Ubc9 and Protein Inhibitor of Activated STAT 1 Activate Chicken Ovalbumin Upstream Promoter-Transcription Factor I-mediated Human CYP11B2 Gene Transcription*

Aldosterone synthase (CYP11B2) is involved in the final steps of aldosterone biosynthesis and expressed exclusively in the adrenal zona glomerulosa cells. Using an electrophoretic mobility shift assay, we demonstrate that COUP-TFI binds to the –129/–114 element (Ad5) of human CYP11B2 promoter. Transient transfection in H295R adrenal cells demonstrated that COUP-TFI enhanced CYP11B2 reporter activity. However, the reporter construct with mutated Ad5 sequences showed reduced basal and COUP-TFI-enhanced activity, suggesting that binding of COUP-TFI to Ad5 is important for CYP11B2 transactivation. To elucidate molecular mechanisms of COUP-TFI-mediated activity, we subsequently screened for COUP-TFI-interacting proteins from a human adrenal cDNA library using a yeast two-hybrid system and identified Ubc9 and PIAS1, which have small ubiquitin-related modifier-1 (SUMO-1) conjugase and ligase activities, respectively. The coimmunoprecipitation assays confirmed that COUP-TFI forms a complex with Ubc9 and PIAS1 in mammalian cells. Immunohistochemistry showed that Ubc9 and PIAS1 are markedly expressed in rat adrenal glomerulosa cells. Coexpression of Ubc9 and PIAS1 synergistically enhanced the COUP-TFI-mediated CYP11B2 reporter activity, indicating that both proteins function as coactivators of COUP-TFI. However, sumoylation-defective mutants, Ubc9 (C93S) and PIAS1 (C351S), continued to function as coactivators of COUP-TFI, indicating that sumoylation activity are separable from coactivator ability. In addition, chromatin immunoprecipitation assays demonstrated that ectopically expressed COUP-TFI, Ubc9, and PIAS1 were recruited to an endogenous CYP11B2 promoter. Moreover, reduction of Ubc9 or PIAS1 protein levels by small interfering RNA inhibited the CYP11B2 transactivation by COUP-TFI. Our data support a physiological role of Ubc9 and PIAS1 as transcriptional coactivators in COUP-TFI-mediated CYP11B2 transcription.

Aldosterone is exclusively produced in adrenal zona glomerulosa cells due to its unique expression of aldosterone synthase cytochrome P450 (CYP11B2), the enzyme required for the final steps of aldosterone biosynthesis. In aldosterone-producing adrenal cortical adenomas of patients with primary aldosteronism, overexpression of CYP11B2 is demonstrated at the transcriptional level (1,2). Although the reason for aberrant expression of CYP11B2 in these adenomas is not known, mutations in the CYP11B2 gene do not appear to be the cause (3,4). We therefore postulated that transcription factors and/or coregulators may play important roles in CYP11B2 overexpression in the tumors.
The trans-acting factors that regulate CYP11B2 expression remain poorly defined. The orphan nuclear receptor, steroidogenic factor-1 (SF-1), 1 is shown to play a crucial regulator of most steroid hydroxylase genes, including CYP17 and CYP11B1 (5,6). However, SF-1 actually represses rather than activates expression of hCYP11B2 (7)(8)(9). In addition, other transcription factors that are expressed in the adrenal cortex include the NGFI-B family of orphan nuclear receptors, such as Nurr1, NGFI-B, and NOR-1. The NGFI-B family receptors are highly expressed in the adrenal zona glomerulosa cells as well as in aldosterone-producing adenomas (8,10,11). These three nuclear receptors are rapidly induced early response genes that enhance transcription by binding to a consensus sequence, named NBRE-1, as well as an Ad5 element of the hCYP11B2 promoter. In addition, CREB and ATF-1 enhance transcription of the hCYP11B2 gene by binding to a CRE (9,12,13). Our previous data (12) showed that human adrenocortical H295R nuclear proteins containing chicken ovalbumin upstream promoter-transcription factors (COUP-TFs) were bound to the Ϫ129/Ϫ114 sequence, designated as the Ad5 element of the hCYP11B2 promoter by electrophoretic mobility shift assays. The COUP-TFI was originally identified as an activator of the chicken ovalbumin gene (14,15); however, COUP-TFs mostly function as transcriptional repressor of many target genes.
COUP-TFs inhibit the transcription of other nuclear receptor such as retinoic acid receptor and thyroid hormone receptor (14). Furthermore, COUP-TFI represses basal transcriptional activity by active repression utilizing transcriptional corepressors, such as N-CoR and SMRT (16). We and other investigators have previously demonstrated that COUP-TFI and SF-1 regulate the bovine CYP17 expression in a mutually exclusive manner (17,18). We have previously reported that COUP-TFI is expressed in the normal adrenal cortex and that expression levels of COUP-TFI is inversely correlated with those of CYP17, but correlated with those of N-CoR in adrenal cortical adenomas (18 -23).
