Block of Nuclear Receptor Ubiquitination

Peroxisome proliferator-activated receptor δ (PPARδ) is a ligand-activated transcription factor involved in many physiological and pathological processes. PPARδ is a promising therapeutic target for metabolic, chronic inflammatory, and neurodegenerative disorders. However, limited information is available about the mechanisms that control the activity of this nuclear receptor. Here, we examined the role of the ubiquitinproteasome system in PPARδ turnover. The receptor was ubiquitinated and subject to rapid degradation by the 26 S proteasome. Unlike most nuclear receptors that are degraded upon ligand binding, PPARδ ligands inhibited the ubiquitination of the receptor, thereby preventing its degradation. Ligand binding was required for inhibition of the ubiquitination since disruption of the ligand binding domain abolished the effect. Site-directed mutagenesis showed that the DNA binding domain was also required, indicating that ligands preferentially stabilized the DNA-bound receptor. In contrast, the activation function-2 domain and co-repressor binding site were not involved in ligand-induced stabilization. Block of ubiquitination by ligands may be an essential step to avoid rapid degradation of a receptor, like PPARδ, with a very short half-life and sustain its transcriptional activity once it is engaged in transcriptional activation complexes.

Peroxisome proliferator-activated receptors (PPARs) 2 belong to the superfamily of nuclear hormone receptors and act as ligand-activated transcription factors (1). PPARs are activated by a diverse group of lipophilic compounds, including long-chain fatty acids, prostaglandins, and leukotrienes (2). PPARs form heterodimers with the retinoic X receptor and bind DNA in correspondence to specific PPAR response elements (PPRE) located in the promoter of target genes (1). Unli-ganded receptors maintain the promoter in a repressive or inactive state (3). Ligand binding induces a conformational remodeling of the receptor, resulting in the release of co-repressor molecules and recruitment of co-activators necessary for transcriptional activation (3). The PPAR subfamily includes three isotypes, named ␣, ␦ (or ␤), and ␥, that share extensive structural homology (1). Although all three isotypes act as lipid sensors and are involved in various aspects of lipid metabolism, they have distinct tissue distribution, ligand specificity, and functions (2,4). PPAR␣ is involved in fatty acid metabolism, and high affinity ligands of this receptor, like fenofibrate and bezafibrate, are effective hypolipidemic drugs (5). PPAR␥ prevalently controls lipid and glucose metabolism, and PPAR␥ agonists, like rosiglitazone and pioglitazone, are widely used antidiabetic drugs (6). PPAR␦, which is the less studied of the three isotypes, has been implicated in wound healing, inflammatory responses, and embryo implantation in addition to lipid metabolism (4,7). PPAR␦ is a potential therapeutic target for diseases such as atherosclerosis and other inflammatory, metabolic, and neurodegenerative disorders (7). PPAR␦ has also been associated with cancer (8). PPAR␦ is overexpressed in colon, endometrial, and head and neck cancers (9 -12). PPAR␦ agonists stimulate proliferation and survival of cancer cells in vitro (13)(14)(15)(16) and promote tumor growth in mice (17)(18)(19). Consistent with a tumor promoting role, PPAR␦ has been shown to increase the expression of anti-apoptotic genes, like ILK (integrin-linked kinase) and PDK1 (3-phosphoinositide-dependent kinase-1), and to activate the pro-survival Akt signaling pathway in keratinocytes (20). However, the involvement of this nuclear receptor in carcinogenesis is still controversial. Somatic knockdown of PPAR␦ decreased growth of human colon cancer xenografts in mice (21), whereas in other studies the frequency of colon polyps was not affected or even increased in PPAR␦ knock-out mice (22,23) and upon treatment with PPAR␦ ligands (24). Thus, this nuclear receptor is involved in many critical cell functions, including proliferation, survival, and differentiation. However, the mechanisms mediating its effects and the outcome in different cells and tissues remain unclear.
Although much is known about the mechanisms that regulate the activity of most nuclear receptors, very limited information is available regarding PPAR␦. At the transcriptional level, PPAR␦ was shown to be a downstream target of the oncogenic Wnt/APC/␤-catenin pathway, showing for the first time a link with colon carcinogenesis (9). Activated oncogenic Ras stimulated PPAR␦ expression in intestinal cells (25), whereas activator protein-1 (26) and C/EBP␣ (CCAAT/enhancer-binding protein) (27) controlled its expression in keratinocytes.
