Differential Regulation of Chemokine Expression by Peroxisome Proliferator-activated Receptor γ Agonists

Chemokine-mediated inflammatory cell infiltration is a hallmark of asthma. We recently demonstrated that glucocorticoids and β2-agonists additively or synergistically suppress tumor necrosis factor-α (TNFα)-induced production of chemokines eotaxin and interleukin-8 (IL-8), respectively, in human airway smooth muscle (HASM) cells, which may partly explain their combined benefits in asthma. Peroxisome proliferator-activated receptors (PPARs) also modulate inflammatory gene expression. We reported here that the PPARγ agonists 15-deoxy-Δ12,14-PGJ2 (15d-PGJ2) and troglitazone, but not PPARα agonist WY-14643, inhibited TNFα-induced production of eotaxin and monocyte chemotactic protein-1 (MCP-1) but not IL-8. Eotaxin inhibition was transcriptional and additively enhanced by the glucocorticoid fluticasone and the β2-agonist salmeterol, whereas MCP-1 inhibition was post-transcriptional and additively and synergistically enhanced by fluticasone and salmeterol, respectively. Coimmunoprecipitation revealed that 15d-PGJ2 induced a protein-protein interaction between PPARγ and the glucocorticoid receptor (GR) in TNFα-treated HASM cells, which was enhanced by fluticasone and salmeterol. 15d-PGJ2, fluticasone, and salmeterol all inhibited TNFα-induced histone H4 acetylation at the eotaxin promoter and NF-κB p65 binding to the eotaxin promoter and induced PPARγ and GR association with the eotaxin promoter, as analyzed by chromatin immunoprecipitation assay. Our data suggest that chemokine expression in HASM cells is differentially regulated by PPARγ agonists and that the interaction between PPARγ and GR may be responsible for the additive and synergistic inhibition of chemokine expression by PPARγ agonists, glucocorticoids, and β2-agonists, particularly the chromatin-dependent suppression of eotaxin gene transcription. The interaction may have wide applications and may provide a potential target for pharmacological and molecular intervention.

Chemokine-mediated inflammatory cell infiltration is a hallmark of asthma. We recently demonstrated that glucocorticoids and ␤ 2 -agonists additively or synergistically suppress tumor necrosis factor-␣ (TNF␣)-induced production of chemokines eotaxin and interleukin-8 (IL-8), respectively, in human airway smooth muscle (HASM) cells, which may partly explain their combined benefits in asthma. Peroxisome proliferator-activated receptors (PPARs) also modulate inflammatory gene expression. We reported here that the PPAR␥ agonists 15-deoxy-⌬ 12,14 -PGJ 2 (15d-PGJ 2 ) and troglitazone, but not PPAR␣ agonist WY-14643, inhibited TNF␣-induced production of eotaxin and monocyte chemotactic protein-1 (MCP-1) but not IL-8. Eotaxin inhibition was transcriptional and additively enhanced by the glucocorticoid fluticasone and the ␤ 2 -agonist salmeterol, whereas MCP-1 inhibition was post-transcriptional and additively and synergistically enhanced by fluticasone and salmeterol, respectively. Coimmunoprecipitation revealed that 15d-PGJ 2 induced a protein-protein interaction between PPAR␥ and the glucocorticoid receptor (GR) in TNF␣-treated HASM cells, which was enhanced by fluticasone and salmeterol. 15d-PGJ 2 , fluticasone, and salmeterol all inhibited TNF␣-induced histone H4 acetylation at the eotaxin promoter and NF-B p65 binding to the eotaxin promoter and induced PPAR␥ and GR association with the eotaxin promoter, as analyzed by chromatin immunoprecipitation assay. Our data suggest that chemokine expression in HASM cells is differentially regulated by PPAR␥ agonists and that the interaction between PPAR␥ and GR may be responsible for the additive and synergistic inhibition of chemokine expression by PPAR␥ agonists, glucocorticoids, and ␤ 2 -agonists, particularly the chromatin-dependent suppression of eotaxin gene transcription. The interaction may have wide applications and may provide a potential target for pharmacological and molecular intervention.
Asthma is a common chronic disease featured by airway inflammation. Inflammatory cells are key players in the in-flammatory process of asthma and are attracted to the airways by a network of chemokines (1). Mast cells and eosinophils are the most important inflammatory cells in asthmatic airways (2), and neutrophil infiltration is associated with asthma exacerbations (3). Eosinophils are attracted by chemokines such as eotaxin, IL-5, 1 and MCP-1 (monocyte chemotactic protein-1), and neutrophils are recruited by chemokines such as IL-8 (4). Eotaxin and IL-8 are also mast cell chemotactic factors (5), and MCP-1 also attracts T-lymphocytes and monocytes (4). We and others have demonstrated that human airway smooth muscle (HASM) cells produce a number of important chemokines such as IL-8 (6,7), eotaxin (8), MCP-1 (9), RANTES, and GM-CSF (10) and is an important source of chemokines in the airways, consequently contributing to the orchestration and perpetuation of airway inflammation. Thus, understanding the molecular mechanisms of chemokine gene expression and regulation in HASM cells will have important implications for the regulation of airway inflammation in asthma.
