Manipulation of Distinct NFκB Proteins Alters Interleukin-1β-induced Human Rheumatoid Synovial Fibroblast Prostaglandin E2 Formation

Interleukin 1β (IL-1β) up-regulates human rheumatoid synovial fibroblast (RSF) 85-kDa phospholipase A2 (PLA2) and mitogen-inducible cyclooxygenase (COX) II. Promoter regions for these genes contain a motif that closely resembles the “classic” NFκB consensus site. Immunoblot analysis identified NFκB1 (p50), RelA (p65), and c-Rel in RSF. Upon IL-1β-stimulation, p65 and c-Rel but not p50 protein levels were reduced suggesting nuclear translocation. IL-1β-induced RSF nuclear extracts contained a p65-containing complex, which bound to the classical NFκB consensus motif. An NFκB classical oligonucleotide decoy produced a concentration-dependent decrease in IL-1-stimulated PGE2 production (IC50 = ∼2 μM), indicating a role of NFκB. Utilization of antisense technology showed that p65 but not p50 or c-Rel mediated IL-1β-stimulated PGE2 formation. Treated RSF could not transcribe COX II or 85-kDa PLA2 mRNA, which reduced their respective proteins. Interestingly, stimulated IL-8 production was not inhibited by the classical NFκB decoy but was reduced by treatment with antisense to both p65 and c-Rel supporting preferential binding of c-Rel-p65 to the “alternative” IL-8 κB motif. Taken together, these data provide the first direct evidence for a role of p65 in COX II and 85-kDa PLA2 gene induction and support the IL-1 activation and participation of distinct NFκB protein dimers in RSF prostanoid and IL-8 formation.

Rheumatoid arthritis is an autoimmune disease characterized by chronic inflammation and hyperproliferation of the synovial lining (1). Enhanced levels of the cytokine, interleukin (IL)-1␤, 1 perpetuate the disease process through up-regulation of a multitude of factors leading to eicosanoid formation, matrix degradation, bone resorption, and proliferation in the joint (2)(3)(4)(5)(6). We and others have demonstrated that human rheumatoid synovial fibroblast (RSF) prostaglandin (PG) E 2 accumulation in response to IL-1␤ is a direct result of the coordinate up-regulation of 85-kDa phospholipase A 2 (PLA 2 ) and the induction of COX II (6 -8). Indeed, we reported that depletion of IL-1␤-induced 85-kDa PLA 2 to basal levels by antisense severely compromised the ability of RSF to make PGE 2 . However, the mechanism(s) by which IL-1␤ regulates 85-kDa PLA 2 and COX II gene induction in this system have not been elucidated.
IL-1␤ is a potent activator of nuclear factor B (NFB) (1,9,10) in other cell systems, and this transcription factor in turn regulates a wide variety of inflammatory and immunoregulatory genes (10 -16). 5Ј-flanking regulatory regions for both the human 85-kDa PLA 2 and COX II genes have recently been isolated (17,18), and sequence analysis has identified a number of possible transcription factor consensus binding motifs, including NFB. The putative NFB motif in the 85-kDa PLA 2 promoter is located at Ϫ1099 base pairs (17), whereas the NFB consensus site in the human COX II promoter is located at Ϫ233 base pairs (18).
NFB is a dimeric DNA binding protein comprised of members of the NFB/Rel/dorsal family of proteins including the mammalian forms, NFB1 (p50), NFB2 (p52), RelA (p65), RelB, and c-Rel (10,11). NFB proteins are capable of forming numerous homodimers and heterodimers with other family members, and this conveys a degree of NFB gene target specificity. In addition, the different dimeric pairs bind to different NFB DNA motifs with varying affinities, again permitting specificity (19,20). These motifs can be subdivided into two main categories, the "classic" B consensus site (GGGRN-NTYCC) and the "alternative" or B-like motif ((T/A/C)GGAR-NYYCC). Reports demonstrated that different homodimer or heterodimer pairs exhibit varied affinity for these two motif classes. The p50-p65 heterodimer has been shown to bind preferentially to the decameric oligonucleotide sequence (GG-GACTTTCC) present in Ig -light chain enhancer (20). The NFB motifs present in the 85-kDa PLA 2 (GGGAATTCTC) and COX II (GGGACTACCC) promoters most closely resemble classic B consensus motifs. In contrast, the alternative motif, present in promoters such as IL-8, intercellular adhesion molecule, and tissue factor, binds c-Rel-p65 heterodimers with higher affinity than p50-p65 (20).
