Coregulator Small Nuclear RING Finger Protein (SNURF) Enhances Sp1- and Steroid Receptor-mediated Transcription by Different Mechanisms*

The small nuclear RING finger protein SNURF is not only a coactivator in steroid receptor-dependent transcription but also activates transcription from steroid-independent promoters. In this work, we show that SNURF, via the RING finger domain, enhances protein binding to Sp1 elements/GC boxes and interacts and cooperates with Sp1 in transcriptional activation. The activation of androgen receptor (AR) function requires regions other than the RING finger of SNURF, and SNURF does not influence binding of AR to cognate DNA elements. The zinc finger region (ZFR) together with the hinge region of AR are sufficient for contacting SNURF. The nuclear localization signal in the boundary between ZFR and the hinge region participates in the association of AR with SNURF, and a receptor mutant lacking the C-terminal part of the bipartite nuclear localization signal shows attenuated response to coexpressed SNURF. Some AR ZFR point mutations observed in patients with partial androgen insensitivity syndrome or male breast cancer impair the interaction of AR with SNURF and also render AR refractory to the transcription-activating effect of SNURF. Collectively, SNURF modulates the transcriptional activities of androgen receptor and Sp1 via different domains, and it may act as a functional link between steroid- and Sp1-regulated transcription.

The androgen receptor (AR), 1 a member of the steroid receptor family, acts as a hormone-regulated transcription factor. The N-terminal region contains a powerful ligand-independent transcription activation function-1 (AF-1). The second but weaker activation function (AF-2) localizes to the ligand binding domain; it requires hormone for activation and perhaps also an intramolecular interaction with the AF-1 region (1)(2)(3)(4)(5)(6). The apo-ligand binding domains of steroid receptors interact with transcriptional corepressors and heat-shock proteins, and the ligand-induced conformation in the ligand binding domain enables interactions between several coactivators and AF-2 (7)(8)(9)(10)(11)(12). The steroid receptor zinc finger region (ZFR) consists of two zinc finger (ZF) structures (13). The first ZF is responsible for contacting the specific hormone response element (HRE), whereas the second ZF stabilizes the receptor-DNA interactions and participates in homodimer formation (13)(14)(15)(16). The bipartite nuclear localization signal (NLS) begins immediately C-terminal to the second ZF and continues through the first amino acid residues of the hinge region (17,18). Besides binding to HREs and mediating the dimerization, the ZFRs of steroid receptors have also other less well characterized but nevertheless important functions. Unlike glucocorticoid receptor (GR) null mice, animals with a GR mutation that prevents dimerization and efficient DNA binding are viable, attesting to the importance in vivo of those GR activities that are independent of DNA binding (19). In human males, mutations in the AR ZFR can lead to either complete or partial androgen insensitivity syndrome (3,20,21). AR ZFR is important in transrepression of AP1-and NF-B-activated genes, and it is able to interact with the coactivator cAMP-response element-binding protein (CREB)-binding protein (22)(23)(24)(25). In support to the role of ZFR/hinge as an interaction interface for heterologous proteins, several proteins capable of associating with ZFR and hinge region of nuclear receptors have been identified over the last few years (26 -33).
Steroid receptors act together with other transcription factors in the regulation of target genes. Promoter specificity protein 1 (Sp1) is a ubiquitously expressed transcription factor that frequently works in concert with other sequence-specific transactivating proteins to control inducible promoters (34,35). Functional and physical interactions between Sp1 and factors such as SREBp-1a (36), Stat1 (37), GATA-1 (38), estrogen receptor (39), progesterone receptor (40), and helicase-like transcription factor (41) often result in synergistic activation of specific target promoters, whereas association with promyelocytic leukemia gene product inhibits Sp1-mediated transcription (42). The Sp family of proteins, including Sp2, Sp3, and Sp4 in addition to Sp1, recognize GC-rich promoter sequences, the GC boxes, via their highly homologous C-terminal zinc finger regions (34,35,(43)(44)(45). The N-terminal regions of Sp proteins contain transactivation domains rich in glutamine and serine/threonine residues, and an additional transactivation region in the C terminus (the D domain) is involved in Sp1 synergistic function (46). Sp3 also contains an inhibitory domain, the function of which is promoter-and cell line-dependent.
