15-Deoxy-Δ12,14-prostaglandin J2-mediated ERK Signaling Inhibits Gram-negative Bacteria-induced RelA Phosphorylation and Interleukin-6 Gene Expression in Intestinal Epithelial Cells through Modulation of Protein Phosphatase 2A Activity*

We have previously shown that non-pathogenic Gram-negative Bacteroides vulgatus induces transient RelA phosphorylation (Ser-536), NF-κB activity, and pro-inflammatory gene expression in native and intestinal epithelial cell (IEC) lines. We now demonstrate that 15-deoxy-Δ12,14-prostaglandin J2 (15d-PGJ2) but not prostaglandin E2 inhibits lipopolysaccharide (LPS) (B. vulgatus)/LPS (Escherichia coli)-induced RelA phosphorylation and interleukin-6 gene expression in the colonic epithelial cell line CMT-93. This inhibitory effect of 15d-PGJ2 was mediated independently of LPS-induced IκBα phosphorylation/degradation and RelA nuclear translocation as well as RelA DNA binding activity. Interestingly, although B. vulgatus induced nuclear expression of peroxisome proliferator-activated receptor γ (PPARγ) in native epithelium of monoassociated Fisher rats, PPARγ-specific knock-down in CMT-93 cells using small interference RNA failed to reverse the inhibitory effects of PPARγ agonist 15d-PGJ2, suggesting PPARγ-independent mechanisms. In addition, 15d-PGJ2 but not the synthetic high affinity PPARγ ligand rosiglitazone triggered ERK1/2 phosphorylation in IEC, and most importantly, MEK1 inhibitor PD98059 reversed the inhibitory effect of 15dPGJ2 on LPS-induced RelA phosphorylation and interleukin-6 gene expression. Calyculin A, a specific phosphoserine/phospho-threonine phosphatase inhibitor increased the basal phosphorylation of RelA and reversed the inhibitory effect of 15d-PGJ2 on LPS-induced RelA phosphorylation. We further demonstrated in co-immunoprecipitation experiments that 15d-PGJ2 triggered protein phosphatase 2A activity, which directly dephosphorylated RelA in LPS-stimulated CMT-93 cells. We concluded that 15d-PGJ2 may help to control NF-κB signaling and normal intestinal homeostasis to the enteric microflora by modulating RelA phosphorylation in IEC through altered protein phosphatase 2A activity.

The mucosal surfaces and cavities of the gastrointestinal tract in humans and animals are populated by a complex mix-ture of non-pathogenic microorganisms of more than 400 species having spatial differences in population size and relative species predominance along the digestive tract (1)(2)(3)(4). The host has evolved various homeostatic mechanisms to acquire tolerance (hyporesponsiveness) to resident enteric microorganisms, whereas protective cell-mediated and humoral immune responses to enteropathogens are maintained. This complex homeostasis toward the normal enteric microflora is broken under conditions of chronic intestinal inflammation including the chronically relapsing, immune-mediated idiopathic disorders ulcerative colitis and Crohn's disease (5,6). The selective effect of microbial factors in initiating and perpetuating chronic intestinal inflammation is extensively supported in comparative studies using germ-free and gnotobiotic rodent models for experimental colitis (7). For example, reconstitution studies with various non-pathogenic bacteria implicate Bacteroides vulgatus as particularly important to the induction of experimental colitis in monoassociated HLA-B27 transgenic rats (8,9).
Increased NF-B activity has been well documented in intestinal epithelial cell (IEC) 1 and lamina propria cells of inflammatory bowel disease patients with active disease (10 -12), and accordingly, pharmacological NF-B blockade may become potentially important in the treatment of chronic intestinal inflammation (13,14). Indeed, local administration of antisense RelA oligonucleotides abrogated clinical and histological signs of trinitrobenzene sulfonic acid-induced experimental colitis, suggesting a mechanistic role for sustained NF-B activity in the pathogenesis of chronic mucosal inflammation (15). On the other hand, blocking NF-B activity with pharmacological inhibitors during the resolution phase of carrageenaninduced acute inflammation is deleterious to the host (16), suggesting dual functions of activated NF-B including protective and detrimental mechanisms during the course of inflammation.