Activity and expression of PPAR␦ were inhibited by nonsteroidal anti-inflammatory drugs and cyclooxygenase-2 inhibitors in colon cancer cells, relating this nuclear receptor to the chemopreventive effects of this class of anti-inflammatory compounds (9,14). Less is known about post-transcriptional mechanisms that control PPAR␦ protein level and activity. A better understanding of the multiple mechanisms that modulate ligand-dependent and -independent activity might provide hints into the involvement of PPAR␦ in human diseases and indicate ways to exploit its therapeutic potential.
In this study we examined the role of the ubiquitin-proteasome system (UPS) in basal and ligand-dependent turnover of PPAR␦. The UPS is the major cellular machinery responsible for degradation of proteins and determines the level of critical regulatory factors, including nuclear receptors and transcription factors (28 -31). We found that PPAR␦ was ubiquitinated and subject to rapid degradation by the 26 S proteasome. However, unlike most nuclear receptors that are degraded upon ligand binding, PPAR␦ was stabilized by its ligands. PPAR␦ agonists inhibited ubiquitination of the receptor with consequent block of its degradation. Thus, PPAR␦ ligands have a dual effect on the receptor. Ligands induce conformational changes, allowing co-activator binding and promoter transactivation, and at the same time prevent ubiquitination of the receptor engaged in transcriptional activation complexes. Block of ubiquitination may be an essential step to avoid rapid degradation of a receptor with very short half-life, like PPAR␦, and sustain its transcriptional activity.
Reporter Assay-Cells (2 ϫ 10 4 cells/well) were plated in 48-well plates and transfected with the PPREx3-tk-Luc, pRL-SV40 vectors and, when indicated, with PPAR␦ expression vectors. Dual-luciferase reporter assay was performed according to manufacturer's instructions (Promega, Catalys AG, Wallisellen, CH) using a Turner luminometer (Turner Design, Sunnyvale, CA). Data were normalized for Renilla luciferase activity used as control for transfection efficiency.
PPAR␦ Half-life-U2OS cells were grown in 100-mm dishes to 90% confluence. Cells were transfected with 3 g of His-PPAR␦ expression vector. The next day cells were plated into 6-well plates at a concentration of 3.5 ϫ 10 5 cells/well. After overnight incubation with or without a PPAR␦ ligand, cells were treated with 50 M puromycin and harvested at the indicated times for Western blotting.
PPAR␦ Ubiquitination-U2OS cells were transfected with His-PPAR␦ and HA-Ub and incubated with proteasome inhibitors and/or PPAR␦ ligands. After 24 h cells were lysed in a denaturing buffer consisting of 8 M urea, 0.1 M Na 2 HPO 4 / NaH 2 PO 4 , and 10 mM imidazole, pH 8.0. His-PPAR␦ was pulled down using His-select nickel affinity gel (Sigma) starting with ϳ100 g of protein from cell lysates. After multiple washes with urea buffer, proteins were eluted from the beads with the same buffer supplemented with 250 mM imidazole, pH 6.0. Aliquots of whole lysate and flow-through samples corresponding to 15 g of proteins and an equivalent fraction of the eluates were loaded on gels and analyzed by Western blotting.
RNA Isolation and Analysis-Cells (1 ϫ 10 6 cells/flask) were incubated with PPAR␦ ligands overnight and proteasome inhibitors for 4 h. For RNA interference studies, cells were transfected with 100 nM small interfering RNA (Ambion Ltd., Huntingdon UK) specific for PPAR␦ (siPPAR␦) or the firefly luciferase gene (siGL3) using Lipofectamine 2000, incubated for 48 h, and then treated with ligands and/or proteasome inhibitors. Total RNA was isolated using Trizol (Invitrogen) and further purified with RNeasy MiniKit (Qiagen). RNA concentration was determined using a NanoDrop spectrophotometer. RT-PCR was performed using the SuperScript One-Step RT-PCR system (Invitrogen) and gene-specific primers for the adipose differentiation-related protein (ADRP), glyceraldehyde-3-phosphate dehydrogenase, and PPAR␦. PCR products were run on 2% agarose gels, stained with ethidium bromide, and visualized using the AlphaImager 3400. Band intensity was determined using the AlphaEase software. For quantitative real time RT-PCR, 1 g of total RNA was reverse-transcribed using the SuperScript First-Strand Synthesis system (Invitrogen). Real time PCR was performed on a 7900HT Fast Real Time PCR System (Applied Biosystems, Foster City, CA) using primer sets for ADRP and ␤-actin and Absolute SYBR Green ROX Mix (ABgene, Epsom, UK). Standard curves were generated for each primer set, and ␤-actin RNA was used as control to quantify ADRP RNA. Sequences of primers and small interfering RNAs are available as supplemental Table S1.

