Troglitazone, a Peroxisome Proliferator-activated Receptor γ (PPARγ) Ligand, Selectively Induces the Early Growth Response-1 Gene Independently of PPARγ A NOVEL MECHANISM FOR ITS ANTI-TUMORIGENIC ACTIVITY

Troglitazone (TGZ) is a peroxisome proliferator-activated receptor γ (PPARγ) ligand that has pro-apoptotic activity in human colon cancer. Although TGZ binds to PPARγ transcription factors as an agonist, emerging evidence suggests that TGZ acts independently of PPARγ in many functions, including apoptosis. Early growth response-1 (Egr-1) transcription factor has been linked to apoptosis and shown to be activated by extracellular signal-regulated kinase (ERK). We investigated whether TGZ-induced apoptosis may be related to Egr-1 induction, because TGZ has been known to induce ERK activity. Our results show that Egr-1 is induced dramatically by TGZ but not by other PPARγ ligands. TGZ affects Egr-1 induction at least by two mechanisms; TGZ increases Egr-1 promoter activity by 2-fold and prolongs Egr-1 mRNA stability by 3-fold. Inhibition of ERK phosphorylation in HCT-116 cells abolishes the Egr-1 induction by TGZ, suggesting its ERK-dependent manner. Further, the TGZ-induced Egr-1 expression results in increased promoter activity using a reporter system containing four copies of Egr-1 binding sites, and TGZ induces Egr-1 binding activity to Egr-1 consensus sites as assessed by gel shift assay. In addition, TGZ induces ERK-dependent phosphorylation of PPARγ, resulting in the down-regulation of PPARγ activity. The fact that TGZ-induced apoptosis is accompanied by the biosynthesis of Egr-1 suggests that Egr-1 plays a pivotal role in TGZ-induced apoptosis in HCT-116 cells. Our results suggest that Egr-1 induction is a unique property of TGZ compared with other PPARγ ligands and is independent of PPARγ activation. Thus, the up-regulation of Egr-1 may provide an explanation for the anti-tumorigenic properties of TGZ.

There are several known ligands for PPAR␥, including the natural prostaglandin 15-deoxy-⌬ 12,14 -prostaglandin J 2 (PGJ 2 ), the synthetic anti-diabetic thiazolidinediones, and certain polyunsaturated fatty acids. PPAR␥ ligands are able to bind to the PPAR␥ transcription factor, which then forms a heterodimeric complex with retinoid X receptor that functions as a central regulator of differentiation, and modulator of cell growth. Many reports present evidence for anti-tumorigenic activity of PPAR␥ ligands (5, 6, 8 -10). Among PPAR␥ ligands, the antitumorigenic activity of troglitazone (TGZ) has been well established. For example, TGZ significantly inhibits tumor growth of human colorectal cancer cells (HCT-116), human breast cancer cells (MCF-7), and human prostate cancer cells (PC-3) in immunodeficient mice (11)(12)(13). However, the molecular mechanism of TGZ effects on anti-tumorigenesis, other than PPAR␥ activation, is not known.
The Egr-1 transcription factor (also known as NGFI-A, TIS8, Krox-24, and Zif268) is a member of the immediate early gene family and encodes a nuclear phosphoprotein involved in the regulation of cell growth and differentiation in response to signals such as mitogens, growth factors, and stress stimuli. However, many reports have recently suggested Egr-1 as a tumor suppressor gene (20). Egr-1 activates PTEN (phosphatase and tensin homolog) tumor suppressor gene during UV irradiation (21), and re-expression of Egr-1 suppresses the growth of transformed cells both in soft agar and in athymic nude mice (22). Egr-1 is induced very early in the apoptotic process, where it mediates the activation of downstream regulators such as p53 (23)(24)(25). However, Egr-1-induced apoptosis has been reported in p53Ϫ/Ϫ cells (26), indicating that Egr-1 induction occurs in both p53-dependent and p53-independent manners. Moreover, Egr-1 is down-regulated in several types of neoplasia, as well as in an array of tumor cell lines (27,28).
These results indicate that Egr-1 plays a consistent role in growth suppression. Therefore, it is reasonable to think that Egr-1 could be regulated, at least in part by TGZ, because both TGZ and Egr-1 have anti-tumorigenic effects.
In the present study, we examine the relationship between TGZ and Egr-1 expression and the effect of TGZ-induced apoptosis. We show that TGZ induces Egr-1 expression by transcriptional and post-transcriptional regulation. Egr-1 induction by TGZ results in the increase of binding affinity and transactivation of the promoter containing Egr-1 consensus sequences, thereby possibly inducing other anti-tumorigenic proteins. Furthermore, Egr-1 induction by TGZ appears to be independent of PPAR␥, because other PPAR␥ ligands do not induce Egr-1. These data provide a novel mechanism for understanding how troglitazone exerts its anti-tumorigenic activity.
