Peroxisome Proliferator-activated Receptor β (δ)-dependent Regulation of Ubiquitin C Expression Contributes to Attenuation of Skin Carcinogenesis

The role of peroxisome proliferator-activated receptor-β (PPARβ) in the molecular regulation of skin carcinogenesis was examined. Increased caspase-3 activity associated with apoptosis was found in the skin of wild-type mice after tumor promotion with 12-O-tetradecanoylphorbol-13-acetate, and this effect was diminished in PPARβ-null mice. The onset of tumor formation, tumor size, and tumor multiplicity induced from a two-stage carcinogen bioassay (7,12-dimethylbenz[a]anthracene/12-O-tetradecanoylphorbol-13-acetate) were significantly enhanced in PPARβ-null mice compared with wild-type mice. To begin to characterize the molecular changes underlying this PPARβ-dependent phenotype, microarray analysis was performed and a number of differentially regulated gene products were identified including ubiquitin C. Subsequent promoter analysis, reporter gene assays, site-directed mutagenesis, and electrophoretic mobility shift assays provide evidence that PPARβ regulates ubiquitin C expression, and that ubiquitination of proteins is influenced by PPARβ. These results strongly suggest that activation of PPARβ-dependent target genes provides a novel strategy to inhibit tumor promotion and carcinogenesis.

Keratinocytes, the major cell type of the epidermis, provide a protective epidermal barrier against the external environment. The regulation of proliferation and the unique process of differentiation in keratinocytes are essential in maintaining epidermal function (1). Keratinocyte proliferation and differentiation are regulated by various biological factors including cytokines and extraneous signals through different signal transduction pathways. For example, activation of transcription factors such as activator protein 1 (AP-1), 1 and signal transducer and activator of transcription, can regulate the expression of genes that in turn influence cell proliferation and differentiation. AP-1 is a major regulator of keratinocyte function, and is activated by phosphorylation via several mechanisms including the protein kinase C and mitogen-activated protein kinase pathways. Alterations in these pathways induced by chemical toxicants can cause disruption of keratinocyte cell cycle control, which could lead to skin cancer (1,2).
Peroxisome proliferator-activated receptors (PPARs) are ligand-activated transcription factors and members of the nuclear hormone receptor superfamily (3)(4)(5). PPARs modulate target gene expression in response to ligand activation after heterodimerization with retinoid X receptor and binding to peroxisome proliferator-responsive elements. Three different PPAR isoforms, each encoded by distinct genes, have been identified and designated PPAR␣, PPAR␤ (also referred to as PPAR␦), and PPAR␥. PPARs exhibit relatively unique tissue distribution and appear to have distinct physiological functions (3)(4)(5). Extensive studies have been performed examining the biological role of PPAR␣, and many of these were facilitated by the use of the PPAR␣-null mouse. Through these studies, it was shown that PPAR␣ mediates the pleiotropic response of peroxisome proliferators, with critical roles in the regulation of target genes that modulate lipid metabolism and hepatocarcinogenesis (6 -10). PPAR␥ is known to have important roles in adipocyte differentiation, glucose homeostasis, lipid metabolism, and immune function (11)(12)(13)(14). Additionally, the recent description of a conditional PPAR␥-null mouse model provided definitive evidence demonstrating a direct role of PPAR␥ in target gene regulation that modulates lipid and cholesterol transport in macrophages (15).
In contrast to PPAR␣ and PPAR␥, considerably less is known about the biological functions of PPAR␤. PPAR␤ is ubiquitously expressed in many tissues with relatively high expression in brain, digestive tract, and skin, suggesting possible developmental or physiological roles in these tissues (16 -18). A number of studies have reported that PPAR␤ is involved in adipocyte proliferation and differentiation (19 -22), brain/nerve cell function (23)(24)(25), lipoprotein metabolism (26,27), and cancer including colon (28 -30), head and neck squamous cell carcinoma (31), and endometrial adenocarcinoma (32). Despite the significant increase in the number of reports correlating expression of PPAR␤ to specific biological functions in the past five years, definitive evidence from a null mouse model is lacking for many of these associations. A striking reduction in adiposity in PPAR␤-null mice compared with wild-type mice strongly supports its functional role in lipid metabolism (33,34). A role for PPAR␤ in regulating expression of adipose differentiation-related protein in macrophages in response to very low density lipoprotein has also been described using PPAR␤null cells, suggesting that PPAR␤ may influence atherosclerotic lesion formation (35). Furthermore, recent studies using a PPAR␤-null mouse model showed that PPAR␤ functions in skin wound healing (36,37) and regulates apoptosis in keratinocytes (38). Whereas PPAR␤ does not directly modulate changes in the expression of genes known to regulate epithelial cell differentiation including transglutaminase, involucrin, and small proline-rich proteins in response to 12-O-tetradecanoylphorbol-13-acetate (TPA), the hyperplastic response to TPA in the epidermis is significantly enhanced in PPAR␤-null mice (33). This observation suggests that PPAR␤ has a critical role in attenuating epidermal cell proliferation, likely because of reduced expression of PPAR␤ target genes. In this study, the role of PPAR␤ in the molecular and biochemical regulation of skin cancer was examined in the PPAR␤-null mouse model.

