RNF20 Functions as a Transcriptional Coactivator for PPARγ by Promoting NCoR1 Degradation in Adipocytes

Adipose tissue plays key roles for systemic energy homeostasis (1). The primary role of adipocytes is to store excess energy in the form of triglycerides and hydrolyze triglycerides during periods of nutritional deprivation. Emerging evidence suggests that adipocytes could also act as endocrine cells that secrete various hormones, lipid metabolites, and chemokines to coordinate systemic glucose and lipid metabolism (2). Thus, both excessive and deficient adipogenesis are closely associated with the dysregulation of energy metabolism, such as obesity, type 2 diabetes, and lipodystrophy (3,4)

Numerous in vivo and in vitro studies have shown that peroxisome proliferator–activated receptor-γ (PPARγ) is the necessary and sufficient transcription factor for adipogenesis (5). At the early phase of adipogenesis, C/EBPs, Krüppel-like factors (KLFs), and early B-cell factors (EBFs) stimulate expression and activity of PPARγ (6). As a nuclear receptor, PPARγ promotes expression of genes that are involved in adipogenesis and maintenance of mature adipocytes (5). PPARγ forms a heterodimer with retinoid X receptor-α (RXRα) and binds to the peroxisome proliferator response elements (PPREs) in the promoters or enhancers of target genes. In the absence of PPARγ ligand, corepressor complexes, including nuclear corepressor 1 (NCoR1) and histone deacetylase 3 (HDAC3), repress the expression of PPARγ target genes (7). In contrast, PPARγ ligands induce a conformational change of PPARγ to recruit coactivators that enhance the transcriptional activity of PPARγ (7). In addition to the ligand-dependent mechanism, PPARγ activity is modulated by various posttranslational modifications, such as SUMOylation, phosphorylation, and acetylation (8).

Ring finger protein 20 (RNF20) is a member of the RING domain E3 ubiquitin ligases, which recognize target substrates and mediate ubiquitin transfer to its substrates. RNF20 is involved in many biological events, such as transcription, cell division, and heat shock responses (9–11). We have demonstrated previously that RNF20 is crucial for fasting-induced degradation of SREBP1c, the key transcription factor for de novo lipogenesis (12,13). In particular, aberrant expression of RNF20 leads to ectopic lipid accumulation in the liver (12) and tumorigenesis in the kidney by elevating lipogenic activity and cell proliferation (13). However, the physiological roles of RNF20 in adipose tissue have not been explored.

In this study, we demonstrate that RNF20 is a crucial factor for adipogenic programming by regulating NCoR1 degradation, which eventually potentiates the transcriptional activity of PPARγ. Compared with wild-type (WT) littermates, Rnf20 defective (Rnf20^+/−) mice exhibited reduced fat mass and downregulated expression of PPARγ target genes in adipose tissue. Furthermore, knockdown of Rnf20 decreased adipogenic capacity, while overexpression of RNF20 enhanced adipogenesis. Mechanistically, RNF20 promoted proteasomal degradation of NCoR1 accompanied by a reduction in HDAC3 recruitment in the PPREs. Collectively, our data suggest that RNF20 would function as a novel coactivator for PPARγ through degradation of NCoR1 in adipocytes.

Mice in which the exons 3–20 of the Rnf20 gene were deleted (C57BL/6, Rnf20^{tm1(KOMP)Vlcg}, RRID:IMSR_KOMP:VG15351–1-Vlcg) were obtained from the Knockout Mouse Project Repository. Mice were housed in a temperature- and humidity-controlled, specific pathogen–free animal facility at 22°C under a 12:12-h light:dark cycle. High-fat diet (HFD) feeding experiments were performed using 7-week-old mice fed a diet consisting of 60% of calories from fat (D12492; Research Diets, New Brunswick, NJ). For the intraperitoneal glucose tolerance test (GTT), mice were fasted for 16 h, and glucose was administered (1 g/kg body weight). For the insulin tolerance test (ITT), mice were fed ad libitum, and insulin was administered (0.75 units/kg body weight, 91077C; Sigma-Aldrich, St. Louis, MO). Energy expenditure and physical activity were measured using an indirect calorimetry system (Minispec LF50; Bruker, Hamburg, Germany). This study was reviewed and approved by the Institutional Animal Care and Use Committee of Seoul National University.