PPAR␦ Ligands Stabilize the Receptor-Ligand-dependent
transactivation of nuclear receptors is often associated with proteasome-mediated degradation of the receptor (33,34). Ligand-induced proteolysis serves as a common mechanism to terminate transcriptional activity of the ligand-activated receptors (33,34). To determine whether PPAR␦ was subject to a similar ligand-dependent control, human lung cancer cell lines expressing different levels of the receptor were incubated overnight with a PPAR␦-selective ligand. PPAR␦ protein level increased after incubation with L165041 in cells expressing high (H441) and low (H358) levels of the receptor, whereas there was no effect in A549 cells in which PPAR␦ was undetectable in the immunoblots both before and after ligand treatment (Fig. 1A). The effect of the ligand on the receptor levels correlated well with the ligand ability to transactivate the PPARresponsive reporter in these cells (Fig. 1B). Ligand treatment increased reporter activity ϳ13and 5-fold in H441 and H358 cells, respectively. The effect on A549 and U2OS cells was minimal (ϳ2-fold), in agreement with the very low levels of PPAR␦ in these cells.
To better differentiate between transcriptional and posttranslational effects, in subsequent experiments we used U2OS cells expressing recombinant PPAR␦ from a heterologous promoter. Increased levels of PPAR␦ were seen when PPAR␦-expressing U2OS cells were treated with synthetic agonists (i.e. GW501516 and L165041) and a stable prostaglandin analogue (cPGI 2 (carboprostacyclin)) that binds to PPAR␦ (Fig. 1, C and D). On the other hand treatment of cells with prostaglandin E 2 , a prostaglandin that activates PPAR␦ indirectly and does not bind to the receptor (19), did not have any effect on the receptor protein level, suggesting that direct binding to the receptor was required to increase its stability. A time course experiment in PPAR␦-expressing U2OS cells showed that the ligand acted rapidly (Fig. 1D). A substantial increase of PPAR␦ (ϳ2-fold) was seen already after 4 h, and the levels continued to increase (up to ϳ4-fold) after 24 h. The effect on the protein level was reversible, and PPAR␦ returned to base-line levels within 4 h after the removal of the ligand (Fig. 1E). The effect seen in cells express-ing PPAR␦ from a heterologous promoter made it unlikely that the increased protein level was a consequence of enhanced transcription. In fact, PPAR␦ transcript levels measured by RT-PCR were identical in PPAR␦-expressing U2OS cells incubated with or without ligand (supplemental Fig. S1).
The data described above suggested that ligands could affect PPAR␦ levels by influencing protein turnover. To estimate PPAR␦ half-life, we transiently transfected U2OS cells with a His-PPAR␦ expression vector and monitored protein level by Western blotting after the addition of the protein synthesis inhibitor puromycin. His-tagged PPAR␦ level was greatly reduced within 1 h of puromycin treatment ( Fig. 2A). The estimated PPAR␦ half-life under these conditions was ϳ30 min (Fig. 2C). Next, His-PPAR␦ expressing U2OS cells were incubated with puromycin in the presence of L165041 to determine the effects of the ligand on the receptor half-life. Incubation with L165041 increased PPAR␦ protein level and completely prevented the decline induced by puromycin (Fig. 2, B and C). Thus, the ligand increased the receptor protein level by extending its half-life considerably.
PPAR␦ Turnover Is Controlled by the Ubiquitin-Proteasome Pathway-The 26 S proteasome is the major cellular complex responsible for the degradation of proteins, including nuclear receptors and transcription factors (28 -31). To determine whether proteasome-mediated degradation played a role in controlling the turnover of PPAR␦, H441 and H358 cells expressing, respectively, high and low levels of the receptor were treated with the proteasome inhibitor PS341. In both cell lines PPAR␦ protein level increased, consistent with reduced degradation of the receptor in the presence of the proteasome inhibitor (Fig. 3A). Similar results were obtained by treating cells with another proteasome inhibitor, MG132 (data not shown). To further study the role of the 26 S proteasome in PPAR␦ turnover, U2OS cells expressing recombinant His-PPAR␦ were treated with PS341, and protein level was determined at different times by Western blotting. As shown in Fig.  3B, treatment with PS341 led to a significant accumulation of PPAR␦ (ϳ3-fold) already after 4 h of incubation. Next, U2OS cells were treated either with puromycin, PS341, or both compounds together (Fig. 3C). PPAR␦ level decreased rapidly in the presence of puromycin, whereas treatment with puromycin and PS341 resulted in a higher level of the receptor, indicating that PPAR␦ turnover was under the control of the proteasome.