EXPERIMENTAL PROCEDURES
Two-stage Chemical Carcinogenesis Bioassay-The generation of wild-type and PPAR␤-null mice used for this work has been previously described (33). Two-stage chemical carcinogen testing was performed with male wild-type (n ϭ 11) and PPAR␤-null (n ϭ 13) mice (7-8 weeks old). Mouse skin was initiated with 50 g of DMBA dissolved in 200 l of acetone. One week after DMBA treatment, mice were treated topically with 5 g of TPA dissolved in 200 l of acetone, three times per week for 24 weeks. The onset of papilloma formation, and the number and size of papillomas were assessed weekly for each mouse.
Caspase-3 Activity-Cytosolic protein samples from wild-type or PPAR␤-null mice (control and TPA-treated) were isolated 8 and 48 h post-TPA application or after 2 weeks of 3ϫ/week applications (5 g/ mouse/application) as described above. Caspase-3 activity was measured using a colorimetric assay (Assay Designs, Inc., Ann Arbor, MI) and data are presented as units of activity per g of protein.
TUNEL Staining-Skin samples from wild-type or PPAR␤-null mice (control and TPA-treated) were isolated after 2 weeks of 3ϫ/week applications (5 g/mouse/application) as described above, fixed in 10% neutral buffered formalin, embedded in paraffin, and sections (4 -6 m) were prepared. Apoptotic cells were detected using a FragEL TM DNA Fragmentation Detection Kit (Oncogene Research Products, Boston, MA). Apoptotic cells were quantified using light microscopy and data are presented as the average number of apoptotic cells/frame/slide examined. A total of four slides per treatment group were analyzed and an average of 55 frames Ϯ 3 were examined per slide.
Bromodeoxyuridine (BrdUrd) Labeling Index-Wild-type or PPAR␤null mice were injected intraperitoneally with BrdUrd (100 mg/kg) and then treated with TPA (5 g/mouse/application), followed 48 h later with a second intraperitoneal injection of BrdUrd and TPA application. Twenty-four hours after the second treatment with BrdUrd and TPA, mice were euthanized and skin samples were fixed in 10% neutral buffered formalin. Fixed skin samples were embedded in paraffin, sections (4 -6 m) were prepared, and detection of BrdUrd-labeled keratinocytes was performed using an immunohistochemical assay (Exalpha Biologicals, Inc., Watertown, MA). BrdUrd-labeled keratinocytes were quantified using light microscopy and the labeling index quantified as described by others (39). A total of five slides per treatment group, and a minimum of 1000 cells per slide were examined.
Keratinocyte Culture-A previously described method (40) was used to culture keratinocytes obtained from 2-day-old neonates from both genotypes. Keratinocytes were cultured in low calcium medium (0.05 mM) until 80% confluent at which time medium was replaced with low calcium medium containing various concentrations of L-165041 in a range known to specifically activate PPAR␤ (41) or used for transient transfections. The treatment period with the PPAR␤ ligand L-165041 was 8 h, and RNA was isolated from cells using TRIzol reagent (Invitrogen) and the manufacturer's protocol.
Skin Histology-After TPA treatment for 2 or 24 weeks, mice were euthanized by overexposure to carbon dioxide. The dorsal skin area was excised, fixed in 10% neutral buffered formalin, embedded in paraffin, and sections (4 -6 m) were stained with hematoxylin and eosin for histological examination. Skin tumors were randomly selected after termination of TPA treatment, fixed and stained as described above, and examined for lesion progression.
RNA Isolation-Skin samples from wild-type or PPAR␤-null (control and TPA-treated) mice were isolated 8 h post-treatment. Tissue samples were placed in liquid nitrogen and ground with a mortar and pestle. Ground tissue was further homogenized in TRIzol reagent using a tissuemizer. The remaining procedures for the isolation of total RNA from skin were performed using the manufacturer's protocol. Isolation of RNA from cultured keratinocytes was performed using TRIzol extraction and the manufacturer's protocol. Total RNA was resuspended in diethyl pyrocarbonate-treated water and quantified by spectrophotometry.
Microarray Analysis-A commercial microarray containing 2700 cDNAs (Discovery ARRAY 1.1, Display Systems Biotech, Vista, CA) was used to identify novel genes that may be controlled by PPAR␤. The advantage of this microarray is there is no bias toward known genes or previously cloned genes, and there is an emphasis on coding regions rather than 3Ј-untranslated regions. Hybridization and washing were performed following the manufacturer's recommended procedure. Ten micrograms of total RNA isolated from the skin of either one wild-type or PPAR␤-null mouse treated with TPA (8 h post-application, 5 g per mouse) was labeled with either Cy3 or Cy5 dyes and hybridized overnight on the microarray slide. A reciprocal labeling method was used to verify changes in gene expression. After hybridization and washing, the slide was scanned with lasers and the fluorescent images were overlaid. Differences in the Cy5/Cy3 ratio were analyzed for statistically significant differences using Gene Pix Pro (Axon Instrument, Union City, CA). A 99% confidence interval was calculated from the ratios of Cy5/ Cy3, and differentially expressed cDNAs were determined by selecting those cDNAs whose relative Cy5/Cy3 ratio was outside of the 99% confidence interval.
