PIASγ Represses the Transcriptional Activation Induced by the Nuclear Receptor Nurr1*

Nurr1 is a transcription factor essential for the development of ventral dopaminergic neurons. In search for regulatory mechanisms of Nurr1 function, we identified the SUMO (small ubiquitin-like modifier)-E3 ubiquitin-protein isopeptide ligase, PIASγ, as an interaction partner of Nurr1. Overexpressed PIASγ and Nurr1 co-localize in the nuclei of transfected cells, and their interaction is demonstrated through co-immunoprecipitation and glutathione S-transferase pulldown assays. Co-expression of PIASγ with Nurr1 results in a potent repression of Nurr1-dependent transcriptional activation of an artificial NGFI-B response element (NBRE) reporter as well as of a reporter driven by the native tyrosine hydroxylase promoter. We identified two consensus sumoylation sites in Nurr1. The substitution of lysine 91 by arginine in one SUMO site enhanced the transcriptional activity of Nurr1, whereas the substitution of lysine 577 by arginine in the second SUMO site decreased transcriptional activity of Nurr1. Interestingly, PIASγ-induced repression of Nurr1 activity does not require the two sumoylation sites, because each mutant is repressed as efficiently as the wild type Nurr1. In addition, the mutations do not alter Nurr1 nuclear localization. Finally, we provide evidence that Nurr1 and PIASγ co-exist in several nuclei of the rodent central nervous system by demonstrating the co-expression of Nurr1 protein and PIASγ mRNA in the same cells. In conclusion, our studies identified PIASγ as a transcriptional co-regulator of Nurr1 and suggest that this interaction may have a physiological role in regulating the expression of Nurr1 target genes.

Nuclear receptors (NRs) 1 are a family of transcription factors that regulate the expression of genes essential for development, reproduction, and homeostasis. NRs can be divided into two major groups: classical steroid/thyroid receptors, which activate transcription in a ligand-dependent manner, and orphan receptors, for which the corresponding ligands are either unknown or activate transcription in a ligand-independent manner. Nurr1 (NR4A2, RNR-1, TINUR, HZF-3) is an orphan NR with a close structural relationship to the orphan receptors Nur77 (NR4A1, NGFI-B, TR3, NAK-1) and NOR-1 (NR4A1, MINOR, TEC). Nurr1, Nur77, and NOR-1 form the Nur subfamily and, unlike most NRs, are products of immediate early genes (1) and function as constitutively active transcription factors (2). Accordingly, recent crystal structure study has shown that the Nurr1 ligand-binding domain (LBD) lacks a ligand-binding cavity (3).
Nurr1, as well as the other Nur family members, binds as a monomer to the DNA recognition element, NGFI-B response element (NBRE), which consists of the classical estrogen receptor half-site preceded by two adenines (AAAGGTCA) (4). In addition, Nur factors can bind as either homo-or heterodimers formed among subfamily members to NurRE (everted repeats of a related NBRE) elements (5). Furthermore, Nurr1 and Nur77 (but not NOR-1) can heterodimerize with RXR and bind DR-5 elements (6 -8). Nurr1 shares with other NRs the same structural/functional motifs, consisting of a DNA-binding domain (DBD), an LBD, and two transactivating domains. One transactivating domain (AF1) is located in the amino terminus and is structurally divergent among NRs, whereas the other transactivating domain (AF2) is evolutionary conserved among NRs and contains the LBD. Nurr1 acting as a monomer through the NBRE site promotes constitutive transcriptional activation that is dependent on the two transactivating functions, AF1 and AF2 (2,9).
Nurr1 is essential for the genesis and differentiation of midbrain dopaminergic neurons in the central nervous system (10 -12). Thus, mice lacking Nurr1 are devoid of dopaminergic neurons in the midbrain. Based on this evidence, several investigators have suggested that the genes encoding tyrosine hydroxylase (TH), the limiting enzyme in dopamine synthesis, and the dopamine transporter (DAT), both of which are expressed in differentiated dopaminergic neurons, are Nurr1 target genes. Consistently, TH and DAT genes harbor NBRE elements in their promoters (13)(14)(15), and several reports show that Nurr1 regulates the expression of both genes in transcriptional assays performed in cell lines as well as in primary cultures (14,16,17).