The induction of the IB/NF-B system and NF-B-dependent gene expression is a complex process that involves the participation of multiple adaptor proteins and kinases acting in a coordinate fashion to give specificity to the cell surface stimuli. We have previously shown that B. vulgatus and lipopoly-saccharide (LPS) signal through the TLR4 cascade to trigger IRAK1 degradation and IB␣ phosphorylation/degradation as well as NF-B DNA binding activity and NF-B transcriptional activity in native and IEC lines (17). In addition to the activation of the IB/NF-B system and nuclear translocation of transcriptionally active RelA, the modification of NF-B transcriptional activity by phosphorylation of RelA at various serine residues (Ser-276, Ser-529, Ser-536) has been shown to be an important regulatory element of this signaling pathway (18 -21). Potential kinases involved in signal-induced RelA phosphorylation are the casein kinase II, Akt, and IKK. We showed that IKK␤ and the phosphatidylinositol 3-kinase/Akt pathway participate in B. vulgatus-induced phosphorylation of serine 529 and/or 536 of the RelA transactivating domain 1 (TAD1) in IEC (17). Most importantly for the physiological relevance, we showed that monoassociation of wild type rats with B. vulgatus triggered transient nuclear localization of phosphorylated (Ser-536) and transcriptionally active NF-B subunit RelA (p65) in the intestinal epithelium (17,22). The absence of colitis and pathological immune responses in B. vulgatus-monoassociated wild type rats confirmed the non-pathogenic nature of this obligate anaerobic Gram-negative bacterial strain and suggests that the normal host developed mechanisms to control NF-B activity in IEC (23).
Peroxisome proliferator-activated receptor ␥ (PPAR␥) is a member of the steroid receptor superfamily with various cellular functions including differentiation, apoptosis, lipid metabolism, and anti-inflammatory responses (33)(34)(35). Although PPAR␥ is expressed in multiple tissues, the highest levels are found in adipose tissue and colonic epithelium (36). Ligandspecific activation of the PPAR␥ transcription factor has been shown to inhibit pro-inflammatory gene expression and experimental colitis for synthetic anti-diabetic thiazolidinediones (TZDs) including rosiglitazone and troglizitazone as well as the endogenous prostaglandin D 2 metabolite 15-deoxy-⌬ 12,14 -prostaglandin J 2 (15d-PGJ 2 ) (37-42). It appears from several studies that TZD and 15d-PGJ 2 , which is present in vivo during the resolution phase of acute inflammation (43), mediate their anti-inflammatory effects also through PPAR␥-independent mechanisms affecting the NF-B signaling pathway at the level of IKK activity and IB␣ degradation as well as RelA DNA binding activity (37,44,45).
In this study we characterized the molecular mechanism for the inhibitory effect of 15d-PGJ 2 on RelA phosphorylation (Ser-536) and IL-6 gene expression in IEC. Consistent with the transient induction of phospho-RelA (Ser-536) in the intestinal epithelium of B. vulgatus-monoassociated Fisher rats, 15d-PGJ 2 but not the synthetic high affinity PPAR␥ ligand rosiglitazone inhibited LPS (B. vulgatus)/LPS (Escherichia coli)-induced RelA phosphorylation and IL-6 gene expression in CMT-93 cells. Although B. vulgatus triggered nuclear expression of PPAR␥ in native epithelium of monoassociated Fisher rats, PPAR␥-specific knock-down in CMT-93 cells using small interference RNA failed to reverse the inhibitory effects of PPAR␥ agonist 15d-PGJ 2 , suggesting PPAR␥-independent mechanisms. Finally, we could demonstrate that 15d-PGJ 2 inhibits LPS-induced RelA phosphorylation and IL-6 gene expression in IEC through the induction of the ERK signaling cascade by modulating PP2A activity.

MATERIALS AND METHODS
Animals and Bacterial Monoassociation-Germ-free Fisher F344 rats were monoassociated at 10 -12 weeks of age with B. vulgatus (a generous gift from Dr. A. B. Onderdonk, Harvard University, Cambridge, MA) and maintained in the Gnotobiotic Animal Core at the College of Veterinary Medicine, North Carolina State University (Raleigh, NC). Bacterial monoassociation and the absence of contamination by other bacterial species were confirmed by culturing samples from the large intestine at necropsy and culturing serial fecal samples. Animal use protocols were approved by the Institutional Animal Care and Use Committee (IACUC), North Carolina State University. Rats were killed 3, 7, 14, and 35 days after initial bacterial colonization. Germ-free mice were used as controls. Sections of the ileum, cecum, proximal, and distal colon were fixed in 10% neutral buffered formalin. The fixed tissue was embedded in paraffin. Histology scoring was analyzed by blindly assessing the degree of lamina propria mononuclear cell infiltration, crypt hyperplasia, goblet cell depletion, and architectural distortion.
