A Novel Signaling Pathway Mediates the Inhibition of CCL3/4 Expression by Prostaglandin E2*

In response to pathogen-associated molecular patterns, dendritic cells initiate an innate immune response characterized by expression and release of proinflammatory cytokines and chemokines. The extent of the inflammatory response is limited by various endogenous factors, including lipid mediators such as prostaglandin E2 (PGE2). We described previously the inhibitory effect of PGE2 on the expression and release of the inflammatory chemokines CCL3 and CCL4 from activated dendritic cells. In this study we describe a novel PGE2 signaling pathway that proceeds through EP-2 → cAMP → EPAC → phosphatidylinositol 3-kinase → protein kinase B → GSK-3 and results in increased DNA binding of the CCAAT displacement protein (CDP), a potent mammalian transcriptional repressor. The direct link between CDP and CCL3/4 transcription was established in knock-down experiments using CDP small interference RNA.

CCL3 and CCL4 Measurement by ELISA-Purified DCs (or DC2.4) were cultured in 24-well culture plates at 1 ϫ 10 5 cells per well in a final volume of 1 ml and stimulated with LPS (1 g/ml) in the presence or absence of PGE 2 (1 M). Cell-free supernatants were harvested and subjected to CCL3/4 ELISAs. In some experiments, inhibitors of MAPKs or PKA were added to cultures 1 h before LPS and PGE 2 . For the detection of CCL3, we used 0.5 g/ml purified polyclonal antibody as capture Ab, followed by detection with biotinylated anti-mouse CCL3 Ab (0.2 g/ml). CCL4 was assayed with AG5-2 as capture Ab (4 g/ml) followed by detection with the biotinylated anti-mouse CCL4 Ab (2 g/ml). The sensitivity of ELISAs for CCL3 and CCL4 was 20 pg/ml.
Real Time PCR-The SYBR green-based real time PCR technique was used to detect the expression of chemokines, prostanoid receptors (EPs), and the CDP gene. Total RNA was prepared from BM-DCs or DC2.4 cells, and 1 g of RNA was reverse-transcribed into cDNA in the presence of 200 units of Moloney murine leukemia virus-RT, 40 units of RNasin, 1 g of random primers, 0.5 mM dNTPs, 3 g of bovine serum albumin, and 1 ϫ Moloney murine leukemia virus reaction buffer (Promega, Madison, WI) in a total volume of 30 l at 42°C for 1 h. cDNA was diluted 10 times for real time PCR. The PCR mixture consists of 2 l diluted cDNA, 5 l of SYBR green-containing PCR master mixture (2 ϫ), and 150 M of each primer in a total volume of 10 l. The specific primers for real time PCR were designed by using the Primer Express TM software from Applied Biosystems (Foster City, CA) and were as follows: CCL3 sense 5Ј-ACTGCCTGCTGCTTCTCCTACA-3Ј and antisense 5Ј-AGGAAAATGACACCTGGCTGG-3Ј; CCL4 sense 5Ј-AAACC-TAACCCCGAGCAACA-3Ј and antisense 5Ј-CCATTGGTGCTGAGAAC-CCT-3Ј; EP1 sense 5Ј-TTAACCTGAGCCTAGGGGATG-3Ј and antisense 5Ј-CGCTGGTGATGTGCCATTATC-3Ј; EP2 sense 5Ј-TGCAAG-AGTCGTCAGTGGCT-3Ј and antisense 5Ј-AACAGTGCCAGTGCGAT-GAG-3Ј; EP3 sense 5Ј-CAATCAGATGTCGGTTGAGCA-3Ј and antisense 5Ј-AGATCTGGTTCAGCGAAGCC-3Ј; EP4 sense 5Ј-TTTCTTCG-GTCTGTCGGGTC-3Ј and antisense 5Ј-CGCTTGTCCACGTAGTGGCT-3Ј; CDP sense 5Ј-GGACGTAATGTCCGAAGATGC-3Ј and antisense 5Ј-CTCAGAGGGCGAGGAAATTG-3Ј; and GAPDH sense 5Ј-GGAGCGA-GACCCCACTAACA-3Ј and antisense 5Ј-ACATACTCAGCACCGGCCT-C-3Ј. Real time PCR was performed using the ABI PRISM 7900HT sequence detection system (Applied Biosystems), and the cycling conditions used were 95°C for 15 s and 60°C for 1 min for 40 cycles, followed by a melting point determination that results in a single peak if the amplification is specific. The expression level for each gene was shown by the cycle numbers needed for the cDNA to be amplified to reach a threshold. The cycle numbers were also converted into arbitrary amounts of DNA using a standard curve generated for each pair of primers, and the results are normalized to the housekeeping gene GAPDH mRNA in the same sample.