Apoptosis and Soft Agar Cloning Assay in the Presence of TGZ-The DNA content for sub-G 1 population was determined by flow cytometry. HCT-116 cells were plated at 3 ϫ 10 5 cells/well in 6-well plates, incubated for 16 h, and then treated with TGZ at different time points in the presence of serum. The cells (attached and floating cells) were then harvested, washed with phosphate-buffered saline, fixed by the slow addition of cold 70% ethanol to a total of 1 ml, and stored at 4°C overnight. The fixed cells were pelleted, washed with ethanol (50%, then 30%), followed by phosphate-buffered saline, and stained in 1 ml of 20 g/ml propidium iodide containing 1 mg/ml RNase in phosphatebuffered saline for 20 min. 7,500 cells were examined by flow cytometry using BD Biosciences FACSort equipped with CellQuest software by gating on an area versus width dot plot to exclude cell debris and cell aggregates. Apoptosis was measured by the level of sub-diploid DNA contained in cells following treatment with compounds using CellQuest software.
Soft agar assays were performed to compare the clonogenic potential of HCT-116 cells in semisolid medium. HCT-116 cells were resuspended at 10,000 cells in 2 ml of 0.4% agarose-containing TGZ in McCoy's 5A medium and plated on top of 1 ml of 0.8% agarose in 6-well plates. Plates were incubated for 2-3 weeks at 37°C. Cell colonies were visualized by staining with 0.5 ml of p-iodonitrotetrazolium violet staining (Sigma).
Transfection and Luciferase Assay-HCT-116 cells were plated in 6-well plates at 2 ϫ 10 5 cells/well in McCoy's 5A medium supplemented with 10% fetal bovine serum. After growth for 16 h, plasmid mixtures containing 0.5 g of promoter linked to luciferase, 0.5 g of expression vector, and 0.05 g of pRL-null (Promega) were transfected by Lipo-fectAMINE (Invitrogen) according to the manufacturer's protocol. After 48 h of transfection, the cells were harvested in 1ϫ luciferase lysis buffer, and luciferase activity was determined and normalized to the pRL-null luciferase activity using a dual luciferase assay kit (Promega). For the PPAR␥ ligand treatments, cells were treated with PPAR␥ ligand in the absence of serum for 24 h and assayed for luciferase activity.
RNA Stability in the Presence of Actinomycin D-When reaching 60 -80% confluence in 10-cm plates, the cells were grown in the absence of serum for 24 h and then treated with either vehicle or TGZ (5 M) for 2 h. The transcription inhibitor, actinomycin D (5 g/ml), was treated at the indicated time points. Total RNAs were isolated using TRIzol reagent (Invitrogen) according to the manufacturer's protocol. For Northern blot analysis, 20 g of total RNA was denatured at 55°C for 15 min and separated in a 1.0% agarose gel containing 2.2 M formaldehyde and transferred to Hybond-N membrane (Amersham Biosciences). After fixing the membrane by UV, blots were prehybridized in hybridization solution (Rapid-hyb buffer; Amersham Biosciences) for 1 h at 65°C followed by hybridization with cDNA labeled with [␣-32 P]dCTP using random primer extension (DECAprimeII kit; Ambion). The probes used were full-length human Egr-1 fragment. After 4 h of incubation at 65°C, the blots were washed once with 2ϫ SSC/0.1% SDS at room temperature and twice with 0.1ϫ SSC/0.1% SDS at 65°C. Messenger RNA abundance was estimated by intensities of the hybridization bands of autoradiographs using Scion Image (Scion Image Co.). Equivalent loading of RNA samples was confirmed by hybridizing the same blot with a 32 P-labeled glyceraldehyde-3-phosphate dehydrogenase probe, which recognizes RNA of ϳ1.3 kb.
Western Blot Analysis-The level of protein expression was evaluated using Western blot analysis with Egr-1, Egr-2, Egr-3, PPAR␥1, PTEN, and phospho-ERK antibodies. Cells were grown to 60 -80% confluency in 10-cm plates followed by 24 h of additional growth in the absence of serum. Cells were treated with indicated compounds, and total cell lysates were isolated using 0.1 M Tris/pH 8.0 containing proteinase inhibitors (Sigma). After sonication of samples, proteins (30 g) were separated by SDS-PAGE and transferred for 1 h onto nitrocellulose membrane (Schleicher & Schuell). The blots were blocked overnight with 5% skim milk in TBS-T (Tris-buffered saline/Tween 0.05%), and probed with each antibody for 2 h at room temperature. After washing with TBS-T, the blots were treated with horseradish peroxidase-conjugated secondary antibody for 1 h and washed several times. Proteins were detected by the enhanced chemiluminescence system (Amersham Biosciences). Western analysis for p53 was described previously (29).