Immunohistochemistry and Isolation of Primary Rat Intestinal Epithelial Cells-B. vulgatus-monoassociated and germ-free rats were euthanized, and the cecum as well as colon were removed and placed in Dulbecco's modified Eagle's medium (Invitrogen) containing 5% fetal calf serum. Cecum and colon were cut longitudinally, washed three times in calcium/magnesium-free Hanks' balanced salt solution (Invitrogen), cut into pieces 0.5 cm long, and incubated at 37°C in 40 ml of Dulbecco's modified Eagle's medium containing 5% fetal calf serum and 1 mM dithiothreitol for 30 min in an orbital shaker. The supernatant was filtered and centrifuged for 5 min at 400 ϫ g, and the cell pellet was resuspended in Dulbecco's modified Eagle's medium containing 5% fetal calf serum. The remaining tissue was incubated in 30 ml of phosphatebuffered saline (1ϫ) containing 1.5 mM EDTA for an additional 10 min. The supernatant was filtered and centrifuged for 5 min at 400 ϫ g, and the cell pellet was resuspended in Dulbecco's modified Eagle's medium containing 5% fetal calf serum. Finally, primary IEC were collected by centrifugation through a 25/40% discontinuous Percoll gradient at 600 ϫ g for 30 min. Cell viability and purity was assessed by trypan blue exclusion and fluorescence-activated cell sorter analysis using mouse anti-CD3 monoclonal antibody (BD Biosciences Pharmingen, clone G4.18). Cells were Ͼ80% viable and Ͼ90% pure. Primary rat IEC from cecum and colon were combined and collected in sample buffer for subsequent Western blot analysis. Immunohistochemistry on paraffinembedded tissue sections was performed using anti-PPAR␥ Ab (Cell Signaling, Beverly, MA) according to the protocol of the manufacturer, and sections were counterstained with hematoxylin.
RNA Isolation and Real-time Reverse Transcription-PCR-RNA from IEC was extracted using Trizol reagent (Invitrogen) according to the manufacturer's instructions. Extracted RNA was dissolved in 20 l of water containing 0.1% diethyl pyrocarbonate. For reverse transcription, 1 g of total RNA was added to 30 l of reaction buffer containing 8 l of 5ϫ first-strand buffer, 4 l of dithiothreitol (100 mM), and 6 l of deoxyribonucleoside triphosphate mixture (300 M) (all reagents from Invitrogen) and incubated for 5 min at 65°C. After adding 10 l of a solution containing 0.2 g of random hexamers, 40 units of RNase Out, and 200 units of Moloney murine leukemia virus reverse transcriptase (all reagents from Invitrogen), the total mixture was incubated for an additional 60 min at 37°C followed by a final 1-min heating step at 99°C.
Small Interference RNA and Transfection-Synthetic PPAR␥-specific (accession number NM_011146) siRNA was designed and pur-FIG. 1. RelA phosphorylation and PPAR␥ expression in intestinal epithelium of B. vulgatus-monoassociated germ-free Fisher rats. Germ-free Fisher rats (n ϭ 2) were monoassociated with B. vulgatus. Rats were killed at days 3, 7, 14, and 35 after initial bacterial colonization. Native IEC were isolated from cecum and colon. Total protein was extracted, and 20 g protein were subjected to SDS-PAGE followed by phospho-RelA, PPAR␥, and ␤-actin immunoblotting using ECL technique (A). Sections of cecum and colon were fixed in 10% neutral-buffered formalin. Immunohistochemistry was performed on paraffin-embedded tissue sections using anti-PPAR␥ antibody and analyzed at 20ϫ or 40ϫ magnification (B and C). Sections were counterstained with hematoxylin. chased from Qiagen (Hilden, Germany) according to the protocol of the manufacturer. The sequence was as follows: sense, 5Ј-AGACCCAG-CUCUACAACAG(TT)-3Ј; reverse 5Ј-CUGUUGUAGAGCUGGGUCU-(TT)-3Ј. The annealed double stranded ribooligonucleotides were dissolved in RNase-free buffer and stored at Ϫ20°C in a concentration of 20 M. Before performing the experiments, the siRNA was heat-treated for 1 min at 90°C and then incubated for an additional 60 min at 37°C. CMT-93 cells, which were grown to 80 -90% confluency, were transfected with single-stranded siRNA (0.4, 0.8, or 1.6 g of total siRNA) according to the protocol of the manufacturer using TransMessanger reagent (Qiagen). Finally, the transfected CMT-93 cells were cultured for an additional 48 h and then stimulated with LPS (10 g/ml) for 12 h in the presence or absence of 15d-PGJ 2 . Rhodamine-stained siRNA was used to visualize the cellular distribution in transfected CMT-93 epithelial monolayers using fluorescence microscopy.
Nuclear Extracts and NF-B p65 Protein/DNA Binding Activity-CMT-93 cells were stimulated for various times (0 -4 h) with E. coli LPS (10 g/ml), and nuclear extracts were prepared according to the manufacturer's instructions (Active Motif, Carlsbad, CA). Extracts (5 g) were used to determine nuclear RelA binding activity to the B-nucleotide consensus sequence 5Ј-GGGACTTTCC-3Ј by using the TransAM ELISA-based NF-B transcription factor assay (Active Motif). Protein/ oligonucleotide binding activity was quantified by colorimetric analysis using a MultiScan spectrophotometer.