Peroxisome proliferator-activated receptors (PPARs) belong to a subfamily of the nuclear receptor superfamily and are potentially important transcription factors that modulate inflammatory responses (11). PPARs are comprised of three isoforms, ␣, ␤(␦), and ␥, and are activated by a heterogeneous group of structurally dissimilar endogenous and synthetic agonists. Of these PPAR agonists, the prostaglandin D 2 metabolite 15-deoxy-⌬ 12,14 PGJ 2 (15d-PGJ 2 ) is a direct-binding natural ligand for PPAR␥ (12). It has been documented that PPAR␥ agonists may play an important role in the regulation of inflammatory process and may become a new class of anti-inflammatory compounds (11,13,14). Studies have shown that regulation of chemokine expression by PPAR␥ activation is cell-, stimulus-, and gene-specific. For instance, 15d-PGJ 2 induces IL-8 expression, reduces MCP-1 expression, and has no effect on RANTES expression in human monocytes/macrophages (15), whereas it reduces both IL-8 and MCP-1 gene expression in colon epithelial cells (14). We have shown that PPAR␣ and -␥ are constitutively expressed in HASM cells and that activation of PPAR␥, but not PPAR␣, induces the expression of cyclooxygenase-2 (16). PPAR␥ activation also inhibits GM-CSF and G-CSF production, whereas glucocorticoid dexamethasone inhibits only GM-CSF in HASM cells (17). However, the effects of PPAR agonists on other chemokine expression in HASM cells have not been investigated so far.
Glucocorticoid receptors (GR), like PPAR, are ligand-activated nuclear receptors, and glucocorticoids (GCs) are commonly applied in the treatment of inflammatory diseases. ␤ 2 -Agonists promote bronchodilation by increasing intracellular cAMP and stimulating cAMP-dependent protein kinases. We have recently reported that ␤ 2 -agonists synergistically and additively enhance the inhibitory effect of GCs on TNF␣-induced IL-8 and eotaxin production, respectively (7,8), from HASM cells, which may partially explain their combined benefits in asthma therapy (18). However, many patients have persistent symptoms despite these treatments. Approaches seeking new therapeutic targets for asthma treatment are therefore required.
The study of crystal structures has revealed that nuclear receptors share highly conserved ligand-binding domain, denominated activation function-2, which is necessary for transcriptional activation (19). Because steroids estrogen, progesterone, and GCs interact with each other (20) and GCs also interact with ␤ 2 -agonists on chemokine production (7,8), we hypothesized that PPAR␥ agonists, 15d-PGJ 2 in particular, could interact with GCs and ␤ 2 -agonists on chemokine expression.
In unstimulated cells, DNA is packaged into a highly organized and dynamic protein-DNA complex known as chromatin. The fundamental subunit of chromatin is the nucleosome, which consists of 146 bp of DNA wrapped twice around an octomer core of four histones (two molecules each of histones H2A, H2B, H3, and H4) (21). This nucleosomic structure prevents the access of transcription factors and RNA polymerase II to their respective recognition sequences and the initiation of transcription (22). When the cells are stimulated with inflammatory mediators, histones undergo an array of post-transcriptional modifications on N-terminal domains, including acetylation, phosphorylation, and methylation (23), which result in chromatin remodeling and transcription factors binding to specific gene promoters, leading to the initiation of gene transcription (24). The local hyper-acetylation of histone H4 at a specific gene promoter plays a key role in the regulation of gene transcription (24,25). GR activation has been shown to suppress inflammatory gene transcription by inhibiting histone acetylation associated with specific gene promoters (24,26); whether PPAR␥ has a similar effect is not yet known. The purpose of this study was to investigate the regulation of TNF␣-induced eotaxin, MCP-1, and IL-8 expression by PPAR agonists, either alone or in combination with GCs and ␤ 2 -agonists and to explore the underlying molecular mechanisms.
Cell Culture-Primary HASM cells were prepared from explants of HASM as reported previously (27) and cultured in Dulbecco's modified Eagle's medium containing 10% fetal calf serum, penicillin (100 units/ ml), streptomycin (100 g/ml), amphotericin B (2.5 g/ml), and L-glutamine (4 mM) in humidified 5% CO 2 , 95% air at 37°C. This protocol was approved by the Nottingham City Hospital Research Ethics Committee. Cells at passage 6 were used for all experiments. We have shown previously that the cells grown in this manner depict the immunohistochemical and morphological features of typical HASM cells (27). HASM cells were grown to confluence in 24-well plates and growtharrested in serum-free medium for 24 h prior to experiments in fresh serum-free medium unless otherwise stated. Because we have reported previously that HASM cells produce ample amounts of eotaxin (8) and IL-8 (7) after 8 h of stimulation with TNF␣ (10 ng/ml), this time point was chosen to examine the drug effects. In the concentration-response or drug combination experiments, the cells were incubated with increasing concentrations or a single concentration of drug(s) or the drug vehicle Me 2 SO (maximum concentration 0.4%) for 0.5 h and then stimulated with TNF␣ (10 ng/ml) for 8 h. The supernatants were collected and stored at Ϫ20°C until ELISA for chemokines, whereas in some experiments total RNA was extracted from the cells and stored at Ϫ20°C until further detection of mRNA by RT-PCR.
