C-terminus of HSC70-Interacting Protein (CHIP) Inhibits Adipocyte Differentiation via Ubiquitin- and Proteasome-Mediated Degradation of PPARγ

PPARγ (Peroxisome proliferator-activated receptor γ) is a nuclear receptor involved in lipid homeostasis and related metabolic diseases. Acting as a transcription factor, PPARγ is a master regulator for adipocyte differentiation. Here, we reveal that CHIP (C-terminus of HSC70-interacting protein) suppresses adipocyte differentiation by functioning as an E3 ligase of PPARγ. CHIP directly binds to and induces ubiquitylation of the PPARγ protein, leading to proteasome-dependent degradation. Stable overexpression or knockdown of CHIP inhibited or promoted adipogenesis, respectively, in 3T3-L1 cells. On the other hand, a CHIP mutant defective in E3 ligase could neither regulate PPARγ protein levels nor suppress adipogenesis, indicating the importance of CHIP-mediated ubiquitylation of PPARγ in adipocyte differentiation. Lastly, a CHIP null embryo fibroblast exhibited augmented adipocyte differentiation with increases in PPARγ and its target protein levels. In conclusion, CHIP acts as an E3 ligase of PPARγ, suppressing PPARγ-mediated adipogenesis.

CHIP (C-terminus of HSC70-interacting protein) is an E3 ligase with a variety of target proteins, including p53, PTEN (Phosphatase and tensin homolog), Tau, and RIPK3 (Receptor-interacting serine/threonine-protein kinase 3) [32][33][34][35] . It contains a U-box domain responsible for ubiquitylation activities and a tetratricopeptide repeat (TPR) domain required for protein interactions. In particular, the TPR region is involved in HSP70 (heat shock proteins 70) and Hsp90 (heat shock proteins 90) association 36 . By interacting with molecular chaperones that affect the E3 ligase function in negative or positive ways, CHIP seems to be an essential factor for the maintenance of protein homeostasis. In this paper, we identified a new physiological role of CHIP: regulation of adipocyte differentiation. CHIP induces ubiquitylation and degradation of the PPARγ protein through direct interaction.
To confirm this, we show that stable overexpression of CHIP in 3T3-L1 cells suppresses adipocyte differentiation, while CHIP knockdown promotes adipogenesis. In accordance with these observations, CHIP-null mouse embryonic fibroblasts exhibited increased adipocyte differentiation with elevated levels of PPARγ . These results confirm the role of CHIP in suppressing adipogenesis by inducing degradation of the master adipocyte transcription factor PPARγ .

Results
CHIP interacts with PPARγ. In previous reports, CHIP was found to be capable of modulating the activities of several nuclear receptors, including ARs (Androgen Receptors), ERs (Estrogen Receptors), and GRs (glucocorticoid receptors) [37][38][39] . This led us to investigate the impact of CHIP on the nuclear receptor PPARγ . We first found that CHIP was able to interact with PPARγ 2, a PPARγ isoform, under both exogenous conditions ( Fig. 1a,b-d). Endogenous CHIP and PPARγ were also able to bind to each other in both PC-3 and 3T3-L1 cell lines (Fig. 1c,d). Supporting these observations, recombinant GST (glutathione s-transferase)-CHIP was able to bind PPARγ 2 that had been transcribed and translated in reticulocyte lysates, indicating a possible direct interaction between the two proteins (Fig. 1e). PPARγ 1, another isoform of PPARγ , exhibited similar CHIP-binding capabilities, implying that CHIP may not distinguish between the two isoforms of PPARγ (Supplemental Fig. 1a). To identify the domain responsible for these interactions, we generated several deletion mutants of either PPARγ 2 or CHIP and conducted immunoprecipitation analyses. Results showed that CHIP binds to multiple regions of PPARγ 2, including DNA-and ligand-binding domains (Fig. 1f). The CHIP TPR region is required for the interaction with PPARγ 2 (Fig. 1g). Of note, an E3 ligase-defective mutant, H260Q, was able to bind to PPARγ 2, while K30A, a mutant defective in the TPR domain, could not, corroborating the finding that the TPR region is required for the interaction between CHIP and PPARγ 2 ( Fig. 1h) 40 .
