Ligand Binding Reduces SUMOylation of the Peroxisome Proliferator-activated Receptor γ (PPARγ) Activation Function 1 (AF1) Domain

Peroxisome proliferator-activated receptor gamma (PPARγ) is a ligand-activated nuclear receptor regulating adipogenesis, glucose homeostasis and inflammatory responses. The activity of PPARγ is controlled by post-translational modifications including SUMOylation and phosphorylation that affects its biological and molecular functions. Several important aspects of PPARγ SUMOylation including SUMO isoform-specificity and the impact of ligand binding on SUMOylation remain unresolved or contradictory. Here, we present a comprehensive study of PPARγ1 SUMOylation. We show that PPARγ1 can be modified by SUMO1 and SUMO2. Mutational analyses revealed that SUMOylation occurs exclusively within the N-terminal activation function 1 (AF1) domain predominantly at lysines 33 and 77. Ligand binding to the C-terminal ligand-binding domain (LBD) of PPARγ1 reduces SUMOylation of lysine 33 but not of lysine 77. SUMOylation of lysine 33 and lysine 77 represses basal and ligand-induced activation by PPARγ1. We further show that lysine 365 within the LBD is not a target for SUMOylation as suggested in a previous report, but it is essential for full LBD activity. Our results suggest that PPARγ ligands negatively affect SUMOylation by interdomain communication between the C-terminal LBD and the N-terminal AF1 domain. The ability of the LBD to regulate the AF1 domain may have important implications for the evaluation and mechanism of action of therapeutic ligands that bind PPARγ.


Introduction
PPARc (NR1C3) is a ligand-activated transcription factor that plays an important role in various physiological processes including adipogenesis [1,2,3], glucose homeostasis [4] and inflammatory responses [5,6]. PPARc binds to enhancers and promoters of target genes as a heterodimer with retinoid X receptor alpha RXRa [7]. Alternative promoter usage yields two PPARc isoforms (PPARc1 and PPARc2) that differ in their Nterminal extension. PPARc2 contains 30 amino terminal amino acids that are absent in PPARc1 [8]. Expression of PPARc2 is largely restricted to adipocytes whereas PPARc1 is found in several tissues [9] including lower intestine and macrophages.
The modular domain structure of PPARc resembles those of other nuclear receptors and consists of an N-terminal activation function 1 (AF1) domain, a DNA-binding domain (DBD), a Cterminal ligand-binding domain (LBD) and the most C-terminal activation function 2 (AF2) domain ( Figure 1A). PPARc is activated by polyunsaturated fatty acids and certain prostaglandins [10]. Synthetic PPARc agonists include thiazolidinediones such as rosiglitazone and pioglitazone that ameliorate insulin resistance and lower blood glucose in patients with type 2 diabetes.
PPARc is subject to several post-translational modifications (reviewed in [11,12]) including phosphorylation, ubiquitination, O-GlcNAcetylation and SUMOylation that control transcriptional activity and stability. Phosphorylation occurs at serine 112 (S82 in PPARc1) within the AF1 domain by extracellular signal-regulated kinase 1 or 2 [13] resulting in decreased transcription activity in reporter assays and decreased biological activity. Interestingly, phosphorylation of the amino terminal S112 reduces ligand binding to the C-terminus of PPARc indicating an interdomain communication between the N-terminal AF1 and the C-terminal LBD/AF2 domains [14]. Another serine in the PPARc ligandbinding domain (S273) is phosphorylated by cyclin-dependent kinase 5 [15]. Phosphorylation of serine 273 dampens the expression of selected genes such as adiponectin or adipsin and is blocked by rosiglitazone.
Another SUMOylation site has been reported within the LBD of PPARc1 [22]. According to Pascual et al. [22], SUMO1 isoform-specific modification of K365 within the LBD is induced by ligands thereby directing PPARc to the promoters of inflammatory NF-kB target genes where it inhibits transcription [22]. Specificity of SUMO1 isoform-specific conjugation to K365 by PIAS1 was highlighted as a hallmark of PPARc SUMOylation [23,24] thereby demarking it from transrepression mediated by liver x receptors, which are SUMOylated specifically by SUMO2/ 3 promoted by HDAC4 rather than by PIAS1 [24].
