Peroxisome Proliferator-activated Receptor γ Interacts with CIITA·RFX5 Complex to Repress Type I Collagen Gene Expression*

Recent reports demonstrate that peroxisome proliferator-activated receptor γ (PPARγ), a member of the nuclear receptor superfamily, acts as a repressor of type I collagen synthesis. Our data demonstrate that exogenously expressed PPARγ down-regulates collagen expression in a dose-responsive manner in human lung fibroblast cells. Silencing PPARγ using lentiviruses expressing short hairpin RNAs partially reverses interferon-γ (IFN-γ)-induced repression and activates collagen mRNA levels. Previous studies indicate that IFN-γ represses collagen gene expression and induces major histocompatibility complex II (MHC II) expression by activating the formation of a regulatory factor for X-box 5 (RFX5) complex with class II transactivator (CIITA). This report demonstrates that PPARγ is within the RFX5·CIITA complex as judged by co-immunoprecipitation and DNA affinity precipitation studies. Most importantly, occupancy of PPARγ on the collagen transcription start site and MHC II promoter increases with IFN-γ treatment. The PPARγ agonist, troglitazone, sensitizes the cells to IFN-γ treatment by increasing recruitment of PPARγ to collagen gene while repressing collagen expression, and these effects are blocked by the PPARγ antagonist T0070907. PPARγ may mediate IFN-γ-stimulated collagen transcription down-regulation and MHC II up-regulation by interacting with CIITA as well as regulating CIITA expression. Therefore, PPARγ is a critical target for investigations into therapeutics of diseases involving extracellular matrix remodeling and the immune response.

The process of wound healing has been traditionally divided into several stages, including hemostasis, inflammation, cellular proliferation, and remodeling, when collagen is synthesized and fibers are deposited (1). These stages proceed by a series of overlapping, yet coordinated and highly regulated events highlighted by the interplay of cells, extracellular matrix, and an array of soluble mediators that are necessary to regenerate and remodel tissue. The extent of injury and the subsequent repair depend on the balanced control of matrix remodeling and inflammation. When this process is not regulated properly, there is either delayed healing with destruction of tissues or a fibroproliferative process with overproduction of collagen resulting in scarring, fibrosis, or cirrhosis depending on the organ.
Peroxisome proliferator-activated receptor-gamma (PPAR␥) 2 is a member of the nuclear hormone receptor superfamily that regulates the expression of genes involved in a variety of biological processes, including lipid metabolism, insulin sensitivity, and inflammation (2)(3)(4). Ligands such as fatty acids, eiconosanoids and oxidized lipids, have been identified that bind to PPAR␥ and alter its transcriptional activity. The thiazolidinediones class of drugs used in diabetes and dyslipidemias are synthetic ligands that activate PPAR␥. Synthetic antagonists of PPAR␥ are also available to inhibit PPAR␥ activity (5)(6)(7).
Activation of PPAR␥ by agonists attenuates fibrosis in several organs, including kidneys (8,9), liver (10,11), heart (13), and lungs (14,15), and arteries in atherosclerosis (12) . Several studies suggest that PPAR␥ inhibits extracellular matrix synthesis, including collagen (16 -18). PPAR␥ may alter connective tissue target genes by blocking TGF-␤ signaling (16,19,20), an important profibrotic cytokine. Alternatively, PPAR␥ may be directly involved in constitutive expression of collagen as part of transcriptional regulation complexes (17). Recent results suggest that PPAR␥ represses collagen transcription on the minimal COL1A1 promoter (17). Usually PPAR␥ heterodimerizes with retinoid-X-receptor and binds to a specific DNA response element with a direct repeat separated by a single intervening nucleotide (21). There are no PPAR␥ response elements in either collagen type I proximal promoter. On the other hand, PPAR␥ interacts with many proteins in different combinations to alter transcription of several other genes (4).
We have demonstrated that interferon-␥ (IFN-␥), an important inflammatory cytokine, represses collagen transcription, in part, through the induction of regulatory factor for X box 5 (RFX5) complex proteins and Class II transactivator (CIITA), and their recruitment to the collagen transcription start site (22)(23)(24). RFX5 and CIITA are important activators of the immune response through transcriptional activation of major histocompatibility complex II (MHC II). In fact, mutations in these proteins cause bare lymphocyte syndrome. Because PPAR␥ may regulate a subset of IFN-␥ target genes (25), we decided to investigate whether collagen was a target of PPAR␥ especially during IFN-␥ treatment. In this report, we demonstrate that PPAR␥ is recruited to a complex with RFX5 and CIITA at the collagen transcription start site and the MHC II promoter during IFN-␥ treatment. The IFN-␥-induced repression of collagen and activation of MHC II transcription is enhanced in the presence of PPAR␥ agonists and reversed by PPAR␥ antagonists. Finally, silencing PPAR␥ blocks the IFN-␥induced collagen repression.
DNA Affinity Pulldown Assay-The collagen sequence (COL1A2 Ϫ25/ϩ30, GenBank TM accession number AF004877) with a HindIII overhang was synthesized as complementary strands and annealed as previously described (26). Doublestranded collagen DNA was biotin-labeled by incubating with Klenow fragment (New England Biolabs, Ipswich, MA) and biotin-14-dATP (Invitrogen) supplemented with regular dCTP, dTTP, and dGTP at room temperature for 30 min. The reaction mixture was phenol-chloroform-extracted and alcohol-precipitated to remove unincorporated biotin.