We, therefore, have screened COUP-TFI-interacting proteins from a human adrenocortical adenoma cDNA library using a yeast two-hybrid system and identified Ubc9 (24) and PIAS1, which are small ubiquitin-related modifier-1 (SUMO-1)-conjugating enzyme and SUMO-1 ligase, respectively. The SUMO post-translationally modifies many proteins with roles in diverse processes, including regulation of transcription, chromatin structure, and DNA repair (25)(26)(27)(28)(29)(30). The SUMO modification has not been generally associated with increased protein degradation. Rather, similar to non-proteolytic roles of ubiquitin, SUMO modification regulates protein localization and activity. The SUMO E1-activating, E2-conjugating enzymes, and E3-ligase are involved in the sumoylation machinery. In contrast to the ubiquitin system where dozens of E2 enzymes have been identified, Ubc9 is the only known SUMO-E2-conjugating enzyme. Several SUMO-E3 ligases have been identified that promote transfer of SUMO from E2 to specific substrates. To date, three unrelated proteins have been suggested to have SUMO-E3 ligase activity; the protein inhibitors of activated STAT1 (PIAS1) proteins (31)(32)(33), RanBP2 (34,35), and polycomb group protein Pc2 (36). The present study described that both Ubc9 and PIAS1 can function as transcriptional coactivators of COUP-TFI for the hCYP11B2 gene transcription in a sumoylation-independent manner. These proteins are shown to form a complex in the nucleus and exhibit a very unique localization in the adrenal zona glomerulosa cells. We demonstrated here that COUP-TFI, Ubc9, and PIAS1 are recruited to an endogenous CYP11B2 promoter, thus contributing to aldosterone biosynthesis in adrenal zona glomerulosa cells.
Cloning of Ubc9 and PIAS1 by a Yeast Two-hybrid System-Yeast two-hybrid screening was conducted with a MATCHMAKER Two-Hybrid System 3 kit (Clontech) and COUP-TFI (amino acids 55-423) as bait. A human adrenocortical adenoma cDNA library was prepared as shown previously (24). Yeast strain AH109 containing pGBKT7-COUP-TFI-(55-423) was transformed with a human adrenocortical adenoma cDNA library in pGADT7 (Clontech) and plated on synthetic complete medium lacking tryptophan, adenine, leucine, and histidine. His ϩ and Ade ϩ colonies exhibiting ␤-galactosidase activity by filter lift assay were further characterized according to the manufacturer's protocol (Clontech). ␤-Galactosidase activity was determined with chlorophenol red ␤-D-galactopyranoside as described previously (16,24). To recover the library plasmids, total DNA from the yeast was isolated with a Zymoprep TM yeast plasmid Miniprep kit (Zymo Research, Orange, CA) and used to transform Escherichia coli (HB101) in the presence of ampicillin. To ensure that the correct cDNAs were identified, the library plasmids isolated were transformed into Y187 containing pGBKT7-COUP-TFI- , and ␤-galactosidase activity was determined. The specificity of the interaction of #2-3 (PIAS1-(5-651)) and #2-4 (Ubc9-(1-158)), both part of the 20 positive clones, with COUP-TFI was determined by mating with Y187, which contains pGBKT7lamin (Clontech). The ␤-galactosidase activities of these diploids were examined by the filter lift and chlorophenol red ␤-D-galactopyranoside methods. The sequence of the #2-3 and #2-4 clones was identical to the GenBank TM -submitted sequence of PIAS1 and Ubc9, respectively. The yeast two-hybrid system was also used to determine protein-protein interaction between COUP-TFI/COUP-TFII/Ad4BP/DAX-1/TR␤ and these clones.
Western Blot Analysis and Coimmunoprecipitation-The cells were lysed with lysis buffer (10 mM Tris-HCl (pH 8.0), 150 mM NaCl, 1% Triton X-100, 5 mM EDTA, 2 mM phenylmethylsulfonyl fluoride), and Western blots were performed before the immunoprecipitation (IP) steps to confirm protein expression by corresponding antibodies as described previously (24). The same samples for the Western blots were diluted to 1 ml in IP buffer (20 mM Tris-HCl (pH 7.5), 150 mM NaCl, 10 mM dithiothreitol, 5 ng/l aprotinin, 0.5 mM phenylmethylsulfonyl fluoride, 0.1% Tween 20) and precleared with protein G plus-agarose beads (Santa Cruz Biotechnology, Santa Cruz, CA), and antibodies were added for 1 h. Immune complexes were adsorbed to protein G plusagarose beads and washed four times in IP buffer. Proteins were then separated on 12.5% polyacrylamide gels and transferred onto Hybond ECL nitrocellulose membranes (Amersham Biosciences). The primary antibodies used for immunoprecipitation were rabbit polyclonal anti-COUP-TFI antibody (generous gift by Dr. Ming-Jer Tsai) (15), and the antibodies used for the Western blots were anti-COUP-TFI, anti-Xpress (Invitrogen), anti-FLAG (Sigma), anti-Ubc9 (BD Biosciences Pharmingen), anti-PIAS1 (Santa Cruz Biotechnology), and anti-␣-tubulin (Oncogene Research Product) antibodies.
Fluorescence Imaging-The images of EGFP-tagged Ubc9 and The PIAS1 fragments shown were cloned into the pGADT7, pEGFP, and pcDNA3.1/His expression vectors as described under "Experimental Procedures." The N terminus of PIAS1 contains NR box, a coactivator motif that interacts with nuclear receptors, whereas the C terminus of PIAS1 contains two CoRNR motifs, the corepressor of which interacts with nuclear receptors. In the middle of the PIAS1 proteins, the RING finger domain, which is necessary for SUMO-1 ligase activity, is localized.
