Peroxisome proliferator-activated receptor gamma activation modulates cyclin D1 transcription via beta-catenin-independent and cAMP-response element-binding protein-dependent pathways in mouse hepatocytes.

Activation of peroxisome proliferator-activated receptor gamma (PPARgamma) following exposure to PPARgamma-specific ligands resulted in growth inhibition in various carcinoma cell lines. Our aim was to elucidate the pathway of PPARgamma2 activation-mediated modulation of cyclin D1 transcription in mouse hepatocytes. To address this we utilized stable control and PPARgamma hepatocyte cell lines created via retroviral overexpression utilizing AML-12 hepatocytes. Addition of PPARgamma ligand troglitazone (TZD) activated PPARgamma2 in proliferating hepatocytes and resulted in growth arrest accompanied by a down-regulation of proliferating cell nuclear antigen, cyclin D1, and beta-catenin expression. Furthermore activation of PPARgamma2 attenuated cyclin D1 promoter activity indicating a transcriptional regulation of cyclin D1. Since beta-catenin plays a pivotal role in regulating cyclin D1 transcription, we studied whether PPARgamma2-mediated inhibition of cyclin D1 transcription involved beta-catenin. Interestingly overexpression of either wild-type or S37A mutant beta-catenin was unable to rescue PPARgamma2-mediated suppression of cyclin D1 transcription, whereas overexpression of cAMP-response element-binding protein (CREB) was capable of antagonizing this inhibitory effect of PPARgamma2. Additionally pretreatment with okadaic acid antagonized PPARgamma2-mediated inhibition of cyclin D1 transcription without any effect on beta-catenin expression. These studies also showed a TZD-mediated inhibition of total and phospho-CREB(Ser133) levels, CREB promoter activity, and cAMP-response element-mediated transcription in PPARgamma hepatocytes. Pretreatment of PPARgamma hepatocytes with okadaic acid, however, maintained higher total and phospho-CREB(Ser133) levels in the presence of TZD. These results indicated that PPARgamma2 activation inhibited cyclin D1 transcription in hepatocytes via CREB-dependent and beta-catenin-independent pathways.

Peroxisome proliferator-activated receptors (PPARs) 1 belong to the nuclear receptor superfamily and are involved in regulating various cellular processes. The three known members of the PPAR family, PPAR␣, PPAR␦, and PPAR␥, can regulate transcription of genes in response to specific ligands (1)(2)(3). The PPAR␥ subfamily consists of two protein isoforms, PPAR␥1 and PPAR␥2, with the former being expressed in a wide variety of tissues and the latter being preferentially expressed in adipose tissue (4). PPAR-mediated transactivation of target genes involves dimerization of PPARs with the retinoid X receptor and binding of the resulting heterodimer to specific peroxisome proliferator-activated receptor response elements (PPREs) located within the target gene promoters/enhancers (5). This process of transcriptional activation also involves the recruitment of different coactivator molecules to specify select target gene activation, which include p300 (6), the SRC-1 class of coactivators (7), PGC-1 and PGC-2 (8,9), ARA70 (10), and DRIP205 (or TRAP220) (11).
In the preadipocytes PPAR␥ is known to orchestrate the process of differentiation into adipocytes via regulation of the transcription of target genes, which requires a prior exit from the cell cycle (12). In addition, PPAR␥ activation can inhibit proliferation in a wide variety of cell types (for a review, see Ref. 13) including those originating from hepatocellular carcinoma (14 -16). The mechanism(s) by which PPAR␥ regulates cell growth is still unclear, although it mostly involves a G 1 /Sspecific arrest (14,17). In fibroblasts PPAR␥-mediated growth arrest was associated with reduced transcriptional activity of E2F⅐DP, which was a consequence of the down-regulation of serine-threonine phosphatase PP2A expression (18). Similarly, in fibroblastic cell lines, activation of PPAR␥ resulted in induction of the cyclin-dependent kinase inhibitors p21 CIP1 and p18 INK4c (19) accompanied by differentiation into adipocytes. Additionally PPAR␥ activation in vascular smooth muscle cells resulted in G 1 /S arrest associated with reduced Retinoblastoma phosphorylation and modulation of cyclin-dependent kinase inhibitor levels (17). PPAR␥ activation in the MCF-7 breast cancer cells attenuated cyclin D1 transcription due to a decreased availability of p300 (20).