Preparation of Nuclear Extracts and Electrophoretic Mobility Shift Assay (EMSA)-Nuclear extracts were prepared as described previously (30). For the gel shift assay, double stranded oligonucleotides were end-labeled with [␥-32 P]ATP by T4 polynucleotide kinase (New England Biolabs). Assays were performed by incubating 10 g of nuclear extracts in the binding buffer (Geneka Biotechnology) containing 200,000 cpm of labeled probe for 20 min at room temperature. To assure the specific binding of transcription factors to the probe, the probe was chased by 50-fold molar excess of cold wild type or mutant oligonucleotide. For the supershift experiments, Egr-1 antibody (Geneka Biotechnology) was incubated with nuclear extracts on ice for 30 min before adding to the binding reaction. Samples were then electrophoresed on 5% nondenaturing polyacrylamide gels with 0.5ϫ TBE (Tris/borate/ EDTA), and gels were dried and subjected to autoradiography.
Nuclear Run-on-HCT-116 cells were grown in the absence of serum for 24 h and treated with 5 M TGZ or vehicle for 3 h. Nuclei were isolated as described previously (30). In vitro nuclear run-on transcription was carried out using 1 ϫ 10 7 nuclei and 250 Ci of [␣-32 P]UTP in transcription-optimized buffer (P118A; Promega). The reaction was performed at 30°C for 30 min. Labeled transcripts were purified using TRIzol reagent (Invitrogen) according to manufacturer's protocol. A total of 1 ϫ 10 7 cpm elongated nascent RNA per assay was hybridized for 24 h at 65°C to filter-immobilized 5 g of plasmid DNA. The filters were washed with 2ϫ SSC/0.1% SDS for 20 min, followed by 0.1ϫ SSC/0.1% SDS for 40 min. The autoradiographs were subjected to densitometric analysis using Scion Image.

Troglitazone Inhibits Clonogenic Growth and Induces Apoptosis in HCT-116 Cells-
We examined the correlation between TGZ-induced apoptosis and TGZ-induced anti-tumorigenesis in HCT-116 cells. Anti-tumorigenic activity was measured by soft agar assay, whereas apoptosis was measured by flow cytometry. As shown in Fig. 1A, TGZ treatment in HCT-116 cells dramatically inhibited the growth of HCT-116 cellsonsoftagar.ThegrowthinhibitionbyTGZwasconcentrationdependent, and 5 M TGZ completely inhibited the growth of HCT-116 cells in soft agar assays. To determine whether TGZ induces cell cycle arrest and/or apoptosis in HCT-116 cells, flow cytometry analysis was performed (Fig. 1B). Apoptosis and G 1 cell cycle arrest were observed as early as 12 h after treatment with 5 M TGZ. The stimulation of apoptosis and G 1 cell cycle arrest by TGZ are consistent with previous publications (8,31) reporting that TGZ induces apoptosis and cell cycle arrest in human colon and breast cancer cells.
Expression of Anti-tumorigenic Proteins Mediated by TGZ-One logical mechanism by which TGZ exerts anti-tumorigenesis is the up-regulation of anti-tumorigenic proteins. To address this question, we measured the expression of known anti-tumorigenic proteins, Egr-1, PTEN, and p53. HCT-116 cells were treated with 5 M TGZ at indicated time points, and Western analysis was performed. As shown in Fig. 2A, Egr-1 is induced dramatically within 3 h, and longer treatment, up to 24 h, results in a decrease of Egr-1 expression. TGZ-induced Egr-1 expression was also seen in Northern analysis using HCT-116 cells treated with 5 M TGZ for 2 h (data not shown). We also measured PTEN and p53 tumor suppressor gene expression. PTEN is only marginally induced at an early time point, whereas p53 expression is not altered by TGZ in HCT-116 cells. Taken together, these results suggest that Egr-1 induction by TGZ may be pivotal to the anti-tumorigenic activity of TGZ. Because TGZ is a PPAR␥ ligand, we next compared TGZ to other PPAR␥ ligands with regard to Egr-1 induction. HCT-116 cells were treated with several PPAR␥ ligands, BRL, PGJ 2 , PAF, ciglitazone, and 13-hydroxyoctadecadienoic acid, for 3 h. These PPAR␥ ligands are reported to bind and activate PPAR␥ transcription factor (32,33). Consistent with previous data, Egr-1 induction was seen in TGZ-treated cells, but poor or no induction was observed in other PPAR␥ ligand-treated cells (Fig. 2B), suggesting that Egr-1 induction by TGZ may be independent of PPAR␥. In addition, the expression of other Egr family proteins, Egr-2 and Egr-3, was not altered by TGZ and other PPAR␥ ligands, indicating that the effect of TGZ is specific for Egr-1.