Northern Blot Analysis-To confirm altered expression of PPAR␤ target genes detected by microarray analysis or to examine mRNA levels after various treatments, 5 to 10 g of total RNA was electrophoresed on a 1.0% agarose gel containing 0.22 M formaldehyde, transferred to a nylon membrane, and ultraviolet cross-linked to fix the RNA. Membranes were hybridized in ULTRAhyb hybridization buffer (Ambion, Austin, TX) with random primed probes (Display Systems Biotech, Vista, CA) following the manufacturer's protocol, and washed with salt/detergent solutions using standard procedures.
Protein Isolation-Skin samples from wild-type or PPAR␤-null (control and TPA treated) mice, isolated 8 h post-treatment, were used to prepare cytosolic fractions. Samples of skin were placed in liquid nitrogen and then ground to a fine powder before total nitrogen evaporation. Samples were placed in lysis buffer containing protease inhibitors without detergent followed by differential centrifugation at 1,000 and 100,000 ϫ g. The supernatant from the high speed spin was used for the cytosolic fraction and the pellet was used for microsomes. The protein concentration of each sample was quantified using the BCA detection system (Pierce).
Western Blot Analysis-Twenty-five to 50 g of protein from wildtype or PPAR␤-null (control and TPA treated) skin samples were resolved using SDS-PAGE. The samples were transferred onto a nitrocellulose membrane using an electroblotting method. The membrane was incubated overnight at 4°C with primary antibody, followed by an incubation with an horseradish peroxidase-conjugated secondary antibody (Santa Cruz Biotechnologies, Inc., Santa Cruz, CA). An ECL chemiluminescence system (Amersham Biosciences) was used to detect immunoreactive protein. The following primary antibodies were used: anti-ubiquitin (Sigma) or anti-lactate dehydrogenase (Rockland Inc., Gilbertsville, PA). Detection of free ubiquitin was performed as described by others (42).
Cloning the Murine Ubiquitin C Promoter-The PCR primers for amplifying the ubiquitin C promoter were designed using the sequence obtained from the Celera Mouse Genome Data base (CMGD, Release 12). The forward primer sequence was 5Ј-CAGCCTGGTCTA-CATAGCATATTCC-3Ј, and the reverse primer sequence was 5Ј-ACG-GAAACCAAACGTCTGGG-3Ј. A 1092-bp sequence containing the murine ubiquitin C promoter was cloned by PCR into PCRIITOPO vector (Invitrogen), using the C57BL/6N mouse genomic DNA as template, and confirmed by sequencing. Based on the murine polyubiquitin sequences (GenBank TM accession numbers AF285161 and AF285162), the 1092-bp promoter clone contains nucleotides Ϫ963 to ϩ128 (ϩ1 refers to the transcription start site) of the ubiquitin C promoter. The ubiquitin C promoter clone was subsequently subcloned into the pGL3-Basic firefly reporter plasmid (Promega, Madison, WI) in the 5Ј to 3Ј orientation, and this construct was confirmed by sequencing. The human ubiquitin C promoter reporter construct containing nucleotides Ϫ951 to ϩ4 (GenBank TM accession number D63791) was kindly provided by Dr. S. Russ Price.
Transient Transfections-HepG2 cells were transfected using the LipofectAMINE reagent (Invitrogen) with 0.4 g of PPAR␣-, PPAR␤-, or PPAR␥-pSG5 expression vectors, 0.4 g of either the human ubiquitin C promoter reporter construct or the murine ubiquitin C promoter reporter construct, and 0.2 g of a ␤-galactosidase construct as a control for transfection efficiency. Transfected cells were harvested 24 h after transfection and luciferase activity was measured with a luciferase assay system (Promega) using a Turner TD-20e luminometer. Experiments were performed three times using triplicates for each treatment group, and because of some variability between experiments, data are expressed as a percentage of control normalized luciferase activity (luciferase activity/␤-galactosidase activity). For analysis of the PPRE in the mouse ubiquitin C promoter, a 121-bp fragment containing the putative PPRE was amplified by PCR using the ubiquitin C promoter construct as template, subcloned into the pGL3-Promoter firefly reporter plasmid (Promega) in the 5Ј to 3Ј orientation, and confirmed by sequencing; referred to as pGL3P-mUbcPPRE herein. Site-directed mutagenesis was performed (QuikChange TM Kit, Stratagene Inc.) to introduce a single base pair mutation in the mUbcPPRE (AGGTCACAGC-CCT was changed to AAGTCACAGCCCT, mutation in bold), and confirmed by sequencing. Keratinocytes from wild-type and PPAR␤null mice were obtained as described above, cultured to 80% confluency, and transfected with 0.4 g of the pGL3P-mUbcPPRE and 0.2 g of a ␤-galactosidase construct, or 0.4 g of the mutant pGL3P-mUbcPPRE and 0.2 g of a ␤-galactosidase construct. Transfected keratinocytes were allowed to recover for 12 h prior to treatment with L-165041 or vehicle control for 12 h, at which time cells were harvested for analysis of luciferase activity as described.