A necessary step for targeting proteins to the proteasome is the covalent attachment of ubiquitin (Ub) chains catalyzed by Ub ligases (28). To determine whether PPAR␦ was ubiquitinated, U2OS cells were transfected with His-PPAR␦ and HA-Ub expression vectors and then incubated with a proteasome inhibitor for 4 h. His-tagged PPAR␦ was pulled down using nickel affinity gel under denaturing conditions. This strategy avoided the use anti-PPAR␦ antibodies that had  revealed limited specificity in preliminary immunoprecipitation experiments. In addition, it used highly stringent conditions to minimize non-covalent interactions of the receptor with potential ubiquitinated protein partners. Whole cell lysate and flow-through samples were examined along with the eluates to control for quantitative recovery of His-PPAR␦ throughout the procedure (Fig. 3D, lanes 1-4). The specificity of the pulldown was demonstrated using cells transfected with Histag empty vector (lanes 1, 3, and 5). In His-PPAR␦-expressing cells, high molecular weight protein species were detected with the anti-HA antibody indicating the presence of PPAR␦ with covalently linked poly-Ub chains (lane 6). No such bands were detected in the eluate from empty vectortransfected cells (lane 5).
Proteasome activity is required for receptor turnover, and its inhibition generally leads to reduced transcriptional activity of nuclear receptors (34,35). To determine whether a functional proteasome was required for PPAR␦ transcriptional activity, U2OS cells stably expressing recombinant PPAR␦ and H441 cells were transfected with a PPAR-responsive reporter and incubated with PS341 in the presence or absence of a PPAR␦ ligand (Fig. 4A). GW501516 increased reporter activity ϳ 6 and 12-fold in U2OS and H441 cells, respectively. PS341 induced a ϳ2-3-fold increase in the absence of the ligand in both cell lines. The combination of PS341 and ligand increased reporter activity by ϳ13 and 26-fold in U2OS and H441, respectively. Thus, inhibition of proteasome activity did not negatively affect ligand-dependent activation of PPAR␦. Ligands and proteasome inhibitors had a positive and apparently synergistic effect on the activity of both recombinant and endogenous PPAR␦ in the reporter assays. To determine whether ligands and proteasome inhibitors had similar effects on transcription of endogenous PPAR␦ target genes, H358 and H441 cells were treated as described above, and the level of the ADRP RNA was monitored by real time RT-PCR or conventional RT-PCR. The ADRP gene contains a PPRE, and its transcription is activated by PPAR␦ agonists (36). In both cell lines PPAR␦ ligand increased ADRP RNA levels (by ϳ10 and 25-fold in H358 and H441 cells, respectively). This effect was enhanced in the presence of the proteasome inhibitor (up to ϳ25 and 100-fold in H358 and H441, respectively) (Fig. 4B). Neither the ligand nor proteasome inhibitor affected endogenous PPAR␦ RNA levels in H358 and H441 cells, confirming that they acted post-transcriptionally, increasing the protein level of endogenous PPAR␦ ( Fig. 4C and data not shown). The specificity of these effects was further assessed by RNA interference. The increase induced by the PPAR␦ ligand both alone and in combination with PS341 was attenuated in cells in which PPAR␦ had been silenced (up to 75%) by a small interfering RNA against PPAR␦ before the incubation with GW501516 and/or PS341 (Fig. 4C). Despite the evident reduction of PPAR␦ RNA, ligand-dependent activation of ADRP transcription was reduced only partially by RNA interference because of the stabilizing effect of the ligand and proteasome inhibitor at the protein level. Taken together, these data indicate that inhibition of the proteasome did not reduce PPAR␦ activity but, along with PPAR␦ ligands, led to the accumulation of transcriptionally competent receptor. To determine whether the accumulated protein had also the proper subcellular localization, PPAR␦-expressing U2OS cells were treated with PS341 or a selective ligand before isolation of the nuclear and cytoplasmic fractions (Fig. 4D). Although a small amount was found in measured with a Dual-Luciferase assay system using Renilla luciferase as a control for transfection efficiency. Data represent -fold increase (mean Ϯ S.D. of triplicate experiments) relative to control cells. *, p Յ 0.005 compared with control cells. B, H358 and H441 cells were treated with GW501516 and PS341 as described above. Total RNA was isolated, and the levels of ADRP and ␤-actin were determined by quantitative real time RT-PCR. Data are expressed as ADRP/␤-actin ratios and normalized for the value in untreated control cells. *, p Յ 0.01 C, H358 cells were transfected with 100 nM small interfering RNA against PPAR␦ (siPPAR␦) or firefly luciferase (siGL3) and 48 h later treated with GW501516 and/or PS341 as indicated above. Total RNA was isolated, and ADRP, PPAR␦, and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) RNA was measured by RT-PCR. D, U2OS cells stably expressing FLAG-PPAR␦ were incubated with or without PS341 (10 M) for 4 h or L165041 (5 M) overnight. Nuclear (n) and cytoplasmic (c) fractions were separated by sucrose gradient centrifugation. PPAR␦ was detected using an anti-FLAG antibody. the cytoplasm, PPAR␦ was detected predominantly in the cell nucleus of both PS341 and ligand-treated cells.