Double-stranded NFB decoys are an effective approach to specifically modulate expression of NFB-driven reporter genes through successful competition of the decoy with the authentic DNA motifs, preventing dimer translocation and/or DNA binding (21). In addition, antisense technology has been successfully used to modulate NFB proteins to assess their role in the transcriptional control of a variety of genes, such as the elucidation of a role for p65 but not p50 in IL-8 expression (14). Herein, we demonstrate that IL-1␤ induces the activation of NFB in human RSF and that modulation of NFB activation using the tools described above interferes with prostanoid formation and transcriptional up-regulation of 85-kDa PLA 2 and COX II.

MATERIALS AND METHODS
RSF Culture and Eicosanoid Measurement-Primary cultures of human RSF were obtained from 10 adult patients and maintained in culture as described previously (6,22). For some studies, fibroblasts were plated at 5 ϫ 10 4 cells/ml in 16-mm (diameter) 24-well plates (Costar, Cambridge, MA) and used when confluent. In studies using antisense, the medium containing 10% fetal bovine serum was removed and serum-free medium supplemented with insulin, transferrin, and sodium selenite (Sigma) was added. Cells were exposed to IL-1␤ (1 ng/ml, Genzyme, Cambridge, MA) for the designated time (6) after which PGE 2 levels or IL-8 levels in cell-free medium were measured using enzyme immunoassay purchased (Cayman Chemical Co., Ann Arbor, MI and Biosource International, Camarillo, CA) as described previously (6). In some studies cells were first treated with NFB decoys (4 h at 37°C) or antisense (18 h at 37°C) prior to the addition of IL-1␤ for 8 or 24 h.
RSF Subcellular Fractionation and Immunoblot Analysis-Human RSF were trypsinized, resuspended to 1.0 ϫ 10 8 cells/ml in homogenization buffer, and disrupted on ice by sonication, and the homogenate was centrifuged at 100,000 ϫ g for 60 min at 4°C to obtain a supernatant (cytosol) and particulate fraction as described previously (6). Cell fractions (25-40 g protein/lane) were analyzed by SDS-polyacrylamide gel electrophoresis (10% gels, Bio-Rad) and visualized using the ECL Western blotting system (Amersham Corp.). Rabbit polyclonal antiserum against the recombinant human 85-kDa PLA 2 or rabbit anti-human COX II (kindly donated by D. Dewitt (Michigan State University, East Lansing, MI)) was used as described previously (6,22). A 24-h lipopolysaccharide-stimulated monocyte particulate fraction was used as a positive control for the COX II protein. Recombinant human 85-kDa PLA 2 was cloned and expressed as described previously (6) and used as a protein standard. Rabbit polyclonal antisera against NFB proteins p50, p65, and c-Rel were obtained from Santa Cruz Biotechnology (Santa Cruz, CA) and were used according to the manufacturer's instructions.
Preparation of Nuclear Extracts and Electrophoretic Mobility Shift Assay-Nuclear extracts of RSF were prepared according to published methods (23,24) with some modifications. In brief, treated and control cells were removed by addition of trypsin/EDTA and then pelleted on ice. 10 7 cells were resuspended in 20 l of Buffer A (10 mM HEPES, pH 7.9, 10 mM KCl, 1.5 mM MgCl 2 , 0.5 mM dithiothreitol, 0.1% (w/v) Nonidet P-40) and incubated on ice for 10 min, and nuclei were pelleted by microcentrifugation at 3500 rpm for 10 min at 4°C. The pellet was suspended in 15 l of Buffer C (20 mM HEPES, pH 7.9, 0.42 M NaCl, 1.5 mM MgCl 2 , 25% (v/v) glycerol, 0.2 mM EDTA, 0.5 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride) and gently mixed for 20 min at 4°C. The sample was centrifuged at 14,000 rpm for 10 min at 4°C, and the decanted supernatant was diluted to 75 l with Buffer D (20 mM HEPES, pH 7.9, 50 mM KCl, 20% glycerol, 0.2 mM EDTA, 0.5 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride). Samples were stored at Ϫ80°C until analyzed.