We recently identified a novel coactivator of AR-dependent transcription, a small nuclear RING finger protein termed SNURF, which interacts with the AR ZFR (27). SNURF also associates with some other steroid receptors and modulates their transactivating functions. Besides steroid receptor-dependent promoters, SNURF also activates transcription from promoters driven by Sp1 elements/GC boxes. Interestingly, SNURF is able to bind to the TATA-binding protein, and it could thus act as a bridging factor between steroid receptors or other sequence-specific transcription factors and the general transcription machinery (27). To gain a better understanding of the functional domains of SNURF, we have examined the ability of mutated SNURF forms to act as transcriptional coactivators in the context of both androgen-and Sp1-regulated promoters. The requirements for interactions between SNURF and AR were investigated by using a panel of mutated AR forms.

EXPERIMENTAL PROCEDURES
Materials-Protease inhibitors aprotinin, leupeptin, pepstatin, and phenylmethylsulfonyl fluoride were obtained from Sigma. Testosterone was purchased from Makor Chemicals. Mouse monoclonal M2 anti-Flag antibody, mouse monoclonal anti-Lex antibody, rabbit polyclonal anti-Sp1 antibody, and horse radish peroxidase-conjugated anti-mouse IgG were obtained from Eastman Kodak Co., CLONTECH, Santa Cruz, and Zymed Laboratories Inc., respectively. AP1 and Sp1 oligonucleotides were from Promega.
Cell Culture and Transfections-COS-1 and CV-1 cells were obtained from American Type Culture Collection and were maintained in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum and 25 units/ml streptomycin and penicillin. For transient transfections using FuGene reagent (Roche Molecular Biochemicals), 3-3.5 ϫ 10 4 cells were seeded on 12-well plates 24 h before transfections. For transfections by the calcium phosphate precipitation method (47, 48), 1.5 ϫ 10 6 cells were plated on 10-cm dishes. Four h before the addition of DNA, the cells received fresh medium with 10% charcoal-stripped fetal bovine serum. After 18 h, the medium was changed to Dulbecco's modified Eagle's medium supplied with 2% charcoal-stripped fetal bovine serum and 100 nM testosterone or vehicle. Luciferase and ␤-galactosidase activities and the concentration of soluble cell proteins were assayed as described previously (47,49). Statistical analyses were carried out with two-tailed Student's t test.
Electrophoretic Mobility Shift Assay (EMSA)-EMSAs were performed as previously outlined (50). Briefly, whole cell extracts from COS-1 cells transiently transfected by electroporation or by using Fu-Gene reagent were preincubated at 4°C for 30 min before the addition of 20 fmol of 32 P-labeled double-stranded Sp1-binding site oligonucleotide (5Ј-ATTCGATCGGGGCGGGGCGAG-3Ј), AP1-binding site oligonucleotide (5Ј-CGCTTGATGAGTCAGCCGGAA-3Ј), or androgen receptor response element (ARE) oligonucleotide (5Ј-GATCATAGTACGTG-ATGTTCTCAAGATC-3Ј). When antibodies were used, the preincubation was extended to 90 min before the addition of the probe. The binding reaction was performed for 30 min, and protein-DNA complexes were resolved on 4% polyacrylamide gels under nondenaturing conditions at 4°C.
Yeast Two-hybrid Assay-Saccharomyces cerevisiae strain L40 (a gift from S. M. Hollenberg) was transformed with plasmids encoding LexN-a AR and VP16 AD or VP16 AD fused to SNURF amino acids 20 -177. The transformants were plated on a selective medium, and ␤-galactosidase activities were assayed from three separate liquid yeast cultures according to the instructions of the Matchmaker two-hybrid system (CLONTECH).