The normal cell growth cycle is tightly regulated via a complex interplay of multiple regulatory proteins. The transition from G 1 to S phase is regulated by the controlled expression of cyclins, particularly cyclins D1 and E, which can bind and activate the corresponding cyclin-dependent kinases (21). In hepatocytes, ectopic expression of cyclin D1 can drive cells through G 1 to S phase in a growth factor-independent manner (22), and chronic expression of cyclin D1 is sufficient to initiate hepatocellular carcinoma (23). Therapeutic interventions targeted toward regulation of cyclin D1 expression might thus be an effective means of controlling growth. The capability of the PPAR␥ ligands to regulate cyclin D1 expression indicates the possibility of utilizing these ligands as chemotherapeutic drugs. Detailed studies of the mechanism(s) by which PPAR␥ attenuates cyclin D1 expression are required to achieve this goal. Apart from the NFB and AP-1 (24,25) family of transcription factors, transcription from the cyclin D1 gene can also be induced following activation of the Wnt signaling pathway via a ␤-cateninand TCF-mediated transactivation (26,27). The cyclic AMP-response element (CRE) is another key site in the cyclin D1 promoter regulating cyclin D1 expression (28) particularly in hepatocytes (29).
PPAR␣ and PPAR␥ are expressed together in several tissues, and their relative abundance changes in response to physiological effectors and disease (30,31). Although expressed at lower levels in normal liver, hepatic expression of PPAR␥2 is elevated in obese mice compared with lean littermates (32,33). Additionally PPAR␥ levels are elevated in hepatocellular carcinoma cell lines (15,16); activation of PPAR␥ in hepatocellular carcinoma cell lines inhibits growth (14). In a recent study, adenovirus-mediated overexpression of PPAR␥1 resulted in hepatic steatosis in the liver of PPAR␣ Ϫ/Ϫ mice (34). In other studies expression of PPAR␥ and its downstream target genes were increased following a homozygous deletion of cyclin D1 gene (35). These cyclin D1 Ϫ/Ϫ mice developed hepatic steatosis with increased neutral lipid accumulation in the liver indicative of increased PPAR␥ activity. These results clearly indicate a distinct involvement of PPAR␥ in regulating hepatocyte physiology, although the mechanism and effectors involved are largely unclear.
To understand the role of PPAR␥ in regulating hepatocyte growth and to elucidate the pathway involved, we designed studies with control and PPAR␥ hepatocytes (see "Experimental Procedures") in the presence and absence of the PPAR␥specific ligand TZD. Our studies indicated that activation of PPAR␥2 in exponentially growing AML-12 hepatocytes attenuated growth and inhibited expression of proliferating cell nuclear antigen (PCNA), cyclin D1, and ␤-catenin. PPAR␥2-mediated inhibition of cyclin D1 transcription in these cells, however, was completely restored following overexpression of CRE-binding protein (CREB) but not ␤-catenin and involved modulation of total and phospho-CREB Ser133 levels as well as CREB transcription. Additionally pretreatment of the PPAR␥ hepatocytes with okadaic acid maintained higher cyclin D1 transcription despite PPAR␥2 activation via maintaining high total CREB and phospho-CREB Ser133 levels and without restoring ␤-catenin expression. These studies demonstrated that PPAR␥2 activation-mediated attenuation of cyclin D1 transcription in the hepatocytes was independent of ␤-catenin and dependent on CREB-mediated pathways.
Generation of Stable Cell Lines-Stable hepatocyte cell lines ectopically expressing PPAR␥2 (PPAR␥ cells) or the empty control retroviral vector (control cells) were generated using AML-12 hepatocytes (46) as described previously (19) and were obtained from Dr. Stephen Farmer (Department of Biochemistry, Boston University School of Medicine, Boston, MA). Briefly BOSC 23 packaging cells were transiently transfected with either the pBabe-PPAR␥-puro or pBabe-puro retroviral expression vectors as described previously (19). Viral supernatants obtained following this transfection were used to infect AML-12 cells, and retrovirus-expressing cells were selected in growth medium (Dulbecco's modified Eagle's-F12 medium supplemented with 10% fetal bovine serum, 1ϫ ITS, and 100 nM dexamethasone) containing 2 g/ml puromycin. All the studies described here were performed with exponentially growing PPAR␥ or control hepatocytes treated with either Me 2 SO (vehicle) or TZD.
The stable P␥/wt-␤cat and P␥/S37A-␤cat cells were generated by transfection of PPAR␥ cells with the expression vectors wild-type-␤catenin-HA or S37A-␤-catenin-HA expressing either wild-type or S37A mutant ␤-catenin, respectively (41). Stably transfected cells were selected for neomycin resistance, and individual clones were screened for HA expression.
Cell Proliferation Assay-PPAR␥ and control cells were plated at a density of 0.1 ϫ 10 6 cells in each well of a 6-well plate and treated with either Me 2 SO (vehicle) or 10 M TZD. Cells were harvested at the indicated time intervals via trypsinization and counted using a hemocytometer. The cell number in the TZD-treated samples at each time point was calculated as percentage of control considering the corresponding vehicle-treated sample as 100%. Cells for each time point were plated in three separate wells, and each experiment was repeated at least three times.
Western Blot Analysis-Total cellular or nuclear protein was extracted from the cells at different time intervals according to the procedure described previously (47). Western blot analysis was performed with equal amounts of each protein sample and blotted with the respective antibodies.