Nuclear Extracts and Electrophoretic Mobility Shift Assay (EMSA)-Purified CD11cϩ DCs as well as DC2.4 cells were cultured at a concentration of 1 ϫ 10 6 cells/ml and stimulated with LPS (1 g/ml) in the presence or absence of PGE 2 (1 M) for 1 h. Nuclear extracts were prepared using the NucBuster protein extraction kit (Novagen, Inc., Madison, WI) as recommended by the manufacturer. Briefly, cell pellets were resuspended in 150 l of NucBuster extraction reagent 1 for 5 min on ice to release nuclei. The nuclei were harvested by centrifugation (16,000 ϫ g for 5 min at 4°C) and washed with ice-cold 1 ϫ PBS to remove cytoplasmic proteins. The nuclei were resuspended in 50 l of NucBuster extraction reagent 2 for 5 min on ice, and nuclear extracts were separated from cell debris by centrifugation (16,000 ϫ g for 5 min at 4°C). The protein concentration in nuclear extracts was measured using the protein assay (Bio-Rad).
Double-stranded oligonucleotides binding to target transcriptional factors were end-labeled with [␥-32 P]ATP using T4 polynucleotide kinase (Promega). The sequences for the probes used in these experiments were as follows: AP1 5Ј-CTTGATGAGTCAGCCGGAA-3Ј; NF-B 5Ј-AGTTGAGGGGACTTTCCCAGGC-3Ј; CREB 5Ј-AGAGATTGCCTG-ACGTCAGAGAGCTAG-3Ј; CDP 5Ј-ACCCAATGATTATTAGCCAATT-TCTGA-3Ј. The radiolabeled probes (10,000 cpm) were incubated with 10 g of nuclear extracts for 20 min at room temperature in a 10-l total volume of binding solution containing 20 mM HEPES (pH 7.8), 0.1 mM EDTA, 1 mM dithiothreitol, 100 M phenylmethylsulfonyl fluoride, 10% glycerol, 1 g of poly(dI-dC). For CDP supershift assays, 10 g of nuclear extracts were incubated with 1 l of specific CDP Ab (Santa Cruz Biotechnology, Santa Cruz, CA) for 20 min at room temperature, followed by incubation with 1 l of labeled CDP probe for an additional 20 min. The mixtures were separated on 4% native PAGE in 0.5ϫ TBE, and the gels were dried and quantified by using a PhosphorImager (Amersham Biosciences).
Proteins/DNA Assays-Protein/DNA assays were used to screen si-multaneously a large number of transcription factors for DNA binding activity. Nuclear extracts were prepared as described above. The protein/DNA assays were carried out with the TranSignal™ Protein/DNA Array I (Panomics, Redwood City, CA). In brief, 15 g of total nuclear proteins were incubated with biotin-labeled DNA-binding probes (Tran-Signal Probe Mix) to allow the formation of protein-DNA (or TF-DNA) complexes. The protein-DNA complexes were separated from free probes by electrophoresis in agarose gels. The probes present in the complexes were eluted and hybridized to the TranSignal membrane dotted with corresponding nonlabeled probes, and the signals were detected by chemiluminescence. siRNA Transfection-The 21-nucleotide small interfering RNA duplexes with two overhang dT nucleotides at 3Ј-ends were designed according to the method of Elbashir et al. (21) and synthesized by Dharmacon Inc. (Lafayette, CO). The sequence of the sense strand of CDP siRNA is AAGAGACUAAUUGAUGUUCCA. The control siRNA was an irrevelant siRNA with random nucleotides and a similar GC ratio to CDP siRNA (Dharmacon). Transfections were conducted by using the GeneSilencer siRNA Transfection reagent as recommended by the manufacturer (Gene Therapy System, San Diego). Briefly, DC2.4 cells (2 ϫ 10 5 cells) were plated in 6-well plates overnight and changed into 1 ml of serum-free RPMI 1640 medium before transfection. Five l of 40 M CDP siRNA or control siRNA duplexes were incubated with 5 l of siRNA transfection reagent for 5 min at room temperature, and these mixtures were added to the serum-free DC2.4 cells. Mock controls were transfected with 5 l of siRNA transfection reagent alone. Four hours later, 1 ml of RPMI 1640 containing 20% FBS was added to each well. RNA was prepared from DCs 24 -72 h after transfection and subjected to real time PCR for CDP expression. In some experiments, DC2.4 cells were transfected with siRNA for 72 h and then treated with LPS and LPS plus PGE 2 followed by analysis of chemokine expression.