TGZ Induces DNA Binding Affinity of Egr-1-To address whether Egr-1 protein, produced by TGZ-treated HCT-116 cells, has potential binding activity to the Egr-1 consensus sequences, EMSA was performed. Because Egr-1 protein increases in the presence of TGZ, the protein should bind to the Egr-1 consensus binding site. As shown in Fig. 3A, oligonucleotides containing two copies of Egr-1 binding sites were generated and used as a probe. Nuclear extracts prepared from 5 M TGZ treatment at different time points (Fig. 3B, lanes 2-5) and the recombinant proteins, Egr-1, Egr-2, and Egr-3, generated by in vitro translation (Fig. 3B, lanes 6 -8), were used for the EMSA. Using nuclear extracts from TGZ-treated HCT-116 cells and a probe corresponding to the Egr-1 binding site, results show multiple DNA⅐protein complexes with a mobility shift (Fig. 3B, arrows ␣ and ␤). Compared with in vitro synthesized Egr proteins, the shifted band ␣ correspond to Egr-1 proteins, whereas ␤ indicates Egr-3 proteins. Interestingly, a shifted band representing Egr-1 was seen only at the 3-h time point, whereas Egr-3 was constitutively expressed during the time course. A shifted band representing Egr-2 was not detected in TGZ-treated HCT-116 cells. These results are consistent with Egr-1 induction at the 3-h time point after TGZ treatment as assessed by Western analysis. These bands represent a specific protein binding to the Egr-1 sequence elements, because complex formation was diminished by the addition of 50 molar excess of non-radiolabeled identical competitor but not by addition of the identical oligonucleotide in which the Egr-1 sites were point-mutated (Fig. 3C, lanes 2 and 3). To confirm that Egr-1 binds to these sites, we performed a gel shift assay in the presence of Egr-1 antibody to demonstrate supershifting. As shown in Fig. 3C, the Egr-1 antibody supershifted the band (lanes 4 and 6). This indicates that the shifted bands (SS) contain Egr-1 protein. As a control, we examined Egr-1 binding affinity using nuclear extracts from 12-O-tetradecanoylphorbol-13-acetate-treated K562 cells and observed similar results as with the nuclear extracts from TGZ-treated HCT-116 cells (Fig. 3C, lanes 5 and 6). It has been shown that Egr-1 is induced FIG. 2. TGZ induces anti-tumorigenic proteins, Egr-1 and PTEN, but not p53. A, HCT-116 cells were grown in the absence of serum for 18 h and then treated with 5 M TGZ for indicated times. The cell lysates were isolated and subjected to Western analysis. Protein levels of Egr-1, PTEN, and p53 were measured using specific antibodies described under "Materials and Methods." Actin antibody was used for loading control.  B, gel shift assay was performed using a NUSHIFT Egr-1 kit (Geneka Biotechnology) with TGZ-treated HCT-116 nuclear extracts. Nuclear extracts were prepared from HCT-116 cells after TGZ treatment. Ten g of nuclear extracts from the indicated time points were incubated with a 32 P-labeled double-stranded oligonucleotide corresponding to the two Egr-1 binding sites (lanes 2-5). As a control, gel shift assay was performed using in vitro translated (IVT) Egr-1, Egr-2, and Egr-3 proteins (lanes 6 -8). Arrow ␣ indicates Egr-1, whereas arrow ␤ indicates Egr-3. C, competitions were done in the presence of 50 molar excess of non-radiolabeled oligonucleotide corresponding to the wild type (Wt) or mutant Egr-1 oligonucleotide (Mu) shown in A. The binding reactions were resolved by 5% nondenaturing acrylamide electrophoresis. Supershift assays were performed by a 30-min pre-incubation of the reaction mixture with 2 g of Egr-1 antibody (Ab) (Geneka Biotechnology), prior to the addition of radiolabeled probe. The arrows indicate shifted bands, whereas SS indicates supershifted bands. Nuclear extracts (NE) were used from two different cell lines, H for 3 h of TGZ-treated HCT-116 cells, and K for 12-O-tetradecanoylphorbol-13-acetate-treated K562 cells.