Although SUMOylation of PPARc is well documented in the current literature, several important aspects including SUMO isoform-specificity and the impact of ligand binding on SUMOylation remain unresolved or contradictory. In this study, we provide a comprehensive analysis of SUMOylation of PPARc1. We found that PPARc1 can be SUMOylated by SUMO1 as well as SUMO2 arguing against SUMO1 isoform-specificity. SUMOylation occurred exclusively within the N-terminal AF1 domain predominantly at lysines 33 and 77. Ligand treatment reduced SUMOylation of lysine 33 but not of lysine 77. Mutation of the SUMO attachment sites increased basal as well as ligand-induced transcriptional activation by PPARc but did not affect PPARcmediated transrepression. Lysine 365 within the LBD was not a target for SUMOylation, but was essential for ligand-induced reduction of SUMOylation, activation and transrepression. Collectively, our results suggest that PPARc ligands negatively regulate SUMOylation by intramolecular communication between the C-terminal LBD and the N-terminal AF1 domain.
Ni-NTA Pull-down Assays and Western Blotting HEK293 and HeLa cells were cultured under standard conditions. Cells were seeded at a density of 1610 6 cells per 10 cm dish, and after 24 hours transfected with 1.5 mg PPARc1 and 1.5 mg His-SUMO expression plasmids using FuGene HD (Promega). Twenty-four hours after transfection, cells were treated  Figure 1B in the absence and presence of 1 mM GW1929 or 1 mM rosiglitazone. The asterisk indicates a cross-reacting protein. Lower panel: To control for loading, the blot was reprobed with an anti His antibody. S1, untagged SUMO1, His-S1 and His-S2, His-tagged SUMO1 and His-tagged SUMO2; I, Input: 1% of cell lysate; P, Nipulldown: 90% of cell lysate. doi:10.1371/journal.pone.0066947.g001 with 1 mM rosiglitazone (Enzo Life Sciences), 1 mM GW1929 (Tocris Bioscience) or the vehicle as indicated in the figures. Fortyeight hours post transfection, cells were lysed in 1 ml lysis buffer (6 M guanidinium HCl, 0.1 M sodium phosphate buffer pH 8.0, 0.05% Tween 20, 20 mM imidazole), and His-SUMO modified proteins were isolated by incubation with 20 ml of Ni-NTA magnetic agarose beads (Qiagen) for 16 hours at 4uC. Beads were washed three times each with 750 ml buffer A (8 M urea, 0.1 M sodium phosphate buffer pH 8.0, 0.05% Tween 20, 20 mM imidazole) and buffer B (8 M urea, 0.1 M sodium phosphate buffer pH 6.4, 0.05% Tween 20, 20 mM imidazole). After a final washing step with phosphate buffered saline, the beads were boiled in 50 ml SDS sample buffer. Proteins were separated by SDS-PAGE and subsequently transferred on an Immobilon-P membrane (Millipore) for chemiluminescence or on an Immobilon-FL membrane (Millipore) for fluorescence detection according to the manufacturer's instructions. Primary and secondary antibody incubations were carried out in 1% skim milk for 1 hour each at room temperature. The rat anti-HA antibody (3F10, Roche) was used for chemiluminescence (1:2000 dilution) and for fluorescence (1:1000 dilution) detection of HA-PPARc1 proteins. The anti-FLAG M2 (Sigma), 1:1000, antibody was used for detection of FLAG-PPARc (1-256) and FLAG-PPARc (247-475). Visualization of immunoblots by chemiluminescence was performed with horseradish peroxidase-coupled anti-rat or anti-mouse antibodies (GE Healthcare Life Science, 1:15,000) followed by incubation with the Immobilon Western chemiluminescent horseradish peroxidase substrate (Millipore). The IRDye 680-labeled anti-rat secondary antibody (LI-COR Biosciences, 1:2000) was used for quantitative fluorescent detection with the Odyssey Infrared Imager (LI-COR Biosciences).