Nuclear protein extracts from IMR-90 cells were obtained as previously described using 450 mM sodium chloride (22,27). The streptavidin beads (Promega, Madison, WI) were washed three times with ice-cold phosphate-buffered saline supplemented with 1 mM phenylmethylsulfonyl fluoride. Nuclear proteins (100 -200 g) were pre-cleared by incubating with the washed beads for 30 min at 4°C on a shaking platform as described previously (23). Pre-cleared nuclear proteins were prepared by capturing the beads on a magnetic stand and removing the supernatant. The supernatant was then incubated with biotin-labeled collagen DNA probe (Ϫ25/ϩ30) for 1 h at room temperature in binding buffer (60 mM NaCl, 20 mM HEPES, pH 7.9, 0.1 mM EDTA, 4% glycerol, 2 mM dithiothreitol) supplemented with bovine serum albumin, poly(dI-dC), and sonicated salmon sperm DNA to remove nonspecific binding. DNA⅐protein complex formed was captured by the magnetic beads and washed extensively with binding buffer supplemented with 0.01% Triton X and 100 mM KCl. The bound pro-teins were eluted with 1ϫ electrophoresis sample buffer by incubating at 90°C for 10 min and analyzed by SDS-PAGE gels.
Plasmids, Transfections, and Luciferase Assays-The col1a2luciferase construct (pH20) (28) contains sequences from Ϫ221 to ϩ54 bp of mouse col1a2 promoter fused to the luciferase reporter gene. A larger col1a2-luciferase construct (pGL3-Col-Luc) containing sequences from Ϫ357 to ϩ55 bp of mouse col1a2 promoter fused to the luciferase reporter gene was a gift from Dr. Jenny Ting (29). The human COL1A1 promoter plasmids were all cloned into a pGL2-luciferase vector from the Ϫ804hCOL1-luc plasmid described previously (24). Fulllength PPAR␥ construct was generated as previously described (30). Full-length FLAG-RFX5 and three different isoforms of FLAG-CIITA were kindly provided by Dr. Jenny Ting. Cells were plated at the density of 3 ϫ 10 5 cells/well in 6-well tissue culture dishes (for IMR-90 cells) or 5 ϫ 10 6 cells per p100 tissue culture dish (for 293FT cells). Transfections were performed with Lipofectamine 2000 reagent (Invitrogen) according to the manufacturer's protocol. Cells were harvested 48 h post transfections, and luciferase assays were performed with a luciferase reporter assay system (Promega).
RNA Isolation and Real-time PCR-Cells were harvested, and RNA was extracted using an RNeasy RNA isolation kit (Qiagen) according to the manufacturer's protocol. Reverse transcriptase reactions were performed using a SuperScript First-strand synthesis system (Invitrogen) according to the manufacturer's protocol. Real-time PCR reactions were performed on an ABI-Prism 7700 sequence detection PCR machine (Applied Biosystems, Foster City, CA) according to the manufacturer's protocol. The oligonucleotide forward and reverse PCR primers and fluorescent probes are described in previous studies (22,23,31).
PPAR␥ RNA Interference-Five PPAR␥-silencing clones were purchased from Sigma. Upon testing, two were determined to knock down PPAR␥ expression efficiently (clone #2 and clone #4, data not shown). These two clones, along with a non-functional clone (#5) as a control, were transfected into 293FT cells using Lipofectamine 2000 reagent according to the manufacturer's protocol (Invitrogen). Viruses were harvested FIGURE 1. PPAR␥ co-immunoprecipitates with RFX5. A, overexpressed RFX5 interacts with endogenous PPAR␥ in 293FT cells. 293FT cell extracts (500 g) with transfected FLAG-RFX5 were precipitated using FLAG-conjugated beads (40 l) (M2) or pre-immune IgG-conjugated beads (40 l) (IgG) as described under "Experimental Procedures." Eluates were separated by 10% SDS gel, and Western blots were performed with an anti-FLAG, or an anti-PPAR␥ antibody as indicated. 10% of the original lysate was also loaded as input. A representative picture from three independent experiments is shown. M, protein marker; N.S., nonspecific. B and C, RFX5 and PPAR␥ interact with each other in IMR-90 cells. Whole cell extracts (500 g) prepared from IMR-90 cells were precipitated using an anti-RFX5 antibody (5 g) (B), two different sources of anti-PPAR␥ antibodies (C), or anti-pre-immune IgG (5 g) as indicated. Eluates were separated by 10% SDS gel, and Western blots were performed with antibodies as indicated. The original lysate (10%) was also loaded and labeled as input. A representative picture from three independent experiments is shown. M, protein marker; H.C., heavy chain. Proteins were extracted using RIPA buffer (1ϫ phosphate-buffered saline, 0.1% SDS, 1% Nonidet P-40, 0.5% sodium deoxycholate) and analyzed by SDS-PAGE gel followed by Western blot. RNAs were extracted and analyzed as described before.
Statistical Analysis-One-way analysis of variance with posthoc Scheffe analyses were performed using an SPSS package. p values smaller than 0.05 were considered statistically significant.