TGZ Induces the Transactivation of Egr-1-The transactivation activity of Egr-1 in TGZ-treated HCT-116 cells was determined using an Egr-1-responsive reporter. The plasmid pEBS1 4 luc contained four copies of Egr-1 response elements linked to the basal promoter followed by a luciferase reporter gene (Fig. 4A) (35). To determine whether the expression of Egr-1, Egr-2, and Egr-3 proteins could activate the luciferase reporter, the plasmid pEBS1 4 luc was transfected into HCT-116 cells, in combination with empty, Egr-1, Egr-2, or Egr-3 expression vectors. As shown in Fig. 4B, the luciferase activity was increased by overexpression of Egr-1, Egr-2, and Egr-3 compared with vector-transfected cells, suggesting that overexpression of Egr family proteins is able to bind and transactivate the reporter vector. Subsequently, to determine whether TGZ induced Egr-1 expression would also transactivate the reporter vector containing Egr-1 binding sites, we transfected the reporter vector and treated with varying concentrations of TGZ. Indeed, TGZ induced luciferase activity in a concentration-dependent manner, with 4-and 12-fold induction observed with 10 and 20 M TGZ treatment, respectively. Because TGZ is a PPAR␥ ligand, we tested whether other PPAR␥ ligands would also induce luciferase activity. We examined BRL, PGJ 2 , and PAF, which have been known to bind to PPAR␥. TGZ is the strongest Egr-1 inducer of luciferase activity (Fig. 4D), although the other compounds are better PPAR␥ ligands in terms of PPAR␥ binding activity. However, in this system, 10 M TGZ is required to see any significant luciferase induction, compared with 5 M TGZ being used for Western and Northern analyses. This system is an artificial construct and may require higher concentration of TGZ to see effects. Taken together with previous results, these data demonstrate that TGZ not only induces Egr-1 expression and binding activity but also transactivates Egr-1 responsive genes.
TGZ Induces Egr-1 at the Transcription Level but Not Other PPAR␥ Ligands-We performed nuclear run-on experiments to examine whether the Egr-1 induction by TGZ is regulated at a transcriptional level. HCT-116 cells were treated with serumfree medium for 24 h followed by treatment with TGZ or vehicle for 3 h. Radioactive-labeled nascent transcripts were analyzed by hybridization to immobilized DNAs. Egr-1 gene transcription at 3 h after TGZ treatment was increased 2-fold, as determined by triplicate independent experiments. The gene for Sp1 transcription factor was used as an internal control, because Sp1 expression is not altered by TGZ treatment in HCT-116 cells (data not shown). Thus, Egr-1 transcripts were induced by TGZ at the transcriptional level. Fig. 5A is a representative autoradiogram of three experiments. To confirm whether TGZ induces Egr-1 at the transcription level, the Egr-1 promoter was cloned into the luciferase reporter gene. A plasmid, pEgr1260/LUC, was transfected into HCT-116 cells, and several PPAR␥ ligands were treated. As shown in Fig. 5B, TGZ enhances Egr-1 promoter activity 2-fold, whereas the other PPAR␥ ligands do not enhance promoter activity significantly. Taken together with the nuclear run-on experiments, these results suggest that TGZ induces Egr-1 expression at the transcription level by at least 2-fold. However, these data do not fully explain the dramatic induction shown in Western analysis (Fig. 2), indicating that TGZ may be involved with other mechanisms of Egr-1 induction.
TGZ Induces Egr-1 at the Post-transcription Level-Because TGZ affects minimum induction at the Egr-1 promoter, and Egr-1 proteins were dramatically induced in HCT-116 cells, it is possible that TGZ may increase Egr-1 mRNA stability. HCT-116 cells were treated with either TGZ or vehicle for 2 h and then 5 g/ml of actinomycin D was added at the indicated time points. Fig. 6, A and B demonstrates that the half-life of Egr-1 mRNA in control HCT-116 cells was ϳ15 min compared with 48 min in TGZ-treated cells. These results suggest that an increase in stability of Egr-1 mRNA is also a major factor in the induction of Egr-1 proteins by TGZ. Therefore, TGZ-induced Egr-1 promoter activity and increased Egr-1 mRNA stability may explain the dramatic induction of Egr-1 at the protein level.