Adipose tissues were fractionated as described previously (14). For differentiation, attached stromal vascular fraction (SVF) cells were grown to confluence in DMEM supplemented with 10% FBS. After achieving confluent growth, the cells were stimulated with DMEM containing 10% FBS, dexamethasone (1 μmol/L), methylisobutylxanthine (520 μmol/L), insulin (850 nmol/L), and rosiglitazone (1 μmol/L) for 48 h. The culture medium was replaced with DMEM containing 10% FBS, insulin (850 nmol/L), and rosiglitazone (1 μmol/L) for 6 more days.

Protein sampling and proteomic analysis were done as previously reported (15). Protein was extracted from adipose tissue from WT and Rnf20^+/− mice with modified radioimmunoprecipitation assay buffer (150 mmol/L NaCl, 50 mmol/L Tris-HCl [pH 7.4], 1% NP-40, 0.25% sodium deoxycholate, 1 mmol/L EDTA, and protease inhibitor cocktail [P3100; GenDEPOT, Katy, TX]). The protein extracts were precipitated overnight at −20°C, and iTRAQ analysis was performed (AB Sciex, Framingham, MA). Raw spectra were acquired with an Orbitrap Fusion Lumos Tribrid Mass Spectrometer (Thermo Fisher Scientific, Waltham, MA) and EASY-nLC 1200 system (Thermo Fisher Scientific). Raw mass spectrometry spectra were processed with MaxQuant version 1.5.8.3 software (www.maxquant.org) at default settings and with Perseus software (www.biochem.mpg.de/5111810/perseus) using a UniProt Mus musculus database (16,987 reviewed protein sequences). Output files generated by MaxQuant were subjected to Scaffold Q+ software (Scaffold 4.7.5; Proteome Software, Portland, OR) to quantitate iTRAQ peptide and protein identifications.

Cells and tissues were lysed on ice with radioimmunoprecipitation assay buffer. Antibodies against RNF20 (ab32629; Abcam, Cambridge, U.K.), PPARγ (sc-7196; Santa Cruz Biotechnology, Dallas TX), perilipin 1 (PLIN1) (20R-PP004; Fitzgerald, Acton, MA), GAPDH (G8795; Sigma-Aldrich), MYC-tag (05-724; Millipore, Billerica, MA), FLAG-tag (F1804; Sigma-Aldrich), HA (3724; Cell Signaling Technology, Danvers, MA), NCoR1 (ab3482; Abcam), adiponectin (2789; Cell Signaling Technology), ubiquitin (BML-PW0150; Enzo Life Sciences, Farmingdale, NY), and β-actin (A5408; Sigma-Aldrich) were used. The bands were visualized with horseradish peroxidase–conjugated secondary anti-rabbit IgG or anti-mouse IgG antibodies (A0545 and A9044, respectively; Sigma-Aldrich).

siRNA for mouse Rnf20, Ncor1, human RNF20, and negative control siRNA (siNC) were purchased from Bioneer (Daejeon, Korea). Cells were mixed with siRNA and transfected through a single pulse of 1,100 V for 30 ms using an MP-100 Microporator (Digital Bio, Seoul, Korea). HEK293T cells were transfected with Lipofectamine RNAiMAX Reagent (Invitrogen, Carlsbad, CA) according to the manufacturer’s protocol. The sequence information for siRNA is described in Supplementary Table 2.

Adipose tissues were fixed in 4% paraformaldehyde for 10 min. The fixed tissues were washed twice with PBS containing Tween 20. Whole-mounted adipose tissues were incubated with primary antibodies against CD11b (14-0112-81; eBioscience, San Diego, CA). Samples were incubated with fluorescent-labeled secondary antibodies (Thermo Fisher Scientific) and DAPI (Vector Laboratories, Burlingame, CA) and stained with FITC-conjugated BODIPY 493/503 (D3922; Thermo Fisher Scientific).

Yeast two-hybrid screening was performed by Panbionet (www.panbionet.com). A cDNA library was made using cDNA from mouse epididymal adipose tissue. The cDNA inserts were introduced into yeast strain PBN204 with a SmaI-linearized pGADT7-Rec vector. The entire mouse Rnf20 cDNA was used as bait. Each DNA insert was integrated into the pGADT7 vector by yeast homologous recombination. The yeast prey library strain was mixed with RNF20 bait strain. A mixture of two yeast strains was spread on selection medium (medium deficient of leucine, tryptophan, histidine, and adenine), which supports the growth of diploid yeasts in which RNF20 bait and prey proteins physically interact with each other. Prey DNA was reintroduced into PBN204 with or without RNF20 bait. Positive clones were identified by ADE2, URA3, and lacZ under the control of different galactose promoters. The identified clones were sequenced and subjected to Basic Local Alignment Search Tool search.