RESULTS
PPAR␥ Co-immunoprecipitates with RFX5 and CIITA IV-PPAR␥ represses COL1A1 gene transcription even on the minimal promoter (17) raising the possibility that there might be an interaction between PPAR␥ and RFX5, because the latter binds to the collagen transcription start site and represses collagen synthesis (22). First, protein extracts from cells transfected with FLAG-RFX5 were immunoprecipitated with a FLAG antibody (M2) or IgG antibody-conjugated beads. PPAR␥ was co-immunoprecipitated with FLAG-RFX5 using specific epitope antibody but not control IgG (Fig. 1A). Next, proteins from human lung fibroblasts were immunoprecipitated by anti-RFX5 antibodies to examine endogenous PPAR␥ interactions with RFX5. PPAR␥ was co-immunoprecipitated with RFX5 (Fig. 1B). The secondary anti-rabbit IgG reacts with a nonspecific band in the nonimmune control and the RFX5 serum. In the reciprocal experiment, RFX5 was co-immunoprecipitated using two different sources of PPAR␥ antibody (Fig. 1C) suggesting that PPAR␥ and RFX5 are in a similar complex.
CIITA is difficult to detect in un-stimulated mesenchymal cells. Yet, CIITA is largely responsible for the IFN-␥-induced repression of collagen transcription, because IFN-␥ treatment stimulates CIITA expression, nuclear localization, and occupation on the collagen start site (23). Most importantly, silencing CIITA blocks IFN-␥-induced collagen repression (23). Two isoforms, CIITA III and CIITA IV, are inducible by IFN-␥ in human smooth muscle cells (32) and fibroblasts (data not shown).
To determine whether individual isoforms might co-immunoprecipitate with PPAR␥, these two isoforms were overexpressed in 293FT cells and immunoprecipitated with epitope-specific antibody (M2). PPAR␥ was co-immunoprecipitated with CIITA IV isoform, but not with CIITA III isoforms ( Fig. 2A). Experiments were also performed with CIITA I, which was not expressed in these cells as CIITA III or IV. There was no detectable association of CIITA I with PPAR␥ (data not shown).
To examine endogenous interactions, human lung fibroblasts were treated with IFN-␥ and proteins were immunoprecipitated with either an anti-CIITA or an anti-PPAR␥ antibody. CIITA was detected in IFN-␥-treated extracts along with PPAR␥ using CIITA antibody that recognizes all CIITA isoforms (Fig. 2B). In a reciprocal experiment with PPAR␥ antibody, CIITA was co-immunoprecipitated only from IFN-␥treated extracts (Fig. 2C) suggesting that when CIITA is expressed it interacts with PPAR␥ in the same complex.
PPAR␥ Represses Collagen Promoter Activity in a Dose-responsive Manner and Enhances CIITA IV Transcription Activity-PPAR␥ suppresses COL1A1 proximal promoter activity in activated hepatic stellate cells (17). In dermal fibroblasts, PPAR␥ transient transfection had a small inhibitory effect on COL1A2 activity (16). Our data demonstrate that col1a2 promoter activity is repressed by increasing amounts of PPAR␥ expression in human lung fibroblasts (Fig. 3A). Because PPAR␥ interacts with RFX5 and CIITA, we addressed whether RFX5 or the isoforms of CIITA might function to repress collagen promoter activity in an additive fashion with PPAR␥. PPAR␥ (0.1 g) enhanced the repression of col1a2 promoter activity most prominently with CIITA IV (Fig. 3B). This exact pattern was observed using a larger promoter of col1a2 (data not shown) and COL1A1 (Fig. 3C). The collagen genes have RFX5⅐CIITA binding sites at the transcription start site within a potential initiator site not in the promoter. Mutations in this region may block transcription as well as blocking RFX binding, so using mutations as a method to determine whether PPAR␥dependent transcription is mediated through the RFX5 binding site was not performed.
Most importantly, CIITA IV in combination with PPAR␥ was the best activator of the MHC II promoter (Fig. 3D) as well as the best repressing combination for collagen promoter activ-FIGURE 2. PPAR␥ co-immunoprecipitates with CIITA. A, overexpressed CIITA type IV interacts with endogenous PPAR␥ in 293FT cells. 293FT cell extracts (500 g) with transfected CIITA isoforms or an empty vector (EV) were precipitated using FLAG-conjugated beads (40 l) (M2) or pre-immune IgG-conjugated beads (40 l) (IgG) as described under "Experimental Procedures." Eluates were separated by 10% SDS gel, and Western blots were performed with an anti-FLAG, or an anti-PPAR␥ antibody as indicated. 10% of the original lysate was also loaded as input. A representative picture from three independent experiments is shown. B and C, CIITA and PPAR␥ interact with IMR-90 cells when CIITA is stimulated by IFN-␥. Whole cell extracts (500 g) prepared from IMR-90 cells with (ϩ) or without (Ϫ) IFN-␥ treatment were precipitated using an anti-CIITA antibody (5 g) (B), an anti-PPAR␥ antibody (5 g) (C), or anti-pre-immune IgG (5 g) as indicated. Eluates were separated by 10% SDS gel, and Western blots were performed with antibodies as indicated. The original lysate (10%) was also loaded and labeled as input. A representative picture from three independent experiments is shown.
ity. Because CIITA IV is the isoform that co-immunoprecipitates with PPAR␥, CIITA may function with PPAR␥ to both repress collagen and activate MHC II transcription.