Chemokine Assays-The concentrations of eotaxin, MCP-1, and IL-8 in the culture media were measured by ELISA according to the manufacturer's instructions as reported previously (7,8). The sensitivity of the ELISA kits at our hands was at least 5 pg/ml, which was consistent with the manufacturer's specifications. According to the kit insert, there is no significant cross-reactivity or interference with other human cytokines and chemokines.
RNA Stability-Confluent and growth-arrested HASM cells were incubated with or without TNF␣ for 2 h and then treated with the drugs or drug vehicle Me 2 SO for 2 h before adding Act D (5 g/ml) for the times indicated to block new transcript generation. Total RNA was then extracted from the cells, and eotaxin mRNA level was analyzed by RT-PCR using the full-length eotaxin coding region primers described above.
Plasmids and Transient Transfection-The firefly luciferase reporter gene constructs in pGL3-basic vectors containing the full-length eotaxin promoter fragment (Ϫ1363) were obtained from Professor Robert Schleimer (Johns Hopkins University) and were described previously (28). IL-8 promoter fragment (Ϫ1481 to ϩ44) was digested with XhoI and HindIII from the IL-8 PUX-CAT (chloramphenicol acetyltransferase) reporter plasmids (a gift from Professor Kouji Matsushima, Kanazawa University, Kanazawa, Japan, see Ref. 29) and was then ligated into the XhoI-HindIII site of the pGL3-basic plasmid upstream to the translation initiation site of the firefly luciferase gene. The resulting construct was verified by sequencing. Because deletion analysis has shown that the transcription of MCP-1 is manipulated by two 5Ј-flanking regions, the distal enhancer region (ENH) and the proximal promoter region (PRM) (30), we obtained the 167-bp human MCP-1 PRM between Ϫ107 and ϩ60 by digesting with XhoI and HindIII of the pGLM-PRM (a gift from Professor Atsuhisa Ueda, Yokohama City University, Yokohama, Japan, see Ref. 31) and the 230-bp human MCP-1 ENH between Ϫ2742 and Ϫ2513 by digesting with KpnI and XhoI of the pGLM-ENH (also from Professor Atsuhisa Ueda, see Ref. 31). The juxtaposed MCP-1 ENH fragment (Ϫ2742/Ϫ2513) and PRM fragment (Ϫ107/ϩ60) were then ligated into the KpnI-HindIII site of the pGL3-basic plasmid to generate firefly luciferase reporter gene construct pGL3-MCP-1-ENH-PRM. The pCMX vectors expressing wildtype PPAR␥ were the generous gift from Professor Mitchell Lazar (University of Pennsylvania) and were described previously (32).
Transient transfection was performed as described previously (16, 33) with minor modifications. Briefly, HASM cells in 24-well plates were grown to 60 -70% confluence and then transfected with 0.4 g/well of the reporter gene plasmids for 16 h. Transfected cells were pretreated with or without drugs for 0.5 h and then incubated with or without TNF␣ (10 ng/ml) for 6 h. To assess the roles of PPAR␥ agonists, cells were cotransfected with increasing concentrations of pCMX-PPAR␥ together with the reporter gene plasmids for 16 h, and transfected cells were pretreated with or without 15d-PGJ 2 (1 M) for 0.5 h before incubation with or without TNF␣ (10 ng/ml) for 6 h. Firefly luciferase activity was determined as described before (16,33).
ChIP Assay-ChIP assay was performed as described previously (24,33). Briefly, confluent and serum-deprived HASM cells in 90-mm dishes were pretreated with or without drugs for 0.5 h and then incubated with or without TNF␣ (10 ng/ml) for 2.5 h. The cells were then incubated with 1% formaldehyde to fix protein-DNA complexes. Chromatin pellets from these cells sheared by sonication were pre-cleared with salmon sperm DNA-saturated protein A-and G-Sepharose. One portion of the soluble chromatin was used as DNA input control, and the remains were sub-aliquoted and then precipitated using specific antibodies against acetylated histone H4 or transcription factors NF-B p65, PPAR␥, or GR␣ (the most abundant GR isoform). The purified DNA from the immunoprecipitated complexes of antibody-protein-DNA was detected by semi-quantitative PCR (28 cycles) using the following specific primer pairs spanning promoter regions that contain major regulatory elements such as NF-B: for eotaxin promoter, forward, 5Ј-CTTCATGTTGGAGGCTGAAG-3Ј, and reverse, 5Ј-GGATCTGGAA-TCTGGTCAGC-3Ј; for IL-8 promoter, forward, 5Ј-TTCACCAAATTGT-GGAGCTT-3Ј, and reserve, 5Ј-GAAGCTTGTGTGCTCTGCTG-3Ј.