CHIP is an E3 ligase of PPARγ. To observe the effect of CHIP on PPARγ activity, we assessed the activity of a peroxisome proliferator response element (PPRE) promoter-driven luciferase stimulated by PPARγ 2 with or without CHIP expression. When PPARγ 2 was overexpressed, luciferase activity was doubled, an effect that was inhibited by simultaneous overexpression of CHIP. Importantly, the E3 ligase-defective mutant version of CHIP, H260Q, was not able to suppress PPARγ 2 transcription factor activity, implying that post-translational modification processes may be required for the CHIP and PPARγ interaction. The similar effects were observed under the treatment of troglitazone, a artificial ligand of PPARγ (Fig. 2a). The elimination of CHIP using two different CHIP siRNAs in PC3 cells resulted in increased PPARγ protein without altering mRNA levels, indicating that CHIP may in fact regulate PPARγ post-translationally (Fig. 2b).
As CHIP is an E3 ligase and its elimination led to an increase in PPARγ protein, we next tested whether CHIP induced the degradation of PPARγ . Results demonstrated that overexpression of CHIP did indeed induce endogenous and exogenous PPARγ degradation ( Fig. 2c and Supplemental Fig. 2a). CHIP-mediated degradation of endogenous PPARγ appears to be proteasome dependent, as degradation is blocked by MG132, a proteasome inhibitor in both 3T3-L1 and PC3 cell lines (Fig. 2d,e). Furthermore, while H260Q, the E3 ligase-defective mutant of CHIP, was able to bind PPARγ 2, it was incapable of inducing degradation of endogenous PPARγ , indicating that ubiquitylation is also required (Figs 1h and 2d,e). We observed the similar effects with overexpressed proteins including PPARγ 2 and CHIP or H260Q (Supplementary Figure 2b,c). Supporting this, the half-life of PPARγ 2 was reduced only by wild-type (WT) CHIP, not H260Q (Fig. 2f). Results for PPARγ 1 were similar to those for PPARγ 2 (Supplemental Fig. 1b,c). Overall, our results indicate that CHIP may regulate PPARγ via post-translational modification.
Since CHIP is a well-known E3 ligase, we next examined whether CHIP mediates the ubiquitylation of PPARγ . We found that poly-ubiquitylation of PPARγ 2 increased because of CHIP overexpression, with or without MG132 (Fig. 3a). In accordance with these observations, endogenous poly-ubiquitylated forms of PPARγ were reduced by the elimination of CHIP in PC3 cell lines, indicating that CHIP regulates PPARγ endogenously. Here, cells were treated with troglitazone to induce accumulation of PPARγ so that ubiquitylated forms could be detected, as previously reported (Fig. 3b) 29 . As a result, poly-ubiquitylated forms of endogenous PPARγ were also reduced following CHIP knockdown in 3T3-L1 cells that were induced to differentiate into adipocytes by treatment with dexamethasone, IBMX, and insulin (DMI treatment) for two days (Fig. 3c). As with CHIP-mediated degradation of PPARγ 2, the E3 ligase activity of CHIP was required for this regulation, as H260Q was not able to induce ubiquitylation of PPARγ 2 (Fig. 3d). Finally, using recombinant CHIP, H260Q, and PPARγ 2, we confirmed Scientific RepoRts | 7:40023 | DOI: 10.1038/srep40023 that CHIP directly ubiquitylates PPARγ 2 (Fig. 3e). We also observed that CHIP similarly mediates the ubiquitylation of PPARγ 1, supporting the conclusion that CHIP induces degradation of PPARγ (Supplemental Fig. 3). In summary, CHIP functions as an E3 ligase by inducing the ubiquitylation and proteasome-dependent degradation of PPARγ .