PPAR␥ Agonists Enhance the IFN-␥-induced Repression of Collagen-PPAR␥ agonists disrupt TGF-␤ signaling in normal human dermal fibroblasts (16) and in pulmonary myofibroblasts (19). Because IFN-␥ induces CIITA expression and abrogates TGF-␤ activity, we next used the agonist, Trog (0.5 M), in the presence or absence of IFN-␥. Without exogenously added PPAR␥, Trog had minimal affect on collagen promoter activity (Fig. 4, A and B). However, Trog enhanced IFN-␥-induced collagen repression for all collagen type I promoters tested. To determine if the agonist effect was through PPAR␥, cells were pretreated with the antagonist T0070907 for 2 h before IFN-␥ and/or Trog treatment. The antagonist blocked the action of Trog on all collagen promoters in the presence of IFN-␥ suggesting there is a PPAR␥ effect (Fig. 4, A and B). Interestingly, without IFN-␥ treatment the antagonist activated collagen pro-moter activity in the smaller (Fig. 4A) but not the larger col1a2 promoter (Fig. 4B). All collagen promoter activity changes were similar with a second antagonist GW9662 (data not shown), and the COL1A1 promoter (311 bp) was similar to the larger col1a2 promoter (357 bp) (data not shown).
PPAR␥ Agonists Enhance the IFN-␥-induced Repression of Collagen and Activation of MHC II and CIITA mRNA Steadystate Levels-Human lung fibroblasts were treated with PPAR␥ agonists and antagonists either in the presence or absence of IFN-␥ to examine mRNA steady-state levels. Clearly PPAR␥ agonist and antagonists altered collagen steady-state levels only in the presence of IFN-␥ (Fig. 5, A and B). Trog enhanced collagen repression, and T0070907 antagonized this repression suggesting that PPAR␥ was involved.
CIITA steady-state mRNA was not detectable in the absence of IFN-␥, and, neither the agonist nor the antagonist altered the CIITA mRNA level (Fig. 5C). IFN-␥ stimulated the CIITA mRNA, and Trog slightly enhanced the stimulation, which was blocked by the antagonist, T007. ⌴⌯C II expression followed a similar pattern to CIITA, indicating that PPAR␥ might directly impact on CIITA expression and hence MHC II and the inflammatory response.
Agonist and Antagonist Alter the Association of Proteins at the RFX5 Site on the Collagen Transcription Start Site by DNA Affinity Chromatography-RFX5 recruits CIITA to the collagen start site to repress transcription during IFN-␥ treatment (22,23). Because PPAR␥ is associated with both RFX5 and CIITA in the same complex, we next determined whether PPAR␥ could also be recruited by RFX5 and, if so, how the recruitment would be altered by different treatments. The biotinylated DNA oligonucleotide spanning the col1a2 start site (Ϫ25/ϩ30) was incubated with nuclear proteins extracted from lung fibroblasts with different treatments as described earlier; eluates were separated by SDS gels and Western blots were performed to detect recruitment of RFX5, PPAR␥, and CIITA as described under "Experimental Procedures" (22,23). IFN-␥ enhanced the recruitment of all three proteins to the collagen transcriptional start site (Fig. 6, compare lane 2 to lane 1), whereas Trog decreased the binding of both RFX5 and PPAR␥ FIGURE 3. PPAR␥ represses collagen promoter activity. A, PPAR␥ represses col1a2 promoter activity in a dose response manner. A col1a2 promoter construct (pH20, 0.5 g) was co-transfected with different concentrations of PPAR␥ constructs as indicated, along with GFP (0.1 g) for normalization, into IMR-90 cells in triplicate wells as described under "Experimental Procedures." Average luciferase activities were normalized by both protein concentration and GFP fluorescence and were expressed as relative percentage activity compared with the control group in which an empty vector was transfected. This representative experiment was repeated at least three times. At each dose tested, PPAR␥ repressed collagen promoter activity significantly (**, p Ͻ 0.01). B and C, PPAR␥ enhances the repression of collagen promoter activities by CIITA. A col1a2 promoter construct (pH20, 0.5 g) (B) or a COL1A1 promoter construct (COL1A1-311, 0.5 g) (C) was co-transfected with PPAR␥ (0.1 g), RFX5 (0.1 g), CIITA III (0.1 g), or CIITA IV (0.1 g) construct as indicated, along with GFP (0.1 g) for normalization, into IMR-90 cells in triplicate wells as described under "Experimental Procedures." Average luciferase activities were normalized by both protein concentration and GFP fluorescence and were expressed as relative percentage activity compared with the control group in which an empty vector was transfected. This representative experiment was repeated at least three times. Co-transfection of PPAR␥ renders significant (**, p Ͻ 0.01) additional repression of col1a2 promoter activity by CIITA IV. D, PPAR␥ enhances the activation of MHC II promoter activity by CIITA. An MHC promoter construct (pGL3-DRA300, 0.5 g) was co-transfected with PPAR␥ (0.1 g), RFX5 (0.1 g), CIITA III (0.1 g), or CIITA IV (0.1 g) construct as indicated, along with GFP (0.1 g) for normalization, into IMR-90 cells in triplicate wells as described under "Experimental Procedures." Average luciferase activities were normalized by both protein concentration and GFP fluorescence and were expressed as relative percentage activity compared with the control group in which an empty vector was transfected. This representative experiment was repeated at least three times. Co-transfection of PPAR␥ renders significant (**, p Ͻ 0.01) additional activation of HLA-DRA promoter activity by CIITA IV. ( Fig. 6, compared lane 3 to lane 1). Antagonist T0070907 did not affect the association of proteins with the collagen oligonucleotide, but it did reverse the effect of Trog (Fig. 6, compare  lane 5 to lane 3). Trog enhanced the IFN-␥-stimulated recruit-ment of all three proteins to the collagen site (Fig. 6, compare lane 6 to lane 2) whereas the antagonist did not alter the IFN-␥ response, but did abrogate the Trog effect (Fig. 6, compare lane 8 to lane 6). Rather than mutate the RFX5 binding site within the transcription start site, a specific competitor was added to remove the RFX5, which blocked the recruitment of both PPAR␥ and CIITA (Fig. 6, lane 9), suggesting that RFX5 is responsible for recruiting both proteins to the collagen site. A nonspecific competitor failed to block the recruitment of the proteins (Fig. 6, lane 10). These data show that all three proteins are in a complex that can bind to an oligonucleotide corresponding to the col1a2 start site.