Ligands Prevent Proteasomal Degradation of PPAR␦ by Blocking Its Ubiquitination-Ligand-induced stabilization of PPAR␦ might be the consequence of inhibition of proteasomemediated degradation. To exclude the possibility that PPAR␦ ligands might affect activity of the 26 S proteasome, U2OS cells expressing a Ub-enhanced green fluorescent protein fusion protein were treated with ligands or PS341 as a positive control and analyzed by flow cytometry. Although PS341 induced the expected accumulation of Ub-enhanced green fluorescent protein (EGFP), none of the PPAR␦ ligands changed Ub-EGFP levels (supplemental Fig. S2), suggesting that the ligands did not inhibit the 26 S proteasome and might affect PPAR␦ protein turnover by acting directly on the receptor.
To determine whether PPAR␦ ligands modulated the ubiquitination or proteasomal degradation of the receptor, we treated U2OS cells transfected with HA-Ub and His-PPAR␦ expression vectors with L165041 in the presence and absence of PS341. If the ligand blocked proteasomal degradation of the receptor, one would expect accumulation of ubiquitinated PPAR␦. On the contrary, a block of ubiquitination would lead to reduced amounts of ubiquitinated PPAR␦ in the presence of ligand. As expected, L165041 and PS341 alone and in combination increased PPAR␦ in the cell lysates (Fig. 5A, top panel). However, although PS341 led to the accumulation of ubiquitinated PPAR␦, treatment with the ligand both in the presence and absence of PS341 led to a reduction of the amount of ubiquitinated PPAR␦ (Fig. 5A, bottom panel, lanes 2 and 4) compared with control and PS341-treated cells (lanes 1 and 3). Thus, the ligand selectively affected the ubiquitination of the receptor rather than its proteasomal degradation. The effect of the ligand on PPAR␦ ubiquitination was reversible and depended on its continuous presence in the culture medium as seen before with the effect on PPAR␦ protein level (Fig. 1E). Removal of the ligand restored the ability of the cells to ubiquitinate the receptor within 4 h, concomitant with the decrease of the receptor level (Fig. 5B). To confirm that ligand binding was required to prevent PPAR␦ ubiquitination, a similar experiment was done with a truncated form of the receptor (PPAR␦-(1-299)), which missed most of the ligand binding domain. Disruption of this domain abolished the effects of the ligand on PPAR␦ ubiquitination and stabilization (Fig. 5C). Thus, direct interaction of the ligand with the receptor was absolutely required for inhibition of PPAR␦ ubiquitination.
