982.
26
Clancy, R. M., P. F. Gomez, S. B. Abramson.
2004
. Nitric oxide sustains nuclear factor κB activation in cytokine-stimulated chondrocytes.
Osteoarthritis Cartilage
12
:
552
.-558.
27
Alessi, D. R., A. Cuenda, P. Cohen, D. T. Dudley, A. R. Saltiel.
1995
. PD 098059 is a specific inhibitor of the activation of mitogen-activated protein kinase kinase in vitro and in vivo.
J. Biol. Chem.
270
:
27489
.-27494.
28
Poppers, D. M., P. Schwenger, J. Vilcek.
2000
. Persistent tumor necrosis factor signaling in normal human fibroblasts prevents the complete resynthesis of IκB-α.
J. Biol. Chem.
275
:
29587
.-29593.
29
Arend, W. P., M. Malyak, C. F. Bigler, M. F. Smith, Jr, R. W. Janson.
1991
. The biological role of naturally-occurring cytokine inhibitors.
Br. J. Rheumatol.
30
:(Suppl. 2):
49
.-52.
30
Morita, Y., M. Yamamura, M. Kawashima, T. Aita, S. Harada, H. Okamoto, H. Inoue, H. Makino.
2001
. Differential in vitro effects of IL-4, IL-10, and IL-13 on proinflammatory cytokine production and fibroblast proliferation in rheumatoid synovium.
Rheumatol. Int.
20
:
49
.-54.
31
Serhan, C. N., N. Chiang.
2004
. Novel endogenous small molecules as the checkpoint controllers in inflammation and resolution: entree for resoleomics.
Rheum. Dis. Clin. North Am.
30
:
69
.-95.
32
Udalova, I. A., A. Richardson, A. Denys, C. Smith, H. Ackerman, B. Foxwell, D. Kwiatkowski.
2000
. Functional consequences of a polymorphism affecting NF-κB p50–p50 binding to the TNF promoter region.
Mol. Cell. Biol.
20
:
9113
.-9119.
33
Plaksin, D., P. A. Baeuerle, L. Eisenbach.
1993
. KBF1 (p50 NF-κB homodimer) acts as a repressor of H-2Kb gene expression in metastatic tumor cells.
J. Exp. Med.
177
:
1651
.-1662.
34
Kang, S. M., A. C. Tran, M. Grilli, M. J. Lenardo.
1992
. NF-κB subunit regulation in nontransformed CD4+ T lymphocytes.
Science
256
:
1452
.-1456.
35
Bohuslav, J., V. V. Kravchenko, G. C. Parry, J. H. Erlich, S. Gerondakis, N. Mackman, R. J. Ulevitch.
1998
. Regulation of an essential innate immune response by the p50 subunit of NF-κB.
J. Clin. Invest.
102
:
1645
.-1652.
36
Ma, X. Y., H. Wang, B. Ding, H. Zhong, S. Ghosh, P. Lengyel.
2003
. The interferon-inducible p202a protein modulates NF-κB activity by inhibiting the binding to DNA of p50/p65 heterodimers and p65 homodimers while enhancing the binding of p50 homodimers.
J. Biol. Chem.
278
:
23008
.-23019.
37
Nakashima, O., Y. Terada, S. Inoshita, M. Kuwahara, S. Sasaki, F. Marumo.
1999
. Inducible nitric oxide synthase can be induced in the absence of active nuclear factor-κB in rat mesangial cells: involvement of the Janus kinase 2 signaling pathway.
J. Am. Soc. Nephrol.
10
:
721
.-729.
38
D’Acquisto, F., L. Sautebin, T. Iuvone, M. Di Rosa, R. Carnuccio.
1998
. Prostaglandins prevent inducible nitric oxide synthase protein expression by inhibiting nuclear factor-κB activation in J774 macrophages.
FEBS Lett.
440
:
76
.-80.
39
Poligone, B., A. S. Baldwin.
2001
. Positive and negative regulation of NF-κB by COX-2: roles of different prostaglandins.
J. Biol. Chem.
276
:
38658
.-38664.
40
Warner, T. D., F. Giuliano, I. Vojnovic, A. Bukasa, J. A. Mitchell, J. R. Vane.
1999
. Nonsteroid drug selectivities for cyclo-oxygenase-1 rather than cyclo-oxygenase-2 are associated with human gastrointestinal toxicity: a full in vitro analysis.
Proc. Natl. Acad. Sci. USA
96
:
7563
.-7568.
41
Jiang, B., P. Brecher, R. A. Cohen.
2001
. Persistent activation of nuclear factor-κB by interleukin-1β and subsequent inducible NO synthase expression requires extracellular signal-regulated kinase.
Arterioscler. Thromb. Vasc. Biol.
21
:
1915
.-1920.
42
Jiang, B., S. Xu, X. Hou, D. R. Pimentel, P. Brecher, R. A. Cohen.
2004
. Temporal control of NF-κB activation by ERK differentially regulates interleukin-1β-induced gene expression.
J. Biol. Chem.
279
:
1323
.-1329.
43
Pillinger, M. H., N. Marjanovic, S. Y. Kim, J. U. Scher, P. Izmirly, S. Tolani, V. Dinsell, Y. C. Lee, M. J. Blaser, S. B. Abramson.
2005
. Matrix metalloproteinase secretion by gastric epithelial cells is regulated by E prostaglandins and mitogen-activated protein kinases.
J. Biol. Chem.
280
:
9973
.-9979.