Agonist Enhances the IFN-␥-induced Recruitment of Repressors on the Collagen Transcription Start Site and on the MHC II
Promoter-To determine the recruitment of different proteins to the endogenous col1a2 and MHC II promoters under various conditions, chromatin immunoprecipitations were performed. PPAR␥ was present at the collagen transcription start site, and treatment of Trog slightly decreased the occupancy, which was over-ridden by the antagonist T0070907 (Fig. 7A). IFN-␥ increased the occupancy by more than 2-fold and could further the increase in the presence of Trog; the antagonist again blocked the effect of Trog. Strikingly similar patterns were observed for PPAR␥ occupancy on the MHC II promoter as well as for RFX5 and CIITA occupancies on both collagen and MHC II sites (Fig. 7, B-E). In contrast, binding of p300 did not change significantly in response to either Trog or T0709007 treatment, whereas IFN-␥ slightly decreased the p300 occupancy on the collagen transcription start site (Fig. 7F).
Silencing PPAR␥ Partially Reverses IFN-␥-induced Repression of Collagen Messages-Finally, to determine the role of PPAR␥ in IFN-␥ mediated repression of collagen synthesis, lentivirus carrying PPAR␥-silencing plasmids were used to infect lung fibroblasts followed by IFN-␥ treatment. Lentivirus was chosen to increase efficiency of expression in early passage human fibroblasts. The lung fibroblasts used are a cell strain that has a finite life span in cell culture and therefore has low transfection efficiency for plasmids or adenovirus transduction. Infection of cells by clone #2 and #4 led to dramatic decrease in PPAR␥ expression both with (Fig. 8A, upper panel, lane 6) and without IFN-␥ (Fig. 8A, upper panel, lane 3) treatment. In contrast, neither clone #5 (Fig. 8A, upper panel, lanes 2 and 5) nor an empty vector (Fig. 8A, upper panel, EV, lanes 1 and 4) resulted in any change in PPAR␥ expression. As a negative control, expression of mSin3B was not affected by any one of the infections (Fig. 8A, bottom panel). To our surprise, silencing PPAR␥ greatly reduced the IFN-␥ induction of CIITA (Fig. 8A,  lane 6). As expected, no CIITA was detected without IFN-␥. On the other hand, the expression of the co-repressor molecule, Sin3B, was not altered by silencing PPAR␥.
Next, expression of type I collagen and CIITA messages was examined by real-time PCR. Infection by clone #2 and #4 slightly but significantly increased basal level COL1A1 mRNA expression (Fig. 8B) compared with empty vector infection (EV), indicating that PPAR␥ is a repressor of collagen synthesis on its own. IFN-␥ decreased COL1A1 mRNA by more than 60%, which is partially abrogated by clone #2 and #4 infection, suggesting that PPAR␥ is involved in IFN-␥-induced collagen  repression. Infection by clone #5 did not cause any significant change of collagen expression either with or without IFN-␥ treatment. Expression of col1a2 mRNA exhibited a similar pattern when cells were infected with PPAR␥-silencing virus (Fig.  8B, second graph). On the other hand, PPAR␥ clone #2 and #4 infection blocked the IFN-␥ induction of CIITA mRNA, but #5 did not cause any change in IFN-␥-induced CIITA expression (Fig. 8B, third graph). Because, surprisingly, IFN-␥-induced CIITA expression was also greatly reduced by PPAR␥ short hairpin RNA at both the mRNA and protein levels (Fig. 8, A and  B), PPAR␥ might be a direct regulator of CIITA expression. Therefore, the changes in collagen gene expression could be caused by changes in CIITA expression.