HSP70 suppresses CHIP-mediated PPARγ degradation and ubiquitylation. Hsp90 and Hsp70
function as major CHIP partners in the regulation of various target proteins 41,42 . Ubiquitylation analyses with recombinant proteins in the absence of Hsp70 and Hsp90, however, suggest that these two molecular chaperones may not participate in the CHIP-mediated ubiquitylation of PPARγ (Fig. 3e). To examine the roles of both molecular chaperones in PPARγ degradation processes further, the effect of geldanamycin (GA), an Hsp90 inhibitor, on CHIP-mediated degradation of PPARγ was assessed. Treatment with this inhibitor had little effect on the CHIP-mediated degradation of PPARγ , indicating that HSP90 may not be required for the degradation process (Supplemental Fig. 4). In contrast, transient overexpression of HSP70 inhibited CHIP-mediated degradation of PPARγ (Fig. 4a). While it seems that HSP70 has little effect on the binding affinity between CHIP and PPARγ 2, the presence of HSP70 prevents CHIP-mediated PPARγ 2 ubiquitylation (Fig. 4b,c). These results indicate that of the two major partners of CHIP, only Hsp70 functions as a negative regulator by suppressing CHIP-mediated ubiquitylation of PPARγ 2.
CHIP suppresses adipocyte differentiation induced by PPARγ activation. Because CHIP destabilizes the PPARγ protein via ubiquitylation and proteasome-dependent degradation, we next tested whether CHIP is able to regulate adipocyte differentiation in 3T3-L1 cells. First, we generated 3T3-L1 cell lines that stably overexpress CHIP or H260Q using a retroviral system. Along with a control cell line transfected with pBABE, these cell lines were induced to differentiate into adipocytes through DMI treatment. Oil Red O staining data and quantification of lipid content showed that the cell line overexpressing CHIP displayed reduced levels of adipogenesis compared to both the control and H260Q-transfected cells (Fig. 5a,b). We next examined the protein and mRNA levels of PPARγ and its targets in the same cell lysates. As expected, cells overexpressing CHIP exhibited reductions in both PPARγ mRNA and protein levels. We also observed reductions in the PPARγ targets aP2, cofactor C/EBPα and CD36 in cells overexpressing CHIP ( Fig. 5c-d). We further generated 3T3-L1 cell lines with stable knockdown of CHIP (shCHIP#3, shCHIP#4) using a lentiviral vector system. In contrast to the overexpression system, stable cell lines with CHIP knockdown displayed more adipocyte differentiation than that of control cells (shGFP) detected by Oil Red O staining (Fig. 6a), and this was confirmed by quantification of lipid content (Fig. 6b). The cell lines expressing shCHIP#3 and shCHIP #4 were further analyzed by western blot and qRT-PCR. Like PPARγ , aP2, C/EBPα and CD36 were elevated following CHIP knockdown when compared to control cells (Fig. 6c,d). Finally, WT or CHIP-null MEFs were employed to examine the effects of CHIP on adipocyte differentiation induced by DMI plus troglitazone, a PPARγ ligand 29 . Two CHIP knockout MEFs, CHIP KO #6 and CHIP KO #9, exhibited greater adipocyte differentiation than the two wild-type MEFs, WT #3 and (a) CHIP represses the transcriptional activity of PPARγ . H1299 cells were transfected with PPRE, pcDNA3. 