Chromatin immunoprecipitation (ChIP) was performed as described previously (16). Primary adipocytes were isolated from epididymal white adipose tissue (eWAT), subjected to crosslinking with 1% formaldehyde for 10 min, and lysed with lysis buffer (1% SDS, 10 mmol/L EDTA, 50 mmol/L Tris-HCl [pH 8.1], and protease inhibitor cocktail). The samples were diluted with a dilution buffer (0.01% SDS, 1.1% Triton X-100, 1.2 mmol/L EDTA, 16.7 mmol/L Tris-HCl [pH 8.1], 167 mmol/L NaCl, and protease inhibitor cocktail) and sonicated. After being precleared with protein A agarose/salmon sperm DNA mix, the samples were immunoprecipitated overnight with antibodies. Precipitated DNA fragments were analyzed by quantitative real-time PCR (qRT-PCR). The primer sequences for ChIP-qRT-PCR are provided in Supplementary Table 2. Antibodies against PPARγ (sc-7196; Santa Cruz Biotechnology), HDAC3 (sc-8138; Santa Cruz Biotechnology), and mouse IgG (A9044; Sigma-Aldrich) were used.

Data are presented as mean ± SD. In box plots, center lines represent medians, limits represent first and third quartiles (interquartile range), whiskers represent the lowest and highest values within a 1.5× interquartile range ± the first/third quartile, and points represent values outside whiskers. All n values indicated by dots in the figures refer to biological replicates. For comparison between two groups, two-tailed Student t test was used. For comparison between more than two groups, one-way ANOVA with multiple comparisons was used followed by Tukey post hoc test. For comparison between two independent variables, two-way ANOVA with multiple comparisons was used followed by Sidak multiple comparisons test. Statistical analyses were performed using GraphPad Prism 7 software (GraphPad Software, La Jolla, CA).

The data sets generated during the current study are available from the corresponding author upon reasonable request. All noncommercially available resources generated and/or analyzed during the current study are available from the corresponding author upon reasonable request.

To investigate the roles of mammalian RNF20 in vivo, we examined Rnf20 knockout mice. Since Rnf20 whole-body knockout mice died before birth (17), Rnf20^+/− mice were used to study the roles of RNF20 in vivo. Under normal chow diet (NCD) feeding, Rnf20^+/− mice exhibited similar body weights compared with WT littermates (Fig. 1A). However, unexpectedly, fat mass of Rnf20^+/− mice was reduced compared with WT mice. Although lean mass of Rnf20^+/− mice was heavier than that of WT mice, the weights of other tissues, including liver, kidney, and pancreas, were relatively similar between the two genotypes (Fig. 1B–D). Furthermore, histologic analysis revealed that overall adipocyte sizes of inguinal WAT (iWAT) and eWAT were decreased in Rnf20^+/− mice (Fig. 1E and F).

RNF20 is expressed in various tissues, such as the spleen, lymph nodes, and liver (18). However, the expression level of RNF20 in adipose tissue has not been reported. As shown in Supplementary Fig. 1A, RNF20 protein was highly expressed in WATs. Furthermore, the levels of RNF20 protein in WATs were increased in HFD-fed mice (Supplementary Fig. 1B). Together, these results suggest that RNF20 might play certain roles in adipose tissue biology.

