Communicated by: Hisato Kondoh
The transactivating function of peroxisome proliferator-activated receptor γ is negatively regulated by SUMO conjugation in the amino-terminal domain
Article first published online: 27 OCT 2004
© 2004 Blackwell Publishing Limited
Genes to Cells
Volume 9, Issue 11, pages 1017–1029, November 2004
How to Cite
Yamashita, D., Yamaguchi, T., Shimizu, M., Nakata, N., Hirose, F. and Osumi, T. (2004), The transactivating function of peroxisome proliferator-activated receptor γ is negatively regulated by SUMO conjugation in the amino-terminal domain. Genes to Cells, 9: 1017–1029. doi: 10.1111/j.1365-2443.2004.00786.x
- Issue published online: 27 OCT 2004
- Article first published online: 27 OCT 2004
- Received: 27 May 2004 Accepted: 3 August 2004
Peroxisome proliferator-activated receptor (PPAR)-γ2, a member of the nuclear hormone receptor superfamily, plays a key role in adipocyte differentiation. Its amino-terminal region carries a ligand-independent gene-activating function, AF-1, and is composed of activation as well as repression domains. We have found PPARγ2 and its isoform, PPARγ1, to be modified by small ubiquitin-related modifier (SUMO)-1 in vivo, at a lysine residue in the repression domain. In reporter assays, a sumoylation-defective K107R mutant of PPARγ2 exhibited much stronger transactivation than the wild-type, comparable with that of a mutant deleted for the repression domain. A close inverse correlation was observed between the levels of sumoylation and transactivation by PPARγ2, in analyses employing PPARγ2 forms with mutations in the sumoylation motif and a dominant-negative mutant of the SUMO conjugating enzyme, Ubc9. Studies with phosphorylation-defective mutants suggested that phosphorylation at S112 of PPARγ2 promotes K107 sumoylation, and this latter exerts the more potent repressive effects. The K107R mutant PPARγ2, when infected into NIH3T3 cells with a viral vector, promoted differentiation into adipocytes more efficiently than the wild-type. These observations provide evidence that sumoylation is involved in negative regulation of the transactivating function of PPARγ2.
The peroxisome proliferator-activated receptor (PPAR) is a family of nuclear receptors (for review, see Desvergne & Wahli 1999). PPARγ has two isoforms, γ1 and γ2, generated from the same gene by alternative promoter usage, PPARγ2 containing 30 additional amino acids at the amino-terminus. PPARγ1 is widely expressed in many sites, including fat cells and elements of the immune system, whereas PPARγ2 is exclusively found in adipose tissues, playing a key role in adipocyte differentiation. PPARα regulates the expression of several genes involved in fatty acid metabolism and is highly expressed in the liver, while PPARβ/δ is ubiquitously expressed and involved in various physiological processes, including lipid homeostasis. Each PPAR binds to peroxisome proliferator-response elements (PPREs) located in the vicinity of target genes through heterodimerization with another nuclear hormone receptor, the retinoid X receptor (RXR).
Similar to most nuclear receptors, the PPAR consists of four distinct functional regions, A/B, C, D and E, in order from the amino- to the carboxyl-termini (Fig. 1A) (Mangelsdorf et al. 1995). The A/B region varies among the subtypes and has a ligand-independent gene-activating function (AF-1). The C region is the best conserved throughout the nuclear receptor superfamily, containing two zinc finger motifs responsible for DNA binding. The E region is next well conserved and is involved in ligand binding and the ligand-dependent activation function (AF-2). The D region is a hinge domain between the C and E regions and is important for heterodimerization with RXR and AF-2 activity. AF-1 activity has been confirmed in the A/B region of human PPARγ2 (amino acid residues 1–138) (Adams et al. 1997) and a GAL4-PPARγ2 construct containing only residues 1–99 has been shown to exhibit higher activity than the whole A/B region (Werman et al. 1997), implying that the remaining region, residues 100–138, exerts a repressive effect. In addition, it was demonstrated that mitogen-activated protein kinase (MAPK)-dependent phosphorylation of a serine at position 112 (S112) of PPARγ2 reduced its AF-1 (Adams et al. 1997) as well as AF-2 (Hu et al. 1996; Adams et al. 1997) activity. However, the mechanisms of regulation of PPARγ2 AF-1 have yet to be clarified in detail.
The small ubiquitin-related modifier (SUMO) group of peptides of about 100 amino acid residues, covalently attach to target proteins via specific lysine residues. Protein sumoylation has a variety of functions, for example, in regulation of protein stability, nuclear protein localization and gene transcription (for review, see Seeler & Dejean 2003). Of the three forms so far identified, SUMO-2 and -3 are closely related, and are implicated in functions distinct from those of SUMO-1. The conjugating system for SUMO is distinct from, but analogous to, that for ubiquitin conjugation. SUMO peptides are translated as larger precursors and converted to mature forms with a glycine residue (Gly97 in SUMO-1, for example) at the carboxyl-termini, upon cleavage by a SUMO protease. Following exposure of the carboxyl-terminal glycine, SUMO is attached to a heterodimeric E1 activating enzyme, Aos1-Uba2. The activated SUMO is passed from Aos1-Uba2 to Ubc9, an E2 conjugating enzyme, forming a thioester intermediate with an active cysteine residue (Cys93) of Ubc9, and is finally ligated to the lysine residues of target proteins. Multiple E3 ligases have recently been found to act as adaptors between Ubc9 and the substrate proteins. A consensus sumoylation motif, ψKxE/D, has been identified in many target proteins, where ψ represents a large hydrophobic residue, K is the SUMO acceptor site, and x may be any amino acid. A number of transcriptional factors have recently been described as regulated, and in most cases repressed, by sumoylation (Gill 2003; Verger et al. 2003). These include members of the nuclear receptor superfamily, such as the androgen receptor (Poukka et al. 2000), glucocorticoid receptor (Le Drean et al. 2002; Tian et al. 2002), progesterone receptor (Abdel-Hafiz et al. 2002; Chauchereau et al. 2003), and mineralcorticoid receptor (Tallec et al. 2003). Furthermore, co-regulators of nuclear receptors, like HDAC 1, 4 and 6 (David et al. 2002; Kirsh et al. 2002), SRC-1 (Chauchereau et al. 2003) and p300 (Girdwood et al. 2003), have also been shown to be regulated by sumoylation. Therefore, this process is an important regulatory mechanism for nuclear receptor-mediated transcription.