Transient Transfection and Luciferase Reporter Assays-PPAR␥ and control vector cells plated at a density of 0.15 ϫ 10 6 cells/well in 6-well plates were transfected using LipofectAMINE (Invitrogen) with the respective luciferase vectors along with a ␤-galactosidase vector to correct for transfection efficiency. The amount of DNA transfected in any particular experiment was maintained equal by adjusting the amount of empty vector DNA. Following an overnight recovery in the growth medium, the transfected cells were treated with either Me 2 SO (as vehicle) or TZD for the indicated periods of time. Luciferase and ␤-galactosidase assays were performed using a microplate luminometer LB 96V (EG&G Berthhold, Bad Wilbad, Germany) and a plate reader, respectively. Each assay was performed in duplicate, and each transfection was repeated at least six to nine times. The results obtained were calculated as the ratio of relative light units to the ␤-galactosidase values and expressed as percent increase of luciferase activity considering those obtained from the controls as 100%. The relative light units/␤-galactosidase values obtained from the vehicle-treated samples for each set of transfections were considered as controls for that particular experiment.

PPAR␥ Hepatocytes Express Functional PPAR␥2
Protein-To investigate the role of PPAR␥2 in regulating hepatic function, control and PPAR␥ hepatocytes ectopically expressing either a control retroviral vector (pBabe-puro) or PPAR␥2-containing retroviral vector (pBabe-PPAR␥-puro), respectively, were utilized (48). Western analysis at different time intervals with a PPAR␥-specific antibody showed the specific expression of PPAR␥2 protein in the PPAR␥ hepatocytes and not in control cells (Fig. 1A). Luciferase assays designed with the PPAR␥responsive reporter tk-PPREx3-Luc (containing PPRE sites attached to a luciferase vector) (38) showed a low PPRE luciferase activity in the control cells, irrespective of the presence of TZD, that was highly activated in the PPAR␥ cells following exposure to TZD (Fig. 1B). Comparison of the basal level of activity of this promoter also showed a significantly higher activity in the PPAR␥ cells compared with the control cells in the absence of TZD (data not shown). These results indicated that the PPAR␥ hepatocytes express a functional PPAR␥2 protein.
Ectopic Expression and Activation of PPAR␥2 in Proliferating Hepatocytes Arrests Growth and Inhibits Expression of PCNA-To determine whether the ectopic expression and activation of PPAR␥2 in AML-12 hepatocytes can suppress growth, we analyzed the cell proliferation rate in the two cell types. These results showed that activation of PPAR␥2 following addition of TZD resulted in complete inhibition of growth in the PPAR␥ cells ( Fig. 2A) and was associated with an overall decrease in PCNA expression (Fig. 2B). Since PPAR␥ activation inhibits growth at the G 1 /S boundary in other cell types (14,17) and inhibited PCNA expression in AML-12 hepatocytes, we examined the effect of PPAR␥2 on cell cycle distribution. These studies indicated a major population of the PPAR␥ hepatocytes in G 0 -G 1 phase that was further increased following addition of TZD (data not shown). The control hepatocytes, however, multiplied exponentially with a major population in the S phase but not in the G 0 -G 1 phase. This clearly indicated a PPAR␥2-mediated inhibition of hepatocyte growth at G 0 -G 1 phase of the cell cycle.
Activation of PPAR␥2 in Proliferating Hepatocytes Inhibits Cyclin D1 Expression-Since PPAR␥2 activation in hepatocytes resulted in a G 1 /S arrest, we determined whether it inhibited expression of cyclin D1. TZD-mediated activation of PPAR␥2 inhibited expression of cyclin D1 protein in the PPAR␥ hepatocytes without affecting its expression in the control hepatocytes (Fig. 3A). To determine whether PPAR␥2 activation in the hepatocytes attenuated cyclin D1 transcription, we transfected the full-length Ϫ1745 cyclin D1 promoter luciferase construct along with the ␤-galactosidase plasmid (for correction of transfection efficiency) in both PPAR␥ and control cells and performed luciferase and ␤-galactosidase assays following treatment with TZD over a period of 72 h. While addition of TZD to the control cells showed no change in the cyclin D1 promoter activity (Fig. 3B, compare Ϫ and ϩ TZD lanes in control cells), it significantly attenuated cyclin D1 promoter activity in the PPAR␥ cells during the 72-h culture period (compare Ϫ and ϩ lanes in PPAR␥ cells). Cyclin D1 luciferase assays carried out to determine the dose of TZD required for optimal inhibition of cyclin D1 transcription indicated that 1 M TZD significantly inhibited cyclin D1 transcription in PPAR␥ cells (Fig. 3C).