Ligand-induced Stabilization of PPAR␦ Involves the DNA Binding Domain of the Receptor-Like other nuclear receptors, PPAR␦ has well defined structural and functional domains (1). To understand the role played by distinct domains in ligandinduced stabilization of the receptor, we introduced inactivating mutations in the DNA binding domain (C91A/C94A), in the AF-2 transactivating domain (L432A/E435A), and within the region (F270A) known to be responsible for co-repressor binding (Fig. 6A). Mutations in the DNA binding domain (C91A/C94A) were similar to those known to prevent binding of murine PPAR␦ to DNA (37). The double mutant in the AF-2 domain (L432A/E435A) and the F270A mutation blocked the interaction of murine PPAR␦ with coactivators (38) and core-pressors (37), respectively. The ability of the mutated receptors to activate a PPAR-responsive reporter was tested in luciferase assays along with wild type PPAR␦ (Fig. 6B). The DNA binding domain mutant (C91A/C94A) was less active than the wild type receptor in the reporter assay. The mutation in the putative co-repressor binding site (F270A) had a minimal effect on PPAR␦-induced transactivation. On the other hand, the AF-2 mutant (L432A/E435A) had greatly reduced transcriptional activity compared with the wild type receptor, confirming the critical role of this domain for ligand-dependent transactivation. In addition, the AF-2 mutant acted as a dominant negative, reducing ϳ-fold the activation of the reporter by endogenous PPAR␦.
The ability of ligands to stabilize wild type and mutated receptors was tested in transiently transfected U2OS cells. The L432A/E435A and F270A mutants were stabilized by the ligand to a level comparable with wild type PPAR␦ (Fig. 6, C, top panel,  D). The effect of the ligand was reduced with the C91A/C94A double mutant (Fig. 6, C and D). Ligand-induced stabilization of the L432A/E435A-and F270A-mutated receptors was associated with reduced ubiquitination as with the wild type receptor, indicating that mutations in the transactivation domain and co-repressor binding site did not affect the ability of the ligand to block ubiquitination and degradation of the receptor (Fig.  6C, bottom panel). In contrast, the ability of the ligand to affect receptor ubiquitination was considerably reduced in the case of the C91A/C94A mutant (Fig. 6C, bottom panel). This indicated that, although it was not necessary for ubiquitination and degradation of the receptor, DNA binding enhanced the effect of the ligand on receptor stability.

DISCUSSION
The nuclear hormone receptor superfamily includes at least 48 distinct members in the human and mouse genome (39). By controlling the activation state of several genes in response to specific stimuli, nuclear receptors exert fundamental functions in many physiological processes. The activity of nuclear receptors themselves needs to be tightly controlled at multiple levels to ensure rapid and timely ordered responses to the specific stimuli. Furthermore, distinct control mechanisms seem to have evolved for each nuclear receptor depending on its physiological function and ligand properties. In this study we examined basal and ligand-dependent turnover of PPAR␦ and the role of the UPS in this process. PPAR␦ is constitutively expressed at low levels in many tissues. However, a clear understanding of its functions in different cells and tissues is still missing. Understanding the mechanisms and pathways that control PPAR␦ level and activity may provide insights into the diverse functions attributed to this nuclear receptor. Here we show that, like other nuclear receptors, PPAR␦ protein turnover is under the control of the UPS. However, unlike most nuclear receptors that are degraded upon ligand binding, PPAR␦ was more stable in the presence of ligands. PPAR␦ ligands selectively inhibited ubiquitination of the receptor, thereby blocking its degradation by the proteasome. Ligandinduced stabilization may be an essential step to avoid rapid degradation of a receptor with very short half-life, like PPAR␦, and sustain its transcriptional activity once engaged in transcriptional activation complexes. Furthermore, our study reveals important differences between PPAR␦ and the other PPAR isotypes as far as ligand-dependent control of receptor turnover, which may have implications for the diverse functions of PPARs in various cells and tissues.
The Ubiquitin-Proteasome System and PPAR␦ Turnover-The UPS is responsible for degradation of many cellular proteins, including nuclear receptors and transcription factors (28 -31, 33, 34). The first step in proteasome-mediated proteolysis is the covalent modification of proteins by the addition of Ub, a small 76-amino acid polypeptide (28,31). Protein ubiquitination involves three classes of enzymes acting in a highly coordinated fashion: E1 Ub-activating enzymes, E2 Ub-conjugating enzymes, and E3 Ub ligases. E3 Ub ligases have the highest substrate specificity and represent the largest family of Ub-processing enzymes (28,31). Ubiquitinated proteins are generally marked for rapid degradation by the proteasome (28). However, they can be rescued by the action of deubiquitinating enzymes that remove Ub chains before proteolytic attack (40). Deubiquitinating enzymes receive increasing attention since they can participate, along with E3 Ub ligases, in ubiquitination-deubiquitination cycles capable of rapidly modulating the level of target proteins (40). Ligand-induced degradation of nuclear receptors is probably operated by Ub ligases and proteasomal subunits present as integral components of transcription regulatory complexes and acting within the cell nucleus (29,30,33,34). Proteasome-dependent proteolysis serves to ensure both nuclear receptor turnover and timely termination of the transcriptional response (34,35). Accordingly, inhibition of proteasome activity generally leads to reduced transactivation by nuclear receptors (34,35).