A double-stranded oligonucleotide containing the sequence corresponding to the classical NFB consensus site (5Ј-agttgaggggactttcccagcc-3Ј, Santa Cruz Biotechnology, Inc.) was end-labeled with [␥-32 P]ATP using T4 kinase (Life Technologies, Inc.). EMSAs were performed using the classical B motif, which most closely represented the putative NFB sites in the promoters of both COX II and 85-kDa PLA 2 . Unincorporated nucleotides were removed using two Sephadex G-50 columns (Pharmacia Biotech Inc.). Binding reactions were carried out in a final volume of 25 l consisting of 10 mM HEPES, pH 7.9, 4 mM Tris-HCl 60 mM KCl, 1 mM EDTA, 1 mM dithiothreitol, 10% glycerol, 1.5 mg/ml bovine serum albumin, 2 g of poly(dI-dC), 2-10 g of nuclear extract, and 0.5 ng of 32 P-labeled oligonucleotide probe (ϳ50,000 cpm) with or without unlabeled competitor (20 -40-fold excess). The Oct-1 consensus sequence, 5Ј-tgtcgaatgcaaatcactagaa-3Ј, was purchased from Santa Cruz Biotechnology, Inc. Reactions were incubated for 20 min at room temperature. In "supershift" studies, 1 g rabbit anti-p50, anti-p65, or anti-c-Rel serum (Santa Cruz Biotechnology) was added to the reaction and incubated for 45 min at room temperature. Binding reactions were subjected to nondenaturing polyacrylamide electrophoresis through 4% gels in a 1 ϫ Tris borate-EDTA buffer system. Gels were dried and subjected to autoradiography.
Northern Analysis-Total RNA was isolated from RSF using Trizol reagent (Life Technologies, Inc.) according to the manufacturer's protocol. Equal concentrations of RNA (20 g/lane) were subjected to electrophoresis in 1% agarose gel containing formaldehyde. Following electrophoresis, gels were rinsed, transferred to Hybond Nϩ (Amersham Corp.) by vacuum blotting (Bio-Rad) in 10 ϫ SSC according to manufacturer's protocol, and RNA was fixed to the membrane by UV cross-linking (0.12 J/cm 2 ). RNA samples were visualized by staining with 0.02% methylene blue (5 min) and destained in distilled water (15 min). Equal loading was verified by quantitation of ribosomal RNA bands. Hybridizations, following standard prehybridization treatment, were done in prehybridization solution (10 ml/blot) containing 20 ng of denatured specific DNA probe labeled to 1-2 ϫ 10 9 dpm/g with 32 P (see below) at 68°C for 18 h. Following hybridization, blots were washed twice with 2 ϫ SSC, 0.1% SDS at 68°C and and once with 1 ϫ SSC, 0.1% SDS at 68°C and once with 0.2 ϫ SSC, 0.1% SDS at 68°C. Filters were analyzed and quantitated on a PhosphorImager (Molecular Dynamics, Sunnyvale, CA).
The 85-kDa PLA 2 probe was the full-length cDNA (2.87 kilobases) and the COX II probe was a 1.2-kilobase EcoRI fragment of the murine COX II cDNA clone kindly provided by Dr. David DeWitt, Michigan State University. 20 ng of cDNA was labeled using a Rediprime kit and 50 Ci of [ 32 P]dCTP (Amersham Corp.). Unincorporated nucleotides were removed using Quick Spin columns (Boehringer Mannheim).
Calculations and Statistics-All studies were performed using RSF from two to five different individuals. Data represent mean Ϯ standard deviation (S.D.), n ϭ 3, and subjected to one way analysis of variance and Duncan's multiple range test (p Ͻ 0.05) for statistical evaluation where indicated.