Western Blots-Purified CD11cϩ DCs were cultured at a concentration of 1 ϫ 10 6 cells/ml, stimulated with LPS (1 g/ml) in the presence or absence of PGE 2 (1 M), and harvested at the designed time points. Total cell lysates were prepared by using cell lysis buffer (New England Biolabs, Beverly, MA), and equal amounts of total protein were separated on 12% SDS-PAGE. After transfer to nitrocellulose membranes (Bio-Rad), the membranes were probed with specific primary antibodies, followed by incubation with appropriate secondary Abs conjugated to horseradish peroxidase. Peroxidase activity was detected by using enhanced chemiluminescence reagents (Pierce) as recommended by the manufacturer.
Measurement of Intracellular cAMP-Intracellular cAMP production following PGE 2 challenge was measured by using the cAMP Biotrak EIA System as recommended by manufacturer (Amersham Biosciences). Purified DC or DC2.4 cells (1 ϫ 10 5 cells/200 l) were cultured in the presence of the cAMP phosphodiesterase inhibitor, 3-isobutyl-1methylxanthine (Sigma). Cells were treated with or without PGE 2 (1 M) or LPS (1 g/ml) for the designed time points (as indicated in text). The reaction was stopped by brief centrifugation, followed by addition of 200 l of lysis buffer. 100 l of cell lysates were used to determine the intracellular cAMP contents by an enzyme immunoassay method. The amount of cAMP was determined by comparison with standard curves obtained with predetermined concentrations of cAMP. The assay system has a sensitivity of 12.5 fmol of cAMP/100 l.
PKB Assay by FACS-To determine the amount of total and phosphorylated intracellular PKB, we used intracellular flow cytometry. Briefly, the cells were fixed and permeabilized using Cytofix/Cytoperm kit (Pharmingen) according to the manufacturer's instructions. Cells were stained with rabbit polyclonal anti-phospho-PKB (Cell Signaling Technology) or mouse monoclonal anti-PKB (Cell Signaling Technology) antibodies for 45 min, followed by the appropriate FITC-conjugated secondary antibodies for 30 min. Control rabbit and mouse IgG were used to confirm specificity. After extensive washing, cells were analyzed by FACS analysis.
GSK-3 Kinase Assay-Purified BM-DCs were treated with 1 M PGE 2 or with different GSK-3 inhibitors (10 mM LiCl, 20 M SB216763, or 20 M GSK-3 Inhibitor II) for 1 h followed by total cell lysate preparation. Untreated cells were used as control. The GSK-3 kinase activity was assayed by using a kit from Upstate Biotechnology, Inc. (Lake Placid, NY), as recommended by the manufacturer. In brief, 10 g of total protein was incubated for 30 min at 30°C with 1 l of [␥-32 P]ATP (10 Ci/l) and 62.5 M GSM (GSK-3 substrate peptide) in a 40-l total volume containing 2 mM MOPS (pH 7.2), 0.05 mM EDTA, 2.5 mM magnesium acetate, 15 mM MgCl 2 , and 100 M unlabeled ATP. 25 l of the reaction mixture was spotted onto P81 phosphocellulose paper followed by extensive washing with 0.75% phosphoric acid. Then the assay papers were transferred into scintillation mixture, and we measured the radioactivity of each sample in a scintillation counter.