Activation of PPAR␥2 by TZD Inhibits Expression of ␤-Catenin-Since ␤-catenin is known to regulate cyclin D1 transcription (27), we determined whether PPAR␥2-mediated attenua- tion of cyclin D1 transcription and thus hepatocyte growth involved ␤-catenin. Results in Fig. 4A show that activation of PPAR␥2 by exposure to TZD significantly blocked the accumulation of ␤-catenin and reduced its expression to virtually undetectable levels in the PPAR␥ cells without affecting its expression in the control cells. Luciferase assays were then carried out with the ␤-catenin-responsive luciferase vectors pGL3OT or pGL3OF (containing wild-type or mutant TCF sites, respectively). The pGL3OT and pGL3OF vectors were modified versions of the original TCF-4 reporters TOPFLASH and FOPFLASH, respectively (39,40), designed to reduce the levels of background activity via incorporation of transcriptional stop signals surrounding the TCF-4 sequences. These results indicated a significant attenuation of TCF-responsive luciferase activity following activation of PPAR␥2 in the PPAR␥ cells (Fig. 4B, compare lanes 1 and 2 with lanes 5 and 6) without any change in the mutant pGL3OF activity (compare lanes 3 and 4 with lanes 7 and 8). These observations suggested that activation of PPAR␥2 might attenuate cyclin D1 transcription and thus arrest hepatocyte growth in part by attenuating the ␤-catenin-TCF/lymphoid enhancer factor-1 signaling pathway.
␤-Catenin Overexpression Is Unable to Restore PPAR␥2 Activation-mediated Down-regulation of Cyclin D1 Transcription and Growth-To determine whether PPAR␥2 activation reduced cyclin D1 transcription via attenuating ␤-catenin levels, we measured cyclin D1 promoter activity in the PPAR␥ cells (in the presence and absence of TZD) following overexpression of either wild-type or S37A-␤-catenin or the corresponding empty vector. S37A-␤-catenin is a stable mutant form of ␤-catenin where the serine residue at amino acid 37 has been mutated to an alanine (41). Since ␤-catenin degradation by the glycogen synthase kinase-3␤-mediated pathway requires its phosphoryl- FIG. 3. Activation of PPAR␥2 in the hepatocytes inhibits cyclin D1 transcription. A, Western blot analysis of total protein extracts from control and PPAR␥ hepatocytes treated as indicated for different periods of time with antibodies against cyclin D1 and actin. B, exponentially growing control and PPAR␥ hepatocytes were transiently transfected with the full-length cyclin D1 promoter luciferase construct (Ϫ1745 CD1-luc) along with ␤-galactosidase (as control) and treated with (ϩ) or without (Ϫ) 10 M TZD for the indicated periods of time. At the end of incubation cells were harvested, and luciferase and ␤-galactosidase assays were performed. The results obtained were calculated as the ratio of relative light units to the ␤-galactosidase values and expressed as percentage of control considering those obtained from the vehicle-treated controls as 100%. C, subconfluent control and PPAR␥ hepatocytes were transfected with Ϫ1745 CD1-luc and ␤-galactosidase constructs as in B and treated with increasing concentrations of TZD for 72 h, and luciferase and ␤-galactosidase assays were performed. Each transfection (in both B and C) was performed in triplicate, and the data represent the mean Ϯ S.D. of three independent experiments. *, p Ͻ 0.0001 compared with the untreated control.

FIG. 4. Activation of PPAR␥2 inhibits expression of ␤-catenin
and attenuates ␤-catenin/TCF-responsive reporter activity. A, Western analysis of total protein extracts isolated from subconfluent control and PPAR␥ hepatocytes treated as indicated with an antibody against ␤-catenin. The same samples probed with actin served as control. B, exponentially growing control and PPAR␥ hepatocytes were transiently transfected with the ␤-catenin-responsive reporter constructs pGL3OT and pGL3OF along with ␤-galactosidase. Luciferase and ␤-galactosidase assays were performed following treatment without (Ϫ) or with (ϩ) 10 M TZD for 72 h. The ratio of relative light units to the ␤-galactosidase values obtained from the TZD-treated (ϩ) samples for each set of transfection were represented as percentage of control considering those obtained from the corresponding TZD-untreated (Ϫ) controls as 100%. Each transfection was performed in triplicate, and the data represent the mean Ϯ S.D. of three independent experiments. *, p Ͻ 0.0001 compared with the untreated control.  9 and 10 and lanes 11  and 12) were completely unable to reverse the inhibitory effects of PPAR␥2 activation on cyclin D1 transcription and were similar to the empty vector expression (lanes 1 and 2 and lanes  3 and 4). Luciferase assays designed with pGL3OT or pGL3OF luciferase vectors to determine whether the ␤-catenin vectors transfected in the studies of Fig. 5A were functional showed a significant activation of TCF-responsive reporter activity following overexpression of both wt-␤-catenin (Fig. 5B, lanes 5 and 6) as well as S37A-␤-catenin (lanes 9 and 10) compared with the empty vector controls (lanes 1 and 2). The activity of pGL3OF, however, showed no significant variation under these conditions.