The transcriptional activity of NRs relies not only on their ability to enter the nucleus and bind the DNA but also on their interaction with other transcription factors as well as with a number of co-regulator protein complexes (reviewed in Ref. 18). Co-factors that regulate Nur family member-dependent transactivation have recently begun to be elucidated. For instance, it has been shown that Nur77 transcriptional activity is regu-lated by ASC-2 and the co-repressor SMRT/NcoR (19) and by the steroid receptor co-activator 2 (SRC-2) (20). Less is known about how Nurr1 transcriptional activity is regulated. Recently, it was reported that the transcriptional activity induced by both Nurr1 and Nur77 homodimers on a Nur response element is enhanced by the recruitment of SRCs to the AF1 domain (21). Direct interaction between Nurr1 and co-regulators through the AF2 has been questioned by the crystal structure study showing that Nurr1 does not have a classical binding site for co-activators (3).
To gain more insight into the molecular mechanism that governs Nurr1 constitutive transcriptional activating function and to identify regulatory mechanism, we performed a yeast two-hybrid screening using part of the Nurr1 LBD as the bait. We found that Nurr1 interacts with PIAS␥, a member of the family of protein inhibitors of activated STAT (PIAS). In vertebrates, five PIAS proteins (PIAS1/GBP, PIAS3, ARIP3/PI-ASx␣, Miz/PIASx␤, and PIAS␥) have been identified. Biochemical studies indicate that PIAS proteins interact directly with several transcription factors, including STATs (22,23), p53 (24) and steroid receptors (25,26), and can regulate their transcriptional activities both positively and negatively (26). Recent evidence indicates that the mechanism by which PIAS proteins regulate the activity of transcription factors is by functioning as SUMO (small ubiquitin-like modifier)-E3 ligases (27)(28)(29). In this report, we show the in vivo and in vitro interaction between Nurr1 and PIAS␥. Co-expression of PIAS␥ with Nurr1 results in a strong inhibition of Nurr1-dependent transcriptional activation of an artificial NBRE-driven reporter as well as of the rat TH-driven reporter. We identified two consensus sumoylation (⌽KXE (30)) domains in Nurr1. The point mutation of Lys 91 in Nurr1 induces a significant increase in Nurr1dependent transcriptional activation, whereas the point mutation of Lys 577 in Nurr1 decreases transcriptional activation mediated by Nurr1. Finally, we give evidence showing the co-existence of these proteins in neurons in the adult central nervous system of rodents. Taken together, these results show that PIAS␥ is an interacting partner of Nurr1 and suggest that the sumoylation of Nurr1 may act as a regulatory mechanism in the control of Nurr1-mediated transcription over its target genes.

EXPERIMENTAL PROCEDURES
Two-hybrid Screening-LexA-Nurr1-(363-488) was transformed into L-40 together with an 11-day embryonic mouse cDNA library fused to a Gal4 activation domain (Clontech). Library screening was carried out by using HIS3 and LacZ reporters as described previously (31). Clones interacting with Nurr1 were identified by growth on selective medium in the presence of 5 mM 3-aminotriazole and confirmed by assaying for ␤-galactosidase activity. Specificity of the interaction was examined by a mating assay between the positive L-40 transformants and an AMR-70 strain expressing either LexA-Nurr1-(363-488) or the unrelated LexA-laminin fusion protein. Liquid ␤-galactosidase assays were performed as described (32). Positively interacting clones were characterized by sequencing analysis.
Co-immunoprecipitation and Western Blotting-Total extracts from transfected HEK293 cells were obtained by lysing cells in lysis buffer Dotted lines indicate the regions of Nurr1 that were deleted. The resultant LexA-Nurr1 construct was co-transformed into yeast L-40 strain together with an embryonic mouse cDNA library. B, schematic representation of fulllength PIAS␥ and the clones isolated by the two-hybrid screening. The black box represents the variant tail obtained from the isolated clones. C, yeast two-hybrid assay of Nurr1 interaction with PIAS␥ encoding clones. The yeast strain L-40 was co-transformed with LexA-Nurr1-(363-488) (bait), and the indicated clones isolated from the screening. Liquid ␤-galactosidase activity assays were performed on the co-transformants. Control experiments were performed using pGAD empty vector and the vector encoding the pGAD-HSP70 fusion protein.