Statistical Analysis-Results are expressed as the mean Ϯ S.E. Analysis of data was performed using the Student's t test with p Ͻ 0.05 as the minimum significant level.
Screening for Transcription Factors Affected by PGE 2 -Because PGE 2 inhibits CCL3/4 expression at the mRNA level, we investigated the transcription factors (TF) involved in this event. We used first a DNA/protein array to screen for TFs that might be positively or negatively regulated by PGE 2 . The DNA/ protein arrays were performed with nuclear extracts from BM-DCs stimulated with LPS or LPS plus PGE 2. Although LPS induces strong DNA binding activity for AP1, AP2, CREB, NF-B, GATA, Egr, Smad 3/4, and USF-1, PGE 2 does not affect these TFs (Table I). PGE 2 decreased the DNA binding activity of NFAT, Sp1, SRE, TFIID, and Pbx1 and increased the DNA binding activity of CDP/CBF, Est-1/PEA3, and P53 (Table I).
Previous reports indicated that AP1, NF-B, and CREB/ATF are crucial for CCL3/4 transcription. In agreement with our DNA/protein array data, PGE 2 does not affect the DNA binding activity of AP1, NF-B, and CREB/ATF, as determined by EMSA (Fig. 2, A-C). PGE 2 Induces CDP Binding-Since with the exception of members of the CCAAT displacement protein family (CDP), none of the factors whose binding is affected by PGE 2 have been reported to be involved in CCL3/4 expression, we focused on CDP. The DNA/protein array indicated that PGE 2 increases the DNA binding of CDP, a CCAAT displacement protein previously reported to act as a transcriptional repressor. EMSA experiments with nuclear extracts from both BM-DCs and DC2.4 cells show a significant increase in CDP binding in cells treated with PGE 2 , both in the presence and absence of LPS (Fig. 2D). PGE 2 concentrations from 10 Ϫ6 to 10 Ϫ8 M increased CDP binding to a similar degree, and even PGE 2 10 Ϫ9 M induced CDP binding above control levels (Fig. 2E). The CDP binding is specific, as shown by successful competition with a cold specific probe and by supershift with a specific anti-CDP antibody (Fig. 2F).
Knock-down of CDP Reverses the Inhibitory Effect of PGE 2 on CCL3/4 Production-To address the question whether CDP acts as a CCL3/4 repressor, we knocked down CDP expression with siRNA. DC2.4 cells were transfected with CDP siRNA, an siRNA control, or transfection reagent alone (Mock). RNA was prepared 48 h later, and the expression levels of CDP were determined by real time RT-PCR. Compared with untreated cells, the cells transfected with mock or siRNA control express comparable levels of CDP, whereas cells transfected with CDP siRNA exhibit significantly lower levels of CDP (Fig. 3A). The CDP levels were low even 72 h after transfection (data not shown). Next, DC2.4 cells were transfected with CDP siRNA or control for 72 h, followed by treatment with LPS in the presence or absence of PGE 2 . The expression of CCL3/4 was determined 3 h later by real time RT-PCR. In nontransfected or mock-transfected cells, PGE 2 inhibits LPS-induced CCL3/4 expression. This inhibition is partially reversed in cells transfected with CDP siRNA (Fig. 3B). The reversal was also observed at protein level in cells transfected with CDP siRNA (Fig. 3C). Taken together, these results indicate that CDP mediates at least partially the inhibitory effect of PGE 2 on LPS-induced CCL3/4 expression.

FIG. 1. PGE 2 inhibits LPS-induced CCL3/4 production in BM-DCs and DC2.4 cells.