Co-immunoprecipitation and Phosphatase Activity Assay-CMT-93 cells were stimulated with either E. coli LPS (10 g/ml) or 15d-PGJ 2 (20 M). Cells were lysed in lysis buffer (50 mM Tris at pH 8.0, 2 mM EDTA, 0.1% Nonidet P-40, and 10% glycerol supplemented with protease inhibitors. Cell debris was removed by centrifugation at 12,000 ϫ g, and the supernatant was precleared for 1 h with 20 l of protein A/Gagarose (Santa, Cruz, Europe). Total protein concentration was normalized, and immunoprecipitation was carried out overnight at 4°C using 5 l of rabbit anti-phospho-RelA and goat anti-PP2A antibody (Abcam). Immune complexes were collected with 30 l of protein A/G-agarose for 30 min, washed, and resuspended in reaction buffer (100 l) containing 20 mM HEPES (pH 7.4), 10 mM MgCl 2 , and 1 M dithiothreitol. To assay the phosphatase activity, 30 l of beads from LPS-stimulated cells were co-incubated with 50 l of beads from 15d-PGJ 2 -treated cells. Reactions were stopped after 1 and 2 h of incubation at 37°C with the addition 2ϫ Laemmli sample buffer. Western blot analysis for phospho-RelA and PP2A were carried out as described above.
ELISA Analysis-Protein concentrations were determined in spent culture supernatants of IEC cultures using an ELISA technique. IL-6 protein production was determined by mouse-specific ELISA assay kits according to the manufacturer's instructions (R&D Systems, Heidelberg, Germany).
Statistical Analysis-Data are expressed as the mean Ϯ S.D. of triplicates. Statistical analysis was performed by the two-tailed Student's t test for paired data and considered significant if p values were Ͻ0.05 (*) or Ͻ0.01 (**).

B. vulgatus-monoassociated Fisher Rats Trigger Transient
Phosphorylation of RelA and Persistent Expression of PPAR-␥ in Native IEC-We investigated RelA phosphorylation and PPAR-␥ expression in the intestinal epithelium of B. vulgatusmonoassociated and germ-free Fisher rats. The rats were killed after 3, 7, 14, and 35 days of initial bacterial colonization, and native IEC were isolated from cecal and colonic tissue. Phospho-RelA (Ser-536) and PPAR␥ protein expression were measured in isolated native IEC using Western blot analysis. As shown in Fig. 1, although B. vulgatus monoassociation of germfree rats triggered transient phosphorylation of RelA at day 3 in native IEC, PPAR␥ protein expression was persistently induced 3-28 days after bacterial colonization. Nuclear localization of PPAR␥ in cecal and colonic epithelium was confirmed in intestinal tissue sections by performing immunohistochemical analysis (Fig. 1, B and C).
15d-PGJ 2 but Not Rosiglitazone Inhibits RelA Phosphorylation in CMT-93 Cells after Stimulation with B. vulgatus and E. coli LPS-We have previously shown that whole B. vulgatus cells and E. coli-derived LPS trigger RelA phosphorylation, NF-B activity, and pro-inflammatory gene expression in IEC lines. Based on these results, we next asked the question whether PPAR␥ agonists inhibit RelA phosphorylation in CMT-93 cells. We used the endogenous prostaglandin D 2 metabolite 15d-PGJ 2 and the synthetic PPAR␥ ligand rosiglitazone to antagonize B. vulgatus LPS-and E. coli LPS-induced RelA phosphorylation in CMT-93. Fig. 2 shows that 15d-PGJ 2 inhibits RelA phosphorylation in CMT-93 cells after 1 and 2 h of stimulation with B. vulgatus LPS ( Fig. 2A) as well as E. coli LPS (Fig. 2B). Of note, the synthetic high affinity PPAR␥ ligand rosiglitazone did not inhibit LPS-induced RelA phosphorylation. To further characterize the inhibitory mechanisms of 15d-PGJ 2 on LPS-induced NF-B signaling, we used the commercially available E. coli LPS.
15d-PGJ 2 Failed to Inhibit LPS-induced IB␣ Degradation, RelA Nuclear Translocation, and RelA DNA Binding Activity in IEC-We next investigated the effect of 15d-PGJ 2 and rosiglitazone on LPS-induced IB␣ degradation, RelA nuclear translocation, and RelA DNA binding activity in CMT-93 cells. We stimulated the cells with LPS in the presence or absence of 15d-PGJ 2 and rosiglitazone for 1 and 2 h. Fig. 3A shows that

15d-PGJ 2 Triggers PP2A Activity in Intestinal Epithelial Cells
LPS-induced IB␣ phosphorylation as well as IB␣ degradation in CMT-93 cells after 1 h of stimulation followed by complete IB␣ resynthesis after 2 h of stimulation. Of note, 15d-PGJ 2 and rosiglitazone failed to inhibit LPS-induced IB␣ phosphorylation/degradation after 1 h of stimulation. Interestingly and in contrast to rosiglitazone, 15d-PGJ 2 blocked IB␣ phosphorylation after 2 h of stimulation. IB␣ protein resynthesis was not affected in the presence of any of the two PPAR␥ agonists.