DsRed-COUP-TFI were described previously (24). COS-1 cells were transiently transfected with expression vectors of pEGFP-PIAS1, and pDsRed-COUP-TFI. Live cell microscopy of GFP fusion and DsRed fusion proteins was performed on a confocal microscope (Axiovert 100M, Carl Zeiss Co., Ltd.). Imaging for GFP and DsRed was performed by excitation with 488 and 543 nm, respectively, from an argon laser, and the emissions were viewed through band passes ranging from 500 to 550 nm, and 550 to 600 nm, respectively, by band pass regulation with LSM510 (Carl Zeiss Co., Ltd.). All images were processed as TIFF (tagged image file format) files on Photoshop 7.0 using standard imageprocessing techniques.
Northern Blot Analysis-The human tissue Northern blots (Clontech) were hybridized at 42°C overnight with 32 P-labeled cDNA probes of the full-length 1.1-kb hUbc9, 1.9-kb hPIAS-1, full-length 1.3-kb hCOUP-TFI, or 1.1-kb glyceraldehyde-3-phosphate dehydrogenase (Clontech) cDNAs according to the manufacturer's protocol. The membranes were washed at a final stringency of 0.1 ϫ SSC-0.1% SDS at 50°C and analyzed with a BAS 3000 image scanner (Fuji Film Co.). The mRNA levels were determined by comparison with glyceraldehyde-3phosphate dehydrogenase mRNA levels.
Immunohistochemistry-Formalin-fixed tissues were embedded in paraffin, sectioned at 6 m, and mounted on silane-coated slides. For immunohistochemistry, sections were dewaxed, rehydrated, followed by blocking endogenous peroxidase using 3% (v/v) hydrogen peroxidase in phosphate-buffered saline, which were then subjected to microwave antigen retrieval in 0.01 M citrate buffer. Thereafter, they were washed in phosphate-buffered saline and blocked with a blocking solution containing 5% bovine serum albumin in phosphate-buffered saline for 30 min. They were subsequently incubated overnight at 4°C with primary antibodies diluted appropriately with the blocking solution. Primary antibodies for immunohistochemistry included rabbit anti-COUP-TFI, anti-Ubc9 (BD Biosciences, Pharmingen), anti-PIAS1 (Santa Cruz Biotechnology). After two washes in phosphate-buffered saline, immunoreactivities were detected using a Vectastatin ABC Elite kit (Vector Laboratories, CA) and a Vecta DAB substrate kit (Vector Laboratories). As negative controls, sections were incubated with the preimmune or control serum in place of the primary antibody.
Mammalian Cell Culture, Transient Transfections, and Luciferase Assays-H295R cells, derived from human adrenocortical carcinoma cells (8,12,37), were used for luciferase assays. H295R cells were routinely maintained in Dulbecco's modified Eagle's medium/F-12 (Invitrogen) supplemented with 2.5% NuSerum (Collaborative Bio, Bedford, NA) and 1% ITS (insulin-transferrin-selenium) culture supplement (Invitrogen). Twenty-four hours before transfection, 1 ϫ 10 5 cells per well of a 24-well dish were plated in the medium. All transfections into H295R cells were carried out using Lipofectamine 2000 (Invitrogen) with indicated amounts of expression plasmids, according to the manufacturer's protocol. Cells were harvested 48 h after transfection, and cell extracts were assayed for both Firefly and Renilla luciferase activities with a Dual-Luciferase Reporter Assay System (Promega). Relative luciferase activity was determined as ratio of Firefly/Renilla luciferase activities, and data are expressed as the mean (ϮS.D.) of triplicate values obtained from a representative experiment that was independently repeated at least three times.
Statistics-All experiments were performed in triplicate several times. The error bars in the graphs of individual experiments correspond to the S.D. of the triplicate values.

Identification of Ubc9 and PIAS1 as COUP-TFI-interacting
Proteins by Yeast Two-hybrid System-To search for proteins that might regulate the activity of the COUP-TFI, we performed a yeast two-hybrid screen with COUP-TFI encoding amino acids 55-423 as bait and a cDNA library prepared from a human adrenocortical adenoma as described previously (24). In this manner, we identified a full-length clone of Ubc9 and a partial clone of PIAS1, which are SUMO E2-conjugating enzyme and E3-ligase, respectively. This report further describes function and expression of COUP-TFI, Ubc9, and PIAS1 in the transcriptional regulation of the human aldosterone synthase gene (CYP11B2).
We have recently shown that Ubc9 specifically interacted with COUP-TFI and mapped interaction domains that C terminus of Ubc9 encoding amino acids 59 -158 interacted with the C terminus of COUP-TFI encoding 383-403 (24). In the present study we performed yeast two-hybrid assays to demonstrate that PIAS1 interacts specifically with COUP-TFI. Both Ubc9 and PIAS1 interacted with COUP-TFI, and the interactions were specific, as no interaction between Gal4 DBD-COUP-TFI (amino acids 55-423) fusion and Gal4 activation domain (Gal4 AD; empty vector) was observed ( Fig. 2A). In contrast, as a positive control, we showed strong interaction between Gal4 DBD-COUP-TFI and Gal4 AD-COUP-TFI, because COUP-TFI readily forms homodimers (data not shown). In addition, both Ubc9 and PIAS1 did not interact with unrelated bait corresponding to lamin (data not shown). Besides interacting with COUP-TFI, both Ubc9 and PIAS1 also interacted with COUP-TFII and SF-1, but not with DAX-1 or unliganded TR␤ 168 -456 ( Fig. 2A).