Effect of TGZ on ERK1/2 Activation, PPAR␥ Inactivation, and Egr-1 Induction-TGZ activates ERK1/2 activity in smooth muscle cells (19), and the activated ERK pathway induces Egr-1 activity (36,37). In addition, ERK phosphorylates PPAR␥, which results in the inactivation of PPAR␥ activity (38 -40). Therefore, we speculated that TGZ might increase ERK1/2 phosphorylation/activation, resulting in increased expression of Egr-1. In addition, the increase in ERK activity by TGZ would down-regulate PPAR␥ activity. The dual phosphorylation of threonine and tyrosine residues, which is necessary for ERK activation, was evaluated using the anti-phosphorylated ERK antibody. To test whether TGZ induces ERK1/2 activation in human colorectal HCT-116 cells, we treated with 5 M TGZ at the indicated time points and measured the phosphorylation of ERK1/2 using a phosphospecific antibody. Based on Western blot analysis with an anti-phospho-ERK antibody, TGZ activates ERK1/2 as early as 30 min after TGZ treatment in HCT-116 cells and then decreases at later time points (Fig. 7A). However, total levels of ERK proteins did not change. To examine whether PPAR␥ phosphorylation increases concomitant with ERK1/2 activation, PPAR␥ phosphorylation and expression were examined. PPAR␥ antibody recognizes both phosphorylated and nonphosphorylated forms of PPAR␥1 or PPAR␥2. However, HCT-116 cells express only PPAR␥1 (40). TGZ induced maximum PPAR␥1 phosphorylation at 30 min after TGZ treatment, as measured by a shift in PPAR␥1 mobility (Fig. 7B). This result is consistent with a previous report that PPAR␥ is phosphorylated after ERK activation in HCT-116 cells (40). The increase in ERK activation could also be responsible, in part, for the Egr-1 induction. To investigate whether the activation of ERK1/2 is associated with Egr-1 induction, HCT-116 cells were pre-treated with protein kinase inhibitors, PD98059 (mitogen-activated protein kinase/extracellular signal-regulated kinase kinase inhibitor), SB203580 (p38 MAPK inhibitor), and herbimycin (tyrosine kinase inhibitor), for 1 ⁄2 h, followed by 3 h of TGZ treatment. A significant induction of Egr-1 was detected as early as 3 h after the initiation of treatment of HCT-116 cells with 5 M TGZ. However, the Egr-1 induction by TGZ was abolished by pre-treatment with PD98059 (Fig. 7C, lane 6) but not by the pretreatment of SB203580 or herbimycin A (Fig. 7C, lanes 7 and   FIG. 5. Egr-1

is selectively and transcriptionally induced by TGZ in HCT-116 cells.
A, determination of Egr-1 transcription levels by nuclear run-on experiment. HCT-116 cells were grown in the absence of serum for 24 h and treated with either 5 M TGZ or vehicle (0.2% Me 2 SO) for 3 h. The nuclei were isolated, and nuclear run-on assay was performed using 1 ϫ 10 7 nuclei and 250 Ci of [␣-32 P]UTP at 30°C for 30 min. Labeled transcripts were purified and hybridized to membrane containing 5 g of plasmid DNA. Sp1 transcripts were used as internal control. The data shown here are representative of three independent experiments performed in triplicate (p Ͻ 0.003, Student's t test). B, Egr-1 promoter activity in the presence of several PPAR␥ ligands. Human Egr-1 promoter was cloned as described under "Materials and Methods." HCT-116 cells were transfected with pEgr1260/LUC and then serum-starved for 24 h. Several PPAR␥ ligands were treated for 24 h, and luciferase activity was measured. The internal control vector (pRL-null) was used to normalize for transfection efficiency. The data represent means Ϯ S.D. from three different experiments. The concentration of PPAR␥ ligands used were the same as for Fig. 2B. 8), indicating that Egr-1 induction by TGZ is dependent on the ERK/MAPK signaling pathway. Furthermore, the reporter construct containing four copies of Egr-1 responsive elements was used to measure luciferase activity after protein kinase inhibitor and TGZ treatment. As shown in Fig. 7D, PD98059 pretreatment in HCT-116 cells inhibits TGZ-induced Egr-1 activation as assessed by the luciferase assay. This result is consistent with Western analysis for Egr-1 expression as shown in Fig. 7C and indicates that TGZ-induced Egr-1 transactivity is dependent on the ERK pathway.