Coimmunoprecipitation and Western Blotting-Confluent and serum-starved HASM cells were pretreated with or without the drugs for 0.5 h and then incubated with TNF␣ (10 ng/ml) for 1 h. Nuclear proteins were extracted from the cells, and protein concentrations were measured as described previously (16,33). Immunoprecipitation (IP) was conducted as described previously (26,34) with minor modifications. Briefly, the nuclear extracts were precleared with 20 l of protein A/G-agarose. After microcentrifugation, the supernatants were incubated with 5 g/ml anti-PPAR␥, anti-GR␣ rabbit polyclonal antibody, or normal rabbit IgG overnight at 4°C followed by incubation with protein A/G-agarose beads for 3 h. The immune complexes were washed with PBS containing 0.02% Tween 20 and pelleted by gentle centrifugation, and the presence of GR␣ in the resulting immunoprecipitates (IPs) was analyzed by Western blotting (19,38).
Statistical Analysis-Data were expressed as means Ϯ S.E. Statistical analysis was performed with Graphpad Prism 4. Analysis of variance (ANOVA) and unpaired two-tailed student t test were used to determine the significant differences between the means. p Յ 0.05 was considered significant.
Effects of 15d-PGJ 2 and Flut Combination on TNF␣-induced Chemokine Production-As shown in Fig. 2A, 15d-PGJ 2 alone markedly inhibited TNF␣-induced eotaxin production in a concentration-dependent manner (p ϭ 0.0001, ANOVA). Flut alone at 0.001 and 0.01 M also inhibited the production. The concentration-dependent inhibition by 15d-PGJ 2 (0.1-10 M) was significantly enhanced by the addition of Flut at 0.001 and 0.01 M. 15d-PGJ 2 partially inhibited TNF␣-induced MCP-1 production at high concentrations (5 and 10 M, Fig.  2E, p ϭ 0.0304 and p ϭ 0.0029, respectively). Flut alone at 0.001 and 0.01 M also reduced MCP-1 production. An additive inhibition was observed when Flut was used in combination with the highest concentration of 15d-PGJ 2 tested (10 M). 15d-PGJ 2 had no effect on TNF␣-induced IL-8 production (Fig. 2C). Although Flut (0.01 and 0.1 M) significantly inhibited the production, the inhibition was not altered when it was used together with 15d-PGJ 2 (0.1-10 M). The results show that Flut augments the inhibitory effect of 15d-PGJ 2 on eotaxin and MCP-1 production and suggest that that activated PPAR␥ and GR could interact with each other, resulting in additive inhibition on TNF␣-induced chemokine production in HASM cells.
Effects of 15d-PGJ 2 and Salme Combination on TNF␣induced Chemokine Production-To investigate if PPAR␥ agonists, like GCs, could interact with ␤ 2 -agonists to regulate cytokine-induced chemokine production, we compared the effects of 15d-PGJ 2 alone with the effects of 15d-PGJ 2 ϩ Salme on TNF␣-induced chemokine production. As shown in Fig. 2B, TNF␣-induced eotaxin production was markedly inhibited by 15d-PGJ 2 in a concentration-dependent manner (p ϭ 0.0001, ANOVA). Salme at 0.01 M, but not 0.001 M, also significantly inhibited eotaxin production. The inhibition by 15d-PGJ 2 was further enhanced with the addition of Salme at both 0.001 and 0.01 M. TNF␣-induced MCP-1 production was also concentration-dependently inhibited by 15d-PGJ 2 (Fig. 2F); although Salme alone (0.001 and 0.01 M) had no effect, it significantly enhanced the inhibition by 15d-PGJ 2 when they were used in combination. 15d-PGJ 2 and Salme, either alone or in combination, had no effect on TNF␣induced IL-8 production, and even the concentrations of Salme were 10 times higher than those used in eotaxin and MCP-1 production (Fig. 2D). The data strongly suggest that interactions between activated PPAR␥ and signal pathways activated by ␤ 2 -agonists result in additive inhibition on TNF␣-induced eotaxin production and synergistic inhibition on TNF␣-induced MCP-1 production in HASM cells.