Co-treatment of TPA and MG-132-Wild-type or PPAR␤-null mice, 8 weeks of age were shaved of dorsal hair, and 24 h later were treated topically with MG-132 dissolved in acetone. The treatment area of skin was weighed from previous experiments and it was determined that the average weight of skin was approximately 1 g. This weight was used to calculate an estimated effective concentration in skin equal to approximately twice the K I of MG-132 for proteasome inhibition (8 nM). Based on this calculation, 200 l of a solution containing 19 ng of MG-132/ml was applied per mouse. Mice were treated with MG-132 at 0600, and then retreated with MG-132 at 0700, followed by 5 g of TPA dissolved in 200 l of acetone 15 min after the last MG-132 application. Skin samples were collected after 48 h post-TPA for analysis of epithelial hyperplasia after fixation, sectioning in paraffin, and staining with H&E as described above.

Enhanced Skin Carcinogenicity in PPAR␤-null Mice-
The rationale for examining the role of PPAR␤ in skin carcinogenesis is based on the observation of enhanced epidermal hyperplasia and mRNA levels of gene products known to regulate cell cycle progression in PPAR␤-null mouse skin 48 h post-TPA treatment compared with wild-type controls (33). Consistent with these results, topical application of TPA resulted in exacerbated epidermal hyperplasia accompanied with hyperkeratosis in PPAR␤-null mice compared with controls (Fig. 1A). These data suggest that PPAR␤ functions to attenuate epidermal cell proliferation. As it is known that application of TPA can increase epithelial apoptosis (43), caspase-3 activity was measured and TUNEL staining was performed to determine whether alterations in apoptosis contribute to the observed difference in epithelial cell proliferation. Indeed, increased caspase-3 activity and TUNEL-positive cells were observed in response to TPA in wild-type mouse skin (Fig. 1, B and C). In contrast, increased caspase-3 activity and TUNEL-positive cells were lacking in skin samples from similarly treated PPAR␤-null mice. Whereas the BrdUrd labeling index was significantly increased in TPA-treated wild-type mouse skin as compared with control wild-type (Fig. 1D, 2.8-fold increase), the BrdUrd labeling index observed in TPA-treated PPAR␤null mouse skin was considerably greater as compared with similarly treated wild-type mice (Fig. 1D, 4.5-fold increase). To determine whether these alterations in cell proliferation were biologically significant, a skin cancer bioassay was performed, using wild-type and PPAR␤-null mice. Following initiation with DMBA and promotion by weekly applications of TPA, the onset of papilloma formation was significantly sooner in PPAR␤-null mice. The first papillomas were observed in null mice after only 7 weeks of TPA administration, whereas wildtype mice began to develop papillomas after 15 weeks of TPA (Fig. 1E). Whereas 100% of the PPAR␤-null mice developed papillomas after 11 weeks of tumor promotion, a much smaller percentage of wild-type mice had developed papillomas during the same time frame (Fig. 1E). Furthermore, after 24 weeks of treatment, a significantly smaller percentage of wild-type mice had developed papillomas after 24 weeks of TPA treatment (Fig. 1E). The average number and size of papillomas in PPAR␤-null mice were also greater than in wild-type (Fig. 1E). Histological examination of tumors from wild-type and PPAR␤null mice revealed that most lesions in both genotypes were squamous cell papillomas. However, sebaceous gland adenomas and keratocanthomas were found in a small cohort of PPAR␤-null mice, and neither of these lesions were observed in wild-type mice (data not shown). No evidence of early dysplasia, early carcinomas, or carcinomas was found in either genotype.
PPAR␤ Regulates the Expression of Ubiquitin C-To identify possible PPAR␤ target genes that could underlie the enhanced hyperplasia and skin carcinogenicity observed in PPAR␤-null mice, microarray analysis was performed using RNA samples from wild-type and PPAR␤-null mice. A total of 15 of 2700 cDNAs that exhibited statistically significant differential expression between TPA-treated wild-type and PPAR␤-null mouse skin were analyzed further. The identity of these cDNAs were ubiquitin C, pleckstrin homology, and Src homology 3 domain containing Rho GTPase-activating protein (PSGAP), creatine kinase, calmodulin, Src homology 3 domain protein 2A (endophilin), troponin, ubiquitin carboxyl terminus hydrolase, arsenate resistance protein 2, ribosomal protein 9, a sequence with some homology with the human KIAA0537 gene product, and five unidentified gene products that share limited homology with sequenced genes in GenBank TM .