Obesity is characterized by an increase in fat mass. On the basis of the reduced fat mass of Rnf20^+/− mice, we tested whether RNF20 might participate in whole-body energy homeostasis in diet-induced obesity (DIO). Compared with HFD-fed WT mice, HFD-fed Rnf20^+/− mice exhibited reduced body weight gain accompanied by decreased fat mass and fat cell size (Fig. 2A–D). In obesity, expansion of fat tissue with adipocyte hypertrophy is closely associated with elevated proinflammatory responses and dysregulated lipid metabolism (14,19,20). As expected, HFD-fed WT mice showed immune cell infiltration into adipose tissue (Fig. 2E and F). In contrast, inflammatory responses were alleviated in adipose tissue from HFD-fed Rnf20^+/− mice. Consistent with these findings, mRNA levels of proinflammatory genes, such as Tnfa, F4/80, Inos, and Il1b, were downregulated in the adipose tissue from HFD-fed Rnf20^+/− mice compared with HFD-fed WT mice (Fig. 2G). Since adipose tissue plays central roles in whole-body lipid metabolism, we examined serum lipid profiles. As shown in Fig. 2H–J, the levels of free fatty acids (FFA), triglycerides, and cholesterol in serum were downregulated in HFD-fed Rnf20^+/− mice. In addition, basal lipolytic activity was reduced in primary adipocytes from HFD-fed Rnf20^+/− mice (Fig. 2K). Collectively, these data suggest that adipose tissue of Rnf20^+/− mice would be less prone to adipose tissue expansion upon HFD, which might mitigate dysregulation of lipid metabolism in obesity.

Given that lipid metabolism is closely associated with glucose metabolism and insulin sensitivity (1,19,21,22), we tested for glucose tolerance and insulin tolerance. While there were no significant differences in glucose and insulin tolerance, as well as ectopic lipid accumulation, between NCD-fed WT and Rnf20^+/− mice (Supplementary Fig. 2A–J), HFD-fed Rnf20^+/− mice exhibited decreased fasting blood glucose, serum insulin, and HOMA of insulin resistance (HOMA-IR) values (Fig. 2L–N). As shown in Supplementary Fig. 2K–R, HFD-fed Rnf20^+/− mice exhibited slightly increased energy expenditure and reduced ectopic lipid accumulation in liver and muscle compared with HFD-fed WT mice. However, there were no significant differences in physical activity, food intake, and gene expression of fatty acid catabolism in liver. Furthermore, HFD-fed Rnf20^+/− mice displayed alleviated glucose and insulin intolerance (Fig. 2O and P). Similarly, female Rnf20^+/− mice also exhibited reduced adiposity, and both glucose and insulin intolerance were ameliorated in DIO, even though the extents of these phenotypes in female mice were slightly milder than those of male mice in DIO (Supplementary Fig. 3A–E). Taken together, these data propose that RNF20 would play crucial roles in fat tissue biology and systemic energy metabolism in obesity.

To gain further insights into the roles of RNF20 in fat tissue, quantitative proteomic analyses were carried out. In adipose tissue from WT and Rnf20^+/− mice, differentially expressed proteins were primarily involved in metabolic pathways (Fig. 3A and Supplementary Fig. 4A). To elucidate the key effectors in adipose tissue of Rnf20^+/− mice, we analyzed the expression profile through the ChEA tool (23). This analysis revealed that expression levels of PPARγ target proteins were prominently downregulated in adipose tissue of Rnf20^+/− mice (Fig. 3B and C). In addition, reduced expressions of PPARγ target genes in WATs of Rnf20^+/− mice were confirmed by qRT-PCR and Western blot analyses (Fig. 3D and E and Supplementary Fig. 4B). Consistent with these findings, the level of serum adiponectin was decreased in Rnf20^+/− mice (Fig. 3F). In addition, expressions of certain genes that are known to be inversely correlated with PPARγ activation (24) were increased in adipose tissue of Rnf20^+/− mice (Fig. 3G). Together, these findings clearly indicate that adipose tissue RNF20 would modulate PPARγ and its target gene expression.

The observations that fat mass and expression of PPARγ target genes were downregulated in Rnf20^+/− mice prompted us to investigate whether RNF20 might be involved in adipocyte differentiation. In 3T3-L1 cells, RNF20 protein was upregulated during adipogenesis, especially at the early stage of adipocyte differentiation with adipogenic stimuli (Fig. 4A). Furthermore, RNF20 protein was abundantly expressed in adipocyte fractions compared with SVFs (Fig. 4B). To assess the roles of RNF20 in adipocyte differentiation, WAT SVFs were stimulated to differentiate into adipocytes. As shown in Fig. 4C, the degree of lipid accumulation, a hallmark of differentiated adipocytes, was markedly decreased in SVF-derived adipocytes from Rnf20^+/− mice. Consistently, mRNA levels of Pparg and Cebpa, which are key transcription factors for adipogenesis, as well as Fabp4 and Adipoq, which are markers of mature adipocytes, were downregulated in SVF-derived adipocytes of Rnf20^+/− mice compared with those of WT mice (Fig. 4D). Furthermore, knockdown of Rnf20 in 3T3-L1 preadipocytes suppressed lipid accumulation and adipogenic gene expression (Fig. 4E and F), implying that RNF20 would mediate certain roles to execute adipogenic programming.