Effects of PPAR␥ Agonists, Flut and Salme, on TNF␣-induced Chemokine mRNA Expression-To explore the molecular mechanisms by which PPAR␥ agonists GCs and ␤ 2 -agonists differentially regulate TNF␣-induced production of eotaxin, MCP-1, and IL-8, we then examined if the effects of these drugs on chemokine protein production were due to the alteration of the chemokine mRNA expression. As analyzed by RT-PCR, the level of eotaxin mRNA in control cells was low (Fig. 3A, lanes 1 and 12) and was markedly induced by TNF␣ treatment for 8 h (lanes 2 and 13). 15d-PGJ 2 inhibited the mRNA expression in a concentration-dependent manner (Fig. 3A, lanes 5 and 6 and 14 -16), and TRO also reduced the mRNA expression (lanes 7 and 8) as compared with TNF␣ alone (lanes 2 and 13). In contrast, WY-14643 had no effect (Fig. 3A, lane 9). Both Salme (Fig. 3A, lanes  3 and 4) and Flut (lanes 10 and 11) alone inhibited TNF␣-induced eotaxin mRNA expression. An additive inhibition was observed when Salme and Flut were used in combination with increasing concentrations of 15d-PGJ 2 (Fig. 3A, lanes 17-22 and 23-28, respectively) as compared with 15d-PGJ 2 alone (lanes 14 -16).
The results were consistent with the effects of these drugs on eotaxin protein production (Figs. 1, A and B, and 2, A and B), strongly suggesting that PPAR␥ agonists, GCs, and ␤ 2 -agonists when used together, additively and transcriptionally inhibit TNF␣-induced eotaxin gene expression.

2-6)
; 15d-PGJ 2 (lanes 7-10) and Flut (lanes 11-14) did not alter the course, but Salme (lanes 15-18) inhibited the degradation. A time-dependent natural degradation of TNF␣-induced MCP-1 mRNA was also observed after the addition of Act D (Fig. 4D, lanes 2-6), but treatment with the tested drugs did not affect the degradation (lanes 7-18). Collectively, results from Figs. 3 and 4 strongly suggest that the inhibition of TNF␣-induced eotaxin production by PPAR␥ agonists, GCs, and ␤ 2 -agonists is transcriptional and that the inhibition of TNF␣induced IL-8 production by GCs and the inhibition of TNF␣induced MCP-1 production by PPAR␥ agonists and GCs are via post-transcriptional regulations other than mRNA stability.
Effects of PPAR␥ Agonists, Flut, and Salme on TNF␣-induced Eotaxin Promoter Activity-We then focused on the ef- fects of these drugs on the transcriptional regulation of the eotaxin gene, and we performed reporter gene assay to assess whether these drugs alter the eotaxin gene promoter activity. The full-length eotaxin promoter (Ϫ1363 bp) luciferase reporter plasmids were transiently transfected into HASM cells. TNF␣ markedly induced the eotaxin promoter activity by 6-fold, which was markedly inhibited by both 15d-PGJ 2 and TRO, but not WY-14643, in a concentration-dependent manner (Fig. 5A). Flut and Salme alone also significantly inhibited TNF␣-induced eotaxin promoter activity (Fig. 5B). The combined use of low concentration 15d-PGJ 2 (1 M) with Flut or Salme all further augmented the inhibition of the eotaxin promoter activity by the individual drugs alone (Fig. 5B), indicating an additive effect. These results suggest that these drugs suppress eotaxin gene transcription by inhibiting the eotaxin promoter activity and that interactions between these drugs occur upstream of the eotaxin promoter activation.
Because it has been reported that PPAR␥ agonists may exert their effects through PPAR␥-independent mechanisms (35), to clarify the role of PPAR␥ in the inhibition of TNF␣-induced eotaxin transcription by 15d-PGJ 2 and TRO, we cotransfected HASM cells with the eotaxin promoter reporter plasmids and vectors that express human PPAR␥. As shown in Fig. 5C, compared with control (0 g/ml PPAR␥), overexpression of PPAR␥ alone markedly and concentration-dependently inhibited TNF␣induced eotaxin promoter activity (p ϭ 0.0001, ANOVA). The PPAR␥ agonist 15d-PGJ 2 (1 M) alone inhibited TNF␣-induced eotaxin promoter activity and further enhanced the inhibition by PPAR␥ overexpression up to 0.025 g/ml. In contrast, neither 15d-PGJ 2 , TRO, nor WY-14643 had any significant inhibition on TNF␣-induced transcriptional activities of the IL-8 promoter (Fig. 6A) and the MCP-1 enhancer-promoter (Fig. 6B). Taken together, these data demonstrate that 15d-PGJ 2 and TRO do not have any effect on TNF␣-induced IL-8 and MCP-1 transcription, and their inhibition on TNF␣-induced eotaxin expression is transcriptional and PPAR␥-dependent.