To identify interaction domains more precisely, various fragments of COUP-TFI and PIAS1 were cotransformed in yeast, and ␤-galactosidase liquid assays were performed to quantitate the protein-protein interaction (Fig. 2, B and C). Both Gal4 DBD-COUP-TFI encoding amino acids 86 -183 and amino acids 150 -183 strongly interacted with the full-length PIAS1, suggesting that the DNA-binding domain and hinge regions of COUP-TFI were necessary to interact with Ubc9. In contrast, the C-terminal fragments of PIAS1 encoding amino acids 301-651 and amino acids 406 -651 strongly interacted with COUP-TFI, whereas PIAS1 fragments encoding amino acids 1-150, 1-300, 1-405, and 5-73/564 -651 showed minimal interaction with COUP-TFI. Although the N and C termini of PIAS1 contain nuclear receptor-interacting domains, named nuclear receptor box (NR box), and CoRNR motifs, respectively, these domains are not involved in the interaction with COUP-TFI.
Taken together with the interaction data described above, the most C-terminal fragments of PIAS1 encoding amino acids 406 -563 interacted with DNA binding and hinge regions of COUP-TFI in yeast.
Interaction and Subcellular Localization of Ubc9, PIAS1, and COUP-TFI in Mammalian Cells-We previously showed that Ubc9 interacts specifically and directly with COUP-TFI (24). The association between COUP-TFI, Ubc9, and PIAS1 was further investigated by coimmunoprecipitation assays (Fig. 3A). COS-1 cells were transfected with Xpress-tagged Ubc9 (pcDNA3.1/His-Ubc9), FLAG-tagged PIAS1 (p3xFLAG-CMV-10-PIAS1), and RSV-promoter-driven COUP-TFI expression vectors (pRSV-COUP-TFI). Polyclonal anti-COUP-TFI antibody was first used to precipitate the protein complexes containing COUP-TFI, and the presence of PIAS1 protein was detected in lysates from cells transfected with both RSV-COUP-TFI and FLAG-tagged PIAS1, but not with RSV-COUP-TFI or FLAG-tagged PIAS1 alone (Fig. 3A). Similarly, when RSV-COUP-TFI, FLAG-PIAS1, and Xpress-Ubc9 were expressed together, these exogenously overexpressed proteins were shown to form a complex (Fig. 3A). To determine if Ubc9, PIAS1, and COUP-TFI could interact within a cellular environment, COS-1 cells were transfected with EGFP, EGFPtagged PIAS1, EGFP-tagged PIAS1 (C351S), DsRed, DsRedtagged COUP-TFI alone, or in various combinations and photographed using a fluorescence microscope (Fig. 3, B-G). COS-1 cells transfected with EGFP alone displayed a diffuse green fluorescence (data not shown). EGFP-tagged PIAS1 and DsRed-tagged COUP-TFI showed predominantly nuclear localization with occasional dot formation. When cells were then cotransfected with both EGFP-tagged PIAS1 and DsRedtagged COUP-TFI, expression of both proteins was colocalized in the nucleus and was not altered compared with the expression obtained when either was transfected alone. In addition, expression of both EGFP-tagged PIAS1 (C351S) and DsRed-COUP-TFI was colocalized in the nucleus (Fig. 3, E-G), indicating that presence of sumoylation activity did not affect nuclear localization of COUP-TFI.
Tissue Distribution of mRNA of COUP-TFI, Ubc9, and PIAS1 by Northern Blot Analysis-We next examined the expression of COUP-TFI, Ubc9, and PIAS1 mRNA in human endocrine tissues (Fig. 4). Because Ubc9 and PIAS1 were cloned from a human adrenocortical adenoma cDNA library, we confirmed that these mRNAs were expressed in normal human adrenal cortex (Fig. 4). Ubc9 mRNA is widely distributed in many endocrine tissues, and the expression profile of Ubc9 is similar to that of COUP-TFI as shown previously (24). PIAS1 mRNA is also relatively highly expressed in steroidogenic tissues, such as testis, ovary, and adrenal cortex (Fig. 4).
Immunohistochemistry of COUP-TFI, Ubc9, and PIAS1 in Rat Adrenal Glands-We next examined immunohistochemistry of COUP-TFI, Ubc9, and PIAS1 in rat adrenal glands (Fig.  5). Strikingly, both Ubc9 and PIAS1 are exclusively expressed in the nuclei of zona glomerulosa cells of adult rat adrenal cortex (Fig. 5, B and C), whereas COUP-TFI is expressed in the nuclei of all three zones of adrenocortical cells (Fig. 5A). In addition, these expression profiles are shown to be developmentally regulated. At 2 weeks after birth, Ubc9 is expressed in all three zones of the adrenal cortex, and expression of Ubc9 gradually becomes localized in the zona glomerulosa cells as development progresses at 3 weeks after birth (data not shown). At 8 weeks after birth, Ubc9 is specifically expressed in zona glomerulosa cells and slightly expressed in the interstitial stromal cells of zona fasciculata and reticularis (Fig. 5B). These expression profiles of Ubc9 were not altered between male and female rat adrenal glands (data not shown). These results indicate that both Ubc9 and PIAS1 are mainly expressed in the nuclei of adrenal glomerulosa cells, where aldosterone is exclusively synthesized.