To affirm that RNF20 would participate in adipogenesis, RNF20 was overexpressed in 3T3-L1 preadipocytes by lentivirus infection. As indicated in Fig. 4G and H, overexpression of RNF20 in 3T3-L1 preadipocytes stimulated lipid accumulation, even without adipogenic stimuli, including dexamethasone, methylisobutylxanthine, and insulin. Moreover, the expression levels of PPARγ target genes were significantly elevated in RNF20-overexpressing preadipocytes, while no differences were observed in the expression of PPARγ upstream regulators (Fig. 4I). Similarly, SVF-derived preadipocytes from Rnf20^+/− mice or suppression of Rnf20 through siRNA in 3T3-L1 preadipocytes did not show significant changes in mRNA levels of PPARγ upstream regulators compared with each control group (Supplementary Fig. 5A and B). Furthermore, the PPARγ antagonist GW9662 significantly attenuated expression of PPARγ target genes in RNF20-overexpressing 3T3-L1 preadipocytes (Supplementary Fig. 5C). Moreover, inhibition of PPARγ activity using GW9662 nullified the effect of RNF20 overexpression on adipocyte differentiation (Supplementary Fig. 5D), implying that the effect of RNF20 on adipogenesis would be dependent on PPARγ activity. Collectively, these data suggest that RNF20 would be a novel factor to stimulate adipogenic programming by enhancing expression of PPARγ target genes.

Several studies have reported that RNF20 can act as a transcriptional coactivator in certain biological contexts (11,25). The present findings that RNF20 augmented expression of PPARγ target genes and enhanced adipogenesis led us to test whether RNF20 might potentiate the transcriptional activity of PPARγ. This was addressed using a luciferase reporter assay (26) (Fig. 5A). As shown in Fig. 5B, RNF20 dose-dependently promoted the transcriptional activity of PPARγ. In addition, RNF20 overexpression further stimulated the transcriptional activity of PPARγ upon exposure to rosiglitazone, a synthetic ligand of PPARγ (Fig. 5C). Conversely, knockdown of RNF20 through siRNA decreased the transcriptional activity of PPARγ without or with rosiglitazone (Fig. 5D). Next, we addressed whether the effect of RNF20 on PPARγ activity might require the ligand-binding domain of PPARγ using another reporter system in which this PPARγ domain was fused to the GAL4 DNA-binding domain (Fig. 5E). As shown in Fig. 5F–H, RNF20 overexpression augmented the transcriptional activity of PPARγ through the ligand-binding domain, while RNF20 suppression downregulated PPARγ activity, implying that RNF20 could potentiate transcriptional activity of PPARγ in a ligand-binding domain–dependent manner. To affirm these in vivo, lobeglitazone, a more potent thiazolidinedione for PPARγ activation than rosiglitazone (27), was administered to WT and Rnf20^+/− mice. The treatment significantly increased the expression of the PPARγ target genes Fabp4 and Cd36 in eWAT of WT mice, while these effects were attenuated in Rnf20^+/− mice (Fig. 5I). These results indicate that RNF20 would stimulate transcriptional activity of PPARγ in adipose tissue.

To elucidate the mechanisms by which RNF20 would promote transcriptional activity of PPARγ, coimmunoprecipitation was performed to explore the direct binding of RNF20 to PPARγ. Direct physical interaction was not apparent (Supplementary Fig. 6). Although the possibility of a weak or transient interaction between these two proteins cannot be excluded, we postulated that substrates or binding proteins of RNF20 might control PPARγ activity. To identify potential binding proteins of RNF20 in eWAT, yeast two-hybrid screening was performed with the cDNA library of eWAT. Surprisingly, NCoR1 interacted with RNF20 (Fig. 6A and Supplementary Table 1). As shown in Fig. 6B, a coimmunoprecipitation analysis revealed the physical interaction between RNF20 and NCoR1, which was further confirmed in immune complexes in which endogenous RNF20 interacted with endogenous NCoR1 in the adipocyte fractions of eWAT (Fig. 6C).