Reporter Gene Assays
Cells were seeded on 24-well plates (8610 4 cells/well) and cultured for 24 hours prior transfection. For transactivation assays, cells were transfected with 250 ng of reporter plasmid (Aox-tk-luc or 56UAS-luc), 50 ng of expression plasmid (PPARc mutants or Gal4-PPARc-LBD) and 0.5 ng of the Renilla luciferase plasmid pRL-CMV (Promega) as internal reference. Twenty-four hours post transfection, 1 mM rosiglitazone or 1 mM GW1929 was added, and cells were incubated for additional 24 hours. For transrepression assays with the iNOS promoter, RAW264.7 cells received 500 ng of iNOS-luc reporter plasmid, 200 ng of HA-PPARc expression plasmid and 10 ng of phRL-TK Renilla luciferase plasmid. Forty-two hours post transfection, cells were treated for 6 hours with 1 mg/ml LPS (E. coli 0127:B8, Sigma) and/or 1 mM rosiglitazone as indicated in the figures. For transrepression assays with the 36NF-kB promoter, HeLa cells were transfected with 250 ng of p(NF-kB)3-luc reporter plasmid, 50 ng of HA-PPARc1 expression plasmid and 0.5 ng of the pRL-CMV (Promega) Renilla luciferase plasmid. Twenty-four hours after transfection, cells were treated with 1 mM rosiglitazone for 24 hours. Ten ng/ml interleukin-1ß (Thermo Scientific) was added four hours prior cell lysis. Cells were lysed and firefly and Renilla luciferase activities were determined using the Dual Luciferase Kit (Promega) and the Berthold AutoLumat Plus LB953 multi-tube luminometer.

PPARc is SUMOylated by SUMO1 and SUMO2
SUMO1 isoform-specific modification of PPARc is portrayed as a hallmark of PPARc SUMOylation [24,27]. Close inspection of the studies on PPARc SUMOylation, however, revealed that SUMOylation of PPARc by SUMO2 was not addressed. Therefore, we asked whether PPARc could also be SUMOylated by SUMO2. Since only a very small fraction of the PPARc protein is SUMOylated at steady-state, we employed a protocol that relies on the enrichment of SUMO conjugates by purification of 66His-SUMO under denaturing conditions, followed by Western blotting for the protein of interest [28] ( Figure 1B). PPARc was modified by His-SUMO1 as well as by His-SUMO2 ( Figure 1C). The absence of any recovered PPARc upon transfection of untagged SUMO1 confirmed specificity of the PPARc-SUMO signals. PPARc was more efficiently modified by SUMO2 than by SUMO1, and several higher molecular weight PPARc species were visible upon His-SUMO2 transfection indicating multiple SUMO attachment sites or poly SUMO chain formation. We conclude that SUMOylation of PPARc is not SUMO1 isoformspecific, but that PPARc is also efficiently modified by SUMO2.

Ligands Reduce SUMOylation of PPARc
We investigated SUMOylation of PPARc in the presence of its synthetic ligand rosiglitazone and the nonthiazolidinedione PPARc agonist GW1929. Modification of PPARc by SUMO1 as well as by SUMO2 was reduced in the presence of ligands ( Figures 1D and 1E). Reduced SUMOylation of PPARc upon ligand treatment occurred in HEK293 as well as in HeLa cells. This result is in accordance with the observation of Ohshima et al. [18], who found reduced levels of SUMO1-conjugated PPARc2 in HEK293 cells following rosiglitazone treatment. However, reduced SUMOylation of PPARc in the presence of rosiglitazone contradicts the result of Pascual et al. [22], who reported increased SUMO1 conjugation of transiently expressed PPARc1 in HeLa cells.
Ligand Binding to the C-terminal LBD Reduces SUMOylation of Lysine 33 within the N-terminal AF1 Domain PPARc1 contains a perfect SUMOylation consensus sequence (yKXE, y represents a hydrophobic amino acid) at K77 within the N-terminal AF1 domain. We analyzed the PPARc mutant in which K77 is replaced by an arginine residue ( Figure 2A). Compared to wild type PPARc, the amount of SUMO2-modified PPARc K77R protein was much less in the absence of ligand ( Figure 2A) supporting the previous assignment of K77 as a SUMO attachment site [16,18]. However, the PPARc K77R protein was still SUMOylated indicating the existence of an additional SUMO site. Moreover, treatment with rosiglitazone further strongly decreased SUMOylation of the PPARc K77R mutant (Figure 2A) suggesting that SUMOylation of a lysine other than K77 was negatively affected by ligand treatment. This conclusion was further supported by independent quantitative Western blot analyses using fluorescence-labeled secondary antibodies followed by imager quantification ( Figure 2B). Rosiglitazone as well as GW1929 reduced SUMO2-modification of the PPARc K77R mutant. Moreover, also modification of the PPARc K77R mutant by the SUMO1 isoform was reduced in the presence of ligands ( Figure 2B). In conclusion, there is also no SUMO-isoform specificity with respect to ligand-induced reduction of PPARc SUMOylation.