A, purified CD11cϩ BM-DCs or DC2.4 cells were cultured at a concentration of 1 ϫ 10 5 cells/ml and stimulated with LPS (1 g/ml) in the presence or absence of PGE 2 (1 M). Supernatants were harvested 12 h later and subjected to ELISA for CCL3 and CCL4. * indicates statistically significant differences compared with the LPS group (p Ͻ 0.05). B, BM-DCs or DC2.4 cells (1 ϫ 10 6 cells/ml) were treated with LPS (1 g/ml) in the presence or absence of PGE 2 (1 M) for 3 h. Total RNA was prepared and amplified by real time PCR with primers specific for CCL3 and CCL4. The expression levels of CCL3 and CCL4 were normalized to internal GAPDH message. One representative experiment out of three is shown.
cAMP Mediates the Inhibitory Effect of PGE 2 -Previously, we showed that the inhibitory effect of PGE 2 on CCL3/4 production in BM-DCs is mediated through EP-2 and possibly EP-4 receptors. Real time RT-PCR experiments indicate similar patterns for EP-2 and EP-4 expression in BM-DCs and DC2.4 cells (Fig. 4A). To assess the contribution of the EP receptors to the increase in CDP binding observed with PGE 2 , we used three EP receptor agonists, i.e. butaprost, an EP-2specific agonist, misoprostol, a preferential EP-3/EP-2 agonist that also binds EP-4 at high concentrations, and sulprostone, an EP-1/EP-3 agonist. Because butaprost has a 10-fold lower affinity for EP-2 than PGE 2 , we used butaprost concentrations corresponding from 10 Ϫ6 to 10 Ϫ8 M PGE 2 . The two higher concentrations of butaprost induced CDP binding, which indicates the involvement of the EP-2 receptor (Fig. 4B, lower  panel). Misoprostol induced strong CDP binding at high con-centrations, where it acts as an EP-3/EP-2 receptor (Fig. 4B,  upper panel). The possible involvement of the EP-3 receptor in the induction of CDP binding was eliminated by the fact that sulprostone, an EP-1/EP-3 agonist, did not induce CDP binding, even at 10 Ϫ6 M (Fig. 4B, lower panel). These results indicate that the effects of PGE 2 on CDP binding are mediated through the EP-2 receptor.
Because EP-2 receptors are coupled to G␣ resulting in the activation of adenylate cyclase, we investigated whether PGE 2 induces cAMP in BM-DCs and DC2.4 cells. PGE 2 induced a rapid increase in cAMP as early as 5 min following stimulation, reaching a maximum around 10 min, followed by a gradual decrease (results not shown). The effect on cAMP levels is dose-dependent, with significant increases for 10 Ϫ6 -10 Ϫ7 M PGE 2 (Fig. 4C). PGE 2 induced comparable amounts of cAMP in the presence and absence of LPS (Fig. 4C). The involvement of  2 on the DNA binding activity of various TFs BM-DCs were cultured at a concentration of 1 ϫ 10 6 cells/ml and stimulated with LPS (1 g/ml) in the presence or absence of PGE 2 (1 M) for 1 h. Nuclear extracts were prepared and subjected to protein/DNA array as described under "Experimental Procedures." 1 indicates an increase in DNA binding (higher than 3-fold) by PGE 2 ; 2 indicates a decrease in DNA binding (higher than 3-fold) by PGE 2 ; 7 indicates the TFs whose DNA binding activity was not significant affected by PGE 2 . cAMP as a mediator in the inhibition of CCL3/4 by PGE 2 is supported by the inhibitory effect of Bt 2 cAMP, a stable cAMP analog (Fig. 4D).
The CCL3/4 Inhibition by PGE 2 Is Not Mediated by the PKA/CREB Pathway-Because the classical downstream cAMP target is PKA, we pretreated BM-DCs and DC2.4 cells with the PKA inhibitor H89 (10 Ϫ6 -10 Ϫ9 M), followed by LPS and PGE 2 . No reversal of the PGE 2 -induced inhibition was observed (Fig. 5A), suggesting that PKA does not participate in the PGE 2 signaling pathway. To further eliminate the possible involvement of PKA, we examined the phosphorylation of CREB (pCREB) by Western blotting. LPS induced strong CREB phosphorylation, whereas PGE 2 did not induce pCREB and did not affect LPS-induced CREB phosphorylation (Fig.  5B). These results are consistent with the fact that PGE 2 does not affect CREB DNA binding in the EMSA experiments ( Fig. 2A). PGE 2 Does Not Affect the TLR Common Core Response-LPS induces proinflammatory cytokines and chemokines in DCs through the TLR common core response, mediated through the activation of MAPKs and NF-B. PGE 2 did not affect LPSinduced NF-B binding as determined by both DNA/protein arrays and EMSA (Table I and Fig. 2C). To assess the involvement of the ERK1/2 and p38 MAPK pathways in the LPSinduced CCL3/4 expression, we treated BM-DCs and DC2.4 cells with LPS in the presence and absence of MAPK inhibitors. PD98059, an ERK1/2 specific inhibitor, reduced CCL3/4 production in a dose-dependent manner (Fig. 6A, and data not shown), whereas the p38 MAPK inhibitor SB 203580 did not affect chemokine production (data not shown).