To further examine whether expression of ␤-catenin could rescue PPAR␥2-mediated attenuation of cyclin D1 promoter activity, stable cell lines were created utilizing PPAR␥ cells overexpressing either HA-tagged wild-type ␤-catenin or S37A-␤-catenin named P␥/wt-␤cat and P␥/S37A-␤cat cells, respectively. Following selection of the cells with the appropriate antibiotic, clone 22 of P␥/wt-␤cat and clone 15 of P␥/S37A-␤cat showed maximal HA expression (Fig. 5C) and were chosen for further studies. Luciferase assays with pGL3OT and pGL3OF reporters showed an activation of TCF reporter activity (pGL3OT) in the P␥/wt-␤cat and P␥/S37A-␤cat cells compared with the PPAR␥ cells (Fig. 5D,  compare lane 1 with lanes 3 and 5), suggesting that the overexpressed ␤-catenin proteins were functional. Cyclin D1 luciferase assays carried out in these cells showed that stable overexpression of either wild-type or S37A mutant ␤-catenin in the PPAR␥ cells was also unable to antagonize the inhibitory effect of PPAR␥2 on cyclin D1 promoter (Fig. 5E, compare lanes 3 and 4 with lanes 5 and 6 and lanes 7 and 8) as observed earlier with transient ␤-catenin expression (Fig. 5A). Additionally expression of PCNA in both P␥/wt-␤cat and P␥/S37A-␤cat cells showed a TZD-mediated reduction (Fig. 5F, compare Ϫ and ϩ TZD lanes) as in the PPAR␥ cells (Fig. 2B). A comparison of cell number showed a greater reduction following addition of TZD in the P␥/S37A-␤cat cells compared with the PPAR␥ cells (data not shown). These results clearly demonstrated that overexpression of ␤-catenin activated TCF-responsive reporter activity in these cells but was unable to rescue cyclin D1 transcription, PCNA expression, and cell proliferation from the inhibitory effects of TZD indicating that the growth pathway (particularly cyclin D1 transcription) in these cells might be regulated via ␤-cateninindependent mechanisms.  (Fig. 6B), which corresponded to the low cyclin D1 expression at 72 h with similar TZD concentrations (Fig. 6A). These results indicated that PPAR␥ activation could modulate cyclin D1 expression without affecting ␤-catenin expression.

PPAR␥ Activation Can Modulate Cyclin D1 Levels without Affecting ␤-Catenin Steady State Levels in Other Cell Types-
Cyclin D1 Promoter Activity Can Be Restored following Activation of PPAR␥2 in the Absence of ␤-Catenin-To confirm that repression of cyclin D1 transcription following PPAR␥ activation was independent of ␤-catenin expression, we tried to restore cyclin D1 transcription to normal levels following PPAR␥2 activation using pharmacological agents and determined the expression levels of ␤-catenin under the same conditions. Since the phosphatase inhibitor okadaic acid can induce cyclin D1 mRNA expression (51), we performed cyclin D1 luciferase assays in the presence and absence of TZD following a preincubation with okadaic acid. Incubation with okadaic acid not only resulted in a significant induction of basal levels of cyclin D1 transcription (data not shown), it completely blocked the inhibitory effects of TZD on cyclin D1 transcription in PPAR␥ cells (Fig. 7A, compare lanes 5 and 6 with lanes 7 and  8) as well as in P␥/S37A-␤cat cells (Fig. 7B), and overexpression of ␤-catenin (in P␥/S37A-␤cat cells) had no synergistic effect. Okadaic acid-mediated inhibition of phosphatase activity was thus capable of completely attenuating the effect of PPAR␥2 on cyclin D1 transcription. Addition of okadaic acid, however, was unable to maintain higher levels of ␤-catenin expression following addition of TZD in either PPAR␥ (Fig. 7C) or P␥/S37A-␤cat cells (Fig. 7D). In similar experiments, pretreatment of PPAR␥ cells with lithium chloride (LiCl, an activator of CREB (52)) was also capable of partially reversing the inhibitory effects of TZD on cyclin D1 transcription (Fig. 7E). LiCl, however, was completely unable to block PPAR␥2-mediated downregulation of ␤-catenin expression in these cells (data not shown). Since both okadaic acid and LiCl were capable of maintaining high levels of cyclin D1 transcription despite PPAR␥2 activation and in the absence of ␤-catenin expression, we conclude that PPAR␥2-mediated inhibition of cyclin D1 transcription in these cells was mediated through a ␤-cateninindependent pathway.