In the present study, we demonstrated that the carboxyl-terminal part of the A/B region of PPARγ2 is in fact a repression domain and that SUMO is conjugated at a lysine residue, K107, in this domain. Sumoylation is involved in the repression of AF-1 activity, leading to negative regulation of a major physiological function of PPARγ, promotion of adipocyte differentiation. In addition, phosphorylation at S112, another negative regulatory mechanism of PPARγ, enhances sumoylation.
A possible sumoylation motif is important for the repression of PPARγ2 AF-1
To investigate the regulatory mechanism of PPARγ2 AF-1, the A/B region (residues 2–138) was fused to the DNA-binding domain (BD) of yeast GAL4 (residues 1–147) (Fig. 1B). When the expression plasmid of GAL4-PPARγ2[2–138] was transfected into HeLa cells together with a luciferase reporter plasmid, tk-GALpx3-luc, slight but significant ligand-independent transactivation was observed (lanes 1–3). Residues 2–99 fused to GAL4-BD exhibited 15-fold higher reporter gene expression than residues 2–138 (lane 4), as reported for a similar experiment performed with CV-1 cells (Werman et al. 1997). This result supported the notion that the residues 100–138 exert a repressive effect on PPARγ2 AF-1 (see below). Although the S112A mutant of GAL4-PPARγ2[2-138], defective in phosphorylation by MAPK, exhibited a higher level of transactivation (lane 6) than WT, it was lower than that of GAL4-PPARγ2[2–99], suggesting the existence of another mechanism for the regulation of AF-1 activity of PPARγ2.
On inspecting the amino acid sequence of the PPARγ2 A/B region, we noted a putative sumoylation motif (IKVE) in the repression domain, at positions 106–109 (Fig. 1A). A number of transcriptional factors are sumoylated and mutations of sumoylation sites generally increase transactivating function (Gill 2003; Verger et al. 2003). Accordingly, we investigated the possibility that the AF-1 activity of PPARγ2 is regulated by sumoylation. We created a mutant of GAL4-PPARγ2[2–138], in which the lysine at position 107, the potential sumoylation site, was changed to arginine (GAL4-PPARγ2[2–138](K107R)). This mutant showed a high level of transactivation, comparable with that of GAL4-PPARγ2[2–99] (Fig. 1B, compare lanes 4 and 5). Furthermore, a K107R/S112A double mutant of GAL4-PPARγ2[2–138] also exhibited a similar level of transactivation to the constructs, GAL4-PPARγ2[2–99] and GAL4-PPARγ2[2–138](K107R) (lane 7). Expression levels of the GAL4-derivatives were comparable as indicated by Western blotting (data not shown). Similar results were obtained with NIH3T3 cells (data not shown), indicating that K107 is essential for the negative regulation of PPARγ2 AF-1.
To further characterize the repressive function of the region comprising residues 100–138, it was fused to GAL4-BD (Fig. 1C, [100–138](WT)), resulting in a halving of the basal transactivation by GAL4-BD (compare lanes 1 and 3). The S112A mutant of GAL4-PPARγ2[100-138] exhibited greater transactivation than WT (lane 5), whereas the K107R and K107R/S112A mutants showed significantly higher levels of transactivation than both WT and the S112A mutant (lanes 4 and 6). All the protein constructs were expressed at comparable levels, as assessed by Western blotting (data not shown). Hence, it was concluded that the region containing residues 100–138 is in fact a repression domain and K107 is required for the repressive function.
PPARγ2 is modified by SUMO-1 in the repression domain
To examine whether PPARγ2 was modified by sumoylation, myc-tagged PPARγ2 and SUMO-1 fused to CFP were transiently expressed together in HeLa cells (Fig. 2A). We confirmed that the CFP-SUMO-1 protein was functional, based on the observation that it conjugated with promyelocytic leukaemia (PML) protein, an established major target of sumoylation (data not shown). Western blot analysis of cell extracts with an anti-myc antibody exhibited an additional band of a slower migrating cross-reactive material with an apparent molecular mass of 140–150 kDa (lanes 4–8). To examine whether this might represent myc-PPARγ2 covalently conjugated to CFP-SUMO-1, a CFP-SUMO-1(G97A) mutant, which cannot conjugate with target proteins (Kamitani et al. 1997), was used instead of the wild-type CFP-SUMO-1. In this case, the additional band was absent (lanes 9 and 10). Upon treatment of the transfected cells with 1 µm BRL49 653, a specific ligand of PPARγ, the intensity of the additional band was slightly increased (lanes 4 and 5, as well as 6–8; also see Fig. 5A).