Because E3 ubiquitin ligases can regulate protein stability of their substrates, we tested whether RNF20 might affect NCoR1 protein level. As shown in Fig. 6D and E, RNF20 reduced the NCoR1 protein level without affecting Ncor1 mRNA level in differentiated 3T3-L1 adipocytes. In addition, the proteasome inhibitor MG132 abolished the effect of RNF20 overexpression on the reduction of NCoR1 protein (Fig. 6F). Furthermore, RNF20 overexpression promoted polyubiquitination of NCoR1 protein (Fig. 6G), whereas RNF20 suppression conversely reduced polyubiquitination of NCoR1 protein in differentiated 3T3-L1 adipocytes (Fig. 6H), implying that RNF20 would mediate proteasomal degradation of NCoR1. Consistent with these findings, a cycloheximide chase assay showed that ectopic expression of RNF20 increased the degradation rate of NCoR1 protein (Fig. 6I). On the contrary, suppression of RNF20 through siRNA downregulated the degradation rate of NCoR1 protein (Fig. 6J). When the effect of RNF20 on the level of NCoR1 protein in adipocytes was examined, NCoR1 protein levels were elevated in the adipocyte fraction of eWAT from Rnf20^+/− mice independent of Ncor1 mRNA (Fig. 6K and L). Together, these data suggest that RNF20 could act as an E3 ubiquitin ligase of NCoR1 in adipocytes and promote proteasomal degradation of NCoR1 through polyubiquitination.

NCoR1 is a negative regulator for adipogenesis by repressing mainly PPARγ activity (24,28). To test the hypothesis that the effect of RNF20 on adipogenesis might be mediated by NCoR1, we suppressed Rnf20 and/or Ncor1 expression by siRNA in 3T3-L1 preadipocytes. Whereas Ncor1 suppression promoted lipid accumulation and Fabp4 expression, double knockdown of Rnf20 and Ncor1 attenuated the suppressive effect of Ncor1 on adipocyte differentiation (Fig. 7A and B). Furthermore, the expression levels of RNF20 and NCoR1 protein were inversely correlated during adipocyte differentiation (Fig. 7C), indicating that potential mechanisms by which RNF20-stimulated adipogenesis would be dependent on NCoR1.

To further confirm that RNF20 would regulate PPARγ activity through NCoR1, we examined whether RNF20 might affect the interaction between PPARγ and NCoR1. In differentiated 3T3-L1 adipocytes, suppression of Rnf20 through siRNA augmented physical interaction between NCoR1 protein and PPARγ protein (Fig. 7D). Next, we performed a luciferase reporter assay with RNF20 and/or NCoR1 overexpression. Ectopic expression of RNF20 attenuated the repressor activity of NCoR1 (Fig. 7E, lanes 1–4). To address whether RNF20 might potentiate PPARγ activity in the absence of NCoR1 repressor activity, rosiglitazone was then treated in a dose-dependent manner to abrogate the repressor activity of NCoR1. At a high dose of rosiglitazone that nearly abolished the suppressive activity of NCoR1 (Fig. 7E, lanes 13 and 14), RNF20 did not increase PPARγ activity (Fig. 7E, lanes 13 and 15 and 14 and 16), implying that RNF20 would stimulate PPARγ activity by diminishing the repressor function of NCoR1. Moreover, RNF20 overexpression upon NCoR1 suppression did not further promote PPARγ activity (Fig. 7F), indicating that the effect of RNF20 on stimulating PPARγ activity might be NCoR1 dependent. We examined whether RNF20 might also affect the repressor activity of SMRT, another corepressor that functions similarly to NCoR1. However, unlike NCoR1, RNF20 overexpression did not relieve repressor activity of SMRT against PPARγ (Fig. 7G), implying that the effect of RNF20 on transcriptional activity of PPARγ would be selectively mediated through NCoR1.

NCoR1 functions as a corepressor by recruiting HDAC3. To test the idea that the increased level of NCoR1 protein in adipocytes of Rnf20^+/− mice might induce HDAC3 recruitment, the extent of HDAC3 enrichment in the PPREs of PPARγ target genes, such as Fabp4 and Cd36, was assessed (29). As shown in Fig. 7H, the degree of HDAC3 enrichment was elevated in primary adipocytes of Rnf20^+/− mice, implying that RNF20 would modulate the level of NCoR1 protein and eventually influence HDAC3 recruitment in the PPREs in adipocytes.