In this study we observed that PPAR␦ was ubiquitinated and rapidly degraded by the 26 S proteasome. Brief incubation of cells expressing both endogenous and recombinant PPAR␦ with proteasome inhibitors led to a rapid increase of the receptor. Upon proteasome inhibition most of the receptor was in the cell nucleus similar to control and ligand-treated cells. Moreover, it was transcriptionally competent as shown by luciferase reporter assays and direct assessment of the level of an endogenous target gene. PPAR␦ behavior in this regard was quite different compared with other nuclear receptors, including the estrogen-, androgen-, thyroid hormone-, and retinoic acid receptors, whose transcriptional activity is reduced upon proteasome inhibition (34). Furthermore, although in the absence of ligand, PPAR␦ had a very short half-life of ϳ30 min, the addition of ligand considerably increased its half-life. The effect of the ligands was rapid and specific for compounds able to bind and activate the receptor directly. PPAR␦ protein level increased within 4 h after the addition of ligand to cells and remained high as long as the ligand was present in the culture medium. Removal of the ligand was followed by rapid reversal of the effect with return to the base-line level within 4 h. Also in this regard PPAR␦ behavior was almost unique among nuclear receptors, which generally are negatively regulated by their own ligands (34). Only vitamin D3 receptor has been previously shown to be stabilized by the ligand with similar kinetics (41). PPAR␥ was degraded rather rapidly upon exposure to ligands (42), whereas PPAR␣ was stabilized in the presence of ligands but only transiently (43,44). An increase of PPAR␣ protein level was seen within 3 h and was followed by rapid proteolysis upon continuous exposure to ligands (43). Thus, the system in place for PPAR␦ may be geared to prevent both accumulation of high levels and prolonged activation of the receptor. It is possible that overactivity of PPAR␦ may be detrimental to cells, owing perhaps to the anti-apoptotic function and oncogenic potential associated with this nuclear receptor. The level of PPAR␦ must be kept low and under constant control via UPS-dependent proteolysis. Only in the presence of high concentrations of the specific ligands PPAR␦ would be stabilized and activated. Furthermore, the transcriptional response mediated by the ligand-bound receptor would persist only as long as the ligand is present. Under physiological conditions, termination of the transcriptional response would be ensured by the short half-life and low abundance of natural PPAR␦ ligands, like prostaglandin I 2 . This hypothesis would be consistent with the observation that in processes such as wound healing, there is a sharp increase of PPAR␦ level concomitant with increased production of ligands (45). Also in cancer, upregulation of PPAR␦ is apparently coordinated with up-regulation of cyclooxygenase-2 and consequent increased production of prostaglandin metabolites capable of stabilizing and activating PPAR␦ (10,17). In the absence of this coordinated increase of ligand and receptor levels, PPAR␦ would not be able to exert its anti-apoptotic and growth-promoting functions.
Mechanism of Ligand-induced Stabilization of PPAR␦-Ligand-induced stabilization of PPAR␦ was due to a selective block of receptor ubiquitination and not to interference with later steps of the proteolytic process, i.e. binding to and degradation by the proteasome. Like protein stabilization, reduced ubiquitination of PPAR␦ depended on the continuous presence of the ligand and was rapidly reversed after ligand removal. Disruption of the ligand binding domain in PPAR␦-(1-299) abolished the effect of the ligand on PPAR␦ ubiquitination and proteolysis, although the truncated form of the receptor was still ubiquitinated and degraded by the proteasome. This result supported the idea that ligand-induced stabilization of PPAR␦ was due to the direct interaction of the ligand with the receptor. Ligand binding may induce conformational changes that, in addition to allow co-activator binding and transactivation, block the interaction of PPAR␦ with Ub ligases or, alternatively, promote the activity of deubiquitinating enzymes. We are currently testing these hypotheses and attempting to identify the enzymes involved in ligand-dependent control of the PPAR␦ ubiquitination. Ub-processing enzymes involved in this process may represent critical elements in the pathways regulating the activity of this nuclear receptor.