To map the additional SUMO site(s) in PPARc, we analyzed at first the N-and C-terminal domains of PPARc on their own. The N-terminal domain of PPARc comprising the AF1 and the DNAbinding domains (amino acids 1-256) but not the C-terminal domain comprising the LBD and the AF2 domain (amino acids 247-475) was SUMOylated ( Figure 2C). Next, we analyzed a series of N-terminal PPARc1 deletion mutants. These experiments revealed that PPARc1 was SUMOylated exclusively within the Nterminal AF1 domain, exemplified by the PPARc D1-68 K77R mutant that was neither modified by SUMO1 nor by SUMO2 in the absence or in the presence of rosiglitazone ( Figure 2D).
The amino acid sequence 1-68 of PPARc contains three lysines at positions 33 (DIK 33 P), 64 (DYK 64 Y) and 68 (DLK 68 L) that fit the recently uncovered inverted SUMOylation consensus motif D/EXKY/P [29]. We analyzed PPARc mutants in which these lysines were replaced by arginines. By this analysis we identified lysine 33 as an additional SUMO attachment site ( Figure 2E). Most significantly, the PPARc K33/77R double mutant showed only a very weak residual SUMOylation signal that, however, was completely abolished when K64 and K68 are mutated additionally ( Figure 2E and F). Finally, we identified the SUMO sites in PPARc that were affected by ligands. Rosiglitazone treatment reduced SUMOylation of wild type PPARc and of the PPARc K77R mutant (Figures 2A, B and G). This result implied that ligands affected SUMOylation of K33 and potentially also of K64 and K68. To explore whether rosiglitazone also reduced SUMOylation at K77, we compared SUMOylation of the PPARc K33/64/68R triple mutant in the absence and presence of ligand. Rosiglitazone did not affect SUMOylation of the PPARc K33/ 64/68R triple mutant ( Figure 2G) showing that SUMOylation of K77 is not affected upon ligand treatment. In conclusion, our results strongly suggest that ligand binding to the C-terminal LBD of PPARc reduces SUMOylation of the N-terminal AF1 domain at K33 ( Figure 2H).

PPARc Serine 82 Mutations do not Affect SUMOylation
Lysine 77 is located within a phosphorylation-dependent SUMOylation motif [30,31] (IK 77 VEPAS 82 P). Serine 82 (S112 in PPARc2) is phosphorylated by MAP kinases and a previous report provided evidence that phosphorylation of S112 increases SUMOylation at K107 in PPARc2 [16]. We asked whether S82 phosphorylation blocking or mimicking mutations (PPARc1 S82A and PPARc1 S82D) affect SUMOylation of PPARc1 in the absence or presence of ligands. We introduced both types of serine 82 mutations into wild type PPARc and into the PPARc K33R and PPARc K77R mutants, and analyzed the various PPARc lysine/serine mutants for SUMOylation by SUMO1 and SUMO2 (Figure 3). Neither the S82A nor the S82D mutation significantly affected SUMOylation of wild type PPARc, PPARc K33 or PPARc K77 in the absence of ligands ( Figure 3A). Rosiglitazone treatment reduced SUMOylation of wild type PPARc and of all PPARc mutants in which K33 was unchanged, irrespectively whether serine 82 was mutated to alanine or aspartate ( Figure 3B, top and bottom panels). In contrast, rosiglitazone did not affect SUMOylation of the PPARc K33R/S82A and of the PPARc K33R/S82D double mutants ( Figure 3B, middle panels). In summary, these results do not support the possibility that phosphorylation of serine 82 regulates SUMOylation of lysine 77. Importantly, however, the analysis of the various PPARc lysine/serine double mutants further corroborates the conclusion that rosiglitazone specifically regulates SUMOylation of K33 but does not affect SUMOylation of K77.