DISCUSSION
Recent reports demonstrate that PPAR␥, a member of the nuclear receptor superfamily, acts as a repressor of type I collagen synthesis (16,17,33). Our data confirm that exogenously expressed PPAR␥ down-regulates collagen expression in a dose-responsive manner in human lung fibroblast cells as well as human aortic smooth muscle cells. 3 PPAR␥ interacts with CIITA ( Figs. 1 and 2) and is recruited to the collagen start site and to the MHC II promoter site (Figs. 6 and 7), Silencing of PPAR␥ expression activated collagen gene expression and alleviated IFN-␥-induced collagen repression. More importantly, knockdown of PPAR␥ expression led to the abrogation of CIITA activation (Fig. 8), suggesting that PPAR␥ may be part of a regulatory feedback loop, whereby it regulates expression of proteins with which it interacts and functions. As a matter of a fact, this might be a common character shared among PPAR family members, because other investigators have demonstrated that both PPAR␣ and PPAR␦ interact with and regulate the expression of the co-activator PGC-1␣ (34,35). Our observations agree with earlier reports and allude to a broader mechanism whereby PPAR␥ exert its physiological function, which may be unique to the PPAR proteins. The expression of PPAR␥ decreases during hepatic cirrhosis when collagen expression increases (18,36) suggesting that PPAR␥ may be important in   down regulating the profibrotic process. Interestingly, PPAR␥ levels are also dramatically reduced with age in several tissues (37). Therefore, the gene expression of PPAR␥ may be an important mechanism for modulating extracellular matrix production and inflammation through MHC II expression.
PPAR␥ agonists inhibit TGF-␤-induced myofibroblast differentiation and signaling of collagen gene expression (16,19,36). IFN-␥ has been recognized as an antagonist of TGF-␤ signaling (38 -40). Our studies demonstrate that PPAR␥ agonist, Trog, sensitizes the cells to IFN-␥ treatment by further lowering collagen expression levels (Figs. 4 and 5). Combined with the finding that TGF-␤ attenuates CIITA expression (41,42), we propose that CIITA is a key protein, when facilitated by other factors such as RFX5 (22,24), which determines the ultimate consequences of collagen transcription in response to different signaling pathways. Therefore, we investigated whether PPAR␥ agonists alter the IFN-␥ response through the CIITA⅐RFX5 complex.
Previous studies in this laboratory have identified the RFX family of transcription factors that play vital roles in fine-tuning collagen expression and activating MHC II (43) during different physiological and/or pathophysiological scenarios (26,31,44). The RFX5⅐CIITA complex, which mediates IFN-␥-induced collagen repression (22,23), binds to the transcription start site of type I collagen genes and represses collagen gene expression (22,44). PPAR␥ exists in the same complex with RFX5 ( Fig. 1) and type IV CIITA (Fig. 2) in both 293FT cells and IMR-90 cells. Moreover, recruitment of PPAR␥ to the collagen transcription start site was, like that of CIITA, dependent on RFX5 as judged by the removal of RFX5 binding that prevented PPAR␥ from interacting with the collagen DNA (Fig. 6, lane 9).
It should be noted that one of the unique features of the RFX proteins in regulating collagen transcription is that they bind to the first exon of the collagen gene where transcription starts, which coincidentally also serves as the binding site for the preinitiation complex. Our published results indicate that there is a competition of occupancy on the collagen start site between the RFX5⅐CIITA complex and the pre-initiation complex (23). In addition, a ϩ7 C to T mutation mimicking a methylated C at that position increases RFX1 binding and decreases activity of the promoter by 80% (45). Therefore, additional mutations at the RFX site would also deny the access of RNA polymerase II to the promoter, making data interpretation difficult. Therefore, we chose instead to use DNA competitors to demonstrate the specificity and recruitment of RFX binding in DNA affinity pulldown assays. It should be pointed out that this method has been exploited before and has satisfactorily served the same purpose in demonstrating the specificity and recruitment dependence or RFX5⅐CIITA complex to this site (22)(23)(24). Further experiments extending this line of investigation are currently underway and will eventually provide extra evidence to our hypothesis.
More importantly, the physical association correlates with the function in which PPAR␥ enhanced the repression of collagen and activation of MHC II promoter activities by type IV CIITA (Fig. 3, B and C). Interestingly, although it is clear that PPAR␥ and RFX5 co-immunoprecipitate, there was no functional synergism between them on either promoter (Fig. 3, B and C). One reason could be that the interaction between PPAR␥ and RFX5 is not sufficient for transcriptional activity, but rather necessary to recruit CIITA, which then co-operates with PPAR␥ to determine the ultimate transcriptional outcome. Alternatively, this repression may require post-translational modification of the factors through IFN-␥ signaling.
Our data presented in this report indicates that Trog enhances the ⌱FN-␥ repression of collagen and activation of MHC II (Figs. 4B, 4C, and 5). Most importantly, the recruitment of PPAR␥ at the collagen start site and the MHC II promoter is increased with IFN-␥ treatment followed by a further increase in the presence of Trog, which is neutralized by an antagonist (Fig. 7, A and B). There is a strikingly similar pattern of RFX5 and CIITA recruitment on both genes suggesting that IFN-␥ stimulates PPAR␥ recruitment within a complex with CIITA and RFX5. On the other hand, CBP/p300 decreases slightly on the collagen start site with IFN-␥ treatment. PPAR␥ binds to the COL1A2 transcription start site both in vitro (Fig.  6) and ex vivo (Fig. 7A). It has become clear that combinatorial interactions among promoter-bound proteins determine transcriptional outcome in different cellular and experimental contexts on COL1A2 gene (46). Our results in this report demonstrate a unique association of the nuclear receptor, PPAR␥, with transcription factors, RFX5 and CIITA type IV, at the transcriptional start site of both type I collagen genes (Figs. 6 and 7A) and at the MHC II promoter (Fig. 7B) especially during IFN-␥ signaling.