(0.1 M Tris-HCl, pH 7.4, 150 mM NaCl, 1 mM EDTA, 2 mM MgCl 2 , 1% Triton X-100, 10% glycerol, 2 mg/ml phenylmethylsulfonyl fluoride, and 1 mM Na 3 VO 4 ). 1-mg protein aliquots of the extracts were pre-cleared by rotation at 4°C for 5 h with 20 l of protein G-agarose beads (KPL) plus 4 g of preimmune rabbit IgG. The extracts were then recovered by centrifugation and incubated overnight at 4°C with 4 g of the appropriate polyclonal antibodies. Samples were then added to 20 l of fresh protein G-agarose beads and immunoprecipitated by rotation for 3 h at 4°C. Finally, after extensive washing, bound proteins were eluted by boiling in loading sample buffer with 5% ␤-mercaptoethanol. Immunoprecipitated proteins were separated by SDS-PAGE, and Western blotting was performed using a monoclonal anti-HA antibody (Santa Cruz Biotechnology).
GST Pulldown Assays-COS-1 cells were transfected with HA-Nurr1 using FuGENE 6 (Roche Applied Science). 48 h after transfection, whole extracts were prepared as described in the previous paragraph. GST or GST-PIAS␥-(1-158) proteins were expressed in Escherichia coli BL21 following induction with 0.1 mM isopropyl-1-thio-␤-D-galactopyranoside for 1 h at 28°C. Cells were sonicated, and proteins were clarified following the manufacturer's instruction (Amersham Biosciences). 200-l aliquots of the bacterial lysates were then added to 80 l of glutathione-agarose beads. For binding assays, GST-PIAS␥ fusion protein or GST alone (1.5 g) bound to glutathione-agarose beads was incubated with 50 l of the COS-1 cell whole extracts (65 g) for 2 h at 4°C. After extensive washing, the retained proteins were eluted with sample buffer and resolved on 12% SDS-PAGE. Immunoblots were incubated with polyclonal anti-Nurr1 (N-20, Santa Cruz Biotechnology), and the reactive bands were visualized using anti-rabbit immunoglobulin conjugated with horseradish peroxidase (Santa Cruz Biotechnology) followed by ECL chemiluminescence.
Immunofluorescence-Immunofluorescence assays were performed as described (33). In brief, HEK293 or COS-1 cells were plated on poly-L-lysine-coated glass coverslips and transfected with HA and Myc epitope-tagged expression constructs using LipofectAMINE 2000 reagent (Invitrogen). 48 h after transfection, cells were fixed in 3% paraformaldehyde-PBS for 10 min, permeabilized for 10 min in 0.1% Triton X-100-PBS (PBST), and incubated for 1 h in 5% nonfat powdered milk in PBST. Coverslips were then incubated overnight in a 1:500 dilution of polyclonal anti-Myc, monoclonal anti-HA or polyclonal anti-Nurr1 antibodies. After washing, cells were incubated in Alexa 546conjugated goat anti-rabbit and Alexa 488-conjugated goat anti-mouse antibodies (1:1000, Molecular Probes). Cells were visualized by indirect immunofluorescence or confocal microscopy.
Mammalian Reporter Gene Assays-HEK293 and PC12 cells were transfected using LipofectAMINE 2000 reagent (Invitrogen) according to the manufacturer's instructions. HEK293 cells were transfected with 0.125 g of the NBRE-3X-tk-Luciferase reporter gene, whereas PC12 cells were transfected with a chloramphenicol acethyltransferase (CAT) reporter gene driven by a 1.0-kb TH promoter. The reporters were transfected together with 0.5 g of recombinant HA epitope-tagged Nurr1, Nurr1Nmut, or Nurr1Cmut or an equimolar amount of the empty pCGN vector plus either 0.5-2 g of HA-hPIAS␥ (or pcDNA-PIAS␥) or an equimolar amount of the empty vector (pCGN or pcDNA). A reporter gene expressing the ␤-galactosidase cDNA driven by cytomegalovirus promoter was cotransfected (20 ng) in all experi-ments as an internal control for transfection efficiency. The total amount of transfected DNA was kept constant by adding pBluescript SR (Stratagene). Cells were harvested 48 h after transfection. Luciferase (34) and CAT (35) assays were performed as described previously, and the results were normalized by protein content. Statistical analyses were performed using the nonparametric Mann-Whitney test.