To further confirm that the slower migrating band represented PPARγ2 modified by CFP-SUMO-1, immunoprecipitation was performed with an anti-PPARγ antibody (Fig. 2B,C). Western blotting with an anti-SUMO-1 antibody detected a band in the cell extract corresponding to the slower migrating material (Fig. 2B, lane 2, open arrowhead), which was immunoprecipitable with the anti-PPARγ antibody (lane 7, open arrowhead). With an anti-myc antibody, the slower migrating band was detected in both the total cell extract and immunoprecipitate (Fig. 2C, lanes 2 and 7). When the G97A mutant of CFP-SUMO-1 was used instead of WT (Fig. 2B,C, lane 8), the band was not detected with either anti-SUMO-1 or anti-myc antibodies.
We next examined whether K107, located within the sumoylation consensus sequence, IKVE, might be the site of sumoylation. Protein extracts from cells transfected with an expression plasmid bearing a K107R mutant of myc-PPARγ2 were immunoprecipitated with an anti-PPARγ antibody and anti-SUMO-1 antibody failed to detect sumoylated PPARγ2 in either total cell extracts or immunoprecipitates (Fig. 2B, lanes 4 and 9). In addition, SUMO-1-modified PPARγ2 was not detected using this mutant with an anti-myc antibody (Fig. 2C, lanes 4 and 9). Therefore, the mutation of K107 blocked the sumoylation of PPARγ2, indicating that this residue, located in the repression domain of A/B region, is the major sumoylation site of PPARγ2. Although the molecular size of the slower migrating material estimated from the electrophoretic mobility was larger than the sum of the molecular weights of myc-PPARγ2 and CFP-SUMO-1, it is known that sumoylated proteins often migrate on sodium dodecyl sulfate—polyacrylamide gel electrophoresis (SDS–PAGE) abnormally slowly in relation to their real sizes (Poukka et al. 2000; Hietakangas et al. 2003). A single SUMO-1 molecule is attached at a single sumoylation site. Therefore, we concluded the slower migrating material to be PPARγ2 conjugated with monomeric SUMO-1 at K107.
Sumoylation represses the transactivating function of PPARγ2
We next studied the role of SUMO modification in the regulation of PPARγ2 transactivating function. Expression plasmids for WT and the K107R mutant of full-length PPARγ2 were transfected into HeLa cells, together with a luciferase reporter plasmid containing a single or three copies of the peroxisome proliferator-response element (PPRE) of the PEX11α/perilipin gene pair (Shimizu et al. 2004) (Fig. 3A). The K107R mutant activated luciferase expression to a significantly greater extent than WT, in both the presence and absence of the ligand (lanes 3–6 and 9–12). This effect of the mutation was observed with both single and triple PPREs, although the reporter gene expression itself was much higher in the latter case. Strong transactivation by the K107R mutant was also observed when another luciferase reporter plasmid, pAOXPPREluc, was used (Fig. 3B). In addition, similar results were obtained with pGVP-P/P-PPREx3-luc, using NIH3T3 cells (data not shown). Therefore, sumoylation at K107 appeared to regulate not only the AF-1 but also the whole transactivating function of PPARγ2 negatively, independent of the number of binding sites and promoter sequences.
Other than sumoylation, a lysine residue could also be the site for different post-translational modifications, such as ubiquitination and acetylation (Freiman & Tjian 2003). For PPARγ2, ubiquitination is known to be promoted by ligands (Hauser et al. 2000). Therefore, it was necessary to confirm that the K107R mutation disrupted the function of the repression domain by eliminating sumoylation. If sumoylation itself represses the transactivation, the activity should be inversely correlated with the sumoylation level. To examine this, we designed and performed several experiments.
We first determined the role of each amino acid within the IKVE motif of PPARγ2 for sumoylation in vivo, as previously conducted in vitro for other proteins (Rodriguez et al. 2001). Single amino acid substitutions (IKVE (WT) to MKVE (I106M), AKVE (I106A), IRVE (K107R), and IKVA (E109A)) were introduced into myc-PPARγ2 and WT and the mutant constructs were examined for SUMO modification in vivo (Fig. 4A). As shown above, K107R abrogated the conjugation with SUMO-1 (lanes 6 and 7). E109A also lost SUMO modification (lane 8), consistent with the established observation that glutamic acid at the fourth position is essential for sumoylation. Similar to the case of the leucine residue in the RanGAP1 sumoylation consensus (LKSE) (Rodriguez et al. 2001), mutations of the isoleucine preceding K107 decreased sumoylation in the order, WT, I106M and I106A (lanes 3–5). Taken together, the results suggested that each conserved amino acid of the sumoylation motif of PPARγ2 is required for SUMO modification in vivo.
We next examined the transactivation by the mutant PPARγ2 constructs (Fig. 4B) and observed increase in the order, WT, I106M, I106A and K107R (lanes 4–6 and 8). E109A exhibited a comparable level of transactivation with that of K107R (lanes 8 and 9). Therefore, the sumoylation level is inversely correlated with the transactivation, supporting the notion that sumoylation at K107 represses the transactivating function of PPARγ2.