We next studied LPS-induced nuclear translocation and RelA DNA binding activity in the presence or absence of 15d-PGJ 2 and rosiglitazone in CMT-93 cells. As shown in Fig. 3C, whereas LPS-induced RelA DNA binding activity was slightly reduced after 1 h of stimulation, this moderate inhibitory effect was completely abolished after 2 h of stimulation. Rosiglitazone did not affect RelA nuclear translocation or RelA DNA binding activity (Fig. 3B).In summary, these results suggest that 15d-PGJ 2 did not affect LPS-induced IB␣ degradation, IB␣ resynthesis, or RelA nuclear translocation but slightly delayed RelA DNA binding activity in LPS-stimulated IEC.
15d-PGJ 2 Selectively Inhibited IL-6 but Not IB␣ Gene Expression in LPS-stimulated IEC-We next sought to investigate the effect of 15d-PGJ 2 and rosiglitazone on LPS-induced IL-6 and IB␣ gene expression in CMT-93 cells. Therefore, we stimulated the cells for 12 h with LPS in the presence or absence of 15d-PGJ 2 and rosiglitazone. Interestingly, reverse transcription and real-time PCR analysis revealed that 15d-PGJ 2 but not rosiglitazone inhibited IL-6 mRNA expression in LPS-stimulated CMT-93 cells (Fig. 4A). Accordingly, IL-6 protein production was significantly inhibited in the presence of FIG. 3. Effects of 15d-PGJ 2 on LPS-induced IB␣ phosphorylation/degradation, RelA nuclear translocation, and RelA DNA binding activity. CMT-93 cells were stimulated with 10 g/ml LPS from E. coli for 1 and 2 h in the presence or absence of 20 M 15d-PGJ 2 and 20 M rosiglitazone (Rosig). Total protein was extracted, and 20 g of protein were subjected to SDS-PAGE followed by phospho-IB␣, IB␣, and ␤-actin immunoblotting using an ECL technique (A). Nuclear extracts were isolated, and 20 g of nuclear protein were subjected to SDS-PAGE followed by RelA and ␤-actin immunoblotting using ECL technique (B). Nuclear extracts were isolated, and RelA DNA binding activity was quantified using a TransAM ELISA-based NF-B transcription factor assay. Protein/oligonucleotides binding activity was determined by colorimetric analysis using 5 g of nuclear extracts (C). The results are representative three independent experiments. 15d-PGJ 2 Triggers PP2A Activity in Intestinal Epithelial Cells 15d-PGJ 2 but not rosiglitazone (Fig. 4B). Consistent with our previous results, IB␣ mRNA expression was not affected in the presence of 15d-PGJ 2 or rosiglitazone.
Because 15d-PGJ 2 but not the synthetic high affinity PPAR␥ ligand rosiglitazone displayed inhibitory effects on IEC activation, we next transfected CMT-93 cells with PPAR␥-specific siRNA oligonucleotides (0.8 g). Western blot and real-time PCR analysis confirmed PPAR␥-specific knock-down (Ͼ80%) in CMT-93 cells in the presence of siRNA oligonucleotides. Most importantly, functional analysis in the presence PPAR␥-specific siRNA revealed no effect on 15d-PGJ 2 -mediated inhibition of IL-6 gene expression, confirming PPAR␥-independent mechanisms for the inhibitory effects 15d-PGJ 2 in LPS-activated IEC (data not shown).
To further elucidate the specificity of 15d-PGJ 2 -mediated inhibition of RelA phosphorylation, we stimulated CMT-93 cells with LPS in the presence or absence of 15dPGJ 2 and PGE 2 . As shown in Fig. 5, 15d-PGJ 2 but not PGE 2 blocked LPS-induced RelA phosphorylation (Fig. 5A) as well as IL-6 mRNA expression (Fig. 5B).