The RING Finger Region of SNURF Is Necessary for Stimulation of Steroid-independent Minimal
Promoters-To map the functional domains of SNURF, a series of deletion and point mutants was generated (Fig. 1A). Cotransfection of COS-1 cells with a construct encoding wild-type SNURF along with the reporter Sp1 2 -TATA-LUC, which contains two binding sites for Sp1 in front of the TATA box, activated the promoter by ϳ40fold (Fig. 1B). Deletion of the N-terminal 20 amino acids containing a part of the consensus bipartite NLS (mutant ⌬1-20) did not prevent SNURF from entering the nucleus (data not shown) but blunted its ability to stimulate reporter gene transcription (p Ͻ 0.01). Deletion of amino acids 66 -98 encompass-  7 and 12-14), and the total amount of DNA was adjusted to 2 g. Fifteen g of cell extracts were incubated with 32 P-labeled oligonucleotide containing C3(1)-ARE and analyzed by EMSA. D, the amount of AP1 element-binding proteins in COS-1 cells is not affected by SNURF coexpression. The same cell extracts as those in A and B were incubated with 32 P-labeled oligonucleotide containing a consensus AP1 site. Before adding the probe, a 200-fold molar excess unlabeled AP1 oligonucleotide was included in the binding reaction (lane 10). Only the part of the gel containing specific protein-DNA complexes is shown. These experiments were repeated three times with essentially identical results. ing a region with positively charged residues also severely attenuated the activation of the reporter gene (p Ͻ 0.01), whereas deletions of negatively charged regions (⌬31-65 and ⌬99 -118) only weakly influenced the function of SNURF. Any insult in the RING finger domain totally abrogated the ability of SNURF to stimulate transcription from the Sp1 2 -TATA-LUC reporter (mutants ⌬121-143, ⌬178 -194, ⌬157-194, CS1 (C136S/C139S), CS2 (C177S/C180S), and CS3 (C136S/C139S/ C177S/C180S)). Similar results were obtained with the AP1-Sp1-TATA-LUC reporter (27). 2 SNURF Enhances Protein Binding to GC Box Motifs-To elucidate the mechanism(s) underlying SNURF-mediated stimulation of the Sp1-regulated reporter, we examined by using EMSA the possibility that SNURF influences protein binding to Sp1 elements. Extracts from COS-1 cells transfected with an empty expression vector or plasmids encoding SNURF or the RING finger-deficient SNURF-CS1 mutant were incubated with 32 P-labeled double-stranded Sp1 oligonucleotide. Extracts from cells overexpressing wild-type SNURF but not the SNURF-CS1 mutant showed significantly increased binding of proteins to the Sp1 element ( Fig. 2A, lanes 1-9). All detected protein-DNA complexes were specific for the Sp1 element, as they were fully competed for by a 200-fold molar excess of cold Sp1 oligonucleotide but not by the same amount of AP1 oligonucleotide (lanes 10 and 11). Immunoblotting of the same cell extracts with an Sp1-specific antibody ruled out the possibility that expression of SNURF would have increased the amount of cellular Sp1 protein (Fig. 2B). To identify Sp1 protein among the protein-DNA complexes, the cell extract was preincubated with an anti-Sp1 antibody, which resulted in the disappearance of the uppermost band and appearance of a supershifted complex ( Fig. 2A, lane 12). The bands unaffected by the Sp1 antibody probably represent other GC box-binding proteins such as Sp3 and/or Sp4. SNURF stabilized DNA binding of these proteins as well, but not to the same extent as that of Sp1 (cf. Fig.  2A, lanes 5 and 6 versus lane 3). The effect of SNURF appeared to be specific for GC box-binding proteins, as SNURF overexpression did not influence AR-ARE interaction or the amount of proteins binding to an AP1 element (Fig. 2, C and D).
SNURF Associates with Sp1 in COS-1 Cells-We performed coimmunoprecipitation assays on extracts from COS-1 cells expressing Flag-tagged SNURF or SNURF-CS1 to assess whether SNURF and Sp1 interact physically. Protein complexes precipitated by a monoclonal anti-Flag antibody were subjected to immunoblotting with an anti-Sp1 antibody. As illustrated in Fig. 3A, endogenous Sp1 protein coimmunoprecipitated with the wild-type SNURF but not with the RING finger-deficient mutant.