DISCUSSION
Although the anti-tumorigenic activities of PPAR␥ ligands are well established in human cancer, controversy exists in the literature with regard to the relative contributions of nuclear receptor-dependent (13,41,42) and -independent mechanisms (43)(44)(45). In this report, we demonstrate that troglitazone, a PPAR␥ agonist, induces human colorectal cancer cells to undergo apoptosis and inhibits their growth on soft agar. In an effort to better understand the mechanisms responsible for these cellular responses, we measured the expression of several tumor suppressor proteins and found that TGZ but not other PPAR␥ ligands stimulates the expression of Egr-1, a transcription factor involved in cell growth. TGZ uniquely stimulates the ERK pathway that down-regulates the PPAR␥ receptor activity (Fig. 7B), indicating that the increased expression of Egr-1 is not mediated by the activity of PPAR␥ receptor (Fig. 8). Furthermore, TGZ-dependent expression of Egr-1 is attenuated by inhibition of ERK activation indicating that ERK plays a pivotal role to induce TGZ-induced Egr-1 expression.
TGZ differs from other thiazolidinediones in that it has the ability to induce Egr-1 (Fig. 2B). Of the PPAR␥ agonists reported to bind to PPAR␥ and activate target gene expression, TGZ is the only ligand to induce Egr-1 significantly. One explanation for this difference is that TGZ contains a vitamin E moiety, which is a unique feature, compared with the other PPAR␥ ligands (18). One could think that TGZ may exert its Egr-1 induction by virtue of its antioxidant property. However, vitamin E treatment in HCT-116 cells resulted in no significant induction of Egr-1 expression up to 100 M concentration (data not shown), suggesting that the Egr-1 induction by TGZ is not dependent on antioxidant effect of TGZ. Thus, we propose the model illustrated in Fig. 8. TGZ exerts it anti-tumorigenic activity via the activation of the ERK signaling pathway that, in turn, down-regulates the PPAR␥ receptor. However, the ERK pathway up-regulates the expression of the Egr-1 transcription protein that elicits the anti-tumorigenic action. In contrast, other PPAR␥ agonists that do not stimulate ERK pathway but instead activate the PPAR␥ receptor and alter the expression of anti-tumorigenic proteins. Thus, several mechanisms appear to be responsible for the anti-tumorigenic activity of PPAR␥ agonists, both dependent and independent of the PPAR␥ nuclear receptor. It also implies that a different family of proteins may be responsible for the anti-tumorigenic activities of the various PPAR␥ ligands.
Evidence is presented to support that up-regulation of Egr-1 by TGZ occurs by at least two mechanisms; one is at the transcriptional level, and the other is at the post-transcriptional level. The combination of these two mechanisms appears to be responsible for the dramatic increase in the level of Egr-1 protein observed after treatment with TGZ. The fact that Egr-1 induction by TGZ is regulated at the transcriptional level (Fig.  5) suggests that the Egr-1 promoter may contain a binding site downstream of ERK. Indeed, the human Egr-1 promoter contains five serum response elements that mediate signal-induced activation of Egr-1 gene transcription via the ternary response factor Elk1, which is an established substrate for ERK (35). Thus, ERK activation results in the activation of Elk1, which may induce Egr-1 expression. In addition, TGZ also increases mRNA stability of Egr-1 3-fold (Fig. 6). Stability of mRNA can be mediated by several mechanisms. The control of mRNA stability can involve A/U-rich elements (ARE) in the 3Ј untranslated region or specific RNA stem loop motifs (46). Increased stability might be because of a reduction in the level of proteins that destabilize Egr-1 mRNA or specific mRNA-binding proteins, including AUF1 (47). Interacting with ARE may protect Egr-1 mRNA from degradation by endo-and exonuclease, thereby increasing RNA stability. In the present study, TGZ-mediated induction of Egr-1 is dependent on ERK1/2 but not p38 MAPK or tyrosine kinase, because PD98059 inhibits the TGZ-induced Egr-1 expression. Although p38 MAPK has been reported to play a major role in the induction of mRNA stabilization through ARE in the 3Ј untranslated region (48), p38 MAPK pathway may not involve Egr-1 RNA stability in HCT-116 cells, because the inhibition of p38 MAPK pathway does not abolish the TGZ-induced Egr-1 expression (Fig. 7C). Recent studies demonstrate that ERK is also involved in the induction of mRNA stability (49,50). Therefore, ERK or its downstream kinases are considered to affect the proteins that bind to the ARE and induce stabilization. Egr-1 has at least one ARE sequence in the 3Ј untranslated region. However, further studies are required to identify the exact molecular mechanism by which ERK activation induces the stabilization of Egr-1 mRNA and increases the transcriptional activity of the Egr-1 promoter.