Transactivation by PPARγ2 is stimulated by a dominant-negative mutant of Ubc9
We further investigated whether the activity of the SUMO conjugating system affected the transactivating function of PPARγ2. Ubc9(C93S) is defective in SUMO binding (Gong et al. 1997) and has successfully been used as a dominant-negative mutant (Hirano et al. 2003). When HeLa cells were transfected with expression plasmids for myc-PPARγ2 and CFP-Ubc9(C93S), sumoylation of PPARγ2 was reduced significantly by the expression of Ubc9(C93S) (Fig. 4C, lanes 3 and 4). In reporter assays using pGVP-P/P-PPREx3-luc, the ligand-dependent transactivation by WT PPARγ2 was significantly enhanced (up to 2.6-fold) when increasing amounts of Ubc9(C93S) were co-expressed (Fig. 4D, lanes 5–7), whereas the activity of the K107R mutant was much less affected (1.2-fold) (lanes 9–11). Ubc9(C93S) was expressed as expected in all the relevant samples (Fig. 4E). Therefore, repression of the transactivation by PPARγ2 is dependent on the SUMO-conjugating activity of Ubc9, providing further support for the hypothesis that the SUMO pathway represses the transactivating function of PPARγ2.
Relationship between K107 sumoylation and S112 phosphorylation
Previous reports revealed that extracellular signals leading to MAPK activation cause phosphorylation of S112 in the repression domain of the PPARγ2 A/B region, thereby reducing transactivating function (Hu et al. 1996; Adams et al. 1997). In addition to K107R and S112A, we created two more mutants: S112D, mimicking a constitutively phosphorylated form (Shao et al. 1998), and P113Q, a mutant defective in phosphorylation at S112 associated with obesity (Ristow et al. 1998). K107R/S112A and K107R/S112D double mutants were also engineered. WT and the mutant PPARγ2 were expressed with CFP-SUMO-1 in HeLa cells, and cell extracts analysed by Western blotting (Fig. 5A). WT and S112D were sumoylated at comparable levels (lanes 4 and 8), whereas both S112A and P113Q were sumoylated at significantly lower levels (lanes 4, 6 and 10). As expected, all mutants containing K107R demonstrated loss of SUMO-1 conjugation (lanes 5, 7 and 9).
We next examined the transactivation by WT and mutant PPARγ2 (Fig. 5B). S112A and P113Q exhibited higher, whereas S112D displayed similar, levels of transactivation as compared with WT (lanes 4, 7, 9 and 11). All the mutants carrying the K107R mutation activated the reporter gene expression to comparable extents, much higher than with WT, irrespective of the mutations affecting S112 phosphorylation (lanes 6, 8 and 10). These data suggest that phosphorylation at S112 enhances sumoylation at K107 of PPARγ2 and that K107 sumoylation, rather than S112 phosphorylation, exerts the major repressive influence on PPARγ2. In our hands, the S112D mutation did not result in a significant reduction of transactivating function of PPARγ2, consistent with the lack of enhanced sumoylation as described above (Fig. 5A).
Sumoylation is specific for PPARγ among PPARs
Figure 1(A) shows potential sumoylation sites of mouse PPARs: PPARα has one (K185) within the D region and PPARβ/δ one (K104) in the C region. Sumoylation was examined by introducing expression plasmids for myc-PPARs and CFP-SUMO-1 into HeLa cells (Fig. 6A). Western blotting with an anti-myc antibody detected a slower migrating band when PPARγ2 was expressed (lanes 8-10), but not with PPARα (lanes 2–4) or PPARβ/δ (lanes 5–7). In similar experiments, SUMO-2 and -3 were found to conjugate with PPARγ2, but not PPARα or β/δ (data not shown).
PPARγ1 was also conjugated with WT SUMO-1, but not the G97A mutant, at the level apparently lower than with PPARγ2 (Fig. 6B, lanes 3–6 and 7–10). The data indicate that SUMO modification is specific for PPARγ among PPARs. We next examined the transactivation by sumoylation-defective mutants, PPARγ1(K77R) and PPARγ2(K107R) (Fig. 6C). Much higher levels of transactivation (about 3-fold) were found as compared with WT PPARγ1 and PPARγ2, respectively, indicating that SUMO represses transactivation by both PPARγ1 and γ2 to similar extents.
The sumoylation-defective mutation, K107R, enhances PPARγ2-induced adipogenesis
PPARγ2 plays a key role in adipocyte differentiation, as revealed by the observation that exogenously expressed PPARγ2 induced adipogenesis of originally non-adipogenic cells in the presence of an appropriate ligand (Tontonoz et al. 1994). To investigate the role of sumoylation in the adipogenic activity of PPARγ2, we expressed WT and the K107R mutant PPARγ2 in NIH3T3 fibroblasts, using a lentiviral vector. We confirmed that WT and the K107R mutant PPARγ2 were expressed at comparable levels (Fig. 7A) and also that WT PPARγ2, but not the K107R mutant, exogenously expressed with the viral vector, was susceptible to conjugation with CFP-SUMO-1 (data not shown). Exogenously expressed PPARγ2 led to adipogenic conversion of the cells in the presence of a PPARγ-ligand, BRL49 653, and the effect was larger for the K107R mutant than WT, with the following criteria, when assessed 24 h after confluence: neutral lipid accumulation as indicated by Oil Red-O staining (Fig. 7B); specific activity of glycerol-3-phosphate dehydrogenase (GPDH), a key enzyme of triglyceride synthesis (Fig. 7C); and expression of perilipin, a protein enriched on the surface of adipocyte lipid droplets (Fig. 7D). Therefore, the sumoylation-defective PPARγ2 mutant promoted adipogenesis of infected NIH3T3 cells in an early phase of differentiation.