RNF20 is an E3 ubiquitin ligase that participates in various biological events, including lipid metabolism, epigenetic modulations, mitosis, and heat shock response (9–13). However, the roles of RNF20 at the organism level, particularly in whole-body energy metabolism, have not been explored. Here, we demonstrate that RNF20 is a crucial factor in the development of adipocytes, which might influence pathophysiological alteration in obesity. In adipocytes, RNF20 stimulates the degradation of NCoR1 protein, potentiating PPARγ activity in adipocytes.

PPARγ is the most important transcription factor for adipocyte biology. Thus, subtle changes in PPARγ activity could alter adipocyte differentiation and whole-body energy homeostasis. For example, severe dysregulation of PPARγ activity with dominant-negative mutations inhibits adipocyte differentiation and is closely related to metabolic disorders, such as type 2 diabetes (30). In contrast, a slight decrease in PPARγ activity, which does not induce lipodystrophy as in the cases of Pro12Ala and haploinsufficiency of Pparg (Pparg^+/−) mice, is associated with low BMI and improved insulin sensitivity (31–34). On the basis of several observations from mouse and human genetic studies that moderate PPARγ activation can alleviate obesity and insulin resistance (31–34), fine-tuning of PPARγ activity has been suggested as one of the strategies for treating metabolic disorders (35). Thus, the identification of fine-tuners for PPARγ activity could provide important clues for adipocyte biology and for understanding the pathogenesis of metabolic diseases, including insulin resistance and obesity.

Accumulating evidence has shown that coactivators and corepressors determine PPARγ activity (7,24,28,36). For instance, knockdown of Ncor1 stimulates adipogenesis in 3T3-L1 cells (28), and adipocyte-specific Ncor1 knockout mice exhibit enhanced PPARγ activity in adipose tissue, which leads to increased adipogenesis and enhanced insulin sensitivity despite increased adiposity in DIO (24). These observations suggest that quantitative modulations of NCoR1 protein would be crucial for adipogenesis and PPARγ activity. Our data propose that RNF20 is an E3 ubiquitin ligase for NCoR1 and that it stimulates PPARγ activity by promoting proteasomal degradation of NCoR1 in adipocytes. Several lines of evidence support this notion. First, Rnf20^+/− mice exhibited decreased fat mass as a result of the downregulated capacity of adipogenesis at least in part by reduced PPARγ activity. Second, RNF20 potentiated the transcriptional activity of PPARγ. Third, RNF20 physically bound to NCoR1 and promoted proteasomal degradation of NCoR1. Finally, RNF20 modulated HDAC3 recruitment on the PPREs, which eventually influenced the expression of PPARγ target genes. Collectively, our findings suggest that RNF20 would be an important factor to promote adipocyte differentiation by regulating PPARγ activity through NCoR1 degradation (Fig. 8).

PPARγ activation at the early phase of adipogenesis (0–2 days after adipogenic stimuli) is critical for adipocyte differentiation in 3T3-L1 cells. Suppression of PPARγ activity at the early phase disrupts adipogenesis (37), and enhanced PPARγ activity at the early phase promotes adipocyte differentiation (38). However, expression profiles of many coactivators that modulate PPARγ activity and adipogenesis are altered at the late phase of adipogenesis (4 days after adipogenic stimuli), while some are not regulated during adipogenesis (36,39,40). Although these factors are crucial for efficient control of PPARγ activity, they do not seem to provide the environment in which PPARγ is readily activated at the early phase of adipogenesis. In this aspect, our data suggest that RNF20 may serve as a facilitator for PPARγ activation and adipogenesis at the early stage of adipogenesis without affecting the expression levels of PPARγ upstream regulators, such as KLFs, EBFs, and C/EBPs.

Since NCoR1 is a corepressor family member, RNF20 might affect the activity of multiple nuclear receptors by regulating NCoR1 protein stability. However, several studies using NCoR1 knockout mice have suggested that primary targets of NCoR1 appear to be selective, depending on distinct cellular environments (24,41). For example, it has been reported that the dominant function of NCoR1 in adipocytes is to repress the transcriptional activity of PPARγ (24). Similar to this, quantitative proteomic analysis presently revealed that defective RNF20 in adipocytes downregulated expressions of PPARγ target proteins, probably through NCoR1 regulation. In addition, the expression levels of PPARγ target genes were significantly decreased in adipose tissue of Rnf20^+/− mice. Furthermore, responsiveness to PPARγ agonists was downregulated in adipose tissue of Rnf20^+/− mice. Although the possibility that RNF20 might influence the activity of other nuclear receptors or that other RNF20 targets might play a role in adipose tissue cannot be excluded, our data indicate that the RNF20-NCoR1 axis is important to regulate PPARγ activity in adipocytes.