In Situ Hybridization (ISH) and Immunohistochemistry-Adult male mice were deeply anesthetized with chloral hydrate and then killed by perfusion through the left cardiac ventricle with 0.9% NaCl followed by 4% paraformaldehyde in 0.1 M phosphate buffer, pH 9.5. Brains were removed from the skull, cut in blocks, post-fixed in 4% paraformaldehyde for 1 h, and finally maintained in 30% sucrose in phosphate buffer containing 0.9% NaCl (PBS) for 48 h at 4°C. Cryostat serial brain coronal sections (30 m) were collected and washed three times in PBS to be processed for ISH and immunohistochemistry. An oligonucleotide probe of 40 nucleotides (5Ј-gctctgcactcttcttggcatagcgagtctcatacagctc-3Ј) complementary to nucleotides 389 -428 of the mouse PIAS␥ cDNA was used for ISH studies of PIAS␥ mRNA expression in adult rodent brain. The probe was 3Ј end-labeled with digoxygenin-dUTP by using terminal transferase. ISH was carried out essentially as described previously (36). Following ISH, tissue sections were rinsed with SSC (2ϫ and 1ϫ) for 10 min each at 55°C. The immunohistochemical detection of Nurr1like epitopes was carried out as follows. After ISH was completed, tissue sections were blocked (3% goat normal serum, 1% bovine serum albumin, and 0.3% Triton X-100 in PBS) and then incubated with anti-Nurr1 antibody (1:1000) overnight at room temperature. Sections were rinsed twice for 10 min in PBS and mounted on gelatin-coated slides. Then the slices were incubated with Alexa 488-conjugated goat antirabbit (1:200, Molecular Probes) for 1 h in the dark to detect Nurr1 epitopes, together with the anti-digoxygenin antibody conjugated with alkaline phosphatase (Roche Applied Science) to detect the specific hybrids. Finally, ISH reactions were further revealed by incubating with Fast Red solution (one Fast Red tablet (Roche Applied Science) was dissolved in 2 ml of 0.2 M Tris-HCl buffer, pH 8.5, containing 10 mM MgCl 2 ) for 30 min at room temperature. After additional rinsing, the slides were mounted in 90% glycerol diluted in Tris-HCl buffer, pH 9.0, cover-slipped, and examined by confocal microscope. A series of control experiments was performed using substrates for light microscopy, essentially as described previously by our group (36,37).

RESULTS
PIAS␥ Interacts with Nurr1-To identify proteins that might regulate Nurr1 transactivation activity, we performed a yeast two-hybrid screening. The bait was selected after control experiments performed to determine the transcriptional activity of several segments of Nurr1 in yeast (data not shown). Because the full-length LBD (amino acids 363-598) of Nurr1 resulted transcriptionally active, a shorter fragment, LexA-Nurr1-(363-488), which is transcriptionally inactive in yeast (fig 1A), was used as the bait in the genetic screening, together with a mouse embryonic Gal4 activation domain cDNA library. A screening of 7.3 ϫ 10 6 clones yielded 20 positive clones, from which 5 were sequenced. Four of the 5 clones sequenced repre- sented the same cDNA, encoding a 162-amino acid open reading frame. A search of the GenBank TM data base revealed that the isolated cDNAs were identical to the cDNA encoding the mouse PIAS␥ protein from amino acids 2-144, plus a divergent 19-amino acid tail (Fig. 1B). A mouse expressed sequence tag sequence (GenBank TM accession number BB615867) was identical over all the open reading frame, suggesting that the clones found in the two-hybrid assay corresponded to a spliced variant of PIAS␥. The specificity of the interaction between Nurr1 and PIAS␥ was examined, as shown in Fig. 1C, by re-transforming the 4 clones encoding PIAS␥ found in the screening along with the bait in yeast, which gave strong positive interaction. The bait did not interact with pGAD-HSP70 fusion protein in the two-hybrid assay (Fig. 1C). A mating assay in the AMR70 yeast line gave similar results; the bait interacted strongly with clone 32, whereas it did not show interaction with LexA-laminin (data not shown).