In the present study, we obtained evidence that PPARγ is sumoylated in vivo and that the major site, K107, is located within the repression domain that is important for the negative regulation of PPARγ2 AF-1 activity. The sumoylation motif, IKVE, at positions 106–109, is conserved in all mammalian PPARγ2 forms, and a corresponding sumoylation motif, IKLE, has been found in Xenopus PPARγ2. As the sumoylation-defective K107R mutant of PPARγ2 here exhibited enhanced transactivation on three reporter constructs carrying different promoters and PPREs, and taking into consideration the findings using mutant PPARγ2 forms and mutant Ubc9, we conclude that sumoylation at K107 plays a critical role in the function of the repression domain. Importantly, the K107R mutation of PPARγ2 significantly enhanced the adipogenic activity on NIH3T3 cells, indicating a role for sumoylation in regulation of PPARγ2 function in a physiological process.
Members of the nuclear receptor superfamily, such as the androgen receptor, glucocorticoid receptor, progesterone receptor and mineralcorticoid receptor, possess a synergy control (SC) motif (P-X(0,4)-(I/V)-K-(Q/T/S/L/E/P)-E-X(0,3)-P) in the AF-1 domain. This motif has been suggested to be generally involved in negative regulation of synergistic activation, dependent on multiple binding sites (Iñiguez-Lluhí & Pearce 2000). Moreover, all the above nuclear receptors are sumoylated at a site within the SC motif. The sumoylation motif of PPARγ2, including the two downstream proline residues, closely matches the SC motif consensus (106-IKVEPASP-113). In addition, P110 and P113 also constitute a MAPK-site consensus sequence and P113 is known to be a site of mutation associated with obesity (P113Q) (Ristow et al. 1998). Therefore, the putative SC motif of PPARγ2 contains both a sumoylation site and a phosphorylation site. In the glucocorticoid receptor case, the effect of SUMO-1 over-expression (Le Drean et al. 2002) or sumoylation-defective mutation (Tian et al. 2002) varies with the promoter. In contrast, repressive effects of sumoylation of PPARγ2 were consistently observed on different promoters as well as with different numbers of binding sites. Hence, in a functional sense, the SC motif-like region of PPARγ2 may differ from counterpart sequences in other nuclear receptors.
We here found that a phosphorylation-defective mutation of S112 in PPARγ2 decreased the level of sumoylation. Phosphorylation, like sumoylation, represses the transactivating function of PPARγ2, leading naturally to the notion that sumoylation and phosphorylation may function synergistically. However, all the mutants containing K107R mutation exhibited increased transactivation at comparable levels, irrespective of mutations affecting the phosphorylation at S112. Hence, sumoylation at K107 seems more potent than phosphorylation at S112 for repression of transactivating function of PPARγ2. A precedent exists for the regulation of sumoylation by phosphorylation: the transactivating function of HSF1, a heat shock transcriptional factor, is regulated by phosphorylation (Pirkkala et al. 2001). HSF1 is sumoylated at K298, neighbouring two phosphorylation sites, S303 and S307, and levels of sumoylation were found to decrease in the order, WT, S303/307D and S303/307A, in vivo (Hietakangas et al. 2003).
How does K107 sumoylation repress the transactivating function of PPARγ2? The modification of DNA-binding activity is not a likely major mechanism, because the K107R mutation enhanced transactivation through both the DNA-binding domain of PPARγ itself and a heterologous binding domain, GAL4-BD. In addition, possible protein stabilization by sumoylation, shown to occur for IκBα (Desterro et al. 1998), would not account for the repressive effect on PPARγ. It has been reported (Ross et al. 2002) that the activity of Sp3, a GC box-binding transcriptional factor, is repressed by SUMO artificially attached by genetic engineering. The intranuclear distribution of sumoylated Sp3 was found to differ from that of unmodified Sp3. In a similar experiment, we observed that the transactivating function of PPARγ2 was decreased, if SUMO was attached artificially, but in this case the intranuclear distribution was not affected significantly (Yamashita, D. & Osumi, T., unpublished observation).
The possibility that sumoylation alters the local or overall conformation of the AF-1 domain of PPARγ2, thereby modifying its ability to interact with co-regulators, appears most feasible. We speculate that PPARγ2 sumoylation inhibits the binding of co-activator complexes and/or promotes co-repressor binding on to the AF-1 domain. A co-regulator complex that specifically binds to sumoylated transcriptional factors has not been identified so far, but the co-regulators, PGC-2 (Castillo et al. 1999) and p300 (Gelman et al. 1999), are reported as binding partners for the PPARγ2 AF-1 domain. PGC-2 does not seem to modulate AF-1 activity directly, whereas p300 induces activation and is known to be sumoylated itself (Girdwood et al. 2003). Other co-regulators, such as SRC-1 (Chauchereau et al. 2003) and a few HDACs (David et al. 2002; Kirsh et al. 2002), are also modified by SUMO. Their sumoylation may also contribute to the overall transactivating function of PPARγ2-co-regulator complexes.
The present results suggest that, even though the steady state level of sumoylated PPARγ2 is low, it can cause extensive repression of the transactivating function. In fact, we could not detect sumoylation of PPARγ2 by endogenous SUMO-1, under conditions similar to those applied for gene reporter assays. Equivalent results have been documented for the sumoylation of various transcription factors, including the nuclear receptors, androgen receptor (Poukka et al. 2000), progesterone receptor (Abdel-Hafiz et al. 2002; Chauchereau et al. 2003) and mineralcorticoid receptor (Tallec et al. 2003). A possible explanation is that only a small subfraction of the total PPARγ2 pool is involved in transcriptional activation and sumoylation is a major modifier of this subfraction.