To confirm changes in gene expression detected with the microarray, Northern blot analysis was performed using random labeled cDNAs. Whereas two of the identified clones were false positives (data not shown), the majority of differentially expressed gene products between genotypes detected with microarray analysis were those that were significantly higher in TPA-treated PPAR␤-null mouse skin and were undetectable in similarly treated wild-type mouse skin ( Fig. 2A). Included in this group were mRNAs encoding PSGAP, creatine kinase, calmodulin, endophilin, troponin, ubiquitin carboxyl terminus hydrolase, arsenate resistance protein 2, ribosomal protein 9, a sequence with some homology with the human KIAA0537 gene product, and three unidentified gene products. These results demonstrate that the molecular events that occur in response to TPA in PPAR␤-null mouse skin are significantly different from those that occur in similarly treated wild-type mouse skin. The most striking observation made was that the induction of ubiquitin C mRNA was present only in TPA-treated wild-type mouse skin, and not in TPA-treated PPAR␤-null FIG. 1. PPAR␤ attenuates cell proliferation and skin tumor formation. A, significantly enhanced epidermal hyperplasia in PPAR␤-null mice after 2 weeks of tumor promotion with TPA. Skin was collected, fixed, and stained with H&E. Magnification ϫ100. B, caspase-3 activity is increased in wild-type (ϩ/ϩ) mouse skin after TPA treatment and this effect is not found in PPAR␤-null (Ϫ/Ϫ) mouse skin. Asterisk (*), significantly different from control, p Յ 0.05. C, TUNEL-positive cells are increased in wild-type (ϩ/ϩ) mouse skin after TPA treatment and this effect is not found in PPAR␤-null (Ϫ/Ϫ) mouse skin. Asterisk (*), significantly different from control, p Յ 0.05. D, enhanced BrdUrd labeling in PPAR␤-null mouse skin in response to TPA as compared with TPA-treated wild-type mice. Mice were treated twice with TPA and the percentage of BrdUrd-labeled keratinocytes quantified. Asterisk (*), significantly different from control, p Յ 0.05; double asterisk (**), significantly different from TPA-treated wild-type, p Յ 0.05. E, exacerbated skin tumor formation in PPAR␤-null (Ϫ/Ϫ) mice. Mice were initiated with a single application of DMBA followed by weekly treatment with TPA as described under "Experimental Procedures." Onset of papilloma formation is sooner, and the average number of papillomas per mouse and average size of papillomas are higher in PPAR␤-null mice than in wild-type (ϩ/ϩ) controls. Incidence, number, and size of papillomas were measured weekly. Data are presented as the percentage of mice with papillomas (onset of tumor formation), or the mean Ϯ S.E. for average number and size (diameter, mm) per mouse. mouse skin (Fig. 2A). In contrast, TPA treatment induced mRNA encoding ubiquitin B in both genotypes.
To confirm that the change in ubiquitin C mRNA level leads to changes in the amount of free intracellular ubiquitin levels, Western blot analysis was performed. Indeed, TPA treatment resulted in higher levels of free ubiquitin protein in wild-type mouse skin (3.5-fold increase) and this effect was absent in PPAR␤-null mouse skin (Fig. 2B). Furthermore, a higher level of ubiquitinated proteins was detected by Western blot analysis in cytosol from TPA-treated wild-type mice than in PPAR␤-null treated with TPA (Fig. 2C). In addition, keratinocytes obtained from wild-type and PPAR␤-null mouse skin were cultured and incubated in the presence of a PPAR␤ ligand (L-165041). Consistent with the in vivo data, mRNA levels of ubiquitin C were higher (1.5-fold) in wild-type keratinocytes treated with L-165041 (Fig. 2D). In contrast, constitutive levels of mRNA encoding ubiquitin C were significantly lower in PPAR␤-null keratinocytes, and null keratinocytes were refractory to induction by the PPAR␤ ligand (Fig. 2D). To further characterize the direct role of PPAR␤ in the regulation of ubiquitin C, a 1092nucleotide DNA fragment containing the mouse ubiquitin C promoter was cloned and a putative DR-1 element (PPRE) was identified 186 nucleotides upstream from the TATA box (Fig.  3A). The PPRE in the murine ubiquitin C promoter sequence was AGGTCACAGCCCT compared with the consensus PPRE of AGGTCANAGGTCA. A similar PPRE in the human ubiquitin C promoter is also found, and conserved in 5 of the 6 bases representing the PPAR binding half-site (Fig. 3B). A number of other confirmed and putative transcription factor binding sites were also found in the 1092-nucleotide fragment containing the murine ubiquitin C promoter including AP-1, AP-2, c-Jun, NF-1, Sp-1, NF-B, and glucocorticoid receptor. To study the functional relevance of the PPRE, the 1092-nucleotide fragment of the murine ubiquitin C promoter was then subcloned into the pGL3Basic vector and transient transfections were performed in the presence or absence of a PPAR␤ expression vector. Whereas basal luciferase activity was minimal in the absence of PPAR␤, co-transfection of PPAR␤ with either the human or murine ubiquitin C promoter construct resulted in a significant increase in luciferase activity as compared with cells transfected with vector alone in the absence of PPAR␤ (Fig. 3, C and D). Surprisingly, co-transfection of PPAR␣ or PPAR␥ with the human or murine ubiquitin C promoter construct also produced an increase in luciferase activity (data not shown). A 121-bp fragment of the murine ubiquitin C promoter containing the putative PPRE was subcloned into the pGL3 Promoter vector for analysis in wild-type and PPAR␤-null keratinocytes. Luciferase activity was increased significantly in wild-type keratinocytes transfected with pGL3P-mUbcPPRE, and treatment with L-165041 resulted in a further increase in luciferase activity compared with control transfections (Fig.  3E). Both of these effects were not found in similarly treated keratinocytes from PPAR␤-null mice, although treatment with L-165041 did cause a minimal response in the null cells (Fig.  3E); this is likely the result of activation by either PPAR␣ or