Lysine 365 is not SUMOylated but is Essential for Ligandinduced Reduction of SUMOylation
Lysine 365 located within the C-terminal LBD of PPARc1 is also embedded in a SUMO consensus motif (PK 365 FE), and it was previously reported that SUMO1 modification of K365 is induced by rosiglitazone [22]. Our results do not support the assignment of K365 as a SUMO attachment site as rosiglitazone treatment of the PPARc K33/64/68/77 quadruple mutant did not result in any SUMOylation signal ( Figure 2F). However, since ligand binding reduced SUMOylation of the AF1 domain, we asked whether mutation of K365 would affect SUMOylation of the AF1 domain. We analyzed PPARc mutants in which K365 was replaced by arginine ( Figure 4). SUMOylation of the PPARc K365R mutant in the absence of ligands was similar to wild type PPARc ( Figure 4A). Strikingly, however, treatment with rosiglitazone did not reduce SUMO modification of the PPARc K365R mutant by SUMO2 or SUMO1 ( Figure 4A). This result was corroborated by the analysis of the PPARc K77/365R double mutant. PPARc K77/365R was still SUMOylated but SUMOylation did not change upon ligand treatment ( Figure 4B). These results were further substantiated by an independent quantitative Western blot analysis using fluorescence-labeled secondary antibodies. Neither SUMO1 nor SUMO2 modification of the PPARy365R mutant was reduced upon rosiglitazone or GW1929 treatment ( Figure 4C). Collectively, these findings suggest that K365 is not SUMOylated either in the absence or presence of ligands. However, the K365

SUMOylation of PPARc Represses its Activation Function but does not Affect its Transrepression Function
To analyze the impact of the individual SUMOylation sites on the transactivation capacity of PPARc we performed reporter gene assays in HeLa and RAW264.7 cells using a luciferase reporter gene driven by three copies of the acyl CoA oxidase PPARc response element linked to the tk promoter (Aox-tk-luc) [5]. In HeLa cells, wild type PPARc activated the Aox-tk construct approximately 20-fold, which increased up to 40-fold upon rosiglitazone treatment (Figure 5, top). Compared to wild type PPARc, activation by the PPARc K33R and the PPARc K77R was approximately 1.5-fold and 2-fold higher in the absence as well as in the presence of rosiglitazone. Additional 2-fold activation was obtained with the PPARc K33/77R double mutant ( Figure 5A, top). In RAW264.7 cells, the fold-induction rate by rosiglitazone in the presence of wild type PPARc was significantly higher than in HeLa cells ( Figure 5A, bottom). Still the PPARc K33R, PPARc K77R and PPARc K33/77R mutants showed increased activation in the absence and presence of rosiglitazone ( Figure 5A, bottom). Similar results were obtained when we treated the cells with the GW1929 ligand (data not shown). In conclusion, SUMOylation of both lysine residues, K33 and K77, represses PPARc1-dependent activation. We also analyzed the PPARc K33/64/68R triple mutant and the PPARc K33/64/68/ 77R quadruple mutant. Activation by the PPARc K33/64/68R triple mutant was similar to the PPARc K33R mutant and activation by the PPARc K33/64/68/77R quadruple mutant was similar to the PPARc K33/77R double mutant ( Figure 5A). This result suggests that SUMOylation of K64 and K68, which was negligible as compared to SUMOylation of K33 and K77 (see Figure 2) does not markedly influence the activation capacity of PPARc.
We also analyzed the PPARc K365R mutant for activation of the Aox-tk promoter. The PPARc K365R mutant was much less active in the absence as well in the presence of ligand. Strongly reduced activity of the PPARc K365R mutant contradicts the results of Pascual et al. [22] who reported similar activation of the Aox-tk luciferase construct by wild type PPARc and by the PPARc K365R mutant. To finally clarify whether the K365R mutation  Twenty-four hours after transfection, cells were treated with 1 mM rosiglitazone (+) or the vehicle (-), and incubated for additional 24 hours. The reporter activity in the absence of PPARc was arbitrarily set to 1. Error bars are mean +/2 SD. Statistical significance of activation by PPARc mutants compared to wild type PPARc in the absence (*) or presence ( + ) of rosiglitazone was calculated using the Students t-test. * and + , p,0.05; ** and ++ , p,0.005. (B) HEK293 cells were transfected with a 56UAS-driven luciferase reporter along with expression constructs for Gal4 or Gal4-PPARc-LBD fusions as indicated. Twenty-four hours after transfection, cells were treated with 1 mM affects the activation function of PPARc, we analyzed the K365R mutation also in another experimental setting. We fused the PPARc wild type LBD and the PPARc K365R mutant LBD to the DNA binding domain of the yeast transcription factor Gal4, and analyzed the activity of the Gal4-PPARc-LBD fusion proteins on a Gal4-responsive promoter ( Figure 5B). The Gal4-PPARc-LBD-wt protein activated transcription from the 56Gal4-promoter up to 50-fold and 28-fold in the presence of rosiglitazone or GW1929, respectively. Activation by the Gal4-PPARc-LBD-K365R mutant, however, was much lower in the absence and presence of ligands ( Figure 5B). This result supports the conclusion that the K365R mutation within the PPARc-LBD impairs LBD activity.