COUP-TFI Binds to Ad5 Element (Ϫ129/Ϫ114) of the Human CYP11B2 Gene Promoter and Competes with SF-1 for Binding in a Mutually Exclusive Manner-We have previously mapped human CYP11B2 DNA sequences required both for basal transcription and for optimal transcriptional responses to angiotensin II, K ϩ , and cAMP to position Ϫ129/Ϫ114 (12). To define more precisely the DNA-protein interactions occurring within this region, electrophoretic mobility shift analysis (EMSA) was performed using a synthetic 32 P-labeled oligonucleotide probe encompassing this sequence and nuclear extract derived from H295R cell line (Fig. 6A). In the presence of H295R cell nuclear extract, 4 protein-DNA complexes were formed. Formation of complexes 1, 2, and 3 was completely inhibited in the presence of a 200-fold molar excess of non-radiolabeled probe (lane 3 in Fig. 6A). We observed displacement at lower levels of the cold probe, which was 10-fold displaced in about one-half of the radiolabeled binding (data not shown). These data indicate that these three complexes represent specific protein-DNA interactions. Formation of the fourth complex, migrating immediately above complex 3, was unaffected by this maneuver, suggesting that this complex arises through nonspecific DNA binding (Fig.  6, NS). Complex 3 was recognized by an antibody directed against SF-1 (lane 4 in Fig. 6A), whereas complexes 1 and 2 were both recognized by antibodies directed against COUP-TF (lane 5 in Fig. 6A).
To confirm that SF-1 and COUP-TF bind to this element, the binding activity of in vitro translated proteins was assessed. In vitro translated SF-1 bound to the Ad5 probe, producing a complex with mobility similar to that of complex 3 (lane 1 in Fig. 6B). In the presence of either in vitro translated COUP-TFI or COUP-TFII, a complex with mobility similar to that of complex 1 was formed (lanes 3 and 6 in Fig. 6B). All three in vitro synthesized proteins were recognized by their respective antibodies (lanes 2, 4, and 5 in Fig. 6B). These data suggest that complexes 1 and 3 represent binding of COUP-TF and SF-1, respectively. The nature of the protein(s) forming complexes 2 is unknown at present. Because formation of this complex was abolished in the presence of anti-COUP-TF antibody, but not reproduced by recombinant COUP-TFI or II, complex 2 may represent binding either of a heterodimer between COUP-TF and another unidentified protein or binding of a protein related to, but distinct from COUP-TF.
COUP-TFI-mediated Human CYP11B2 Gene Transactivation Is Potentiated by Ubc9 and PIAS1 Independently of the sumoylation Activities-To explore how COUP-TF regulates human CYP11B2 gene transcription, we transiently transfected exogenous COUP-TFI in H295R cells and measured the luciferase activity of the CYP11B2 reporter gene. Overexpression of COUP-TFI activated the CYP11B2 gene transcription in a dose-dependent manner (lanes 2-4 in Fig. 7A). The mutation of Ad5 sequences, to which COUP-TFI binding was disrupted, abrogated basal and COUP-TFI-mediated transactivation of the gene, thus suggesting that Ad5 sequences are crucial for COUP-TFI-mediated CYP11B2 transactivation (Fig. 7B). In addition, overexpression of COUP-TFI⌬35, in which 35 amino acids are deleted from the C terminus, effectively removing repressor domain, did not activate this reporter activity (lanes 5-7 in Fig. 7A), indicating that the deleted region of the COUP-TFI C terminus is indispensable for this activation of CYP11B2 gene.
Furthermore, cotransfection of Ubc9 or PIAS1 with COUP-TFI transactivated COUP-TFI-mediated activation of the CYP11B2 gene transcription (lanes 5 and 8 in Fig. 8A), whereas Ubc9 or PIAS1 alone did not influence the reporter gene activity (lanes 2 and 3 in Fig. 8A). Cotransfection of both Ubc9 and PIAS1 with COUP-TFI showed that their transactivating effect on the COUP-TFI-mediated CYP11B2 transcription was synergistic (Fig. 8B). In addition, coexpression of Ubc9-(1-58) or PIAS1-(1-150), which does not contain COUP-TFI-interacting domain, had no effects on the COUP-TFI-mediated transactivation (lanes 7 and 10 in Fig. 8A), indicating that interaction between COUP-TFI and Ubc9 or PIAS1 is crucial for COUP-TFI-mediated transactivation of the CYP11B2 promoter. These findings indicate that both Ubc9 and PIAS1 can function as transcriptional coactivators of COUP-TFI for the CYP11B2 gene transcription. Although Ubc9 and PIAS1 are enzymes that are responsible for SUMO modification, the sumoylationdefective mutants Ubc9 (C93S) or PIAS1 (C351S) continued to function as activators of COUP-TFI (lanes 6 and 9 in Fig. 8A). These findings suggest that both Ubc9 and PIAS1 function as coactivators for the COUP-TFI-mediated CYP11B2 gene transcription in a sumoylation-independent manner.