SNURF Cooperates with Sp1-To study the role of SNURF in Sp1-mediated transcription further, COS-1 cells were cotransfected with the Sp1 2 -TATA-LUC reporter along with Sp1 and/or SNURF expression vectors. Coexpressed Sp1 stimulated the reporter gene activity only minimally over the activity seen with endogenous Sp1 (Fig. 4). However, a strong synergistic activation was observed when SNURF was cotransfected with Sp1 (p Ͻ 0.05). It is of note that the amount of SNURF expression plasmid used in this experiment (100 ng/well) yielded only ϳ10% of the maximal activity obtained with a saturating SNURF concentration (300 ng/well, see Fig. 1B). Ectopic expression of Sp3, an inhibitory member of the Sp1 family (44), slightly diminished the reporter activity achieved with SNURF or SNURF together with Sp1, and no synergism between Sp3 and SNURF was observed (Fig. 4). The RING finger of SNURF was necessary for the synergistic activation to occur, as the SNURF-CS1 mutant was unable to cooperate with coexpressed Sp1. The enhancement of reporter gene activity was dependent on the Sp1-binding sites on the promoter, since a reporter construct devoid of these sites did not respond to ectopically expressed Sp1. 2 SNURF Domains in the Stimulation of AR-mediated Transcription and Cooperation between AR and Sp1-Wild-type SNURF activates AR-mediated transcription more efficiently than the CS1 mutant, but it also enhances the activity of reporter genes in the absence of hormone. SNURF-CS1 exhibits ϳ50 -80% of the wild-type SNURF activity without influencing the basal activity of androgen-regulated promoters (27). To examine the importance of various SNURF regions in the enhancement of AR-dependent transactivation, we transfected COS-1 cells with pARE 2 -TATA-LUC and pSG5-rAR along with vectors encoding SNURF deletion mutants (Fig. 5A). To eliminate the influence of SNURF on basal reporter activity, the CS1 mutant was used as a reference, and the CS1 mutation (C136S/C139S) was included in all the deletion constructs shown in Fig. 5. None of the mutants studied activated the reporter gene in the absence of AR or testosterone (data not shown). Interestingly, the loss of the N-terminal 20 amino acids, a region rich in positively charged residues, and deletion of amino acids 31-65, a negatively charged region, abolished the ability of SNURF to enhance AR-dependent transactivation (p Ͻ 0.01). Deletion of another region containing a stretch of positively charged residues (CS1⌬66 -98) also resulted in a marked reduction in the coactivator function (p Ͻ 0.01), whereas the more C-terminal stretch of negatively charged residues (CS1⌬99 -118) had a smaller effect on AR function. When the same N-terminal deletions of SNURF were studied in CV-1 cells, a similar pattern was seen (Fig. 5B). Deletions affecting the RING finger structure tended to diminish the activity of SNURF in both cells lines, whereas point mutations of the two most C-terminal cysteines (CS2) were tolerated in CV-1 cells but, surprisingly, not in COS-1 cells (Fig. 5), suggesting that cell-specific factors are involved in the RING finger interactions.
To investigate the ability of SNURF to integrate AR-and Sp1-mediated transcriptional responses, pARE 2 -tk-LUC reporter containing two AREs in front of the herpes simplex virus thymidine kinase proximal promoter (Ϫ105 to ϩ51) was transfected to COS-1 cells (Fig. 5C). This portion of herpes simplex virus thymidine kinase promoter contains two GC/GT box sites that are separated by 50 nucleotides. The experiments were carried out in the presence of AR expression, both without and with 100 nM testosterone. Ectopic expression of Sp1 stimulated the reporter gene activity almost 2-fold, and wild-type SNURF and the addition of androgen increased it by 3.8-and 3.3-fold, respectively. There was a more than additive response of pARE 2 -tk-LUC to coexpression Sp1 with SNURF in the presence of AR without androgen (ϳ6.7-fold compared with AR alone, p Ͻ 0.01). In the presence of androgen, ectopic expression of Sp1 more than doubled the activity of the reporter over that achieved by ligand-activated AR alone. Likewise, coexpression of AR and Sp1 together with SNURF resulted in a greater reporter gene activation than could be produced if one of the factors was omitted, suggesting that SNURF is able to enhance the synergistic activity of AR and Sp1. Furthermore, with the SNURF-CS1 mutant, the functional cooperativity was significantly impaired (p Ͻ 0.05) from that with wild-type SNURF, supporting a role for the RING finger in the synergistic response. Very similar results were obtained when CV-1 cells instead of COS-1 cells were used (Fig. 5D).