Characterization of IL-1␤-induced Activation of NFB in
RSF-Confluent RSF in T-75 flasks were placed in serum-free Eagle's minimum essential medium for 24 h prior to incubation in the presence of vehicle or IL-1␤ for 0, 15, or 30 min at 37°C. Nuclear extracts were prepared and analyzed by EMSA as described under "Materials and Methods." Fig. 1A shows that a protein complex that binds to the 32 P-labeled NFB classical sequence is present in nuclear extracts of unstimulated and stimulated RSF. IL-1␤ stimulation of the RSF increased the nuclear levels of the protein complex in a time-dependent manner. The 15-min time point was chosen for subsequent NFB activation studies because it routinely displayed a 2-3-fold increase in NFB binding activity as assessed by scanning gel densitometry (area pixel values for Fig. 1,: control, 945; IL-1␤ (15 min), 1909; IL-1␤ (30 min), 1100: area pixel values for a similar study: control, 1179; IL-1␤ (15 min), 3773). Specific binding to the NFB classical motif was demonstrated when labeled probe binding was inhibited by incubation with excess unlabeled NFB consensus oligonucleotide (20ϫ) but not by an excess of the unrelated Oct-1 oligonucleotide (20ϫ) (Fig. 1B). Use of the COX II B motif gave identical EMSA results (data not shown). Therefore, the classical motif was used for all subsequent studies.
To determine the presence of p50, p65, or c-Rel in RSF, confluent cells were placed in serum-free medium for 24 h prior to exposure to vehicle or IL-1␤ (1 ng/ml) for 15 min. After the indicated time, cells were pelleted, and cytosolic fractions or nuclear extracts were prepared as described. Western blot analysis of cytosolic fractions (30 g) confirmed the presence all three NFB proteins in unstimulated control RSF (Fig. 2). Exposure to IL-1␤ resulted in 75 and 43% reductions in cyto-solic levels of p65 and c-Rel, respectively, as measured by scanning gel densitometry (area pixel values: p65 nontreated control, 591; p65 IL-1 treated, 157: c-Rel nontreated control, 675; c-Rel IL-1 treated, 388). However, stimulation with IL-1 did not alter p50 over this time frame because there was no detectable change in p50 immunoreactive protein levels (p50 untreated control, 864; p50 IL-1 treated, 938).
In order to evaluate the presence of NFB proteins in control and treated RSF nuclear extracts, supershift EMSA analysis was performed using radiolabeled classic NFB motif and antiserum specific for p50, p65, or c-Rel. Fig. 3 shows one representative supershift EMSA. Exposure of RSF to IL-1 resulted in the appearance of an NFB binding complex, which could be supershifted by antiserum to p65. Neither antiserum specific for p50 nor that specific for c-Rel induced a supershift in either control or stimulated nuclear extracts using the conditions described in this study. In the case of c-Rel, the lack of binding affinity for c-Rel containing dimers to classical NFB could also account for the lack of a supershift (20). The possibility that the antibodies recognize epitopes that are concealed by DNA binding cannot be ruled out.
The Effect of NFB Decoys on IL-1␤-induced PGE 2 Formation-Double-stranded oligonucleotide decoys (22-mers) containing the classic NFB consensus site or a mutant NFB consensus site were prepared by phosphoramidite chemistry, annealed, and complexed to Lipofectin as described under "Materials and Methods." RSF in Eagle's minimum essential medium supplemented with insulin, transferrin, and sodium selenite were incubated with the NFB decoy (0.3-3.0 M), the mutant decoy (3.0 M), or Lipofectin vehicle alone for 4 h at 37°C (21) prior to the addition of IL-1␤ (18 h, 37°C). Cells were monitored throughout the study, and no greater cytotoxity was observed in the treated groups compared with the untreated controls as assessed by morphology, adherence, and trypan blue exclusion. Fig. 4A shows that the NFB decoy but not the mutant decoy concentration-dependently reduced IL-1␤-induced PGE 2 formation. Pretreatment of IL-1␤-stimulated cells with either 1.0 or 3.0 M NFB decoy resulted in a 43% or a 57% reduction in PGE 2 production, whereas preincubation with the mutant decoy failed to attenuate PGE 2 levels. In contrast, IL-1-induced IL-8 production by the same RSF was

FIG. 1. NFB is present in RSF and is activated by IL-1␤.