To determine whether the inhibitory effect of PGE 2 is mediated through the inhibition of ERK1/2, we treated BM-DCs with LPS in the presence or absence of PGE 2 , followed by Western blotting for phosphorylated ERK1/2. In contrast to PD98059, PGE 2 did not affect LPS-induced ERK phosphorylation (Fig. 6B). The effect of PGE 2 on the phosphorylation of JNK and p38 was also analyzed by Western blots. Similar to ERK1/2, PGE 2 did not affect the LPS-induced phosphorylation of JNK or p38 (Fig. 6C).
Activation of EPAC Mediates the Inhibitory Effect of PGE 2 on CCL3/4 -Our previous results showed that the inhibition of chemokines by PGE 2 was mediated in a cAMP-dependent but PKA-independent manner. Recently, other cAMP-dependent pathways have been reported, including cAMP-gated ion channels and EPAC. To determine whether EPAC is involved in the inhibition of chemokine production, BM-DCs were treated with LPS in the presence or absence of different concentrations of the EPAC-specific activator, 8-CPT-2Ј-OMe-cAMP. Similar to Bt 2 cAMP and PGE 2 , 8-CPT-2Ј-OMe-cAMP inhibits chemokine production at both the protein and mRNA level (Fig. 7,  A and B).
Next, we investigated if activation of EPAC by 8-CPT-2Ј-OMe-cAMP induces CDP binding. BM-DCs were treated with PGE 2 , Bt 2 cAMP, or 8-CPT-2Ј-OMe-cAMP and CDP binding was analyzed by EMSA. PGE 2 , Bt 2 cAMP, and 8-CPT-2Ј-OMe-cAMP induced comparable levels of CDP binding (Fig. 7C). These results suggest that activation of EPAC mediates the inhibitory effect of PGE 2 . PGE 2 Induces PKB Phosphorylation-Next, we investigated the downstream events regulated by EPAC. There is evidence that activation of EPAC by cAMP signals through the PI3K-PKB pathway. Because the PI3K inhibitors wortmannin and LY294002 were cytotoxic for DC except at low concentrations that proved to be inefficient in terms of PI3K inhibition, we examined whether PGE 2 activates PKB in untreated or LPStreated BM-DCs. BM-DCs were treated with PGE 2 in the presence or absence of LPS. and the levels of phosphorylated PKB were determined by flow cytometry. We did not detect phosphorylated PKB in cells cultured with medium or stimulated with LPS. However, phosphorylated PKB was present in the majority (Ͼ94%) of the medium-or LPS-stimulated BM-DCs treated with PGE 2 (Fig. 8).
Involvement of GSK-3 in Chemokine Production-Because PGE 2 activates PKB, which in turn phosphorylates and inactivates GSK-3, we considered the possibility that GSK-3 might be involved in CCL3/4 expression. BM-DCs were treated with LPS in the presence of different concentrations of the GSK-3 inhibitors SB216763 and GSK-3 inhibitor II. Both inhibitors decrease LPS-induced CCL3/CCL4 production in a dose-dependent manner (Fig. 9). The inhibitory effect on CCL3/4 release is paralleled by a decrease in CCL3/4 transcription (data not shown). Similar to SB216763 and GSK-3 inhibitor II, a third GSK-3 inhibitor, LiCl, inhibits chemokine production in a dose-dependent manner at both the protein and mRNA level (Fig. 10, A and B).