Effects of 15d-PGJ 2 , Flut, and Salme on TNF␣-induced Histone H4 Acetylation and p65 Binding with the Eotaxin Promoter-Chromatin remodeling following histone acetylation at specific gene promoter sites is a major transcriptional regulatory mechanism that allows transcription factors to bind to specific gene promoters, initiating gene transcription. As studies have shown that GCs inhibit histone H4 acetylation (24,26) and we have demonstrated that TNF␣-induced eotaxin transcription is NF-B-dependent in HASM cells, 2 we then applied ChIP assay to assess directly whether PPAR␥ agonists, GCs, and ␤ 2 -agonists affect TNF␣-induced histone H4 acetylation at 2 M. Nie, A. J. Knox, and L. Pang, unpublished data. the eotaxin promoter and NF-B p65 in vivo binding with the eotaxin promoter in a chromatin context. As shown in Fig. 7A, after TNF␣ treatment, IPs with antibody against acetylated histone H4 revealed a marked enrichment of the eotaxin pro-moter DNA (197 bp containing STAT6 and NF-B-binding sites, lane 2) compared with the control (lane 1), indicating that histone H4 was acetylated specifically at the eotaxin promoter site. Pretreatment of the cells with 15d-PGJ 2 (Fig. 7A, lanes 3  and 4), Flut (lanes 5-7), and Salme (lanes 8 -10) markedly inhibited histone H4 acetylation in a concentration-dependent manner. Similarly, p65 IPs also showed a marked enrichment of the eotaxin promoter DNA after TNF␣ treatment (Fig. 7A,  lane 2) compared with the control (lane 1), indicating p65 binding to the eotaxin promoter. The binding was also inhibited by 15d-PGJ 2 (Fig. 7A, lanes 3 and 4), Flut (lanes 5-7), and Salme (lanes 8 -10), suggesting that the changes in histone H4 acetylation are correlated with those of p65 binding to the eotaxin promoter.
Taken together, these results are consistent with the findings on protein production, mRNA expression, and promoter activity of eotaxin and IL-8 and directly demonstrate that PPAR␥ agonists and ␤ 2 -agonists, like GCs, suppress TNF␣-induced eotaxin gene transcription in a chromatin-dependent manner through inhibition of histone H4 acetylation and in vivo NF-B p65 binding to its promoter. The results also suggest that these drugs have no effect on TNF␣-induced IL-8 gene transcription.
Effects of 15d-PGJ 2 , Flut, and Salme on the Physical Interaction between PPAR␥ and GR and Its Association with the Eotaxin Promoter-We then conducted coimmunoprecipitation (IP) and ChIP assays to explore whether activated PPAR␥ can direct interact with GR and whether PPAR␥ and GR are associated with the eotaxin promoter. As shown in Fig. 8, in control cell (treated with TNF␣ only) nuclear samples, GR␣ was detected only with GR␣ IPs (Fig. 8, lane 1) and not with PPAR␥ (lane 2) or normal rabbit IgG (lane 3) IPs. In cells treated with 15d-PGJ 2 (5 M) in addition to TNF␣, GR␣ was detected not only with GR␣ IPs (Fig. 8, lane 4) but also with PPAR␥ IPs (lane 5), compared with normal rabbit IgG IPs (lane 6). In cells treated with both 15d-PGJ 2 and Flut (0.01 M), more GR␣ in GR␣ IPs (Fig. 8, lane 7) and PPAR␥ IPs (lane 8) were detected compared with control cells (lanes 1 and 2) and cells treated with 15d-PGJ 2 only (lane 4 and 5). Treatment with 15d-PGJ 2 and Salme (0.01 M, Fig. 8, lanes 10 and 11) produced similar results as treatment with 15d-PGJ 2 and Flut (lanes 7 and 8). These data provide direct evidence that PPAR␥ activation with 15d-PGJ 2 results in physical interaction between PPAR␥ and GR␣ and that GR activation with Flut and signaling pathways activated by ␤ 2 -agonists enhance the interaction. It is also of note that ␤ 2 -agonists, together with PPAR␥ agonists, stimulate GR␣ nuclear translocation (activation).
Furthermore, by ChIP assay, PPAR␥ IPs revealed a marked enrichment of the eotaxin promoter DNA in cells treated with 15d-PGJ 2 (Fig. 9, lanes 2 and 3), Flut (lanes 4 and 5), and Salme (lanes 6 and 7) compared with cells treated with TNF␣ alone (lane 1). GR␣ IPs showed a similar enrichment of the eotaxin promoter DNA in cells treated with these drugs (Fig. 9,  lanes 2-7) compared with cells treated with TNF␣ alone (lane 1). These results suggest that the protein-protein interaction between PPAR␥ and GR␣ induced by PPAR␥ agonists alone or in combination with GCs and ␤ 2 -agonists is associated with the eotaxin promoter, which may lead to the inhibition of histone H4 acetylation and p65 binding to the eotaxin promoter, resulting in the suppression of eotaxin gene transcription.