CIITA has been termed a scaffold protein that interacts with several proteins, including NF-Y/CBF, RFX5, CREB (47), and CBP/p300 (29). The collagen type I promoters have a reverse CAAT box where a heterotrimeric CAAT binding factor NF-Y/ CBF interacts (48). On the other hand, this site in the COL1A1 promoter has been identified as a nuclear factor-1 (NF-1) binding site in hepatic stellate cells (49). Interestingly, exogenously transfected PPAR␥ blocks the function of NF-1 and its interaction with the COL1A1 promoter in hepatic stellate cells (17). It is possible that the NF-1 is replaced by NF-Y⅐RFX5 complex (50).
Two col1a2 promoters of different sizes were compared to determine whether a 136-bp region alters the PPAR␥ response. PPAR␥ antagonists activate the shorter promoter but not the longer promoter with the additional 136 bases (Fig. 4). As shown in the model (Fig. 9), the 136 bases contain multiple binding sites for known activators of collagen transcription that interact with CBP/p300. The action of antagonist-bound PPAR␥ may be blocked by these upstream binding proteins. The larger promoter pattern mimics the mRNA pattern (Fig. 5). Studies are ongoing to determine the upstream interactions of proteins with CIITA and PPAR␥.
Earlier results indicate that two CIITA domains (N-terminal and PST domain) are required for repression on the larger promoter, whereas the PST domain alone represses the small basal promoter (23). Others demonstrated that the N-terminal CIITA is required for repression through CBP/p300 (29). In addition, CBP/p300 counteracts PPAR␥ repression (17) and CIITA repression of collagen transcription (29). Our results suggest that PPAR␥ and CIITA are in the same complex and CIITA IV, which has a different N-terminal domain than the other isoforms, is necessary for transcriptional repression possibly by inactivating CBP/p300. CPB/p300 occupancy is decreased in the presence of IFN-␥ (Fig. 7G). A model (Fig. 9) shows the interactions of PPAR␥ with both RFX5 and CIITA. On the collagen genes, this complex containing NF-Y may replace the NF-1⅐CBP⅐p300 complex.
To our surprise, IFN-␥-induced expression of CIITA dramatically decreased in the absence of PPAR␥ when the cells were infected by PPAR␥ short hairpin RNAs. However, expression of several early response genes in the IFN-␥ signaling pathway was not affected (data not shown), suggesting that PPAR␥ might directly bind to the CIITA promoter and activate its transcription. Because CIITA is the critical factor in IFN-␥-mediated connective tissue remodeling (down-regulation of collagen expression) and inflammatory response (up-regulation of MHC II expression), this observation further highlights the importance of PPAR␥ during the regulation of these two events, because it is able to not only functionally interact with CIITA, but also control the expression level of CIITA.
Nuclear receptors are certainly among the proteins that can interact with multiple families of transcription factors, co-activators, and co-repressors (4). PPAR␥ forms a heterodimer with retinoic acid X receptor when interacting with its peroxisome proliferator response element consisting of direct repeats of hexanucleotides separated by a single nucleotide spacer (see reviews (4,51,52)). There is no peroxisome proliferator response element in the proximal collagen promoters used in this study, although there are putative untested upstream sites in col1a2 promoter (Ϫ581 to Ϫ530 and from Ϫ1364 to Ϫ1352) (18). Usually PPAR␥ interacts with co-activators (SRC1 and CBP/p300) when ligands are available. Without ligand, PPAR␥ interacts with co-repressor complexes containing NCoR⅐SMRT and HDAC3 and represses genes. In addition, PPAR␥ when it is sumoylated interacts with NCoR⅐SMRT and HDAC3 without interacting with DNA in a transrepression mechanism (3,53). Although PPAR␥ interacts with the collagen gene and the MHC II gene both in vitro (Fig. 6) and ex vivo (Fig. 7) in the presence of IFN-␥, this is not accompanied by HDAC3 as indicated in our previous report, which demonstrates that Sin3B with HDAC2, not HDAC3, dramatically interacts with collagen promoters in the presence of IFN-␥ (31). In addition, HDAC2 deacetylates RFX5 as well as histones (31). Our unpublished results indicate that no NCoR⅐SMRT is detected in DNA affinity precipitations (data not shown) and that Sin3B and HDAC2 interact with CIITA in a phosphorylation-dependent manner (data not shown). It is not clear yet which proteins interact directly with PPAR␥ in this complex and whether other known PPAR␥-interacting proteins, such as retinoid-X-receptor, are in this complex. Our results demonstrate a hitherto un-described complex with PPAR␥ containing CIITA⅐RFX5⅐HDAC2⅐Sin3B that assembles on the collagen gene to repress collagen gene transcription by transrepression.
CIITA, RFX5, and PPAR␥ are recruited to the MHC II promoter during IFN-␥ treatment, but this gene is activated not repressed. More experiments are required to determine why similar complexes on two different genes can repress or activate. This could be through promoter context or by recruitment of different co-activator/co-repressor complexes on different promoters. We have preliminary data, gained by chromatin immunoprecipitation analysis, that CIITA⅐RFX5⅐PPAR␥ on the MHC II promoter recruits less Sin3B and HDAC2 in the presence of IFN-␥ for 24 h (data not shown). However, the activation of MHC II may be downregulated by similar molecules at a later time (54).