One of the prominent phenotypes of Rnf20^+/− mice was a reduction of fat mass accompanied by downregulated adipogenic programming. Severe dysregulation of adipogenesis often results in lipodystrophy, leading to ectopic lipid accumulation and insulin resistance (42). However, despite the decrease in adipose tissue mass, NCD-fed Rnf20^+/− mice displayed no significant ectopic fat accumulation in liver and muscle as well as no insulin resistance, at least as evaluated by GTT, ITT, and HOMA-IR (Supplementary Fig. 2). On the contrary, HFD-fed Rnf20^+/− mice were resistant to DIO and exhibited repressed adipocyte hypertrophy (Fig. 2). It has been suggested that adipocyte hypertrophy would be one of the major causes of fat tissue dysfunction in obese animals (19,20). For instance, adipocyte hypertrophy per se decreases glucose uptake upon insulin stimulation (43,44) and is closely associated with proinflammatory responses (45). Furthermore, basal lipolysis is increased in hypertrophic adipocytes, probably because of leakage of FFA (46,47). Rnf20^+/− mice exhibited decreased basal lipolysis in primary adipocytes, reduced serum FFA concentration, and decreased proinflammatory gene expression accompanied by alleviated adipocyte hypertrophy. Given that Rnf20^+/− mice displayed downregulated expression of PPARγ target genes in adipose tissue and that HFD-fed Rnf20^+/− mice exhibited alleviated adipocyte hypertrophy, it seems that these phenotypes would be the phenocopy of Pparg^+/− mice that downregulate the expression of PPARγ target genes compared with WT mice (31). Pparg^+/− mice exhibit reduced adipocyte hypertrophy and ameliorated insulin resistance in DIO. Nonetheless, we cannot exclude the possibility that RNF20 defects of other metabolic tissues might contribute to the improvement of insulin resistance in HFD-fed Rnf20^+/− mice, which needs to be analyzed with tissue-specific Rnf20 knockout mice in the future.

It has been reported that haploinsufficiency of several transcriptional modulators could cause a significant effect on lipid and glucose metabolism (48,49). Moreover, a recent study has reported that haploinsufficiency of >660 genes is tightly associated with human diseases and that at least 87 genes of these genes are curated as transcriptional modulators (50). Thus, it is feasible that haploinsufficiency of Rnf20 could alter metabolic regulation, accompanied by decreased adiposity and PPARγ activity.

In summary, we demonstrate that the RNF20-NCoR1 axis is crucial to regulate PPARγ activity in adipocytes. Adipocyte RNF20 could facilitate PPARγ activity by promoting NCoR1 degradation. Given that adipocyte biology is closely associated with metabolic diseases, such as obesity and insulin resistance, the regulation of RNF20 might be a potential target against obesity-induced metabolic disorders.

Acknowledgments. Yeon Su Kim and Won-Ki Huh at Seoul National University helped to find binding proteins of yeast Bre1. Sung Hee Baek at Seoul National University provided FLAG-NCoR1 vectors.

Funding. This work was supported by the National Creative Research Initiative Program of the National Research Foundation of Korea (Ministry of Science and ICT, 2011-0018312 to J.B.K.) and by the Korea Mouse Phenotyping Project of the National Research Foundation of Korea (2014M3A9D5A01073598 to J.-Y.C. and 2013M3A9D5072550 to J.K.S.). Y.G.J., Y.J., J.H.S., K.C.S., and J.P. were supported by the BK21 Plus program (21A20131212006).

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. Y.G.J. designed and conducted the study, performed experiments, and wrote the manuscript. J.H.L., Y.J., J.H.S., K.C.S., and J.P. performed experiments and contributed to the writing of the manuscript. D.L., D.W.K., and J.-Y.C. performed and analyzed proteomics experiments. S.G.Y. and J.K.S. performed the body composition analysis. S.S.C. designed and discussed the study and contributed to the writing of the manuscript. J.B.K. supervised the whole project, discussed the data, and edited the manuscript. J.B.K. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Parts of this study were presented in abstract form at the Annual Meeting of American Society for Biochemistry and Molecular Biology, San Diego, CA, 21–25 April 2018.