Nurr1 and PIAS␥ Interact in Vitro and in Vivo-The interaction between Nurr1 and PIAS␥ observed in the yeast was confirmed in mammalian cells using two approaches. In the first approach, the interaction between Nurr1 and PIAS␥ was demonstrated by using a GST pulldown assay. Whole extracts from COS-1 cells transfected with HA-Nurr1 were incubated with GST-PIAS␥ (amino acids 1-158) or GST alone. Nurr1 protein associated specifically with GST-PIAS␥ as shown in Western blots performed with eluted proteins from the columns ( Fig. 2A). In the second approach, co-immunoprecipitation experiments were performed using extracts from HEK293 cells overexpressing HA-hPIAS␥ and Myc-C-Nurr1. HA epitopes were present in immunoprecipitates using the anti-Myc polyclonal antibody but not the preimmune IgG (Fig. 2B), indicating the specific interaction between Nurr1 and PIAS␥ in cells. Finally, immunofluorescent detection of HA and Nurr1 epitopes in COS-1 cells transfected with Myc-Nurr1 (1-583) and HA-hPIAS␥ showed that these two proteins co-localize in the nucleus (Fig. 2C). Taken together, these results are consistent with the experiments performed in yeast demonstrating the interaction between Nurr1 and PIAS␥.
PIAS␥ Represses Nurr1-dependent Transactivation-The finding that PIAS␥ interacts with Nurr1 suggest that PIAS␥ could function as a co-regulator of Nurr1. To test this hypothesis, functional experiments were performed in mammalian cells. As shown in Fig. 3A, co-expression of HA-hPIAS␥ inhibited significantly the transactivation of the NBRE-3X-tk-luciferase reporter induced by Nurr1 in HEK293 cells. The inhibitory effect induced by PIAS␥ was concentration-dependent (Fig. 3, A and B), and it was not due to a decrease of Nurr1 expression (Fig. 3B). The presence of PIAS␥ did not change the basal activity significantly.
Several investigators have suggested that TH, the limiting enzyme in dopamine biosynthesis, is one of the target genes of Nurr1. To analyze the effect of PIAS␥ on a physiological target of Nurr1, a 1.0-kb TH promoter driving a CAT reporter was transfected in PC12 cells together with Nurr1 with or without PIAS␥. As shown in Fig. 3C, Nurr1 induces the transcription of the TH-driven CAT reporter. This effect was significantly inhibited by PIAS␥. Taken together, these results demonstrate that PIAS␥ represses the constitutive transactivation activity induced by Nurr1 and suggest that PIAS␥ may contribute to the regulation of Nurr1-mediated TH transcription.  scriptional activity and/or cellular localization, we generated two mutant forms of Nurr1 (as shown in Fig. 4A), one in which the lysine 91 was replaced with an arginine (K91R) and another in which lysine 577 was replaced with an arginine (K577R). Using these constructs in co-transfection experiments, we observed that the Nurr1 mutants behave in opposite ways in transcriptional assays. Thus, although the K91R mutant increases the transcription activity significantly as compared with wild type Nurr1, K577R mutant loses about 50% of the Nurr1 transcriptional activity (Fig. 4B).

Mutation of Consensus SUMO Domains in Nurr1 Changes
The ability of K91R mutant to increase transcriptional activation as compared with wild type Nurr1 suggests that sumoylation at Lys 91 might normally repress Nurr1 activity. To test whether PIAS␥ repression over Nurr1 activity is by sumoylation, we examined the contribution of the Nurr1 SUMO conjugation sites for PIAS␥-mediated repression. Our results show that PIAS␥ repressed the transcriptional activation of mutants K91R and K577R almost as efficiently as that of wild type Nurr1 (Fig. 4B) without changing their expression level, as shown by Western blot (Fig. 4C). Thus, the repression of Nurr1 activity by PIAS␥ does not require the two Nurr1consensus sumoylation sites. Because sumoylation of proteins is also known to affect cellular localization, we transfected HEK293 cells with HA epitope-tagged expression vectors for either wild type Nurr1 or the mutants. As shown in Fig. 4D, the staining pattern of the wild type Nurr1 (panel a) mirrored that of the lysine to arginine mutants (panels b and c). Hoechst staining demonstrates that the localization of Nurr1 and K91R and K577R mutants is nuclear (Fig. 4D, d-f). Thus, mutation of lysines in the SUMO consensus domain of Nurr1 does not affect its nuclear localization.