Binding of Ubc9 and PIAS1 to COUP-TFI, but Not to SF-1, Are Crucial for COUP-TFI-mediated CYP11B2 Gene Transactivation-The Ad5 sequence contains a direct repeat of a consensus (AGGTCA) and variant (AGGCTG) nuclear receptor half-site (DR0) on the non-coding strand. To define the precise nucleotides required for binding of SF-1 and COUP-TFI, a series of oligonucleotides containing progressive stepwise 2-bp mutations across these putative half-sites (m1-m9) were radiolabeled and used in EMSA (Fig. 9A). When H295R nuclear extracts were used as the source of protein (top panel of Fig.  9A), m1 through m4 formed specific complexes similar to those formed using the wild-type probe. The m5 through m8 failed to form complexes 1 or 2, although complex 3 was still observed using m7 and m8 as probe. Similar results were observed using in vitro translated proteins (Fig. 9A). SF-1 was bound to all mutant oligonucleotides except m5 and m6, whereas COUP-TFI and COUP-TFII failed to bind to m5 through m8. These data further support the hypothesis that complexes 1 and 3 represent binding of COUP-TF and SF-1, respectively, because the abilities of in vitro synthesized proteins and H295R nuclear extracts to bind to the various mutant oligonucleotides were identical. Further, because only mutations within the consensus nuclear receptor half-site (AGGTCA, m5-m8) disrupt binding of COUP-TF, it is likely that this sequence represents the key site for the COUP-TF binding site. Utilizing Ad5 luciferase reporters containing the wild-type, m5 or m7 sequences, we investigated how Ubc9 and PIAS1 affect the COUP-TFI-regulated CYP11B2 promoter activities. In compared with the finding that Ubc9 and PIAS1 potentiated the COUP-TFI-mediated transactivation of the wild-type Ad5 promoter (lanes 4 -7 in Fig. 9B), both proteins activated neither m5 nor m7 Ad5 reporter activities (lanes 8 -21 in Fig. 9B). These findings indicate that both Ubc9 and PIAS1 enhance the Ad5 reporter activity mediated by COUP-TFI, but not SF-1.
Endogenous Ubc9 and PIAS1 Are Required for the CYP11B2 Transcriptional Activation by COUP-TFI-If Ubc9 and PIAS1 are coactivators of COUP-TFI, reducing the endogenous level of Ubc9 or PIAS1 should decrease the transcriptional activity by COUP-TFI in transient transfection assays. As described previously (Figs. 7 and 8), overexpression of COUP-TFI activated human CYP11B2 reporter activities by 6-fold (lane 3 in Fig.  10A). Cotransfection of siRNA of two sets of Ubc9 (Ubc9a (lane 5) and Ubc9b (lane 6) in Fig. 10A) or two sets of PIAS1 (PIAS1a (lane 7) and PIAS1b (lane 8) in Fig. 10A), but not negative control (lane 4 in Fig. 10A), effectively reduced the endogenous levels of Ubc9 or PIAS1 proteins but had no effect on the ␣-tubulin protein level seen in Western blot (Fig. 10B). Reduction of endogenous Ubc9 or PIAS1 protein level decreased the COUP-TFI-mediated transactivation by ϳ30 -50%. These findings indicate that endogenous Ubc9 and PIAS1 normally function as transcriptional coactivators for the COUP-TFI-mediated transactivation.
COUP-TFI, Ubc9, and PIAS1 Are Specifically Recruited to the Ad5 Element of Human CYP11B2 Gene Promoter-As mentioned above, COUP-TFI-Ubc9-PIAS1 complex activated the human CYP11B2 gene transcription. Chromatin immunoprecipitation (ChIP) assays were used to test whether Ubc9 and PIAS1 are recruited to the endogenous CYP11B2 gene promoter in H295R cells. The cross-linked, sheared chromatin preparations were subjected to immunoprecipitation with various antibodies, and the precipitated DNA was analyzed by PCR amplification of the Ad5 element of the CYP11B2 promoter. We have confirmed the size of the sonicated DNA is ϳ300 -600 bp and these bands looked uniform (data not shown). When COUP-TFI, Xpress-Ubc9, and FLAG-PIAS1 were aberrantly overexpressed in H295R cells, antibodies against COUP-TFI, Xpress, or FLAG efficiently immunoprecipitated the Ad5 element of the CYP11B2 promoter. Normal IgG and no antibody failed to precipitate the CYP11B2 promoter (Fig. 11). In contrast to the Ad5 element of the CYP11B2 promoter, the 3Ј-untranslated control region of the CYP11B2 gene was not detected in association with COUP-TFI, Ubc9, or PIAS1 (Fig. 11). Thus Ubc9, PIAS1, and COUP-TFI were recruited to a native COUP-TFI-regulated CYP11B2 promoter, demonstrating a functional interaction between COUP-TFI and Ubc9-PIAS1 occurring in an in vivo setting. The results from ChIP and RNA interference experiments strongly support a physiological role of Ubc9 and PIAS1 in COUP-TFI-dependent transcription of the human CYP11B2 gene.

DISCUSSION
In this report, we have identified and described Ubc9 and PIAS1, which interact with the COUP-TFI and can function as coactivators for the COUP-TFI-mediated transcription of the human CYP11B2 gene. We detected that COUP-TFI, Ubc9, and PIAS1 form a complex in the nuclei of mammalian cells. Expression of Ubc9 and PIAS1 was markedly detected in rat adrenal zona glomerulosa cells, in which CYP11B2 is exclusively expressed. Transient transfection assays together with small interfering RNA and ChIP assays indicated a physiological role of Ubc9 and PIAS1 in COUP-TFI-dependent transactivation of the human CYP11B2 gene in normal adrenal glomerulosa cells. Both Ubc9 and PIAS1 have SUMO-1conjugating enzyme (E2) and SUMO-1 ligase (E3) activity, respectively (25)(26)(27)(28)(29)(30). The C93S substitution of Ubc9 and the C351S substitution of PIAS1, which abrogate sumoylation activity, continued to interact with COUP-TFI and to potentiate transcriptional activation mediated by COUP-TFI, indicating that their sumoylation enzyme activities are not required for function as coactivators of COUP-TFI.