Mutations in the ZFR and the Hinge Region of AR Weaken the Interaction with SNURF-To identify the residues in the zinc finger and hinge regions of AR important in forming the interaction interface for SNURF, we performed yeast two-hybrid assays by using a panel of AR ZFR/hinge mutants fused to the Lex DNA binding domain together with SNURF (amino acid residues 21-177) tethered to the transcriptional activation domain of VP16 protein. This region of SNURF has previously been shown to encompass the interface for AR interaction (27). The schematic structure of AR ZFR and the mutations studied are illustrated in Fig. 6A. Point mutations in the first and second ZF had varying effects on the recognition of the ZFR by SNURF, and several of the mutations did not markedly impair the interaction. The human AR G568V substitution at the tip of the first ZF causes a partial androgen insensitivity syndrome (51), and the corresponding rat mutation (G551V) rendered the ZFR/hinge unresponsive to SNURF (Fig. 6B). A point mutation R590Q (R607Q in hAR) in the tip of the second ZF, associated with male breast cancer and partial androgen insensitivity syndrome (52)(53)(54), also weakened the SNURF-AR interaction (p Ͻ 0.01). When an additional Y603C substitution was introduced into the sequence (R590Q/Y603C), further attenuation in the interaction was observed. The mutation of the most Cterminal zinc-coordinating cysteine of the second zinc finger (C597G) completely abolished the SNURF-AR interaction. Deletion of the C-terminal part of the bipartite NLS in the mutant ⌬RKLKK (17,18) clearly impaired the interaction (p Ͻ 0.05), and when combined with ZFR substitutions R590Q or R600G/ K601A, the effect became even more profound (Fig. 6B). The expression levels of Lex-AR DNA-binding domain/hinge constructs were controlled by immunoblotting with a monoclonal anti-Lex antibody. No mutant construct was expressed to a level below that of the wild-type AR DNA-binding domain/ hinge construct; in fact, several mutants showed higher protein levels than the wild-type (Fig. 6C). In view of this, it is unlikely that the results depicted in Fig. 6B merely reflect the variable amounts of fusion proteins produced in yeast cells.
GST pull-down experiments assessing physical interaction of SNURF with selected AR ZFR mutants were performed to corroborate the yeast two-hybrid data. As shown in Fig. 7A, in vitro translated SNURF bound efficiently to wild-type AR ZFR. The binding was specific, as SNURF was not retained by GST alone, and control protein (luciferase) did not bind to GST-AR ZFR. In contrast to wild-type ZFR, the mutants ⌬RKLKK and R600G/K601A/⌬RKLKK showed very weak affinity for SNURF in vitro. Also point mutations in the first (G551V) and the second (R590Q) ZF clearly reduced SNURF binding. These results are in line with the yeast two-hybrid data, supporting the notion that the yeast two-hybrid assay reflects direct physical interaction between AR ZFR and SNURF. Collectively, our results indicate that certain AR ZF and hinge region mutations can considerably weaken SNURF-ZFR interaction.
AR ZFR Mutations Attenuate the Transcriptional Response to SNURF-To elucidate the importance of the mutations in-FIG. 6. Interaction of SNURF and AR in yeast is affected by mutations in the AR ZFR and hinge region. A, schematic representation of the rat AR ZFR including 20 residues of the hinge region (amino acids 537-627) used as the target sequence in yeast two-hybrid assays. The P-box involved in sequence-specific DNA binding is indicated as light gray box, and the D-box participating in the dimerization is depicted by cross-hatching. The bipartite NLS is underlined. Mutations analyzed in this study are indicated by arrowheads, and the substituting amino acids are shown. TAD, transcription activation domain. B, two-hybrid analysis of the ZFR and the hinge region with SNURF (amino acids 21-177) in yeast cells. The AR mutants were cloned into LexN-a fusion protein expression vector and transformed into S. cerevisiae strain L40 together with plasmids encoding for VP16 AD or VP16 AD fused to SNURF. ␤-Galactosidase activities were assayed from liquid cultures in three separate experiments, and the activity of yeast transformed with wild-type AR ZFR/hinge construct (LexN-a-ZFR) together with VP16-SNURF was set as 100. The numbers in parentheses depict fold enhancement of SNURF interactions; the activity of ␤-galactosidase in the presence of LexN-a and VP16-SNURF was set as 1. With VP16 alone, the mutated ARs exhibited the same background activity as wild-type AR. C, immunoblot analysis of Lex fusion protein constructs was performed as described under "Experi- terfering with the interaction between AR ZFR and SNURF for activation of AR-dependent