In A, human RSF were stimulated with IL-1␤ (1 ng/ml) for 0, 15, or 30 min prior to preparation of nuclear extracts as described under "Materials and Methods." Nuclear extracts (10 g) were incubated with radiolabeled classic NFB oligonucleotide probe (0.5 ng) in the presence or the absence of unlabeled classic NFB oligonucleotide (40ϫ) and subjected to an EMSA. In B, nuclear extracts of IL-1␤-stimulated cells (10 g) were incubated with radiolabeled classic NFB oligonucleotide probe (0.5 ng) and a 20-fold excess of either the classic NFB motif or the unrelated oligonucleotide Oct-1 for 30 min and then subjected to EMSA.

FIG. 2. Western blot analysis of NFB proteins in cytosol of
control and IL-1␤-stimulated RSF. RSF were exposed to medium vehicle or IL-1␤ (1 ng/ml) for 15 min prior to preparation of subcellular fractions as described under "Materials and Methods." Cytosolic fractions (30 g) from control and IL-1␤-stimulated RSF were electrophoresed through 10% SDS-polyacrylimide gels and subjected to Western blot analysis using anti-p50, anti-p65, or anti-c-Rel rabbit polyclonal serum.

FIG. 3. Supershift analysis using the classic B motif confirms
IL-1␤-induced nuclear translocation of p65. RSF were treated with medium vehicle or IL-1␤ (1 ng/ml) for 15 min prior to preparation of nuclear extracts as described under "Materials and Methods." Control and stimulated nuclear extracts (5 g) were incubated with radiolabeled classic NFB oligonucleotide probe (0.5 ng) for 20 min at 27°C after which time 1 g of anti-p50, anti-p65, and/or anti-c-Rel serum was added to the binding reaction of stimulated nuclear extracts for an additional 45 min. Binding reactions were analyzed by EMSA as described under "Materials and Methods." not significantly altered by pretreatment with either the NFB decoy or the mutant decoy (Fig. 4B).
Effect of NFB Antisense on IL-1␤-induced RSF PGE 2 and IL-8 Formation-To specifically modulate individual NFB proteins, fibroblasts in serum-free Eagle's minimum essential medium were incubated (18 h at 37°C) in the presence of Lipofectin vehicle or increasing concentrations (0.03-0.3 M) of antisense to p65 (SB68), p50 (SB70), c-Rel (SB7721), or a scrambled oligonucleotide control (SB7222) as described under "Materials and Methods." After the indicated time, cells were exposed to IL-1␤ (8 h, 37°C), and cell-free medium was analyzed for PGE 2 levels as described under "Materials and Methods." No toxicity compared with the Lipofectin control cultures was observed. Fig. 5A shows that prostaglandin levels of IL-1stimulated RSF pretreated with 0.03, 0.1, or 0.3 M p65 antisense, SB68, were significantly (p Ͻ 0.05) reduced by Ϫ23, Ϫ35, or Ϫ66%, respectively. In contrast, RSF pretreated with p50 antisense, c-Rel antisense, or the control oligonucleotide produced the same level of PGE 2 in response to IL-1␤ as the unstimulated control cells. Fig. 5B shows that in the same cells pretreatment with p65 antisense and c-Rel antisense dose-dependently reduced stimulated IL-8 formation, whereas antisense to p50 and the control oligonucleotide failed to modulate IL-8 levels. RSF from some individual donors displayed sensitivity to nonspecific oligonucleotide treatment as was discovered using antisense to 85-kDa PLA 2 in previous studies (6), e.g. modest nonspecific reductions or potentiations. Nonetheless, all donors consistently responded to p65 antisense and subsequent studies examining the effect of p65 antisense on PGE 2 formation were performed at 0.3 M.