PIAS␥ and Nurr1 Are Co-expressed in the Same Cells in Different Brain Regions-One crucial requisite to conceiving a physiological role for Nurr1 and PIAS␥ interaction is a temporal as well as physical co-localization of both proteins in the same tissue compartments. To address this issue we analyzed the localization of Nurr1 and PIAS␥ in adult rodent brain. Nurr1-like immunostaining co-localized with a positive ISH signal for PIAS␥ mRNA in several cells in substantia nigra compacta (Fig. 5C) and substantia nigra reticulata (Fig. 5B). This co-localization was also observed in different brain nuclei (see Table I). Nurr1 immunostaining was completely blocked with the co-incubation of antigen peptide (data not shown and Ref. 37) and positive ISH for PIAS␥ mRNA was abrogated with 100 times excess of cold probe (data not shown). This result strongly suggests that Nurr1 and PIAS␥ co-localize in the same cells in the adult rodent brain. Fortyeight hours post-transfection, cells were harvested and the extracts assayed for luciferase activity. The nonparametric MannWhitney test was applied to determine statistical significance of the differences. *, p Ͻ 0.05 (wild type Nurr1 versus mutants forms); ϩ, p Ͻ 0.05 (PIAS␥ ϩ Nurr1 forms versus their respective controls). C, Western blot analysis of whole extracts from HEK293 cells transfected with the constructs indicated in B. PIAS␥ was assayed with an anti-HA antibody and Nurr1 was assayed with an anti-Nurr1 antibody. The amounts of expressed proteins show that inhibition of the transcriptional activation induced by the presence of PIAS␥ is not due to instability of Nurr1 protein. D, HEK293 cells were transfected with HA epitope-tagged versions of Nurr1, Nurr1Nmut, or Nurr1Cmut. After fixation and labeling with anti-HA antibody, cells were examined by indirect immunofluorescence microscopy. Red stains repre- sents HA epitopes (a, b, c). Lower panels (d, e, f) show the nuclei stained with Hoechst.

DISCUSSION
Classical NRs control the expression of target genes by recruiting co-activator or co-repressor complexes in response to ligand presence or absence, respectively. Nurr1 and the other members of the Nur subfamily are orphan NRs that function as constitutive transcriptional activators. The mechanism that triggers the induction of Nur target gene expression seems to be the Nur factor expression increment itself. Indeed, Nur subfamily members are immediate early genes, in which expression is induced by different kinds of external stimuli. Unlike classical NRs, for which the ligand disappearance determines the end of the action and warrant a rapid and transient effect over the target genes, the mechanisms that mediate and ensure an appropriate control over Nurr1 target gene expression is unknown. In this study, for the first time we provide evidence showing that PIAS␥ interacts with and represses Nurr1-dependent transcriptional activation. Several lines of evidence indicate that PIAS␥ is a Nurr1 transcriptional coregulator. First, PIAS␥ interacts with Nurr1 in vitro and in vivo in yeast as well as in mammalian cells. Both proteins are found in the nuclei of transfected cells and co-exist in the same cells in several nuclei of the rodent central nervous system. Second, PIAS␥ inhibits, in a dose-dependent manner, Nurr1 transactivation activity. This effect was observed both with an NBRE artificially driven reporter as well as with a reporter driven by the natural TH promoter.