15d-PGJ 2 -mediated Inhibition of LPS-induced RelA Phosphorylation and IL-6 Gene Expression Is Reversed in the Presence of MEK1 Inhibitor PD98059 -
The mechanisms of 15d-PGJ 2 to inhibit RelA phosphorylation and IL-6 gene expression were independent from the presence of PPAR␥. It has been previously shown that 15d-PGJ 2 triggers ERK1/2 phosphorylation in murine myoblast cell line C2C12 (48). Based on these results, we next investigated the role of the MAP kinase pathway ERK1/2 to mediate the inhibitory effect of 15d-PGJ 2 on LPS-induced IEC activation. Interestingly, 15d-PGJ 2 and LPS but not rosiglitazone (data not shown) induced ERK1/2 phosphorylation in CMT-93 cells after 1 h of stimulation, whereas a weak expression of phospho-ERK1/2 remained after 2 h of stimulation in the presence of 15d-PGJ 2 but not LPS or LPS/ 15d-PGJ 2 (data not shown). We then used the MEK1 inhibitor PD98059 to inhibit ERK1/2 phosphorylation in CMT-93 cells after the stimulation with LPS for 1 h. Fig. 6A shows that the addition of PD98059 inhibited LPS-induced ERK phosphoryl- To further elucidate the role of the ERK signaling pathway on the inhibitory effects of 15d-PGJ 2 , we measured LPS-induced IL-6 gene expression in the presence or absence of 15d-PGJ 2 after 12 h of stimulation. Real-time RT-PCR analysis revealed that the presence of PD98059 completely reversed 15d-PGJ2-mediated inhibition of LPS-induced IL-6 mRNA expression, whereas PD98059 alone did not affect LPS-induced IL-6 gene expression (Fig. 6B). In conclusion, these results suggest that 15d-PGJ 2 inhibits LPS-induced RelA phosphorylation as well as IL-6 gene expression in IEC through induction of the ERK-signaling cascade.
15d-PGJ 2 Triggers Protein Phosphatase 2A Activity, Which Directly Dephosphorylates RelA-Protein serine/threonine phosphatase activity plays an important role in the regulation of transcription factor activity (24,25). To further elucidate the role of protein serine/threonine phosphatase activity in 15d-PGJ 2 -mediated inhibition of RelA phosphorylation and IL-6 gene expression, we used the specific inhibitor calyculin A. Interestingly, the treatment of CMT-93 cells with calyculin A (5 nM) dramatically increased the basal level RelA phosphorylation after 1 h of stimulation (Fig. 7A, lane 2). The additional treatment of CMT-93 cells with LPS did not further increase RelA phosphorylation (Fig. 7A, lane 5). In addition and consistent with our previous results, LPS-induced RelA phosphorylation was inhibited in the presence of 15d-PGJ 2 (Fig. 7A, compare lane 4 and 6), and most importantly, calyculin A completely reversed this inhibitory effect of 15d-PGJ 2 (Fig. 7A,  lane 7). Total PP2A was similar in all samples, suggesting that PP2A expression remained unchanged during the different treatments. As shown in Fig. 7B, LPS-induced IL-6 mRNA expression was significantly increased in the presence of calyculin A. Similar to our previous results, 15d-PGJ 2 inhibited LPS-induced IL-6 gene expression. This inhibitory effect was reversed in the presence of calyculin A, suggesting a role of protein phosphatases for the inhibitory effects of 15d-PGJ 2 on RelA phosphorylation as well as IL-6 gene expression in CMT-93 cells.
To further specify the role of 15d-PGJ 2 in triggering phosphatase activity, we co-immunoprecipitated endogenous phospho-RelA and PP2A after the treatment of CMT-93 cells with FIG. 5. 15d-PGJ 2 but not PGE 2 inhibit LPS-induced RelA phosphorylation and IL-6 gene expression. CMT-93 cells were stimulated with 10 g/ml LPS from E. coli in the presence or absence of 20 M 15d-PGJ 2 and 20 M PGE 2 . A, total protein was extracted after 1 and 2 h of stimulation, and 20 g of protein were subjected to SDS-PAGE followed by phospho-RelA and ␤-actin immunoblotting using ECL technique. B, total RNA was extracted after 12 h of stimulation, reverse-transcribed, and real-time PCR was performed using the Light Cycler system with specific primers for murine IL-6 and GAPDH. The induction of IL-6 mRNA expression was calculated relative to untreated controls (fold increase) using the crossing point of the log-linear portion of the amplification curve after normalization with GAPDH. The bars represent the combined mean values (ϮS.D.) of triplicate samples. *, p value Ͻ0.05.
LPS and 15d-PGJ 2 , respectively. Immunoprecipitated phospho-RelA was then co-incubated with PP2A-loaded A/G-agarose beads for additional 1, 2, and 4 h. As shown in Fig. 7C, PP2A directly dephosphorylated RelA after 1, 2, and 4 h of co-incubation, suggesting that 15d-PGJ 2 triggers PP2A activity in CMT-93 cells. Of note, total RelA and PP2A confirmed equal loading of the samples. In conclusion, we demonstrated that 15d-PGJ 2 triggered PP2A activity, which directly dephosphorylated RelA in LPS-stimulated CMT-93 cells. DISCUSSION In this study, we demonstrate that 15d-PGJ 2 inhibits LPS (B. vulgatus)/LPS (E. coli)-induced RelA phosphorylation and IL-6 gene expression in the colonic epithelial cell line CMT-93 through induction of the MEK/ERK signaling cascade. Consistent with the transient expression of phospho-RelA in native IEC of B. vulgatus-monoassociated Fisher rats, 15d-PGJ 2 triggers PP2A activity, which directly dephosphorylates RelA in LPS-stimulated CMT-93 cells.