In summary, we have presented here evidence that the transactivating function of PPARγ is negatively regulated by sumoylation in the amino-terminal region, through modification of AF-1 activity. Functional and mechanistic links between phosphorylation and sumoylation are interesting issues for future studies. During preparation of this manuscript, a paper reporting the sumoylation of PPARγ appeared (Ohshima et al. 2004), in which involvement of PIAS1 and PIASxβ in the sumoylation process was described, as well as suppression by sumoylation of the apoptosis-promoting activity of PPARγ. Here, we emphasize the importance of sumoylation in the regulation of PPARγ AF-1, including the relationship to phosphorylation, with suppressive effects demonstrated for PPARγ sumoylation with regard to adipogenesis. The results of the papers complement each other in helping understand the physiological significance of PPARγ sumoylation.
Mammalian expression plasmids of myc-tagged PPARγ2 (amino acid residues 2–505), PPARα (2–468) and PPARβ/δ (2–440), CFP-tagged SUMO-1, -2 and -3 and CFP-Ubc9 were constructed as follows: cDNAs encoding mouse PPARγ2, PPARα and PPARβ/δ were amplified from the plasmids, pCMX-PPARγ2, pNCMV-PPARα and pCMX-PPARβ/δ (Shimizu et al. 2004), by PCR using KOD-Plus-DNA polymerase (Toyobo) and inserted into pCMV5-myc (Yamaguchi et al. 2002). cDNAs encoding full-length SUMO-1, -2 and -3 and Ubc9 were amplified by PCR from a human foetal liver Matchmaker cDNA library (BD Biosciences Clontech) and inserted into pECFP-C1 (BD Biosciences Clontech). All the mutations employed in this study were created using a QuickChange mutagenesis kit (Stratagene) according to the manufacturer's protocol and confirmed by DNA sequencing.
For the assays of transactivation by full-length PPARγ constructs, a reporter plasmid, pGVP-P/P-PPREx3-luc, containing three copies of PPRE of the mouse PEX11α/perilipin gene pair (Shimizu et al. 2004), in pGVP (Toyo Ink), a luciferase reporter vector driven by the SV40 basal promoter, was mostly used. pGVP-P/P-PPREx1-luc, carrying a single copy of the PEX11α/perilipin PPRE, and pAOXPPREluc, containing a single copy of the rat acyl-CoA oxidase PPRE and the basal promoter (Osada et al. 1997), were also employed.
For the assay of AF-1 activity of PPARγ2, wild-type and mutant sequences coding for various parts of the A/B region of mouse PPARγ2 (residues 2–138, 2–99 or 100–138) were amplified by PCR from appropriate plasmids and inserted in-frame into pCMX-GAL4-N, a vector containing a coding sequence of the yeast GAL4-BD (residues 1–147), driven by the human cytomegalovirus promoter. The resulting plasmids were pCMX-GAL4-PPARγ2[2–138], [2–99] and [100–138], and their derivatives. The activities of GAL4-fusion constructs were assayed as described earlier (Hi et al. 1999), using a reporter plasmid, tk-GALpx3-luc containing three copies of the GAL4-binding site and the herpes simplex virus thymidine kinase (tk) promoter.
Cell culture and DNA transfection
For DNA transfection, HeLa cells were generally used with culture in Ham's F-12 medium containing 10% foetal bovine serum, at 37 °C under 5% CO2. Transfection was carried out by the calcium phosphate method (Osada et al. 1997). For luciferase assays, 2 × 104 cells were seeded in 24-well plates and cultured overnight. Next day, DNA/calcium phosphate precipitates containing 0.1 µg of the expression vector of a PPARγ construct, 0.75 µg of an appropriate luciferase reporter plasmid, 0.1 µg of a β-galactosidase expression plasmid (pCMVβ) as a reference and 0.55 µg of an empty vector (pCMX), were added to each well. After 4 h, the precipitates were removed, and cells were cultured for 24 h in the medium supplemented with 1 µm BRL49 653, a specific PPARγ ligand, or the vehicle, dimethylsulfoxide (DMSO).
For Western blotting, 3.5 × 105 cells were seeded in 6-cm dishes and cultured overnight. Next day, transfection was performed with 12 µg of DNA in total, containing appropriate amounts of the expression plasmids for PPAR and CFP-SUMO (or their mutants), and an empty vector, pCMX. Other expression plasmids were also included as necessary. After transfection, cells were cultured for 24 h in medium supplemented with ligands (100 µm Wy14 643, 0.1 or 1 µm BRL49 653 and 10 µm carbaprostacyclin, for PPARα, γ, and β/δ, respectively) or DMSO.
For immunoprecipitation, 1 × 106 cells were seeded in 10-cm dishes and cultured overnight. Next day, DNA/calcium phosphate precipitates containing 12 µg each of the expression plasmids of myc-PPARγ2 and CFP-SUMO-1 (or their mutants) were added.
Cells were lysed with cell lysis buffer (Osada et al. 1997) and luciferase activities were measured using a PicaGene reagent kit (Toyo Ink) in a Lumat LB9501 luminometer (Berthold) or a Lucy2 microplate luminometer (Anthos). β-Galactosidase activity derived from the pCMVβ plasmid was measured as previously described (Osada et al. 1997) and employed to normalize luciferase activity for transfection efficiency. The assays were carried out in triplicate and the averages are shown, together with SD.