The mechanism underlying the ability of PIAS␥ to repress Nurr1-dependent transcription is not known. One potential mechanism is that PIAS␥ elicits its action through Nurr1 sumoylation. Thus, we identified two sumoylation consensus sites in Nurr1 that are also present in the other Nur family members. However, the replacement of critical lysines by arginines at positions 91 and 577 did not inhibit the repressive effect of PIAS␥ over Nurr1-dependent transcription, suggesting that the SUMO consensus sites are not necessary for PIAS␥ repression over Nurr1 activity. These results are consistent with several reports showing that PIAS proteins can regulate the activity of other transcription factors in a SUMO-independent fashion. For instance, it has been reported that PIAS␥ inhibits LEF1 independently of its two consensus SUMO sites (29). Schmidt and Muller (24) reported similar results for p53, demonstrating that PIAS/SUMO is able to repress p53 as well as SUMO mutant forms of p53. A possible explanation for our results could be that Nurr1 might be sumoylated by PIAS␥ at additional non-consensus SUMO conjugation sites. Alternatively, SUMO modification of another critical target in the PIAS␥ pathway might be required to inhibit transcription independently of SUMO modification of Nurr1. Consistent with this hypothesis, it was recently demonstrated that SUMO-1 overexpression markedly enhances progesterone receptor-mediated gene activation. However, despite the fact that the progesterone receptor was sumoylated, the sumoylation of the SRC-1 co-activator increased progesterone receptor-SRC-1 interaction and prolonged SRC-1 retention in the nucleus (39). Another attractive explanation comes from a recent study in which it was demonstrated that PIAS␥ interacts with histone deacetylase 1 and represses transforming growth factor-␤/ Smad-mediated transcription in an histone deacetylasedependent manner (40). It is known that PIAS proteins positively or negatively regulate the activity of several transcription factors by modifying either the DNA binding properties, the subcellular and subnuclear localization, or the stability of the transcription factors with which they interact (reviewed in Ref. 41). For instance, PIAS1 inhibits STAT1 transactivation by blocking its DNA binding activity (22), whereas PIAS␥ represses STAT1-and androgen receptor-dependent transcriptional activation without affecting their DNA binding activity (42). Some PIAS proteins may sequester transcription factors and target them to the nuclear matrix, as demonstrated for the PIAS␥-induced LEF1 (lymphoid-enhancing factor 1) sequestration into nuclear bodies (29). Our immunofluorescence experiments show that Nurr1 as well as the K91R and K577R mutants stay in the cell nuclei. However, we cannot exclude an intranuclear redistribution, as in the case of LEF1 that is targeted to the nuclear matrix by PIAS␥ (29). Sumoylation of Nurr1 may also alter the association of this NR with other transcriptional co-regulators, its DNA binding ability, and/or its stability. Further studies will be necessary to address these issues. FIG. 5. PIAS␥ and Nurr1 co-localize in the susbstantia nigra of the mouse central nervous system. Nurr1 and PIAS␥ expression detected in mouse brain tissue using immunohistochemistry and ISH, respectively. Nurr1 protein (green stain) and PIAS␥ mRNA (red stain) co-localize in cells of the substantia nigra (SN) (A). Double labeled cells in higher magnification from substantia nigra reticulata (SNr) (B) and substantia nigra compacta (SNc) (C). The analysis of the transcriptional activity of the mutants Nurr1 K91R and K577R is quite interesting. On one hand, the K91R mutant displays a significant augmentation in transcriptional activity. This effect is similar to what has been reported for other transcription factors such as C/EBP␣ (43), androgen receptor (44) and p53 (24), for which the replacement of lysine by arginine in the SUMO sites increases the transcription activity. Interestingly, it has been reported that the SUMO site in C/EBP␣ is part of a synergy control motif and sumoylation of the critical lysine inhibits transcriptional synergy (45). Thus, it is tempting to suggest that the sumoylation of lysine 91 in Nurr1 may be a control mechanism of Nurr1 synergy. On the other hand, mutant K577R displays a decreased transcriptional activity compared with wild type Nurr1. As proposed recently, lysine 577 of helix 11 of the LBD forms a salt bridge with aspartate 589 of helix 12, playing an important role in stabilizing helix 12 in a transcriptionally active conformation of Nurr1 in the absence of ligand (3). However, our results suggest that sumoylation of this lysine could be required for Nurr1 transcriptional activity. Currently, studies are being done on whether the mechanism by which PIAS␥ regulates Nurr1-dependent transcription is through Nurr1 sumoylation.
SUMO-E3 ligases as well as SUMO modification of proteins have recently been implicated in regulatory pathways in which dysfunction could produce Parkinson's disease (38,46). Nurr1 is an essential factor for the genesis of midbrain dopaminergic neurons (10 -12), the neurons that are lost in this devastating illness. Whether changes in the repressive effect induced by PIAS␥ over Nurr1-dependent transcriptional activation could be related to Parkinson's disease remains an open question. In summary, our results indicate that PIAS␥ is a co-regulator of Nurr1 that could play an important role in ending the transcriptional activation mediated by this early gene.