transcription, we introduced several of the mutations examined in Fig. 6B to the full-length AR and cotransfected the mutated forms with SNURF-CS1 into COS-1 cells. The mutants R590Q, R590Q/Y603C, and ⌬RKLKK, which had impaired ability to interact with SNURF in the yeast twohybrid or GST pull-down assay, also failed to respond to coexpressed SNURF with the rat probasin promoter as the reporter (Fig. 8A). Interestingly, the R590Q/⌬RKLKK mutant was totally unable to transactivate this promoter. When pARE 2 -TATA-LUC was used as the reporter construct, the mutants R590Q, R590Q/ Y603C, and ⌬RKLKK showed weaker response to SNURF-CS1 than the wild-type AR in that the relative inductions (compared with AR ϩ testosterone) were ϳ1.8-2.7-fold for the mutants and 4.9-fold for wild-type AR (Fig. 8B). The expression levels of all the ZFR/hinge mutants were comparable to that of wild-type AR. 2

DISCUSSION
We have investigated the structural and functional characteristics of a recently identified nuclear protein, SNURF. The protein was discovered in a yeast two-hybrid screen by virtue of its ability to interact with the AR ZFR/hinge region (27). Ectopically expressed SNURF stimulates both steroid receptor-mediated transactivation and, even more efficiently, transcription from promoters driven by Sp1 elements/GC box motifs. However, SNURF does not appear to possess an intrinsic transcription activation function (27). The C-terminal region of SNURF harbors a consensus sequence for a C 3 HC 4 -type zinc finger, the so-called RING finger, which is present in several transcriptional regulatory proteins and has been proposed to participate in protein-protein interactions (55). Interestingly, the RING finger region of MSL2 protein has been shown to possess multiple interaction surfaces, as mutations in the coordination sites for the first (CS1) and the second zinc (CS2) atoms have distinct effects on its interactions and functions (56). Our results suggest that similar to MSL2, the RING finger region of SNURF may not act as a single functional unit, as mutations in the first and the second zinc coordination sites had dissimilar consequences in COS-1 and CV-1 cells.
An interaction between SNURF and Sp1 was detected by coimmunoprecipitation, and this association was strictly dependent on the intact RING finger. SNURF enhanced the transcriptional activity of Sp1 in a synergistic and RING finger-dependent fashion. In accordance with this, SNURF, but not its RING finger mutated form, enhanced the DNA binding activity of GC box-recognizing proteins, especially that of Sp1. This effect was selective for proteins interacting with GC box motifs, as binding of AP1 components or AR was not affected. However, when purified recombinant Sp1 and SNURF proteins were examined by EMSA using an Sp1-binding site oligonucleotide, no clear stimulation of Sp1 DNA binding by SNURF was observed, 3 suggesting that a third partner present in the cell lysate or a post-synthetic modification is involved in the process.
SNURF is highly hydrophilic, and about 30% of its residues are charged. These residues are asymmetrically distributed, possessing potential for strong electrostatic interactions. These features resemble the electrostatic properties of Ubc9, another relatively small nuclear protein that it capable of modulating AR function via interacting with the ZFR/hinge region (57). On the Sp1-regulated minimal promoter, the role of the N-terminal negatively charged regions of SNURF appears to be weakly modulatory, as deletions of these stretches (⌬31-65 and ⌬99 -118) did not cause profound effects. By contrast, the loss of the first 20 amino acids harboring most of the bipartite putative NLS or the deletion of another cluster of basic residues in the ⌬66 -98 mutant resulted in severely impaired transcriptional responses.
In contrast to Sp1-dependent transcription, stimulation of AR-mediated transactivation was not dependent on the Nterminal part of RING finger domain in SNURF. Loss of a cluster of negatively amino acid residues (CS1⌬31-65) was a destructive deletion for the AR coactivator function, whereas this acidic region appeared to have only a small effect on Sp1 function. However, similar to Sp1, deletion of positively charged clusters in SNURF sequence (mutants CS1⌬1-20 and CS1⌬66 -98) also strongly compromised the activity of SNURF on AR. Even though the AR ZFR and the hinge region are sufficient for the interaction of SNURF with AR (27), other regions of the receptor have auxiliary roles; deletion of the core AF-1 region of AR (residues 149 -295) attenuates the response to SNURF, and a larger deletion (⌬46 -408) abolishes it completely. 3 Thus, SNURF cannot substitute for the principal Nterminal transactivation region of AR.