It appears from several studies that PPAR␥ agonists including synthetic TZD and 15d-PGJ 2 attenuate colonic inflammation in different models of experimental colitis (41, 42, 49 -51) and antagonize NF-B signaling through covalent modification and inhibition of IKK␤ in PPAR␥-independent mechanisms (37,44,45). Indeed, we show PPAR␥-independent mechanisms for the inhibitory effect of 15d-PGJ 2 in CMT-93 cells using siRNA-mediated knock-down of PPAR␥. In addition, several studies now point toward ligand-independent anti-inflammatory effects of PPAR␥. For example, studies with mice heterozygous for a deficiency of PPAR␥ (PPAR␥ Ϫ/ϩ ) were significantly more susceptible to the development of experimental colitis when compared with wild type mice (42,50,51), and accordingly, PPAR␥ expression in colonic epithelium was substantially reduced in patients with ulcerative colitis (52) as well as dextran sodium sulfate-treated mice (53). Interestingly, Bacteroides thetaiotaomicron triggered PPAR␥-mediated nuclear export of transcriptionally active RelA and directly abolished Salmonella enteriditis-induced inflammatory effects in IEC (54). These results provide compelling evidence that PPAR␥ plays an important role in the regulation of mucosal inflammation. Taken together, two mechanisms are apparent from these studies. First, PPAR␥ inhibits inflammatory processes in a ligand-independent manner, and second, PPAR␥ agonists inhibit NF-B activity as well as experimental colitis through PPAR␥-independent mechanisms. Consistent with previously published results (54), we demonstrate that B. vulgatus monoassociation of germ-free Fisher rats triggered persistent nuclear but not cytoplasmic expression of the transcription factor PPAR␥ in the middle/upper part of the crypts of IEC from large intestinal epithelium. In addition, we show that 15d-PGJ 2 but not synthetic TZD inhibits RelA phosphorylation and IL-6 gene expression, confirming previously published results that demonstrate different effects between 15d-PGJ 2 and synthetic TZD (38,55,56). It remains to be seen whether PPAR␥ expression is directly involved in the negative regulation of B. vulgatus-induced NF-B signal transduction in the intestinal epithelium.
We also demonstrate that 15d-PGJ 2 but not the synthetic high affinity ligand rosiglitazone triggered ERK1/2 phosphorylation in CMT-93 cells and, most importantly, MEK1 inhibitor PD98059 reversed LPS-induced RelA phosphorylation and IL-6 gene expression in CMT-93 cells. Gilroy et al. (43) show in a model for carrageenan-induced acute inflammation that healing and survival in these mice during the resolution phase of inflammation were associated with increased levels of 15d-PGJ 2 and decreased levels of PGE 2 (43). Concordently, we show no inhibitory effects of PGE 2 on LPS-induced RelA phosphorylation and IL-6 gene expression in CMT-93 cells. Furthermore, previous studies demonstrated that 15d-PGJ 2 , which has no known plasma membrane receptor, activates the MEK/ERK signaling cascade through different mechanisms in various cell types including superoxide anion-dependent induction of the Raf/MEK-signaling pathway in myoblasts (48) as well as induction of the phosphatidylinositol 3-kinase pathway in smooth Where indicated the cells were pretreated for 1 h with 50 M PD98059. Total protein was extracted, and 20 g of protein were subjected to SDS-PAGE followed by phospho-ERK1/2, ERK1/2, phospho-RelA, and RelA immunoblotting using an ECL technique (A). CMT-93 cells were stimulated with E. coli LPS (10 g/ml) for 12 h in the presence or absence of 20 M 15d-PGJ 2 . Where indicated the cells were pretreated for 1 h with 50 M PD98059. Total RNA was extracted after 12 h of stimulation and reverse-transcribed, and real-time PCR was performed using the Light Cycler system with specific primers for murine IL-6 and GAPDH. The induction of IL-6 mRNA expression was calculated relative to untreated controls (fold increase) using the crossing point of the log-linear portion of the amplification curve after normalization with GAPDH. The bars represent the combined mean values (ϮS.D.) of triplicate samples. *, p value Ͻ0.05. muscle cells (57). Inhibition of LPS-induced NF-B activity in the presence of 15d-PGJ 2 was demonstrated in native and macrophage cell lines at the level of IKK␤ activity, IB␣ degradation, and NF-B DNA binding activity as well as chemokine expression (55,56,58). Interestingly, we show that 15d-PGJ 2 does not affect LPS-induced IB␣ phosphorylation/ degradation, IB␣ resynthesis, and RelA nuclear translocation in CMT-93 cells. It seems also unlikely that the slight delay in LPS-induced RelA DNA binding activity fully accounts for the complete inhibition of IL-6 gene expression in IEC but, rather, suggests that the 15d-PGJ 2 -mediated inhibition of RelA phosphorylation directly affects IL-6 gene expression.