Using site-directed mutagenesis we investigated the role of distinct receptor domains in the ligand-dependent regulation of PPAR␦ stability. This analysis also revealed additional differences between PPAR␦ and the other PPAR isotypes as far as the mechanisms involved. Mutations in the DNA binding domain reduced the effect of the ligand on receptor ubiquitination and protein level. This indicated that the ligand acted preferentially on DNA-bound receptor molecules preventing their ubiquitination. Interestingly, mutations in DNA binding domain of PPAR␥ had no effect, and DNA binding of this receptor was not a pre-requisite for ligand-induced degradation (42). Furthermore, we show that the AF-2 domain and the co-repressor binding site were not required for ligand-induced changes in PPAR␦ ubiquitination, indicating that the ligand effect was independent of co-activator and co-repressor binding. It is intriguing that the AF-2 domain of PPAR␦ did not have any role in this process since for most nuclear receptors the trans-activation function is tightly linked to proteolytic degradation and mutations in the trans-activating domain are expected to affect both ubiquitination and proteolysis (34,35). This is also in contrast with the established role of the AF-2 domain in ligand-dependent degradation of PPAR␥ (42). The AF-2 domain of PPAR␥ mediated ligand-induced degradation of the receptor, suggesting that conformational changes induced by the ligand favored the interaction with both co-activators and the UPS (42). Using a different approach, co-expression of the transcriptional co-activator CBP (cAMP-response element-binding protein (CREB)-binding protein) led to a decrease of PPAR␣ level in the presence of ligand, showing that the interaction with co-activators via the AF-2 domain promoted proteolysis of the PPAR␣ isotype (46). Thus, in the case of PPAR␣, the initial stabilization induced by ligands is probably followed by recruitment of co-activators along with other co-factors that trigger proteolysis of the receptor (43,46). In contrast, our data show that transactivation and receptor ubiquitination are physically and functionally separated in the case of PPAR␦. It is probably the absence of a physical link between these processes that permits independent regulation of the transactivating function and ubiquitination of PPAR␦ upon ligand binding.
Taken together, our data support a model of ligand-dependent regulation of PPAR␦ activity that is in part different from what has been proposed for PPAR␥ and -␣ (Fig. 7). Under normal conditions, both unbound and DNA-bound PPAR␦ may undergo constitutive ubiquitination and degradation to maintain low levels of the receptor in the absence of ligands. UPSmediated proteolysis of the unliganded receptor may serve to control overall receptor level and, particularly, ligand-independent functions. In the absence of ligands, PPAR␦ can regulate transcription by forming complexes with transcriptional repressors, like SMRT (silencing mediator of retinoic acid and thyroid hormone receptor) (37) and Bcl-6 (36). We propose that upon ligand binding the DNA-bound receptor is protected from proteasome-dependent proteolysis by selective inhibition of receptor ubiquitination. This can be accomplished by blocking Ub ligases or, alternatively, activating deubiquitinating enzymes. The increased half-life of the DNA-bound receptor in the presence of ligand would allow time for transactivation of target genes to occur. Therefore, it is important that transactivation is functionally dissociated from ubiquitination of the receptor. In fact, this separation allows ligand binding, possibly via a conformational change independent of the AF2 domain and co-activator binding, to block ubiquitination of DNAbound PPAR␦. This condition may be essential to prevent degradation of the DNA-bound receptor and elicit a proper transcriptional response. Receptor ubiquitination may also have effects independent of the proteolytic pathway and may assist in switching between the repressive and activated state of the receptor, perhaps favoring either dissociation or assembly of co-repressor and co-activator complexes. FIGURE 7. Model of ligand-dependent regulation of PPAR␦ activity. In the absence of ligand, PPAR␦ is bound to DNA as a heterodimer with retinoid X receptor forming a complex with co-repressor molecules and thereby functioning as a transcriptional repressor. PPAR␦ can be ubiquitinated by ubiquitin ligases (Ub ligase) probably residing within the transcription regulatory complex, and its turnover is regulated by proteasome-dependent proteolysis. PPAR␦-selective ligands induce a conformational remodeling of the DNAbound receptor that allows binding of co-activator molecules and displacement of co-repressors. Ligand-induced remodeling and block of ubiquitin ligases lead to increased stability of DNA-bound PPAR␦ and promote transcriptional activation. Alternatively, deubiquitinating enzymes (DUB) may be activated upon ligand binding, resulting in reduced ubiquitination of the receptor. Deubiquitination of PPAR␦ may also assist in switching between the distinct states of the receptor, favoring dissociation of co-repressor complexes or assembly of co-activator complexes.