Overexpression of CREB Can Overcome PPAR␥2-mediated Attenuation of Cyclin D1 Promoter Activity in the Absence of ␤-Catenin-The transcription from the cyclin D1 promoter can be modulated by a variety of transcription factors including CREB (28,29). To determine any possible participation of CREB in PPAR␥2-mediated attenuation of cyclin D1 transcription, we performed cyclin D1 luciferase assays in PPAR␥ cells following cotransfection of increasing concentrations of wildtype CREB and estimated the effect of PPAR␥2 activation on the promoter activity. The results described in Fig. 8A showed that overexpression of as low as 50 ng of wild-type CREB was capable of significantly restoring the promoter activity following stimulation by TZD (compare lanes 1 and 2 with lanes 9 and  10) and completely abolished the inhibitory effect of PPAR␥2 at higher concentration (compare lanes 5 and 6 with lanes 13 and   FIG. 7. Pretreatment with okadaic acid maintains higher levels of cyclin D1 transcription in the presence of PPAR␥2 activation. A, Ϫ1745 CD1-luc assay was performed in control and PPAR␥ cells in the absence (Ϫ) or presence (ϩ) of 10 M TZD following a 1-h pretreatment with 3 nM okadaic acid. B, Ϫ1745 CD1-luc assays similar to those described in A were performed in P␥/S37A-␤-catenin cells following incubation with okadaic acid and TZD. C, Western analysis of total extracts from PPAR␥ cells treated with TZD for 72 h in the presence (ϩ) or absence (Ϫ) of a pretreatment with 3 nM okadaic acid using an antibody against ␤-catenin. The same samples probed with actin antibody served as control. D, Western analysis similar to that described in C with P␥/S37A-␤-catenin cells. E, Ϫ1745 CD1-luc assay was performed with PPAR␥ cells treated with TZD following an overnight pretreatment with 20 mM LiCl. The data in A, B, and E represent the mean Ϯ S.D. of three independent experiments performed in triplicate. OA, okadaic acid.
14 and compare lanes 7 and 8 with lanes 15 and 16). The cyclin D1 luciferase studies of Fig. 8B showed that only wt-CREB (lanes 3 and 4) and not the empty vector (lanes 1 and 2) or dominant negative CREB (43) (KCREB, lanes 5 and 6) could activate cyclin D1 transcription following PPAR␥2 activation, confirming the participation of CREB in these events. Cyclin D1 luciferase assays designed with wild-type CREB and increasing concentrations of KCREB showed that addition of KCREB significantly inhibited the capability of wt-CREB to reactivate cyclin D1 promoter following addition of TZD (Fig.  8C, compare lane 4 with lanes 6, 7, 8, and 9), indicating that KCREB functioned as a dominant negative vector. Additionally overexpression of S37A-␤-catenin in the presence of wt-CREB in the PPAR␥ cells produced no synergistic effect on cyclin D1 promoter activation (Fig. 8D, lanes 7 and 8) following addition of TZD and was equivalent to the level of activation by wt-CREB alone (lanes 5 and 6). These studies indicated that overexpression of CREB was capable of antagonizing the inhibitory effects of PPAR␥2 on cyclin D1 transcription in the absence of ␤-catenin and overexpression of ␤-catenin had no additional effect.
PPAR␥2 Activation Reduces the Levels of Total and Activated Form of CREB-Since overexpression of CREB reactivated cyclin D1 promoter following PPAR␥2 activation, it was possible that PPAR␥2 activation involved modulation of CREB levels.
To determine any effect of PPAR␥2 on endogenous CREB, we estimated changes in the levels of phosphorylated CREB (Ser 133 ) or total CREB following activation of PPAR␥ by West-  1-8) or wild-type CREB (lanes 9 -16). Luciferase and ␤-galactosidase assays were performed following a 72-h treatment in the absence (Ϫ) or presence (ϩ) of 10 M TZD. B, Ϫ1745 CD1-luc assays were performed as described in A with PPAR␥ cells in the presence of either empty vector, wild-type CREB, or dominant negative CREB (KCREB). C, Ϫ1745 CD1-luc assays were performed in PPAR␥ cells as in A following overexpression of empty vector (lanes 1 and 2), wt-CREB (lanes 3 and 4), or wt-CREB in combination with increasing concentrations of KCREB (lanes 5-9). D, Ϫ1745 CD1-luc assays were performed in PPAR␥ cells following cotransfection of S37A-␤-catenin or wild-type CREB alone (lanes 3 and 4 and lanes 5 and 6, respectively) or in combination (lanes 7 and 8). Similar luciferase studies were also carried out following cotransfection of the corresponding empty vectors (lanes 1 and 2). Each transfection (in A-D) was performed in triplicate, and the data represent the mean Ϯ S.D. of three independent experiments.