1-PPARγ 2, pcDNA3-FLAG-CHIP, and pcDNA3-FLAG-CHIP H260Q plasmids with or without troglitazone. Cells were measured by luminometer. Data are presented means ± SD; n = 3; **P < 0.01, and ***P < 0.001 compared to each lane. (b) The CHIP protein level was increased by CHIP knockdown using siRNAs. Western blots of lysates of PC-3 cells transfected with the indicated plasmids using siRNA IMAX were performed using the indicated antibodies. (c) Endogenous PPARγ protein was degraded by CHIP. Preadipocyte, 3T3-L1 cells were treated with DMI cocktail for 2 days, were transfected with pcDNA3-HA-CHIP plasmid dose dependent manner. Indicated protein was measured by western blotting. (d,e) CHIP degrades PPARγ protein through the E3-ligase function and proteasomal manner. (d) DMI treated 3T3-L1 cells were transfected with HA-CHIP and HA-CHIP H260Q mutant expressing vectors with or without MG132. Each protein was detected by western blotting indicated antibodies. (e) PC-3 cells were transfected with pcDNA3-HA-CHIP and pcDNA3-HA-CHIP H260Q mutant with or without MG132. Lyzed cells were analyzed by western blotting indicated antibodies. (f) The PPARγ protein was destabilized by wild-type CHIP but not by the H260Q mutant. Western blots of H1299 cells transfected with plasmids expressing PPARγ , FLAG-CHIP, and FLAG-CHIP H260Q in the presence or absence of 50 μ g/mL CHX (cycloheximide) treatment were performed using the indicated antibodies and were measured with the Image J program. Asterisk indicates an actin band. #8 (Fig. 7a). The data showing quantification of lipid contents confirmed these results (Fig. 7b). Similarly, mRNA or protein levels of PPARγ , aP2, C/EBPα and CD36 were all elevated in CHIP KO #6 and CHIP KO #9 MEFs compared with those of controls (Fig. 7c,d). In summary, these data suggest that CHIP functions as a negative modulator of adipocyte differentiation in 3T3-L1 and primary MEF cells, possibly by suppressing PPARγ .

Discussion
In this study, we identified a new role of CHIP in adipocyte differentiation. CHIP interacts with and mediates the ubiquitylation of PPARγ , which results in negative effects on adipogenesis. CHIP is commonly known to exert its E3 ligase effects in association with the molecular chaperones Hsp90 and Hsp70 43 . For example, Hsp70 facilitates the CHIP-mediated degradation of p53, HIF-1α , and AR 32,37,41 , while Hsp90 is known to suppress CHIP-mediated degradation by promoting the stability of CHIP target proteins 44,45 . Notably, PPARγ degradation induced by CHIP does not seem to require the presence of either Hsp90 or Hsp70. The poly-ubiquitylation of PPARγ occurs independently of the molecular chaperones, as demonstrated by ubiquitylation analyses with recombinant proteins. The inhibition of Hsp90 by geldanamycin does not have a strong effect on CHIP-mediated PPARγ degradation. Interestingly, it has been reported that Hsp90 inhibition leads to suppression of adipogenesis 46 . Based on our results, it appears that Hsp90 may regulate adipocyte differentiation via a CHIP-independent regulatory pathway. In contrast, Hsp70 appears to function as a negative regulator of CHIP-dependent PPARγ degradation. Upon expression of Hsp70, both the ubiquitylation and degradation of PPARγ were suppressed. A similar pathway, in which Hsp70 functions as a negative regulator of CHIP, can be found in the CHIP-dependent degradation of Tau 34   (c) Western blots were performed on the above lysates using the indicated antibodies. (d) PPARγ 2 protein was ubiquitylated by wild-type CHIP but not the H260Q mutant. Western blots of H1299 cells transfected with asdescribe above all, with His-ubiquitin-conjugated proteins pulled down using Ni 2+ -NTA beads. The indicated antibodies were used. (e) CHIP directly ubiquitinates PPARγ 2 in vitro. Western blots of purified GST-PPARγ 2 recombinant protein incubated with ATP, E1, E2, and ubiquitin in the presence or absence of wild-type GST-CHIP or GST-CHIP H260Q mutant were performed using the indicated antibodies.