PPARc ligands can modulate inflammatory signaling by repressing the induction of inflammatory genes without directly binding to their promoters [5]. This transrepression activity of PPARc ligands can be monitored by their ability to repress lipopolysaccharide (LPS)-induced activation of the mouse inducible nitric oxide synthase (iNOS) promoter in RAW264.7 macrophages [5,22]. We tested PPARc mutants for their ligand-dependent transrepression activity by cotransfecting an iNOS promoter-driven luciferase construct along with PPARc expression constructs in RAW264.7 cells ( Figure 6A). LPS treatment activated the iNOS promoter up to 13-fold. Rosiglitazone did not affect activation in absence of PPARc but inhibited activation by approximately 36% in the presence of PPARc. All PPARc variants with K33R or K77R mutations including the SUMOylation-deficient PPARc K33/64/68/77R quadruple mutant exerted also transrepression activities although to a slightly lesser extent ( Figure 6A). This result suggests that SUMOylation of PPARc is not absolutely essential for repressing LPS-induced activation of the iNOS promoter. We also tested the PPARc K365R mutant in this transrepression assay. The PPARc K365R mutant failed to significantly mediate repression of the iNOS promoter further supporting the conclusion that the K365R mutation impairs PPARc-LBD activity.
Previously, it was shown that rosiglitazone inhibits promoter activation by NF-kB in the presence of PPARc [32]. Therefore, we also analyzed the SUMOylation-deficient PPARc K33/64/ 68/77R and the PPARc K365R mutants for their ability to mediate repression of an NF-kB-specific reporter construct that was activated by interleukin-1ß treatment ( Figure 6B). Wild type PPARc as well as the SUMOylation-deficient PPARc K33/64/ 68/77R mutant inhibited NF-kB activation by approximately 40% upon rosiglitazone treatment. The PPARc K365R mutant, however, retained only residual repression activity ( Figure 6B).
Collectively, these results strongly suggest that SUMOylation negatively affects activation functions but is largely dispensable for transrepression activity of PPARc. Lysine 365 is not SUMOylated. However, it is essential for PPARcs activation and transrepression functions.

Discussion
Previous studies on SUMOylation of PPARc were rather incomplete or yielded contradictory results. In this study, we showed that PPARc can be modified by SUMO1 and by SUMO2 within the N-terminal AF1 domain, and we finally mapped the major SUMOylation sites to lysine 33 and 77. SUMOylation within the N-terminal AF1 domain was negatively regulated by ligand binding to the C-terminal LBD affecting activation but not transrepression functions of PPARc. Interestingly, ligand binding to PPARc reduced specifically SUMOylation of lysine 33 embedded in the inverted SUMO consensus site [D/E]xKY/P [29], but not of lysine 77 embedded in the classical SUMO consensus site yKXE. Whether reduced SUMOylation of lysine 33 reflects impaired SUMOylation or, alternatively, enhanced de-SUMOylation remains unclear.
Reduced SUMOylation of PPARc after ligand treatment is in line with the report of Ohshima et al. [18] who found that the amount of SUMO1-conjugated PPARc2 is lower in HEK293 lysates of rosiglitazone-treated cells. Yet, Pascual et al. [22] reported increased SUMOylation of transfected PPARc1 in HeLa cells following rosiglitazone treatment. Importantly, both studies had not entirely mapped the PPARc SUMO attachment sites and therefore could not include appropriate controls in their analysis. We believe that our mutational analysis finally clarified the effect of ligands on SUMOylation. Ligand binding to the C-terminal PPARc LBD reduces SUMOylation of the N-terminal AF1 domain. A future goal should be to define the SUMOylation pattern of PPARc in primary cells under relevant physiologic and pathologic conditions.