PPAR␥, which are expressed at considerably lower levels in keratinocytes. Luciferase activity was not increased in wildtype keratinocytes transfected with the reporter construct containing the mutant PPRE, and luciferase activity was not changed by the addition of ligand (Fig. 3E). Lastly, electrophoretic mobility shift analysis shows that PPAR␤ can indeed bind to the newly identified PPRE located in the murine ubiquitin C promoter (Fig. 3F). Significant binding was only detected when both PPAR␤ and RXR␣ were added to the reaction mixture, and this binding was inhibited by the addition of non-radiolabeled competitor. When the PPRE was mutated, no binding was detected. Co-incubation with an anti-PPAR␤ antibody resulted in a supershift of the complex, demonstrating FIG. 2. PPAR␤-dependent induction of ubiquitin C. A, confirmation of differential gene expression detected by microarray analysis using Northern blot analysis. mRNA encoding ubiquitin C is higher in TPA-treated wild-type (ϩ/ϩ) mice than in TPA-treated PPAR␤-null (Ϫ/Ϫ) mice. The majority of differentially regulated mRNAs are those that are significantly higher in TPA-treated PPAR␤null mice and not detected in similarly treated wild-type mice. B, representative Western blot of intracellular free ubiquitin levels, showing that levels of free ubiquitin in skin are increased in wild-type (ϩ/ϩ) mice in response to TPA, but not in PPAR␤-null (Ϫ/Ϫ) mice. -Fold change (⌬) represents the increase in free ubiquitin (normalized to lactate dehydrogenase loading control) as compared with control wildtype, based on analysis from three to four samples per group. Asterisk (*), significantly greater than control, p Յ 0.05. C, higher molecular weight proteins that are immunoreactive for ubiquitin, are increased by TPA treatment in wild-type (ϩ/ϩ) mouse skin and not in PPAR␤-null (Ϫ/Ϫ). D, PPAR␤-specific ligand (L-165041) induces ubiquitin C mRNA in cultured keratinocytes from wild-type (ϩ/ϩ) mice but not in PPAR␤-null (Ϫ/Ϫ) keratinocytes. -Fold change (⌬) represents the increase in ubiquitin C mRNA (normalized to glyceraldehyde-3-phosphate dehydrogenase (GAPDH) loading control) as compared with control wild-type. Representative Northern blot from two independent sets of samples. Asterisk (*), significantly greater than control, p Յ 0.05. further specificity. Similar results were obtained from electrophoretic mobility shift analysis using PPAR␣ and PPAR␥ showing significant binding in the presence of both, PPAR and RXR␣, inhibition with addition of non-radiolabeled competitor and no binding with the mutated PPRE (data not shown). Combined, these data demonstrate that expression of ubiquitin C in skin is dependent on PPAR␤, and represents a new target gene for this nuclear receptor.
If the regulation of ubiquitin synthesis and subsequent degradation of target proteins in response to TPA treatment contribute to the enhanced hyperplasia observed in PPAR␤-null mouse skin, then treatment with a proteasome inhibitor in addition to TPA should cause a similar exacerbation of epithelial cell proliferation in wild-type mice. Indeed, the epidermis of wild-type mice treated with TPA and a proteasome inhibitor (MG-132) showed enhanced hyperplastic response compared with control, similar to that observed in the epidermis of PPAR␤-null mouse (Fig. 4). MG-132 treatment in conjunction with TPA application in PPAR␤-null mice did not result in any further increase in epithelial cell proliferation (Fig. 4). These results provide evidence that PPAR␤ regulates ubiquitin C expression and subsequent protein turnover resulting from ubiquitination of target proteins in response to phorbol ester. DISCUSSION These studies clearly demonstrate a novel function for PPAR␤ in attenuating epithelial cell proliferation stimulated by phorbol esters that has a profound effect on tumorigenesis.  bold and underlined). B, sequence comparison between murine and human ubiquitin C PPRE (GenBank TM accession number D63791). The PPAR binding half-site (5Ј region) is conserved in the human sequence in 5 of the 6 bases. The PPRE and the TATA box are highlighted in bold and underlined. C, transfection of HepG2 cells with human (hUbc) ubiquitin C promoter reporter constructs in the presence or absence of PPAR␤. Note the increased luciferase activity accompanying transfection with PPAR␤ with both constructs. Values represent the mean Ϯ S.E. Asterisk (*), significantly different from control, p Յ 0.05. D, transfection of HepG2 cells with murine (mUbc) ubiquitin C promoter reporter constructs in the presence or absence of PPAR␤. Note the increased luciferase activity accompanying transfection with PPAR␤ with both constructs. Values represent the mean Ϯ S.E. Asterisk (*), significantly different from control, p Յ 0.05. E, PPAR␤-dependent increase in reporter activity in wild-type (ϩ/ϩ) primary keratinocytes transfected with pGL3P-mUbcPPRE in response to L-165041 (10 M), not found in PPAR␤-null (Ϫ/Ϫ) keratinocytes. A single point mutation in the PPAR␤ half-site of the PPRE significantly reduces induction of luciferase activity by PPAR␤ and L-165041 (mutant PPRE, see "Experimental Procedures"). Values represent the mean Ϯ S.E. Asterisk (*), significantly different from control, p Յ 0.05. F, PPAR␤ binds to PPRE in murine ubiquitin C promoter in the presence of RXR␣. In vitro translated receptors (PPAR␤ or RXR␣) were incubated in the presence of radiolabeled mUbc PPRE (lanes 1-5), and a competition assay was performed (lanes 6 -8) using the indicated molar excess of unlabeled PPRE (ϫ10 -100). Four bases in the mUbc PPRE half-site (5Ј region of the DR-1) were mutated from 5Ј-AGGTCACAGCCCT-3Ј, to 5Ј-AAGCTTCAGCCCT-3Ј, with mutated bases indicated in bold. In vitro translated receptors incubated in the presence of radiolabeled mutant PPRE show no binding (lane 9). Co-incubation with an anti-PPAR␤ antibody results in a supershift (lane 11). The consensus PPRE from the reported literature and the PPRE located in the murine Ubc promoter are shown, with differences in the murine sequence underlined.