Effect of p65 Antisense on COX II and 85-kDa PLA 2 - Fig. 6A shows one of two representative experiments demonstrating that exposure to IL-1␤ caused the accumulation of COX II immunoreactive protein, whereas little or no COX II was evident in nonstimulated RSF as has been previously reported (6 -8). Pretreatment with p65 antisense significantly inhibited the IL-1␤-induced increase in COX II immunoreactive protein levels. RSF of this particular donor displayed some sensitivity to oligonucleotide treatment as a marginal reduction in COX II protein expression occurred with the control oligonucleotide. This was mirrored in the PGE 2 levels measured in the study (unstimulated PGE 2 , 0.4 ng/ml; IL-1␤-stimulated PGE 2 , 10.1 ng/ml; SB68 ϩ IL-1␤ PGE 2 , 1.1 ng/ml; SB71 ϩ IL-1␤ PGE 2 , 8.9 ng/ml). Western analysis of the 85-kDa PLA 2 revealed no change in protein levels with either p65 antisense or control oligonucleotide after 8 h (data not shown). However, evaluation of samples treated identically but exposed to IL-1␤ for 24 h rather than 8 h showed that 85-kDa PLA 2 protein was reduced Human RSF were exposed to Lipofectin vehicle or increasing amounts of p65 antisense (SB68), p50 antisense (SB70), c-Rel antisense (SB7221), or control oligonucleotide (SB7222) for 18 h prior to 8 h of stimulation with IL-1␤ (1 ng/ml). Cell-free medium was removed and analyzed for PGE 2 (A) levels or IL-8 levels (B) as described under "Materials and Methods." Data are expressed as a mean of the percentage of stimulated control level Ϯ S.D. (stimulated PGE 2 range, 3.0 -27.7 ng/ml; stimulated IL-8 range 1.1-9.4 ng/ml) of triplicate determinations from three to five donors. * indicates significant differences from IL-1␤ control at p Ͻ 0.05. by ϳ40% as assessed by scanning gel densitometry (pixel values: nontreated control, 30; IL-1␤-stimulated control, 320; 0.3 M SB68 198) (Fig. 6B).
The effect of p65 antisense treatment on the mRNA levels for COX II and 85-kDa PLA 2 in IL-1␤ stimulated RSF was analyzed by Northern blotting. Fig. 7A shows one representative of two studies where COX II mRNA was undetectable in unstimulated control cells and highly induced following 8 h of stimulation with IL-1␤ (Fig. 7A) as previously reported (8). p65 antisense treatment reduced this induction 13-fold, whereas the control oligonucleotide caused a minor reduction in this particular donor (ϳ2-fold). A significant level of 85-kDa PLA 2 mRNA was detectable in the absence of IL-1␤ stimulation and increased 14-fold following an 8-h treatment with IL-1␤ (Fig. 7B). This induction was totally blocked by p65 antisense, whereas only a marginal decrease was observed with the control oligonucleotide, SB71. PGE 2 levels measured in cell-free media reflect the Northern blot data (Fig. 7C). DISCUSSION We were interested in assessing the possibility that NFB participates in IL-1-induced PGE 2 formation. NFB proteins, c-Rel, p65, and p50 were identified in RSF and exposure to IL-1␤ clearly induced an early and time-dependent nuclear translocation of c-Rel and p65 but not p50. These results agree with the immunohistochemical localization of p65 in the nuclei of synovial cells from rheumatoid arthritis patients recently reported by Handel et al. (25). The possible role of other NFB proteins such as RelB was not evaluated at this time, and therefore their involvement cannot be ruled out. However, expression of RelB appears to be restricted primarily to lymphoid tissues (26), and RelB homodimers or heterodimers with p65 and c-Rel have not been observed either in vitro or in vivo (27,28). Therefore, it is unlikely that this family member is in-volved in synovial fibroblast IL-1-driven prostanoid formation.
It is well established that many inflammatory inducers such as IL-1␤, tumor necrosis factor, and lipopolysaccharide stimulate PGE 2 formation from a wide variety of cell types including monocyte/macrophage, fibroblast, and endothelial cells (29). Further, these ligands are known to mediate the activation of NFB (30). As such, a role for this transcription factor in the regulation of stimulated prostanoid formation could be envisioned. Little is known about the transcriptional control of either the COX II or the 85-kDa PLA 2 gene expression. The cyclic AMP response element has been shown to be necessary for COX II induction in phorbol ester-differentiated human U937 cells (18), and the importance of both the cyclic AMP response element and NF-IL6 sites for COX II expression in lipopolysaccharide and phorbol ester-stimulated human vascular smooth muscle cells has been described (31). In a murine osteoblast cell line, NFB and NF-IL-6 were important in tumor necrosis factor-stimulated PGE 2 production (32). Here we provide evidence for a role of NFB in IL-1-stimulated PGE 2 formation. This is in line with recent observations that glu- cocorticoids, which inhibit prostanoid formation act, in part, through regulation of NFB activation (33).