Certain point mutations in both the first and the second ZF, such as G551V and C597G, severely impaired the interaction of FIG. 8. Effects of AR ZFR and hinge region mutations on the transcriptional responsiveness to SNURF. A, COS-1 cells were transiently transfected by the calcium phosphate precipitation method. Five g of the rat probasin reporter pPB(Ϫ285/ϩ32)-LUC, 2 g of pCMV␤, 0.25 g of pSG5-AR, and 5 g of SNURF-CS1, or empty expression vector was used for a 10-cm dish. The reporter gene activity obtained with rAR in the presence of 25 nM testosterone was set as 100. Mutants ⌬RKLKK and R590Q/⌬RKLKK were compared with human AR (hAR), since they were constructed in hAR cDNA. B, experimental conditions were as in panel A, except that pARE 2 -TATA-LUC was used as the reporter. The mean Ϯ S.E. values of at least four experiments are shown. These experiments were also performed by using FuGene reagent with comparable results. AR with SNURF. The G551V substitution, situated in the nonhelical region of ZFR and next to the residues making contact with the phosphate backbone, as predicted for the structure of AR ZFR from the crystal structure of GR ZFR (14), results in impaired DNA binding and transactivation by the receptor (25). The R590Q substitution, situated in the beginning of the second ␣-helix of ZFR and in a close proximity of the dimer interface, impaired the AR-SNURF interaction and attenuated the response to cotransfected SNURF in mammalian cells. Also the C-terminal part of the bipartite nuclear localization signal in the hinge region (deleted in the mutant ⌬RKLKK), which corresponds to a disordered area in the GR ZFR crystal structure, appears to be important for AR-SNURF interaction. The latter mutation also weakened the stimulation of AR transactivation by SNURF in the context of minimal ARE-containing promoter and abolished the response entirely when the natural AR-regulated rat probasin promoter was used. Even though the bipartite NLS region of AR is involved in AR-SNURF interaction, the coregulatory role of SNURF is probably not linked to nuclear transport of AR, since ectopic SNURF expression does not increase the amount of nuclear AR protein in mammalian cells (27).
The DNA binding-independent role of steroid receptor ZFRs in transcriptional regulation has gained more attention during the last few years, as these domains have been shown to exhibit the potential for multiple protein interactions. Concomitantly, the importance of the adjacent hinge region has evolved from a passive spacer element to an active participant in making contacts with auxiliary proteins. The ZFR and hinge region are involved in interactions with corepressors SMRT and N-CoR (26,58), as well as in contacts with octamer transcription factors 1 and 2 (Oct-1/2), Brn-3a/3b, and several novel coregulators (26 -33, 57). The histone acetylase PCAF that associates with coactivators CREB-binding protein and p300 was demonstrated to bind directly to the ZFRs of nuclear receptors (59), suggesting that coactivators interacting with distinct domains of nuclear receptors can participate in the formation of common protein complexes. Interestingly, N-CoR binding to the aporetinoic acid receptor ␣ hinge region prevents PCAF association, suggesting that this corepressor either sterically blocks the access of PCAF to the ZFR or induces a conformation that does not allow the ZFR-PCAF contacts. SNURF may play an analogous role in that it may act in concert with other coactivators and activate steroid receptor-mediated transcription by preventing corepressor binding to the hinge region. Recent results have demonstrated that nuclear receptors and Sp1 can utilize common coactivator complexes (60 -63).
In conclusion, the results of the present work indicate that SNURF, a coregulator recognizing the ZFR/hinge region of AR and other steroid receptors (27), participates in the Sp1-dependent transcription and cooperates with AR and Sp1, suggesting a novel transcriptional link between the function of Sp1 and steroid receptors. Identification of other interaction partners of SNURF and genetic studies are, however, needed for better understanding the role of this protein in transcriptional regulation.