In addition to sumoylation and acetylation, PPAR␥ (55) and RFX5 (56) are also phosphorylated, which could alter their transcriptional outcome if IFN-␥ causes post-translational modifications. Therefore, it is quite possible that combinations of different cytokine and/or growth factor treatments result in variations in modifications, which in turn, determine different transcriptional outcomes. This strongly suggests that the interactions demonstrated with CIITA⅐RFX5 are unique with repression of collagen being caused by Sin3⅐HDAC2 associated co-repressor.
Structurally diverse molecules bind to and activate PPAR␥ (2). The thiazolidinedione compounds and certain prostaglan- dins have been used to demonstrate that activation of PPAR␥ inhibits collagen synthesis (18) and fibrosis in lung (14,15,19,57), kidney (8,9,58), liver (10,11,33,36,59), cardiac fibroblasts (60), and skin (61). The issue of endogenous ligand for PPAR␥ in cells has been controversial. PPAR␥ ligands, such as fatty acids, eicosanoids, and oxidized lipids, have been identified that bind to and alter PPAR␥ transcriptional activity. Prostaglandin J2 derivative, often considered the endogenous ligand, does not naturally exist at sufficient concentrations to activate PPAR␥ in mammalian cells (62,63) and affects multiple pathways other than PPAR␥. An unidentified ligand is produced only transiently during adipocyte differentiation (64). Overall, it remains to be determined whether PPAR␥ has a highly specific natural ligand or whether it operates as a physiological lipid sensor activated by the combined concentrations of weakly activating fatty acids and eicosanoids (65). In our observation, there is little or no effect on collagen promoter activity (Fig. 4, B and C) or collagen gene expression (Fig. 5) by Trog alone in lung fibroblasts similar to dermal fibroblasts (16). It is possible that the cells produce an endogenous ligand so that exogenously added Trog does not alter collagen gene expression. Antagonists might be able to reverse the effects of an endogenous ligand. The antagonist T0070907 and GW9662 (data not shown) activated the minimal col1a2 promoter activity (Fig. 4A) but had no effect on the larger col1a2 (Fig. 4B) or COL1A1 (data not shown) promoters or collagen mRNA levels (Fig. 5). The difference in the long and minimal promoter might suggest that PPAR␥ interacts with a protein recruited to an upstream region between Ϫ350 and Ϫ221 independent of ligand.
Alternatively, part of the mechanism of collagen transcriptional regulation by PPAR␥ might not depend on ligand, but rather through interactions with other factors such as the RFX5⅐CIITA complex in a facilitative fashion. Decreased levels of PPAR␥ in these cells could lead to impaired transrepression complex formation on the collagen promoter and partially relieve the repression of collagen transcription, which is irrelevant to the presence or absence of ligand. Another probable explanation is that endogenous ligand might exist in a wide range of concentrations in different cell types. Therefore, one given synthetic ligand at certain concentrations may or may not cause any change to a particular PPAR␥ target gene. Our observation that Trog represses collagen transcription only in the presence of IFN-␥ (Figs. 4 and 5) raises yet another possibility that PPAR␥ activity partially depends on its post-translational modification (such as sumoylation), which is controlled by its ligands in conjuncture with IFN-␥ (and/or other cytokines); increasing the ligand level alone would not change PPAR␥ modification or alter its activity, hence no effect of its target genes could be detected.
PPAR␥ is involved with inflammatory response (4, 51, 52) and in atherosclerosis (66). Activation by agonists makes it an anti-inflammatory molecule in part by down-regulation of cytokines like IFN-␥. PPAR␥ activation by agonist negatively regulates a specific subset of IFN-␥ target genes in macrophages (25). This negative repression is most likely through sumoylation and transrepression (53). CIITA and MHC II, investigated in this report, are clearly IFN-␥ target genes involved in inflammation, because MHC II is critical for activation of T cells and CIITA⅐RFX5 are critical for activation of MHC II. However, these genes are not negatively regulated in fibroblasts by PPAR␥ agonists. In fact, agonists enhance the IFN-␥ induction of CIITA and MHC II expression, which contrasts to a previous report using atheroma cells (67). One explanation is that cell type differences exist or the agonists have PPAR␥-independent effects (67). In our studies, antagonists reversed the activation by agonist (Figs. 5 and 7) suggesting agonist-induced activation is through PPAR␥.
In summary, we demonstrate that PPAR␥ associates with RFX5 and type IV CIITA in the same complex. PPAR␥ is recruited to the collagen transcription start site by RFX5⅐CIITA complex, where it represses collagen expression, and more importantly, is involved in mediating and enhancing the IFN-␥-stimulated collagen transcription down-regulation and MHC II up-regulation. Without IFN-␥ treatment, there is little recruitment of RFX5⅐PPAR␥ to the collagen start site or to MHC II promoter. In addition, silencing PPAR␥ lowers IFN-␥ induction of CIITA and attenuates repression of collagen expression. Therefore, PPAR␥ may be involved in a regulatory loop by regulating the expression of a protein that it interacts with. This report highlights the importance of both connective tissue remodeling and inflammatory response by bringing the transcription factors, RFX5⅐CIITA/PPAR␥ complex, which regulate transcriptional events in a coordinated and cooperative manner. Therefore, PPAR␥ is a critical target for investigations into therapeutics of diseases to whose etiologies both extra matrix remodeling and immune response contribute.