Interestingly, it was reported that, vice versa, the AF1 domain can also affect the LBD domain as phosphorylation of the PPARc AF1 domain at serine 112 by MAP kinase reduced ligand binding affinity to the C-terminal part [14]. Thus, our results lend further credence to the concept of an intramolecular communication between the C-and N-terminal PPARc domains. How the interplay between the N-terminal AF1 domain and the C-terminal LBD is achieved mechanistically is unknown. The structure of the AF1 domain encompassing the SUMOylation sites was not resolved in the crystallized intact PPARc-RXRa nuclear receptor complex on DNA [33], and no direct interaction between the AF1 domain and the LBD of PPARc was detected [14]. Potentially, ligand binding induces allosteric changes that may affect the accessibility of K33 for SUMO-modifying enzymes. An alternative intriguing idea would be that SUMO modification of the AF1 domain mediates a direct interaction between the N-terminal and the C-terminal PPARc domains. In line with this idea, inspection of the PPARc LBD revealed several SUMO-interaction motifs. Unfortunately, inefficient in vitro SUMOylation of PPARc impeded interaction studies of SUMOylated N-terminal PPARc fragments with the C-terminal LBD domain.
Ligand-activated PPARc is recruited to promoters of inflammatory genes where it inhibits transcription by preventing proteasome-mediated clearance of repressive nuclear receptor corepressor (N-CoR) complexes. It was reported that the initial rosiglitazone (Rosi) or 1 mM GW1929 for additional 24 hours. The reporter activity in the absence of Gal4 fusions was arbitrarily set to 1. Error bars are mean +/2 SD. Statistical significance of activation by Gal4-LBD and Gal4-LBD-K365R compared to Gal4 was calculated by the Students t-test. *, p,0.05; **, p,0.005. doi:10.1371/journal.pone.0066947.g005 step of this pathway involves ligand-induced SUMO1 conjugation to K365 within the PPARc ligand-binding domain [22]. Accordingly, ligand-induced SUMOylation of PPARc at lysine 365 specifically by the SUMO1 isoform is repeatedly portrayed in many reviews [1,11,12,23,34,35,36,37,38,39]. Our results do not support the assignment of K365 as a SUMOylation target site as we did not detect any residual SUMOylation of the PPARc K33/ 64/68/77R mutant protein neither in the absence nor in the presence of ligands. However, K365 affected SUMOylation of PPARc indirectly as it prevented ligand-induced reduction of SUMOylation at K33. The PPARc K365R mutant was much less responsive to rosiglitazone indicating that the K365R mutation affects LBD activity. In line with this finding, it was reported that mutation of K395 in PPARc2 (corresponding to K365 in the PPARc1 isoform) also impaired rosiglitazone-induced positive transcriptional activity of PPARc [19]. Impaired LBD activity of the PPARc K365R mutant readily explains why (i) rosiglitazone treatment did not affect SUMOylation, (ii) did only weakly activate PPARc-dependent transcription and (iii) did barely mediate rosiglitazone-induced transrepression. Our results imply that SUMOylation of K365 is not involved in transrepression by PPARc, but do not necessarily challenge the conclusion of Pascual et al. [22] that the SUMOylation machinery is generally required for PPARc-dependent transrepression. However, how SUMOylation acts in this pathway remains to be uncovered. Notably, several proteins involved in this pathway such as N-CoR [40] and the transducin beta-like proteins TBL1-TBLR1 [41] are also targets of SUMOylation.
Taken together, in this study we unambiguously assigned the SUMOylation sites of PPARc to lysine residues within the AF1 domain and provide evidence that ligand binding to the Cterminal LBD affects the function of the N-terminal AF1 domain by altering SUMOylation. Thus, our results may have important implications for the evaluation and mechanism of action of therapeutic agonists and antagonists that bind PPARc. Figure 6. Transrepression activity of PPARc mutants. (A) RAW264.7 macrophages were transfected with the iNOS luciferase reporter plasmid along with PPARc mutants. Forty-two hours after transfection, cells were treated for 6 hours with 1 mg/ml LPS and 1 mM rosiglitazone (Rosi) as indicated. The reporter activities in the presence of LPS were set to 100% promoter activity. (B) Hela cells were transfected with the 3xNF-kB luciferase reporter plasmid along with PPARc mutants. Twenty-four hours after transfection, cells were treated with 1 mM rosiglitazone (Rosi). Four hours prior lysis, 10 ng/ml interleukin-1b (IL-1ß) was added as indicated. The reporter activities obtained by interleukin-1ß stimulation were set to 100% promoter activity. Error bars are mean +/2 SD. Statistics was performed using Students t-test. *, p,0.05; **, p,0.005. doi:10.1371/journal.pone.0066947.g006