Ubc9 and PIAS1 Function as Transcriptional Coactivators of COUP-TFI in CYP11B2 Transcription-Both
Ubc9 and PIAS1 meet all the criteria for transcriptional coactivator proteins in the modulation of COUP-TFI transcriptional properties. First, as shown in Figs. 2 and 3, Ubc9 and PIAS1 specifically interacted with COUP-TFI in yeast and mammalian cells. The Ubc9 was previously shown to interact with the C-terminal putative ligand-binding domain of COUP-TFI-(388 -403) that is described as a transcriptional repressor domain (24). The present study showed that PIAS1 interacted with the DNA-binding domain and hinge region of COUP-TFI-(86 -183). Coimmunoprecipitation and subcellular localization experiments showed that Ubc9 and PIAS1 are colocalized with COUP-TFI in transfected COS-1 cells (Fig. 3). These findings suggest that Ubc9 and PIAS1 specifically interact with COUP-TFI and form a complex in mammalian cells. Second, overexpression of Ubc9 or PIAS1 had no effect on the reporter activities in the absence of transfected COU-TFI (Fig. 8A). However, Ubc9 or PIAS1 potentiated the transactivation mediated by COUP-TFI. Subsequently, coexpression of the Ubc9 or PIAS1 deletion mutant, Ubc9-(1-58) or PIAS1-(1-150), which impairs COUP-TFI binding, did not affect the transactivation mediated by COUP-TFI, indicating that interaction of COUP-TFI with Ubc9 or PIAS1 is required for COUP-TFI-mediated transactivation. In addition, coexpression of Ubc9 and PIAS1 synergistically enhanced the COUP-TFI-mediated transactivation. Third, reduction of endogenous Ubc9 or PIAS1 by small interfering RNA decreased COUP-TFI-mediated transactivation of the human CYP11B2 promoter activity, indicating that endogenous Ubc9 and PIAS1 normally contributes to COUP-TFI-mediated transactivation. Fourth, ChIP assays clearly showed that COUP-TFI, Ubc9, and PIAS1 were recruited to a native COUP-TFI-regulated CYP11B2 promoter, demonstrating a functional coupling between COUP-TFI and Ubc9-PIAS1 in vivo. Therefore, Ubc9 and PIAS1 possess all the characteristics expected for transcriptional coactivator proteins of COUP-TF in vivo.
To confirm further that Ubc9 and PIAS1 are coactivators of COUP-TFI, we ruled out several other possible ways in which these proteins might enhance COUP-TFI-mediated transactivation. First, as SUMO-1 conjugation plays an important role in protein modification, the effects of Ubc9 and PIAS1 on COUP-TFI transactivation might be the result of effects of Ubc9 and PIAS1 on COUP-TFI protein concentrations. Our preliminary experiments showed that overexpression of Ubc9 or PIAS1 did not alter COUP-TFI protein concentration in H295R cells (data not shown). Second, it was also possible that overexpression of Ubc9 or PIAS1 increases the concentrations of some coactivators or decreases the concentrations of some corepressors, which have been shown to interact with COUP-TFI, but the results showed that overexpression of these proteins did not alter the protein concentrations of SRC-1, GRIP-1, or SMRT in the cells (data not shown). Because these experiments were performed by transient transfection, we are not able to conclude unequivocally that Ubc9 and PIAS1 have no effects on these coregulator concentrations, and further investigation is required. Third, another possibility is that overexpression of Ubc9 or PIAS1 increased the DNA-binding affinity of COUP-TFI. To exclude the possibility, we performed electrophoretic mobility shift assays to determine whether bacterially or in vitro transcription-translated Ubc9 or PIAS1 proteins affect the binding of COUP-TFI to its response element DNA (Ad5) of the human CYP11B2 promoter region. The results showed that Ubc9 or PIAS1 have no effects on COUP-TFI binding to the Ad5 element (data not shown). Fourth, it has been proposed that SUMO-1 conjugation targets proteins to different cellular localizations. SF-1 can be directed into nuclear speckles and sequestered from the nucleolus in the presence of SUMO-1, thus resulting in transcriptional repression (39). It is therefore possible that coexpression of Ubc9 and PIAS1 alters subcellular localization of COUP-TFI. However, based on the subcellular localization data (Fig. 3), localization of COUP-TFI continues to be in the nucleus without re-localization. Taken together with the abovementioned findings, Ubc9 and PIAS1 clearly function as novel coactivators of COUP-TFI in vivo. However, detailed molecular mechanisms are largely unknown and further investigation is required.
Role of Ubc9 and PIAS1 in Sumoylation-SUMO-1 conjugation (sumoylation) has been reported to play an important role in many cellular processes (25)(26)(27)(28)(29)(30). Sumoylation resembles ubiquitination, but the enzymes involved in these three processes are distinct. SUMO-1 is conjugated to target proteins at the consensus sequence KXE ( is any hydrophobic amino acid, and X is any amino acid). COUP-TFI has no such SUMO consensus sequence; however, several proteins, such as Mdm2 and CREB, are also modified at sites other than KXE (26). Further investigation is required to elucidate whether COUP-TFI is sumoylated in vivo.