We propose that upon LPS stimulation, GSK-3 phosphoryl-ates CDP, which results in reduced DNA binding and loss of repressor activity. In contrast, when DCs are treated with PGE 2 , the activation of PKB results in GSK-3 phosphorylation and inactivation. Subsequently, nonphosphorylated CDP binds to the CCL3/4 promoters and represses transcription. The link between GSK-3 activity and CDP DNA binding was confirmed by the fact that nuclear extracts from LiCl-treated DCs showed increased CDP binding activity (Fig. 10C).
In addition, we compared GSK-3 activity following treatment with PGE 2 , LiCl, SB216763, or GSK-3 inhibitor II. BM-DCs were treated with PGE 2 or GSK-3 inhibitors for 1 h, followed by kinase assays with GSM as substrate. PGE 2 and LiCl decreased GSK-3 activity by 25%, and the two specific GSK-3 inhibitors resulted in 40% inhibition. This indicates that PGE 2 acts as a GSK-3 inhibitor, similar to LiCl, SB216763, and the GSK-3 inhibitor II. DISCUSSION Activated DC are major producers of proinflammatory chemokines, such as CCL3 and CCL4. Prostaglandins of the E series inhibit the expression of CCL3 and CCL4 in LPS-stimulated macrophages and LPS or PGN-stimulated DC (20,22). Although we have shown previously that PGE 2 affects CCL3/4 expression through the EP-2 and possibly EP-4 receptors (20), there is no information on the intracellular signaling. In this study we identified the CCAAT displacement protein CDP as the major player in the effect of PGE 2 on CCL3/4 gene expression. In addition, we established that the PGE 2 -induced DNA binding of the CDP repressor is mediated through the EP-2 3 cAMP 3 EPAC 3 PKB 3 GSK-3 pathway. A, BM-DCs were treated with different concentrations of PD98059 for 1 h followed by stimulation with LPS (1 g/ml) for another 12 h. Supernatants were subjected to CCL3 and CCL4 ELISA. * indicates statistically significant differences compared with the LPS-treated group (p Ͻ 0.01). One representative experiment out of three is shown. B, BM-DCs were treated with LPS (1 g/ml) in the presence of 1 M PGE 2 (upper panel) or different concentrations of PD98059 (lower panel) for 20 min. The phosphorylated ERK1/2 were visualized by Western blotting. One experiment out of two is shown. C, BM-DCs were treated with LPS Ϯ PGE 2 for 20 min, and total cell lysates were subjected to Western blotting. The blots were probed with Abs for p-p38, p-JNK, and total p38. One experiment out of two is shown. M, medium; L, LPS; P, PGE 2 expression. However, PGE 2 increases the binding of the CCAAT displacement protein CDP, a negative regulator that represses the transcription of a variety of genes, including the pro-inflammatory chemokine CXCL1 (28). The increase in CDP binding is apparent for PGE 2 concentrations that are within the range observed in inflammatory conditions (10 Ϫ7 -10 Ϫ8 M).
The CDP proteins exert their repressor activity by competing with transcriptional activators for occupancy of DNA-binding sites and by direct recruitment of deacetylases (reviewed in Ref. 29). The DNA binding capacity of CDP is modulated through phosphorylation and/or proteolytic processing. In fibroblasts, the CDP DNA binding is increased during the G 1 3 S transition through dephosphorylation and proteolytic cleavage and decreased during the S 3 G 2 transition through phosphorylation (30).
The direct involvement of CDP in CCL3/4 expression was demonstrated by the reversal of CCL3/4 inhibition following CDP siRNA transfection of DC2.4 cells, a dendritic cell line developed by Shen et al. (31). We used the DC2.4 cell line because primary BM-DC are quite difficult to transfect. The DC2.4 cells have been reported to up-regulate major histocompatibility complex class II and costimulatory molecules following LPS/interferon-␥ stimulation and to secrete IL-12p40 (32). However, there is no information regarding their capacity to produce chemokines or to respond to PGE 2 . Our results show that the DC2.4 cells behave similar to BM-DC in terms of EP-2 and EP-4 receptor expression, induction of cAMP by PGE 2 , and effects of PGE 2 on CCL3/4 expression. In addition, PGE 2 treat- FIG. 9. Active GSK-3 kinase is required for the induction of CCL3/4 by LPS. BM-DCs were treated with different concentrations of SB216763 (A) or GSK-3 inhibitor II (B) for 1 h, followed by stimulation with LPS for an additional 12 h. The levels of CCL3 and CCL4 were determined by ELISA. * indicates statistically significant differences compared with the LPS-treated group (p Ͻ 0.01). One representative experiment out of three is shown.