FIG. 9. PPAR␥2 activation modulates CREB levels and attenuates pCRE-luc activity. A, Western analysis of control and PPAR␥ hepatocytes treated without (Ϫ) or with (ϩ) TZD with phospho-CREB Ser133 , total CREB, and actin antibodies. B, luciferase assays with control and PPAR␥ cells following transfection of CREB-responsive reporter pCRE-luc and ␤-galactosidase vectors and treatment in the absence (Ϫ) and presence (ϩ) of 10 M TZD. C, luciferase assays similar to those described in B were carried out following transfection of the CREB promoter (Ϫ1264 CREB-luc) and ␤-galactosidase vectors. D, a functional Ser 133 site of CREB is required for restoring cyclin D1 transcription following activation of PPAR␥2. Ϫ1745 CD1-luc assays were performed in PPAR␥ cells treated with (ϩ) or without (Ϫ) TZD following cotransfection of either empty vector (lanes 1 and 2), wt-CREB (lanes 3 and 4), increasing concentrations of CREBM1 (lanes 5 and 6 and lanes 7 and 8), or a combination of wt-CREB and CREBM1 (lanes 9 and 10). E, p300 can synergize with CREB to activate cyclin D1 transcription following activation of PPAR␥2. Luciferase assays similar to those described in D were performed in PPAR␥ cells following cotransfection of either empty vector (lanes 1 and 2), wt-CREB (lanes 3 and 4), or wt-CREB in combination with increasing concentrations of p300 ( lanes 5 and 6 and lanes 7 and 8). Each transfection (in B-E) was performed in triplicate, and the data represent the mean Ϯ S.D. of two independent experiments. ern blot analysis with antibodies against phospho-CREB Ser133 or total CREB, respectively. These results indicated a gradual TZD-dependent decrease in the total CREB (Fig. 9A) as well as phospho-CREB Ser133 levels (phospho-CREB panel) in the PPAR␥ cells. This was also associated with a similar reduction in CREB-responsive reporter activity (pCRE-luc) following activation of PPAR␥2 (Fig. 9B), suggesting that PPAR␥2 activation inhibited the CREB transactivation pathway. Luciferase assays designed with the CREB promoter (Ϫ1264 CREB-luc) showed a significant reduction of CREB promoter activity following activation of PPAR␥2 (Fig. 9C, compare Ϫ and ϩ TZD lanes in PPAR␥ cells), indicating a transcriptional regulation of CREB by PPAR␥.
To elucidate whether wt-CREB-mediated reactivation of cyclin D1 transcription following activation of PPAR␥2 involved the Ser 133 phosphorylation of CREB, we attempted to restore cyclin D1 transcription in the presence of either wt-CREB or a Ser 133 phosphorylation-defective CREB mutant (CREBM1) (42). Results from these studies indicated that overexpression of CREBM1 alone was unable to restore cyclin D1 transcription to wt-CREB levels (Fig. 9D, compare lanes 3 and 4 with lanes  5 and 6 and lanes 7 and 8) and showed no synergistic effect when overexpressed in combination with wt-CREB (lanes 9 and 10). This indicated that phosphorylation of CREB at Ser 133 was required for restoration of cyclin D1 transcription following PPAR␥2 activation. This was supported by the observation that overexpression of p300 synergized with wt-CREB to reactivate cyclin D1 transcription (Fig. 9E, compare lanes 3 and 4 with  lanes 5 and 6 and lanes 7 and 8) since Ser 133 phosphorylation of CREB enables it to interact with CREB-binding protein/p300 resulting in CRE activation.
Okadaic Acid-mediated Reactivation of Cyclin D1 Transcription Involves CREB-Western blot studies carried out with PPAR␥ hepatocytes to determine whether okadaic acid-mediated reactivation of cyclin D1 transcription involved CREB indicated that addition of okadaic acid in the presence of TZD could maintain higher levels of total as well as phospho-CREB-Ser133 compared with TZD alone (Fig. 10A), which might be responsible for reactivating cyclin D1 transcription. Cyclin D1 luciferase assays carried out in the presence of okadaic acid as well as KCREB showed that okadaic acid was unable to restore cyclin D1 promoter activity following exposure to TZD in the presence of KCREB (Fig. 10B, compare lanes 5 and 6 with lanes 9 and 10), confirming the participation of CREB in okadaic acid-mediated activation of cyclin D1 transcription. These studies showed that PPAR␥2 activation-related inhibition of cyclin D1 transcription was mediated via inhibition of CREB.

DISCUSSION
Several recent studies have demonstrated that activation of PPAR␥ can suppress growth in proliferating cells (for a review, see Ref. 13) including those originating from hepatocellular carcinoma (14,16). In this report we demonstrate that activation of PPAR␥2 in proliferating mouse hepatocytes can arrest growth and reduce PCNA expression. Earlier studies with fibroblasts induced to differentiate into adipocytes by PPAR␥2 showed that the corresponding growth arrest involved induction of the cyclin-dependent kinase inhibitors p18 INK4c and p21 CIP1 (19). In other studies PPAR␥-induced cell cycle withdrawal was associated with inhibition of DNA binding and transcriptional activity of the E2F⅐DP complex due to a downregulation of the catalytic subunit of the serine-threonine phosphatase PP2A (18). In our studies, PPAR␥2-mediated suppression of growth in hepatocytes involved a corresponding TZDdependent inhibition of cyclin D1 and ␤-catenin production suggesting that the molecular mechanisms responsible for this response might be operating in mid-G 1 phase of the cell cycle. In fact, the flow cytometric analysis demonstrated that activation of PPAR␥2 resulted in the accumulation of hepatocytes in G 1 phase of the cell cycle (data not shown) as reported in other cell types (14,17). PPAR␥2-mediated attenuation of cyclin D1 expression might contribute toward this G 1 /S arrest in hepatocytes as shown in breast cancer cells (20) since cyclin D1 via stimulating the activity of select cyclin-dependent kinases facilitates G 1 /S transition.