Exposure to the NFB decoy, mimicking the classical motif, but not the mutant decoy specifically reduced RSF PGE 2 levels in response to IL-1. This indicated that dimers displaying high affinity for the classical B motif were important in IL-1-induced PGE 2 formation. Interestingly, IL-1-induced IL-8 production was not affected, which suggested involvement of a distinct NFB, one that had little or no affinity for the classical motif. This was not unexpected because the IL-8 promoter contains an NFB element, which represents an alternative B motif. Because c-Rel-p65 binds to the alternative site, the data could suggest that this heterodimer is not involved in PGE 2 regulation.
To specifically delineate which NFB proteins are involved in PGE 2 biosynthesis, NFB antisense oligonucleotides to p65, p50, and c-Rel were utilized. Antisense has been successfully used to demonstrate specific NFB regulation of proinflammatory mediators and models of cell adhesion such as CD11b expression in differentiated HL-60 leukemia cells (34) and murine embryonic stem cell adhesion (35). Kunsch and Rosen successfully demonstrated that antisense oligonucleotides to p65 but not p50 reduced phorbol myristate acetate-induced Jurkat T lymphocytes IL-8 production (14).
In RSF, utilization of antisense demonstrated the importance of p65 but not p50 or c-Rel in IL-1␤ induced PGE 2 formation. The lack of effect of p50 antisense in this system could be due to p50 protein stability over the time course of our studies, or it is possible that an alternative heterodimer can compensate for the loss of p50. This particular p50 antisense oligonucleotide was effective in inhibiting cell to substratum adhesion in differentiated embryonic stem cells but not in undifferentiated embryonic stem cells where p65 antisense was shown to be effective (35). The lack of effect of p50 antisense in our system therefore indicates that p50 is not a dimeric partner in the induction of PGE 2 . c-Rel antisense also had no effect on stimulated PGE 2 formation. Alternatively, IL-1-stimulated IL-8 production was reduced by antisense to both p65 and c-Rel consistent with the preferential binding of a c-Rel-p65 heterodimer to the IL-8 B-like motif and the lack of affinity of the classical B motif for c-Rel containing dimers. This strongly suggests that c-Rel is not the dimeric partner participating with p65 in the regulation of COX II and 85-kDa PLA 2 gene expression.
The reduced capability of p65 antisense-treated RSF to produce PGE 2 correlated with a reduction in both COX II and 85-kDa PLA 2 mRNA levels. At 8 h of IL-1 exposure, a time point where COX II protein is clearly present (6 -8), p65 antisense treatment resulted in total depletion of COX II protein.
The 85-kDa PLA 2 protein depletion required a longer time possibly due to a long protein half-life, as suggested by Hulkower et al. (8), or the involvement of more complex posttranscriptional regulatory events (36,37). Because of the temporal discrepancy in protein depletion, it would appear that the reduction in PGE 2 observed in the first 8 h was primarily due to the lack of COX II. Nonetheless, these data support NFB participation in the regulation of both COX II and 85-kDa PLA 2 expression. Indeed, the recent co-localization of both human genes to the same region of chromosone 1 (1q25) (38,39) suggests the intriguing possibility for coordinate gene regulation.
To our knowledge, these data provide the first direct evidence for a role of the NFB protein, p65, in regulation of COX II and 85-kDa PLA 2 gene expression. Because both are re-quired for IL-1-induced PGE 2 formation in RSF, it appears that NFB activation plays a regulatory role in PGE 2 biosynthesis. Finally, we show that PGE 2 formation requires NFB dimers that appear to be different from those that participate in IL-1driven IL-8 expression.