Transfected cells were washed once with PBS containing 20 mm N-ethylmaleimide and immediately dissolved in heated SDS–PAGE sample buffer. For immunoprecipitation, the cells were lysed in a lysis buffer [50 mm Tris-HCl (pH 7.4), 150 mm NaCl, 1% sodium deoxycholate, 0.1% Triton X-100, 1% NP-40, 1 mm phenylmethanesulphonyl fluoride, and 20 mm N-ethylmaleimide], supplemented with a protease inhibitor cocktail (Sigma). The lysates were separated by centrifugation and the supernatants were incubated with a polyclonal anti-PPARγ antibody and rProtein A Sepharose Fast Flow (Amersham Biosciences) at 4 °C. The immunoprecipitates were washed five times with the same buffer and eluted with the SDS–PAGE sample buffer at 100 °C.
Samples were separated on SDS–polyacrylamide gels and then transferred to nitrocellulose membranes (Millipore). Detection was performed using the ECL method according to the manufacturer's instructions (Amersham Biosciences), using Scientific Imaging Film (Kodak). The following antibodies were used: mouse anti-myc monoclonal antibody (9E10) and rabbit polyclonal and mouse monoclonal antibodies to PPARγ (H-100 and E-8) (SantaCruz); rabbit anti-GFP polyclonal antibody (BD Biosciences Clontech); mouse anti-GMP-1 (SUMO-1) monoclonal antibody (21C7; Zymed); guinea-pig polyclonal anti-perilipin antibody (Progen); anti-mouse and anti-rabbit IgG species-specific antibodies linked to horseradish peroxidase (Amersham Biosciences); and anti-guinea-pig IgG species-specific antibody linked to horseradish peroxidase (Chemicon). Rabbit anti-LDH antibody was generously donated by Dr N. Usuda.
Preparation and infection of recombinant lentiviruses
The following constructs were kindly provided by Drs H. Miyoshi and A. Miyawaki: CSII-EF-MCS-IRES2-Venus, a self-inactivating lentiviral construct (Miyoshi et al. 1998); pCAG-HIVgp, a packaging construct expressing the Gag and Pol proteins; and pCMV-VSV-G-RSV-Rev, a construct expressing the vesicular stomatitis virus G glycoprotein (VSV-G) and Rev. This lentiviral system is designed to express a desired gene under the direction of the elongation factor-1 promoter, and Venus, a derivative of YFP (Nagai et al. 2002), as a marker for monitoring the infection efficiency. Recombinant lentiviruses constitutively expressing myc-PPARγ2 and the K107R mutant were produced as follows: cDNAs encoding myc-PPARγ2 (WT and K107R) were inserted into the multicloning site of the lentiviral expression vector. 293FT cells were cultured according to a standard protocol (Invitrogen). To 10-cm dishes, 3.5 × 106 cells were seeded and, next day, 7 µg of the CSII-EF-MCS-IRES2-Venus with a myc-PPARγ2 cDNA insert, 5 µg of the pCAG-HIVgp and 4 µg of the pCMV-VSV-G-RSV-Rev, were co-transfected using Lipofectamine 2000 (Invitrogen), according to the manufacturer's instructions. The culture supernatants containing the recombinant lentiviruses were collected 48 h after transfection, passed through a 0.45-µm filter, and used for infection experiments. NIH3T3 cells seeded in 6-well plates were infected at 30% confluence with each recombinant virus and 6 µg/mL polybrene. We confirmed that virtually all the cells (more than 95%) evenly exhibited Venus protein fluorescence. They were then used in differentiation experiments.
Induction and estimation of adipocyte differentiation
Cells infected with recombinant viruses were cultured to confluence in differentiation medium [Dulbecco's modified Eagle's medium (high glucose) containing 10% foetal bovine serum and 5 µg/mL insulin], then further cultured for 24 h in medium supplemented with 5 nm BRL49 653 or the vehicle, DMSO. Oil Red-O staining and GPDH assays were performed as previously described (Tominaga et al. 2002).
We are grateful to Drs H. Miyoshi and A. Miyawaki for providing the DNA constructs of the lentivirus system, Dr N. Usuda for the anti-LDH antibody, and Dr M. Moore for critical reading of the manuscript. This work was supported in part by a Grant-in-Aid for Scientific Research from the Japan Society for the Promotion of Science, and the 21st Century COE Program.