PPAR␥ activation has been shown to regulate cyclin D1 expression at the level of both transcription as well as posttranscription (20,53). In the studies described here PPAR␥ activation in hepatocytes resulted in a significant attenuation of cyclin D1 promoter activity suggesting a regulation at the level of transcription. In these studies our aim was to identify the downstream events following PPAR␥2 activation in the hepatocytes responsible for regulating cyclin D1 transcription. A detailed analysis of this pathway will provide useful information regarding the effectors involved in PPAR␥-mediated growth arrest and indicate whether this pathway is modulated in a cell type-specific manner. To identify the effectors the initial focus was on ␤-catenin since ␤-catenin is a major regulator of cyclin D1 transcription (27) and can activate cyclin D1 gene through a lymphoid enhancer factor-1 binding site in its promoter. Since PPAR␥2 activation in the hepatocytes resulted in a dramatic down-regulation of ␤-catenin expression, it was conceivable that this event might contribute to the accompanying growth arrest via inhibiting expression of key regulators of G 1 progression such as c-Myc and cyclin D1. Interestingly our experiments indicated that activation of PPAR␥2 in the presence of ectopically expressed ␤-catenin was unable to restore cyclin D1 transcription and growth. Our studies also showed that cyclin D1 transcription can be restored in these hepatocytes following activation of PPAR␥2 in the absence of ␤-catenin and in the presence of ectopically expressed CREB. These observations provided a clear indication that PPAR␥2mediated suppression of ␤-catenin levels and cyclin D1 transcription are independent of each other. This hypothesis is strengthened by other studies that showed that overexpression of an oncogenic mutant form of ␤-catenin in liver despite inducing hepatomegaly was unable to induce the levels of ␤-catenin target genes cyclin D1 and c-myc (54).
The synthesis of cyclin D1 is regulated in the early phase of G 1 by extracellular effectors that activate a variety of signaling pathways, which converge on various transcription factors, many of which involve CREB (28). In response to stimulation by cyclic AMP, CREB is activated via phosphorylation at serine residue 133, which enables it to bind to the transcriptional coactivator CREB-binding protein and its homologue p300 (55). This CREB⅐CREB-binding protein complex then activates transcription of target genes via binding to the CRE site. Additionally CREB can activate CRE-specific activity via binding to other transcription factors (for example Oct-1 and Lim-only protein) in a phosphorylation-independent manner (56,57). Results from our studies demonstrated that PPAR␥2 activation reduced the levels of total as well as phospho-CREB ser133 , which was rescued in part following preincubation with okadaic acid. It thus seemed likely that PPAR␥2 activation regulated the transcription of cyclin D1 gene and thus growth via attenuation of CREB activity. Our studies also indicated that PPAR␥2 activation in the hepatocytes regulated CREB levels via modulating its transcription, although it is unclear whether additional regulations exist at the level of CREB Ser133 phosphorylation as well. Studies designed with the phosphorylation-deficient mutant CREBM1 showed that CREB-mediated reactivation of the cyclin D1 promoter following PPAR␥2 activation required a functional CREB Ser 133 site. In similar experiments cotransfection of p300 together with wt-CREB showed a synergistic effect in reactivating cyclin D1 promoter. Earlier studies by Wang et al. (20) showed that activation of PPAR␥ resulted in increased interaction between PPAR␥ and p300 thus reducing interaction between p300 and c-Fos. The authors concluded that this limited availability of p300 was responsible for reduction of cyclin D1 transcription following PPAR␥ activation in those cells that was rescued in part following overexpression of p300. CREB and p300 might thus be the two key molecules mediating the growth-inhibitory responses following activation of PPAR␥.
In our studies, since okadaic acid was capable of restoring CREB levels as well as cyclin D1 transcription, it is very likely that PPAR␥2 activation resulted in activation of a phosphatase, which might be an important regulator of this growth pathway. Okadaic acid is capable of inhibiting the protein phosphatases PP1 and PP2A. It would also indicate that PPAR␥-mediated growth arrest in the liver cells deviates from the other cell types since this growth arrest in fibroblasts was shown to be associated with inhibition of PP2A (18). Further studies are required to confirm this hypothesis. Regulation of cyclin D1 expression by therapeutic intervention is an attractive means of treating breast and gastrointestinal tumorigen-esis. The ability of PPAR␥ ligands to effectively attenuate cyclin D1 expression suggests that these ligands have a great potential to be utilized as cancer chemotherapeutic drugs in the future. Elucidation of the pathway and identification of the molecules involved in mediating these effects might contribute toward the enhancement of the potency of these potential chemotherapeutic drugs.