Indeed, in the absence of a functional PPAR␤, the onset of papilloma formation, papilloma size, and tumor multiplicity are exacerbated. No evidence of early dysplasia, early carcinomas, or carcinomas was found in either genotype. This was not unexpected because these studies were not carried out beyond 6 months of promotion when metastatic growth is typically observed. It remains a possibility that the increased cell pro-liferative events that occur in the absence of PPAR␤ activation could contribute to the enhanced conversion of papillomas to adenocarcinomas and/or carcinomas. Further work is required to determine whether PPAR␤ can also function in controlling the extent of metastasis. The enhanced epithelial cell proliferation in PPAR␤-null mice treated topically with TPA is functionally significant, and likely contributes to the observed differences in skin tumor formation between wild-type and PPAR␤-null mice in this model system. Results from these studies indicate that the enhanced epithelial hyperplasia observed in PPAR␤-null mouse skin in response to tumor promotion is because of changes in both cell proliferation, as shown by enhanced BrdUrd labeling, and to the absence of increased signaling for apoptosis as found in similarly treated wild-type mouse skin, as indicated by the lack of a TPA-induced increase in caspase-3 activity, an enzyme that has a central role in modulating apoptosis (44), and increased TUNEL staining of apoptotic cells; both known to occur in wild-type mice in response to TPA (43).
Results from these studies showing that PPAR␤ functions to promote apoptosis are of interest, because others have recently provided evidence that PPAR␤ functions to suppress apoptosis (38). This disparity in results is likely because of differences in model systems as the present studies focused on in vivo analysis of skin, whereas the Di-Poi et al. (38) study used cultured keratinocytes that had undergone multiple passages. Additionally, the role of PPAR␤ in an anti-apoptotic response was established with a model system where a role for this receptor in response to tumor necrosis factor-␣ signaling was evaluated, whereas the present study establishes a link between PPAR␤ Treatment with proteasome inhibitor (MG-132) prior to TPA application causes enhanced epithelial hyperplasia in wild-type (ϩ/ϩ) mice, a phenotype similar to that observed in PPAR␤-null (Ϫ/Ϫ) mice treated with TPA and no proteasome inhibitor. and the stimulation of apoptosis in a tumor promotion model. It is worth noting that PPAR␤-null mice used for the present studies as well as those from the research group showing an anti-apoptotic role of PPAR␤ in keratinocytes (38) have both provided evidence of enhanced epithelial cell proliferation in response to TPA (33,36). Furthermore, the current finding that PPAR␤ functions in the promotion of apoptosis in skin in response to tumor promotion is more consistent with the observed phenotype of enhanced epithelial cell hyperplasia and exacerbated skin carcinogenesis observed in PPAR␤-null mice, in contrast to the reported anti-apoptotic role for PPAR␤ in a different model system (38).
Previous work by others showed that ubiquitin B and ubiquitin C mRNAs are higher in mouse epidermis and squamous cell carcinomas in response to TPA (45), and the present studies demonstrate that induction of ubiquitin C is mediated via activation of PPAR␤. Whereas PPAR␤ does not influence ubiquitin B expression in response to TPA, several lines of evidence demonstrate that ubiquitin C represents a putative novel target gene that can be modulated by PPAR␤. This conclusion is based on combined observations including: 1) TPA induces ubiquitin C mRNA and protein in wild-type mouse skin and a PPAR␤ ligand can induce ubiquitin C mRNA in keratinocytes, both events that are absent in similarly treated PPAR␤-null samples; and 2) identification and characterization of a functional PPRE in the mouse ubiquitin C promoter. Whereas the PPAR␣ and PPAR␥ were also capable of activating the ubiquitin C promoter reporter construct in HepG2 cells, their contribution to the regulation of ubiquitin C expression in skin is likely less than that for PPAR␤ because only the latter PPAR isoform is expressed at significant levels in this tissue (18). However, possible roles for the regulation of ubiquitin C by PPAR␣ and PPAR␥ in other tissues deserve further consideration, especially because it was recently demonstrated that a PPAR␥ ligand can induce proteasome degradation of cyclin D1 in breast cancer cells that express high levels of PPAR␥ (46). This is consistent with previous work by others, demonstrating that regulation of fatty acid transporters is mediated by PPAR␥ in adipose, whereas regulation of the same target genes is mediated by PPAR␣ in liver, suggesting that expression patterns of PPARs may in part determine target gene regulation via similar response elements (47,48).