Egr-1 is a member of the immediate early gene response family and encodes a nuclear phosphoprotein involved in the FIG. 6. Effect of TGZ on the stability of Egr-1 mRNA. A, HCT-116 cells were treated with TGZ (5 M) or vehicle (Me 2 SO) for 2 h and subsequently with actinomycin D (5 g/ml). At the indicated times, total RNAs were isolated and examined by Northern blot analysis with radiolabeled probe for human Egr-1 and glyceraldehyde-3-phosphate dehydrogenase (GAPDH). Membrane from vehicle-treated samples (left panel) was exposed to x-ray film for 20 h, whereas membrane from TGZ-treated samples (right panel) was exposed for 2 h. B, hybridization signals were quantitated with the Scion Image program. The relative level of Egr-1 mRNA (relative to the level of glyceraldehyde-3-phosphate dehydrogenase) was calculated, and the results were plotted as the ratio of the mRNA level present at time 0 of actinomycin D treatment.
regulation of cell growth and differentiation in response to signals such as mitogens, growth factors, and stress stimuli. However, evidence was presented recently (20,21,35,51) that Egr-1 is a pro-apoptotic protein, and the overexpression of Egr-1 results in increased susceptibility to apoptosis when induced by an apoptotic agent (25). The molecular mechanism by which Egr-1 induces apoptosis and/or anti-tumorigenesis has not been studied, but Egr-1 can be considered as a tumor suppressor gene (52). One logical mechanism by which Egr-1 exerts anti-tumorigenesis is the transcriptional up-regulation of anti-tumorigenic genes. Several known anti-tumorigenic proteins including PTEN are induced by Egr-1 expression (21). PTEN is induced modestly by TGZ along with robust Egr-1 expression (Fig. 2A). Egr-1 up-regulates transcription of the p53 tumor suppressor directly (25), but we do not observe the p53 induction after Egr-1 induction by TGZ in HCT-116 cells. FIG. 8. Schematic diagram of TGZ effects on anti-tumorigenic activity. TGZ affects several pathways shown in this study. TGZ activates PPAR␥ as a PPAR␥ agonist, thereby many PPAR␥ responsive genes including anti-tumorigenic genes are induced. TGZ also activates ERK phosphorylation, followed by the phosphorylation of PPAR␥. The phosphorylation of PPAR␥ leads to an inactivation of PPAR␥ activity. In addition, TGZ induces Egr-1 expression by ERK activation, which leads to induction of the Egr-1 promoter activity and stabilization of the Egr-1 mRNA. Therefore, the balance of these pathways may be an important factor to determine the TGZ effects on cancer.
Thus, the increased expression of this transcription factor, Egr-1, appears to play a role in the anti-tumorigenic activity of TGZ, but further evidence is required to fully support that conclusion.
The anti-tumorigenic effect of TGZ is not well understood, and there appears to be some dependence on species studied. Whereas TGZ significantly inhibits tumor growth of human tumors in, for example, HCT-116 colorectal cancer cells, MCF-7 breast cancer cells, and PC-3 prostate cancer cells in immunodeficient mice (11)(12)(13), TGZ stimulates colon polyp formation in Min mouse (53,54). In our study, TGZ also induces Egr-1 in mouse colorectal cancer cell, CMT-93 (data not shown), suggesting that Egr-1 induction by TGZ may be, at least, a common effect in both human and mouse cells. There are several ongoing attempts to unravel this puzzle. One hypothesis is that human and mouse cells use different downstream signaling molecules, resulting in these two distinct outcomes. TGZ-induced apoptosis is mediated by ERK activation (55), whereas TGZ treatment in skeletal muscle cells does not alter ERK activity (56). Although the ERK pathway is linked to cell proliferation and tumorigenic activity, recent studies have shown that ERK activation can lead to arrest of cell growth by the activation of p21 cyclin-dependent kinase inhibitor. Because Egr-1 expression can be pro-tumorigenic or anti-tumorigenic, depending on target genes, TGZ may have opposite effects depending on the downstream targets of the Egr-1 pathway. Another hypothesis is that the cells are actually using the same signaling pathways, but then the cells selectively utilize different cofactors, which direct the basic signaling molecules to the appropriate final targets. In this regard, there are two Egr-1binding proteins, NAB1 and NAB2 (57,58), that function as transcriptional repressors of Egr-1. Therefore, either pro-apoptotic or anti-apoptotic effects of TGZ may be determined by cofactor expression. The competition and usage of these two proteins in the same region may be one of the determinations for the TGZ-induced anti-tumorigenic activity.
In conclusion, we have shown that the PPAR␥ ligand TGZ, independent of the nuclear receptor, stimulates the expression of the transcription factor, Egr-1, a protein with established tumor suppressor activity. The expression is mediated by enhancement of ERK signaling pathway and occurs via both transcriptional and post-transcriptional mechanisms. Further investigations are required to elucidate an understanding of how changes in Egr-1 expression mediate the anti-tumorigenic activity of TGZ.