- 2002) The inhibitory function in human progesterone receptor N termini binds SUMO-1 protein to regulate autoinhibition and transrepression. J. Biol. Chem. 277, 33950–33956. , , & (
- 1997) Transcriptional activation by peroxisome proliferator-activated receptor γ is inhibited by phosphorylation at a consensus mitogen-activated protein kinase site. J. Biol. Chem. 272, 5128–5132. , , , & (
- 1999) An adipogenic cofactor bound by the differentiation domain of PPARγ. EMBO J. 18, 3676–3687. , , , et al. (
- 2003) Sumoylation of the progesterone receptor and of the steroid receptor coactivator SRC-1. J. Biol. Chem. 278, 12335–12343. , , , & (
- 2002) SUMO-1 modification of histone deacetylase 1 (HDAC1) modulates its biological activities. J. Biol. Chem. 277, 23658–23663. , & (
- 1998) SUMO-1 modification of IκBα inhibits NF-κB activation. Mol. Cell 2, 233–239. , & (
- 1999) Peroxisome proliferator-activated receptors: nuclear control of metabolism. Endocr. Rev. 20, 649–688. & (
- 2003) Regulating the regulators: lysine modifications make their mark. Cell 112, 11–17. & (
- 1999) p300 interacts with the N- and C-terminal part of PPARγ2 in a ligand-independent and -dependent manner, respectively. J. Biol. Chem. 274, 7681–7688. , , , et al. (
- 2003) Post-translational modification by the small ubiquitin-related modifier SUMO has big effects on transcription factor activity. Curr. Opin. Genet. Dev. 13, 108–113. (
- 2003) p300 transcriptional repression is mediated by SUMO modification. Mol. Cell 11, 1043–1054. , , , et al. (
- 1997) Preferential interaction of sentrin with a ubiquitin-conjugating enzyme, Ubc9. J. Biol. Chem. 272, 28198–28201. , , , & (
- 2000) Degradation of the peroxisome proliferator-activated receptor γ is linked to ligand-dependent activation. J. Biol. Chem. 275, 18527–18533. , , , et al. (
- 1999) Characterization of the amino-terminal activation domain of peroxisome proliferator-activated receptor α. Importance of α-helical structure in the transactivating function. J. Biol. Chem. 274, 35152–35158. , , & (
- 2003) Phosphorylation of serine 303 is a prerequisite for the stress-inducible SUMO modification of heat shock factor 1. Mol. Cell. Biol. 23, 2953–2968. , , , et al. (
- 2003) Sterol regulatory element-binding proteins are negatively regulated through SUMO-1 modification independent of the ubiquitin/26 S proteasome pathway. J. Biol. Chem. 278, 16809–16819. , , , & (
- 1996) Inhibition of adipogenesis through MAP kinase-mediated phosphorylation of PPARγ. Science 274, 2100–2103. , , & (
- 2000) A common motif within the negative regulatory regions of multiple factors inhibits their transcriptional synergy. Mol. Cell. Biol. 20, 6040–6050. & (
- 1997) Preferential modification of nuclear proteins by a novel ubiquitin-like molecule. J. Biol. Chem. 272, 14001–14004. , & (
- 2002) The SUMO E3 ligase RanBP2 promotes modification of the HDAC4 deacetylase. EMBO J. 21, 2682–2691. , , , et al. (
- 2002) Potentiation of glucocorticoid receptor transcriptional activity by sumoylation. Endocrinology 143, 3482–3489. , , & (
- 1995) The nuclear receptor superfamily: the second decade. Cell 83, 835–839. , , , et al. (
- 1998) Development of a self-inactivating lentivirus vector. J. Virol. 72, 8150–8157. , , , & (
- 2002) A variant of yellow fluorescent protein with fast and efficient maturation for cell-biological applications. Nat. Biotechnol. 20, 87–90. , , , et al. (
- 2004) Transcriptional activity of peroxisome proliferator-activated receptor γ is modulated by SUMO-1 modification. J. Biol. Chem. 279, 29551–29557. , & (
- 1997) Identification of an extended half-site motif required for the function of peroxisome proliferator-activated receptor α. Genes Cells 2, 315–327. , , , & (
- 2001) Roles of the heat shock transcription factors in regulation of the heat shock response and beyond. FASEB J. 15, 1118–1131. , & (
- 2000) Covalent modification of the androgen receptor by small ubiquitin-like modifier 1 (SUMO-1). Proc. Natl. Acad. Sci. USA 97, 14145–14150. , , & (
- 1998) Obesity associated with a mutation in a genetic regulator of adipocyte differentiation. N. Engl. J. Med. 339, 953–959. , , , & (
- 2001) SUMO-1 conjugation in vivo requires both a consensus modification motif and nuclear targeting. J. Biol. Chem. 276, 12654–12659. , & (
- 2002) SUMO-1 modification represses Sp3 transcriptional activation and modulates its subnuclear localization. Mol. Cell 10, 831–842. , , & (
- 2003) Nuclear and unclear functions of SUMO. Nat. Rev. Mol. Cell Biol. 4, 690–699. & (
- 1998) Interdomain communication regulating ligand binding by PPAR-γ. Nature 396, 377–380. , , , et al. (
- 2004) Tissue-selective, bidirectional regulation of PEX11α and perilipin genes through a common peroxisome proliferator response element. Mol. Cell. Biol. 24, 1313–1323. , , , & (
- 2003) Protein inhibitor of activated signal transducer and activator of transcription 1 interacts with the N-terminal domain of mineralocorticoid receptor and represses its transcriptional activity: implication of small ubiquitin-related modifier 1 modification. Mol. Endocrinol. 17, 2529–2542. , , , et al. (
- 2002) Small ubiquitin-related modifier-1 (SUMO-1) modification of the glucocorticoid receptor. Biochem. J. 367, 907–911. , , & (
- 2002) Growth hormone has dual stage-specific effects on the differentiation of 3T3-L1 preadipocytes. J. Biochem. 132, 881–889. , & (
- 1994) Stimulation of adipogenesis in fibroblasts by PPARγ2, a lipid-activated transcription factor. Cell 79, 1147–1156. , & (
- 2003) Modification with SUMO. A role in transcriptional regulation. EMBO Report 4, 137–142. , & (
- 1997) Ligand-independent activation domain in the N terminus of peroxisome proliferator-activated receptor γ (PPARγ). Differential activity of PPARγ1 and -2 isoforms and influence of insulin. J. Biol. Chem. 272, 20230–20235. , , , et al. (
- 2002) Sly1 binds to Golgi and ER syntaxins via a conserved N-terminal peptide motif. Dev. Cell 2, 295–305. , , , et al. (