Phosphorylation and dephosphorylation of transcription factors regulates their DNA binding properties as well as their transactivating potential, and apparently, complexes containing both protein kinases and phosphoprotein phosphatases are important in maintaining the phosphorylation state of intracellular substrates. PP2A, which regulates a diverse set of cellular processes including signal transduction and transcription, constitutively associates with RelA in the cytoplasm (24,25). Calyculin A, which is a serine/threonine phosphatase inhibitor, increased base levels of phospho-RelA in untreated CMT-93 cells, suggesting that the basal phosphatase activity has to be relatively active to maintain a low state of phosphorylated RelA in IEC. Upon stimulation of CMT-93 cells with LPS, the constitutive cycle of phosphorylation/dephosphorylation is disrupted presumably through induction of IKK␤, casein kinase, or phosphatidylinositol 3-kinase/Akt activity, resulting in RelA serine phosphorylation (Ser-536). Most importantly, calyculin A, which has higher inhibitory efficiency for PP2A than for PP1 or PP2B, reversed the inhibitory effects of 15d-PGJ 2 on LPS-induced RelA dephosphorylation and IL-6 gene expression in CMT-93 cells. Based on these results, we then show for the first time that immunoprecipitated PP2A from 15d-PGJ 2 -treated CMT-93 cells directly dephosphorylates endogenous RelA (Ser-536) from LPS-treated cells, suggesting that 15d-PGJ 2 triggered PP2A activity in epithelial cells. It is not clear whether PP2A binds to the RelA serine phosphorylation sites, but studies on the interaction of PP2A with protein kinase C, casein kinase-2␣, and CXCR2 suggest that PP2A binds to other sites than the phosphorylation sites in these proteins (27)(28)(29).
Non-pathogenic enteric bacteria play an important role in initiating and perpetuating chronic intestinal inflammation in the susceptible host. It appears from our gnotobiotic studies in normal animals that enteric non-pathogenic bacteria including Gram-negative B. vulgatus (17,22) and Gram-positive Enterococcus faecalis (59) trigger transient NF-B signaling in the intestinal epithelium in normal rats and mice, respectively. In these studies the physiological importance for the induction of protective TGF-␤/Smad signaling was clearly demonstrated in native IEC (59). Although TGF-␤/Smad signaling blocked nonpathogenic bacteria-induced NF-B transcriptional activity and IL-6 gene expression in IEC (22), the molecular mechanism for this inhibitory effect was identified at the level of altered of the amplification curve after normalization with GAPDH. The bars represent the combined mean values (ϮS.D.) of triplicate samples. *, reduction; #, increase; p valueϽ0.05. B, CMT-93 cells were stimulated for 1 h with either E. coli LPS (10 g/ml) or 20 M 15d-PGJ 2 . We then immunoprecipitated phospho-RelA and PP2A from lysed cells using A/G-agarose beads. To assay the phosphatase activity, 30 l of beads from LPS-stimulated cells were co-incubated with 50 l of beads from 15d-PGJ 2 -treated cells. Reactions were stopped after 1, 2, and 4 h of incubation at 37°C with the addition 2ϫ Laemmli sample buffer. Total protein was extracted from beads and subjected to SDS-PAGE followed by phospho-RelA, RelA, and PP2A immunoblotting using ECL technique (C). Where indicated the cells were pretreated for 1 h with 5 nM calyculin A. Total protein was extracted, and 20 g of protein were subjected to SDS-PAGE followed by phospho-RelA, RelA, PP2A, and ␤-actin immunoblotting using ECL technique (A). CMT-93 cells were stimulated with E. coli LPS (10 g/ml) for 12 h in the presence or absence of 20 M 15d-PGJ 2 . Where indicated the cells were pretreated for 1 h with 1 nM calyculin A. Total RNA was extracted after 12 h of stimulation and reverse-transcribed, and real-time PCR was performed using the Light Cycler system with specific primers for murine IL-6 and GAPDH. The induction of IL-6 mRNA expression was calculated relative to untreated controls (-fold increase) using the crossing point of the log-linear portion histone acetylation/phosphorylation rather than inhibition of RelA phosphorylation. Based on the results of this study, we may hypothesize that 15d-PGJ 2 -mediated ERK1/2 signaling contributes to the inhibition of NF-B-dependent gene expression in IEC through directly acting at the level of RelA phosphorylation by modulating PP2A activity. An attractive hypothesis is that under normal physiological conditions the coordinate induction of various protective mechanisms in IEC including TGF-␤/Smad as well as 15d-PGJ 2 /ERK signaling helps to control NF-B activity after initial bacterial colonization to maintain normal epithelial cell homeostasis in the interplay with commensal bacteria.