It should be noted that the induction of ubiquitin C by the PPAR␤ ligand in keratinocytes is not robust. This may reflect the presence of endogenous ligands for PPAR␤, or the possibility that the receptor is active on the ubiquitin C promoter in the absence of ligand. In fact, the higher constitutive level of ubiquitin C mRNA in keratinocytes (Fig. 2D) that is not found in skin ( Fig. 2A), supports the notion that the presence of an endogenous ligand accounts for the lack of a large increase in ubiquitin C mRNA expression in keratinocytes treated with L-165,041. Additionally, PPAR␤ could potentiate the activity of other transcription factors that regulate the mouse ubiquitin C promoter. Indeed, a number of other known and putative transcription factor binding sites were found in the mouse ubiquitin C promoter, including sites for AP-1, AP-2, c-Jun, NF-1, Sp-1, NF-B, and the glucocorticoid receptor. It is also worth noting that the level of ubiquitin C mRNA induction and increase in reporter activity are consistent with observations made in other models examining similar responses of the human and rat ubiquitin C promoter to glucocorticoids (49,50). In this case, the level of induction is known to have significant biological effects including lower levels of free ubiquitin and reduced levels of ubiquitinated proteins (49). Furthermore, the fact that glucocorticoids can significantly up-regulate ubiquitin C expression (49 -52) suggests that a functional glucocorticoid re-ceptor response element may contribute to gene modulation and provides evidence for an important role for nuclear receptor-mediated regulation of ubiquitin C expression. This is of interest given the putative glucocorticoid receptor response element identified in the mouse ubiquitin C promoter fragment analyzed in this study. Lastly, two additional DR-1s that are potential PPREs are located ϳ3.0 kb upstream of the TATA box in the mouse ubiquitin C promoter but their functional significance has not been established. Combined, these results strongly support the notion that ubiquitin C is a PPAR␤ target gene in skin that significantly influences the ubiquitin-proteasome proteolytic pathway. However, given the complex nature of the ubiquitin C promoter, we cannot completely rule out the possibility that the effect of PPAR␤ on ubiquitin C gene expression is indirect.
The ubiquitin-proteasome degradation pathway is a major mechanism for modulation of protein turnover, and has a critical role in maintaining cellular homeostasis. A number of critical proteins that modulate progression through the cell cycle including protein kinases, cyclins, transcription factors (NFB, IB, and E2F-1), tumor suppressor protein (p53), and oncoproteins (Jun, ␤-catenin), are degraded/regulated by the ubiquitin-proteasome pathway (53)(54)(55)(56). Thus, because this pathway is indispensable in maintaining cellular balance and regulation of cell division, its modulation could play a role in the prevention or progression of cancer. Disruption of the ubiquitin-proteasome pathway can significantly alter molecular and biochemical events because of imbalanced activity of proteins including cell cycle regulatory components and tumor suppressors, resulting in cell proliferation and cancer (54). Consistent with a critical impact on cell cycle regulation, data from the current studies suggest that the induction of ubiquitin C mediated by PPAR␤ may attenuate epithelial cellular proliferation in response to phorbol ester. Immunoblots revealed that while the relative amount of ubiquitinated proteins is increased in TPA-treated wild-type mouse skin, this effect is absent in PPAR␤-null mice where ubiquitin C expression is diminished. Additionally, co-treatment of TPA with a proteasome inhibitor in wild-type mice resulted in an exacerbated epithelial hyperplastic response leading to a phenotype similar to that observed in the TPA-treated PPAR␤-null mouse epidermis. Whereas there are at least two different ubiquitin genes (ubiquitin B and ubiquitin C) that respond to TPA by elevated expression, Western blot analysis demonstrates that the induction of ubiquitin C appears to be critical for generating appreciable increases in free, intracellular ubiquitin. This may be because of the fact that mouse ubiquitin C encodes for 10 separate ubiquitin moieties (57), whereas ubiquitin B encodes for only 4 (45).
It is also of interest to note that administration of TPA to PPAR␤-null mouse skin resulted in significant up-regulation of a number of mRNAs that were not found in similarly treated wild-type mice. There are at least two possible mechanisms that could underlie this effect. First, it is possible the lack of protein turnover resulting from reduced ubiquitin C expression in PPAR␤-null mice leads to altered activity of other signaling pathways leading to enhanced activity of other transcription factors. Second, this effect could be the result of enhanced activity of other transcription factors because of the absence of PPAR␤, as there is evidence that PPAR␤ may inhibit transcription by binding to PPREs and/or modulating intracellular levels of co-repressors (58). Further work is necessary to determine whether either of these mechanisms contributes to the large difference in the molecular regulation of gene expression observed in PPAR␤-null mouse skin in response to TPA as compared with controls, and whether or not these changes contribute to the observed differences in skin carcinogenesis.
The finding that PPAR␤ regulates ubiquitin C that in turn likely modulates protein turnover of critical effector proteins that could influence cell cycle progression, identifies a novel role for this nuclear hormone receptor that is not directly linked to lipid metabolism per se, a general role described for all three PPAR isoforms to date. Thus, it is possible that PPAR␤ activation could be targeted for chemoprevention of skin cancer through its ability to inhibit cell proliferation. However, to induce ubiquitin C expression in skin, there must be production of an endogenous ligand for PPAR␤ and to date, this remains to be identified.