Several other E3 ligases have been previously reported to ubiquitylate PPARγ . MKRN1 was the first E3 ligase known to mediate PPARγ ubiquitylation and degradation 29 . SIAH2 and NEDD4-1 were subsequently identified as E3 ligases of PPARγ 30,31 . With the addition of CHIP based on our results, it appears that PPARγ may be regulated by several E3 ligases. However, most of these observations occurred while investigating cellular differentiation. To determine the role of CHIP in other physiological settings, we generated CHIP-null MEFs and compared differentiation in WT and CHIP-null MEFs. The results demonstrated that CHIP-null MEFs differentiated into adipocytes more effectively than WT MEFs, supporting our in vitro data (Figs 1-3). Based on these observations, the systemic effects of CHIP should be further assessed in the future. Since the whole-body CHIP knockout mouse of the C57BL/6 strain does not survive for more than one month, a conditional knockout system must be established to investigate the physiological roles of CHIP in association with PPARγ 35,47 . Additional studies should examine whether the E3 ligases identified so far are expressed in adipocyte tissues and whether their expression levels are regulated by a high-fat diet. Any E3 ligase whose expression is negatively correlated with that of PPARγ under high-fat diet conditions may be a good candidate for drug targeting 48 . Further in vivo studies utilizing high-fat-diet induced mice may indicate the main E3 ligase of PPARγ .
TZD and its analogs have been clinically tested for treating obesity and diabetes mellitus through the agonistic activation of PPARγ 49,50 . Unfortunately, these drugs exhibited various side effects, including heart and liver failure, weight gain, fluid retention, and bone fractures 51,52 . As these drugs have been withdrawn from the market or removed from general usage due to their severe side effects, the concept of developing novel, non-agonistic drugs has been pursued 53 . As various post-translational modifications could regulate the function of PPARγ in possibly non-agonistic ways, targeting post-translational regulators may be a good strategy for discovering non-agonistic drugs with reduced side effects 54 . Since the suppression of CHIP may stabilize and activate PPARγ , CHIP suppression may represent a good target for clinical trials for the treatment of diabetes and obesity.

Generation of mouse fibroblast cells. All of mouse experiments were approved by The Institutional
Animal Care and Use Committees of Yonsei University (IACUC-A-201409-294-01). The mice were maintained in accordance with the approved guidelines and regulations for experimental animals provided by Yonsei Laboratory Animal Research Center. Using C57BL/6 J mice, heterogeneous CHIP male and female mice were mated to produce wild type and knockout CHIP mice, as previously reported 35 . Mouse embryos were obtained 13.5 days after verifying plug formation. Embryos were minced with a blade in 1 ml trypsin-EDTA. Cells were transferred to 150-mm plates containing DMEM medium with 10% FBS. After 16 hours, cells were maintained in DMEM medium.
Translated protein was incubated with GST or GST-PPARγ 2 for two hours and then with sepharose beads for one hour. Complexes were then washed and eluted with reduced glutathione (10 mM), followed by western blot analysis.
Quantification of lipid content. The lipid contents were measured with triglyceride quantification colorimetric/fluorometric kits (Biovision, Milpitas, CA, K622-100). Analysis of lipid contents was performed by Biovision as described. Differentiated cells were collected in PBS. Cells were homogenized with 5% NP-40 diluted with water. Lysates and triglycerides were slowly boiled at 80 to 100 °C in a water bath and then cooled down to room temperature. Triglyceride standards were used to make a standard curve. Lysed samples were incubated with lipase in the assay buffer for 20 min, and a mixture of the triglyceride probe was added to the buffer, followed by incubation for 30 to 60 min. The absorbance of the samples was measured with a microtiter plate reader at 570 nm. The values were entered into the provided formula to obtain the final results.
Statistical analyses. Statistical analyses were conducted using Prism (GraphPad Software Inc., CA, USA), and results are presented as means ± SD. All statistical results in Prism were performed using unpaired two-tailed t-test to compare two groups (n ≥ 3).