DISCUSSION
This study is the first to demonstrate that TNF␣-induced chemokine production in HASM cells is differentially regulated by PPAR␥ agonists and that GCs and ␤ 2 -agonists interact with PPAR␥ agonists in this process. PPAR␥ agonists inhibit TNF␣induced eotaxin production transcriptionally, which is additively enhanced by GCs and ␤ 2 -agonists; PPAR␥ agonists inhibit TNF␣-induced MCP-1 production post-transcriptionally, which is additively enhanced by GCs but synergistically enhanced by ␤ 2 -agonists; PPAR␥ agonists have no effect on IL-8 production. The novel finding is that activated PPAR␥ physically interacts with GR and that GR activation and the signaling pathways activated by ␤ 2 -agonists potentiate the interaction, which may provide an explanation for the observed additive and synergistic inhibition on chemokine production by these drugs. The interaction between PPAR␥ and GR␣ may be associated with the eotaxin promoter, resulting in the inhibition of histone H4 acetylation and p65 binding to the eotaxin promoter and the suppression of eotaxin gene transcription. These results are consistent with the view that PPAR␥ can act FIG. 9. Effects of 15d-PGJ 2 , Flut, or Salme on the association of PPAR␥ and GR with the eotaxin promoter. Confluent and serumdeprived HASM cells were pretreated with or without the drugs at the indicated concentrations for 30 min and then incubated with or without TNF␣ for 2.5 h. ChIP assay was conducted as described under "Experimental Procedures" with specific antibodies to PPAR␥ and GR␣. The results are representative of two independent experiments with similar outcomes. in combination with GR to inhibit inflammatory responses (36).
PPARs have complex regulatory effects on inflammatory responses. Studies so far have shown cell-and stimulus-specific effects of PPAR␥ agonists on MCP-1 and IL-8 production (14,15), but their effects on eotaxin expression have not been explored. We have demonstrated previously that PPAR␥ is expressed in HASM cells, and PPAR␥ agonists up-regulate cyclooxygenase-2 expression (16). The present study is the first to demonstrate that PPAR␥ agonists differentially regulate TNF␣-induced chemokine production in HASM cells, providing further evidence for the complex regulatory effects of PPAR␥ on chemokine production. Although PPAR␣ is also expressed in HASM cells (16), no effect of the PPAR␥ agonist WY-14643 on TNF␣-induced chemokine production in these cells has been detected in the present study, which is consistent with previous findings that WY-14643 has no effect on cyclooxygenase-2 expression in these cells (16) and on IL-1␤-induced IL-8 and MCP-1 expression in colonic epithelial cells (14).
More recent evidence has shown that PPAR␥ agonists also promote their biological effects through PPAR␥-independent mechanisms. For example, 15d-PGJ 2 induces IL-8 mRNA and protein through a mitogen-activated protein kinase and NF-B pathway rather than PPAR␥ activation in human T-cells as synthetic PPAR␥ agonists do not mimic the effect of 15d-PGJ 2 (37). Imaizumi et al. (38) demonstrated that 15d-PGJ 2 inhibits lipopolysaccharide-induced GM-CSF expression through a mechanism unrelated to PPAR␥. In this study, however, the effects of the synthetic PPAR␥ agonist TRO are consistent with those of the natural PPAR␥ agonist 15d-PGJ 2 , and overexpression of human PPAR␥ mimics the effect of 15d-PGJ 2 on the inhibition of TNF␣-induced eotaxin promoter activity, and the addition of 15d-PGJ 2 further augments the inhibition. These data are also consistent with the observation that PPAR␥ overexpression inhibits NF-B-mediated gene expression, and the inhibition is further enhanced by the PPAR␥ agonist TRO (39). Our results therefore indicate that the inhibitory effects of 15d-PGJ 2 and TRO on TNF␣-induced eotaxin expression are PPAR␥-dependent.
PPAR␥ and GR are both nuclear receptors that share highly conserved ligand-binding domain activation function-2, essential for transcriptional regulatory effects (19). Activated nuclear receptors induce transcriptional activity by recruiting a variety of coactivators. It has been shown that PPAR␥ and GR may interact with each other by sharing the same coactivators such as the steroid receptor coactivator-1 (SRC-1) (40, 41) and cAMP-response element-binding protein-binding protein (42). It has been observed that estrogen, progesterone, and GCs could interfere with the functions of each other (20) and that PPAR␥ activation enhances GC-mediated transcription and GCs in turn modulate PPAR␥-mediated gene expression in osteoblasts, suggesting complex interactions between PPAR␥ and GR signaling pathways (43). Our current study is the first to provide concrete evidence that PPAR␥ and GR physically interact with each other to regulate chemokine expression in any cell system. ␤ 2 -Agonists are mainly used as bronchodilators in asthma therapy. However, we have demonstrated recently that they synergistically and additively enhance the inhibition by GCs on TNF␣-induced IL-8 (8) and eotaxin production (7), respectively, in HASM cells in a cAMP-dependent manner. Increasing evidence has shown that GCs can interact with ␤ 2 -agonists through the cross-talk between GR and the cAMP signaling pathways activated by ␤ 2 -agonists. For instance, GR can upregulate ␤ 2 -receptor expression (44), whereas ␤ 2 -agonists can cause GR nuclear translocation (activation) even in the absence of GCs in vascular smooth muscle cells (45) and HASM cells (46). The finding in the present study that Salme enhances the effect of 15d-PGJ 2 on increasing GR nuclear presence is consistent with the previous finding that ␤ 2 -agonists can cause GR activation in HASM cells (46). Because of the structural and functional similarities between PPAR␥ and GR, we proposed that ␤ 2 -agonists could also interact with PPAR␥ agonists to regulate chemokine production. Indeed, we have found in the current study that the ␤ 2 -agonist Salme additively enhances the inhibition of TNF␣-induced eotaxin production by 15d-PGJ 2 , in a similar way as it enhances the inhibition of eotaxin production by GCs (8), and that Salme enhances 15d-PGJ 2induced physical interaction between PPAR␥ and GR. We have also found that Salme synergistically enhances the inhibition of TNF␣-induced MCP-1 production by 15d-PGJ 2 when it has no effect on its own. We have shown in the current study that PPAR␥ activation induces physical interaction between PPAR␥ and GR even in the absence of GCs, and that this interaction is further enhanced by GR activation with GCs as well as the signaling pathways activated by ␤ 2 -agonists. This may have important clinical implications as airway inflammation is a main feature of asthma.