Our data showed that both wild type and sumoylation-defective mutants Ubc9 (C93S) and PIAS1 (C351S) similarly enhanced COUP-TFI-regulated CYP11B2 promoter activities, indicating that these proteins possess dual distinct functions, SUMO-dependent and SUMO-independent pathways, such as coactivator function. However, it is possible that ectopically produced Ubc9 and PIAS1 regulate COUP-TFI-mediated transactivation through not only sumoylation of COUP-TFI but also conjugation of SUMO-1 to one or more other cellular factors involved in transcriptional regulation. Recent data have raised the intriguing possibility that SUMO modification may have a specific impact on the ability of some transcription factors to function synergistically (27,40,41). Previous studies of the glucocorticoid receptor (GR) had identified a region referred to as a synergy control motif, mutation of which led to a selective increase in the activity of the GR from promoters bearing multiple, but not single sites. The synergy control motif contains a consensus SUMO acceptor site, and recent data have shown that this is, in fact, the major site of addition of SUMO in the GR. Considerable numbers of transcription factors, including GR (41)(42)(43)(44), androgen receptor (32,33,(45)(46)(47)(48)(49)(50)(51), progesterone receptor (47,52), mineralocorticoid receptor (53,54), peroxisome proliferator-activated receptor ␥ (55), and steroidogenic factor-1 (SF-1) (39,56), are modulated by SUMO-1 attachment, and SUMO-modified transcription factors mostly resulted in attenuation of transcription. Recent studies demonstrated that one of the molecular mechanisms of sumoylation-mediated repression is protein modified by SUMO-1 recruited histone deacetylases and transcriptional corepressors, thus repressing transcription (29,38,(57)(58)(59). However, there are also opposite examples of SUMO-modified proteins, which lead to transcriptional activation rather than repression. Transcriptional coactivators, such as SRC-1 and GRIP1, are shown to be modified by SUMO-1, thus resulting in enhanced transcriptional coactivator capacities (60,61). Therefore, sumoylation does not necessarily induce transcriptional repression depending on the substrates. The molecular mechanisms that could explain how SUMO modification affects transcriptional regulation are largely unknown.
COUP-TFI-Ubc9-PIAS1 Complexes Are Crucial for Aldosterone Synthesis-We have identified three important cis-elements in the hCYP11B2 promoter: a CRE at Ϫ71/Ϫ64, a ciselement termed Ad5 at Ϫ129/Ϫ114, and a third cis-element termed NBRE-1 (Ϫ766/Ϫ759) (8,9,12). The CRE is common to both hCYP11B1 and hCYP11B2 and is regulated by both protein kinase A-and calmodulin-dependent kinase-dependent mechanisms. Neither the Ad5 nor NBRE-1 cis-elements found in the 5Ј-flanking sequence of hCYP11B2 are present in hCYP11B1 (8,9,12). The present results confirmed Ad5 element is crucial for CYP11B2 transactivation, because deletion of the element dramatically reduces basal and COUP-TFImediated transactivation (Fig. 7B). Based on the present and previous reports (8,9,12), COUP-TFI and/or Nurr1/NGFI-B play important roles in transactivation of human CYP11B2 gene through binding to the Ad5 element. The physiological importance of these particular transcription factors should be investigated in vivo. Our preliminary data demonstrated that levels of expression of Ubc9 and PIAS1 are not altered in aldosterone-producing adenomas of patients with primary al-dosteronism compared with those in normal adrenals. 2 However, the significance of Ubc9 and PIAS1 in aldosterone-producing adenomas remains to be further investigated.
We have recently shown that the effects of K ϩ and angiotensin II (Ang II) on hCYP11B2 transcription occur through two pathways; increased expression of Nurr1/NGFI-B and phosphorylation of ATF-1/CREB (13). Ang II treatment rapidly induces levels of mRNA and protein of Nurr1/NGFI-B, thereby transactivating hCYP11B2 gene (8,9). Both Nurr1 and NGFI-B markedly increased transcription of hCYP11B2 through binding to the NBRE-1 and Ad5 sites, which are unique to hCYP11B2. Buholzer et al. (62) have very recently reported that COUP-TF is a negative regulator of steroidogenesis in bovine adrenal glomerulosa cells. They also showed that Ang II treatment dramatically decreased levels of COUP-TF in the cells, thus activating StAR gene expression. Based on these findings, it is tempting to speculate that levels of expression of COUP-TFI, Ubc9, or PIAS1 are regulated by Ang II treatment. We therefore investigated these possibilities in H295R cells; however, Ang II treatment did not affect expression levels of COUP-TFI, Ubc9, or PIAS1 in real-time reverse transcription-PCR and Western blot analysis (data not shown). The reason why Ang II treatment showed different effects on COUP-TFI levels between other report and ours is not known; however, different cellular context, such as bovine adrenal zona glomerulosa cells and human H295R cells, may be one reason for that. We do not know a physiological role of COUP-TFI and Ubc9-PIAS1 in Ang II and K ϩ regulation of hCYP11B2 expression in this study, and this should be investigated.
In conclusion, we identified novel COUP-TFI-interacting proteins, Ubc9 and PIAS1, and these proteins function as coactivators of COUP-TFI for human CYP11B2 transactivation. The unique localization profiles of these proteins in adrenal zona glomerulosa cells are consistent with a crucial physiological role in aldosterone biosynthesis. Therefore, the COUP-TFI-Ubc9-PIAS1 complexes shed new light in controlling the long term capacity of the adrenal gland to produce aldosterone. In addition, these studies provide new mechanisms through which COUP-TFI can act as a transcriptional activator through the novel interaction with Ubc9 and PIAS1.