FIG. 10. LiCl inhibits CCL3/4 induction by LPS and increases CDP DNA binding.
A, BM-DCs were treated with LPS in the presence of different concentrations of LiCl for 12 h, and the levels of CCL3 and CCL4 released in the supernatants were determined by ELISA. B, BM-DCs were treated with LPS together with different concentrations of LiCl for 3 h, followed by RNA preparation. The expression of CCL3 and CCL4 was measured by real time PCR. C, BM-DCs were treated with LiCl for 1 h, and nuclear extracts were prepared. EMSA was carried out with labeled CDP probe. One representative experiment out of three (for A and B) and one out of two (for C) are shown. ment of either BM-DC or DC2.4 results in increased CDP binding. Therefore, the results obtained with siRNA in the DC2.4 cells are representative for BM-DC as well.
Because the PGE 2 -induced regulation of CDP binding turned out to be essential for CCL3/4 expression, we proceeded to identify the intracellular mediators involved in signaling from the EP receptors to CDP. Agonist studies indicate the involvement of the EP-2 receptors in the effect of PGE 2 on CDP binding. The EP-2 receptors activate adenylate cyclase, resulting in an increase in intracellular cAMP (reviewed in Ref. 33). The involvement of cAMP in the inhibition of CCL3/4 expression was confirmed by the fact that the stable cAMP analog Bt 2 cAMP reduces CCL3/4 expression and increases CDP binding similar to PGE 2 . One of the classical cAMP effects is the subsequent activation of PKA, followed by phosphorylation of CREB. However, PKA is not involved in the PGE 2 -induced inhibition of CCL3/4 because H89, a specific PKA inhibitor, does not reverse the inhibition. In addition, both Western blots and EMSA assays indicate that PGE 2 does not affect CREB phosphorylation or DNA binding.
LPS treatment of BM-DC results in the activation of all three MAPK pathways, but PGE 2 does not affect the phosphorylation of ERK1/2, JNK, or p38. This indicates that signaling through the EP-2/EP-4 receptors does not interfere with the classical MAPK pathways.
Recently, a second cAMP-activated signaling pathway has been identified. Several of the growth-promoting activities of cAMP are associated with the EPAC-mediated activation of PI3K 3 PKB (34,35). Indeed, 8-(4-chlorophenylthio)-2Ј-Omethyl-cAMP, a specific EPAC activator (36), inhibits CCL3/4 expression and promotes CDP binding, similar to PGE 2 and Bt 2 cAMP. Because of the cytotoxicity of the PI3K inhibitors wortmannin and LY294002, we could not directly ascertain the involvement of PI3K. However, we were able to show sustained intracellular PKB phosphorylation in DC treated with PGE 2 , in the presence or absence of LPS.
One of the well characterized PKB substrates is GSK-3, a kinase that is inactivated upon phosphorylation at serine residues (reviewed in Ref. 37). A recent study established the connection between PGE 2 signaling and the phosphorylation of PKB and GSK-3 in cells transfected with EP-2/EP-4 receptors (38). We propose the model presented in Fig. 11 for the PGE 2 inhibition of LPS-induced CCL3/4 expression. In LPS-stimulated DCs, active GSK-3 phosphorylates CDP, which is then unable to bind to DNA and repress CCL3/4 transcription. This is supported by the fact that GSK-3 inhibitors prevent CCL3/4 induction by LPS and increase CDP binding. In contrast, PGE 2 signaling through the EP-2/EP-4 receptors activates the adenylate cyclase, followed by the activation of the EPAC 3 PI3K 3 PKB pathway. PKB phosphorylates and inactivates GSK-3. As a result, nonphosphorylated CDP binds to the CCL3/4 promoter and prevents transcription of the chemokine genes.