Formation of a permissive chromatin environment by hyperacetylation of histone is a prerequisite for gene trans-activation, whereas hypoacetylation is correlated with reduced transcription or gene silencing (21,24,25,47). Targeted acetylation of histone H4 plays an important role in allowing regulatory proteins to access DNA and is likely to be a major factor in the regulation of gene transcription (24,25). It has been demonstrated that activated transcription factors such as NF-B p65 form complexes with cAMP-response element-binding proteinbinding protein, which has intrinsic histone acetyltransferase (HAT) activity (26,48,49), and induce histone acetylation of relevant lysine residues, resulting in local unwinding of DNA, increased transcription factor binding to the promoter, and gene transcription (24,26). GCs have been shown to repress p65-activated HAT activity and consequently inflammatory gene expression (26), but whether PPAR␥ agonists have similar effects has not been known. NF-B is a major transcription factor involved in the regulation of many genes and is implicated in the pathogenesis of a large number of diseases, particularly inflammatory diseases such as asthma and arthritis (50). We have recently identified that TNF␣-induced eotaxin expression in HASM cells is NF-B-dependent. 2 p65 is also involved in TNF␣-induced IL-8 expression (51). In this study, we have demonstrated that TNF␣ induces histone H4 acetylation and p65 binding to the eotaxin and IL-8 promoters and that the effects on eotaxin, but not those on IL-8, are suppressed by 15d-PGJ 2 , Flut, and Salme. We have also found that these drugs stimulated the association of both PPAR␥ and GR␣ with the eotaxin promoter, which may be explained by the physical interactions between PPAR␥ and GR␣ induced by these drugs. To our best knowledge, this is the first demonstration that shows that PPAR␥ and GR␣ are associated with the eotaxin promoter even though there is no peroxisome proliferator-response element and glucocorticoid-response element within the region of the eotaxin promoter we detected in the study. This association could result in the inhibition of TNF␣induced histone H4 acetylation and p65 binding to the eotaxin promoter. As NF-B p65 is likely to be responsible for histone H4 acetylation and the transcription of the eotaxin gene, it is reasonable to speculate that PPAR␥ and GR␣ may directly affect p65 transactivation as there is evidence that the PPAR␥ agonist ciglitazone increases the physical interaction of PPAR␥ with p65, leading to the inhibition of NF-B in colon cancer cells (52). However, we have not found any physical interaction of PPAR␥ and GR␣ with p65 in our current study (data not shown). It is likely that PPAR␥, like GR (26), suppresses eotaxin gene expression by direct inhibition of p65-HAT activity and/or by recruiting histone deacetylase to the p65-HAT complex. But why PPAR␥ agonists and GCs have no effects on TNF␣-induced HAT activity in the IL-8 promoter site is unknown, and the detailed mechanisms, particularly how PPAR␥ and GR␣ physically interact with each other and associate with the eotaxin promoter, remain to be elucidated.
A recent study (53) has demonstrated that PPAR␥ activation reduces antigen-induced lung inflammation, eosinophilia as well as cytokine production, in a murine model of human asthma. It has also demonstrated that PPAR␥ is expressed in eosinophils, and its activation inhibits in vitro chemotaxis of eosinophils (53). Because airway inflammation, particularly eosinophilia, is the main feature of human asthma, these data together with ours strongly suggest that PPAR␥ agonists can act as modulators of asthma and lung inflammation targeting both regulatory and effector cells. Our findings also suggest that the combined use of PPAR␥ agonists with GCs or ␤ 2agonists enhances the overall anti-inflammatory effects of these agents, which may allow the doses of individual drug to be reduced, hence minimizing their side effects. In addition, the interaction between PPAR␥ and GR and its effect on chromatin remodeling in the regulation of inflammatory gene expression may offer novel opportunities for new therapies in asthma and other inflammatory diseases.