SUMO Modification of Repression Domains Modulates Function of Nuclear Receptor 5A1 (Steroidogenic Factor-1)*

Steroidogenic factor 1 (SF-1 or NR5A1), is a Ftz-F1 member of the nuclear receptor superfamily that plays essential roles in endocrine development, steroidogene-sis, and gonad differentiation. We investigated modifi-cations that control SF-1 function and found that SF-1 could be conjugated by SUMO-1 both in vitro and in vivo . SF-1 was modified predominantly at Lys 194 and much less at Lys 119 when free SUMO-1 was supplied. Mutations of Lys 194 and Lys 119 enhanced transcriptional activity of SF-1, although the DNA binding activity of SF-1 was not affected. Sequences around Lys 194 and Lys 119 both repressed transcription intrinsically. The Lys 194 motif repressed transcription more efficiently than the Lys 119 domain, consistent with its ability to be a better substrate for SUMO conjugation. Thus, SUMO modification of SF-1 correlates with transcriptional repression. Wild-type but not conjugation-deficient SF-1 was localized at the nuclear speckles together with SUMO-1. Thus, SUMO-1 conjugation could also target SF-1 into nuclear speckles. Collectively, these results suggest that SUMO modification at the repression domains targets SF-1 to nuclear speckles; this could be an important mechanism by which SF-1 is regulated. Steroidogenic factor 1 (SF-1), 1 Ad4BP, the NR5A1 sumoylation-deficient SF-1 mutant, to nuclear speckles. These results indicate that sumoylation at repression domains is an important posttranslational modification for the transcriptional activity of SF-1. described previously (4). Rat anti-HA antibody (clone 3F10) was obtained from Roche Applied Science. Transcriptional Activity Assay— For transfection into NIH-3T3 or Y1 cells in 24-well culture plates, 100 ng of SF-1 expression plasmid was co-transfected with 100 ng of pS2.3H-Luc or 5xGAL4-E1B-Luc , respec- tively. Luciferase activities were assayed 24 h after transfection. The firefly luciferase activities were normalized with Renilla luciferase ac- tivities from phRLuc, used as an internal control. The results are the average of three independent transfection experiments, with error bars representing standard deviations. In Vitro Protein Binding Assay—In vitro GST pull-down assays to measure protein binding were performed as described previously (8), with some modifications. Briefly, in vitro transcribed/translated SF-1- FH(S) fragment or full-length SF-1 was incubated with 1 (cid:4) g of purified GST, GST-UBC9, or GST-TFIIB in NENT buffer (20 m M Tris-HCl, pH 8.0, 0.5% Nonidet P-40, 5 m M EDTA, and 150 m M NaCl). The protein complexes were then incubated with glutathione-coated beads. After extensive washing, the samples were separated by 12% polyacrylamide gel and detected by autoradiography. In Vitro and in Vivo Sumoylation Assay— The substrate proteins, luciferase and SF-1-HA, were in vitro transcribed/translated to generate radiolabeled proteins. Two microliters of in vitro translation products were added to sumoylation buffer (4 m M MgCl 2 , 1 m M dithiothreitol, and 2 m M ATP) containing 100 ng of SAE1/SAE2 and 1 (cid:4) g of His-SUMO-1 in the presence or absence of 1 (cid:4) g of His-UBC9 at 30 °C for 30 min. The reaction samples were separated by 10% polyacrylamide gel followed by autoradiography for 24 h. To Subsequently, the cells were immunostained with an anti-SF-1 anti- serum (1:100) in blocking buffer at 4 °C for 16 h. The cells were washed with PBST and stained with Cy3-conjugated goat anti-rabbit IgG in blocking buffer for 1 h atroom temperature. The coverslips were washed three times with PBST and stained with 10 mg/ml Hoechst 33258 for 15 min. After three extensive washes, the coverslips were mounted on glass slides in 50% glycerol/phosphate-buffered saline. Fluorescent cells were examined with a Zeiss LSM 510 confocal microscope.

Steroidogenic factor 1 (SF-1 or NR5A1), is a Ftz-F1 member of the nuclear receptor superfamily that plays essential roles in endocrine development, steroidogenesis, and gonad differentiation. We investigated modifications that control SF-1 function and found that SF-1 could be conjugated by SUMO-1 both in vitro and in vivo. SF-1 was modified predominantly at Lys 194 and much less at Lys 119 when free SUMO-1 was supplied. Mutations of Lys 194 and Lys 119 enhanced transcriptional activity of SF-1, although the DNA binding activity of SF-1 was not affected. Sequences around Lys 194 and Lys 119 both repressed transcription intrinsically. The Lys 194 motif repressed transcription more efficiently than the Lys 119 domain, consistent with its ability to be a better substrate for SUMO conjugation. Thus, SUMO modification of SF-1 correlates with transcriptional repression. Wild-type but not conjugation-deficient SF-1 was localized at the nuclear speckles together with SUMO-1. Thus, SUMO-1 conjugation could also target SF-1 into nuclear speckles. Collectively, these results suggest that SUMO modification at the repression domains targets SF-1 to nuclear speckles; this could be an important mechanism by which SF-1 is regulated.
Steroidogenic factor 1 (SF-1), 1 also known as Ad4BP, is the NR5A1 member of the nuclear receptor superfamily (1). In mice, it is expressed in the pituitary, hypothalamus, gonads, and adrenal glands, contributing to their development and differentiation (2); in zebrafish, SF-1 also plays a role in the development of interrenal glands (3). SF-1 regulates the expression of steroidogenic genes such as CYP11A1, CYP19, Mü llerian inhibitory substance, and ␣and ␤-subunit of gonadotropins (2,4). In addition, SF-1 possesses the ability to direct cultured embryonic stem cells toward to the steroidogenic cell lineage (5). Due to its functional importance, the transcriptional regulation of SF-1 is of considerable interest. SF-1 contains an N-terminal DNA-binding domain, a proline-rich domain, and a C-terminal ligand-binding domain. The C-terminal region of its DNA-binding domain is a conserved 30-amino acid basic Ftz-F1 (Fushi-tarazu factor 1) box shared by all members of the NR5 (FTZ-F1) subfamily (6,7). An FP domain (aa 78 -172), including the Ftz-F1 box and a prolinerich sequence, contains a nuclear localization signal and interacting regions for TFIIB and c-Jun and is important for the transactivation function of SF-1 (8). Unlike other nuclear receptors, SF-1 lacks the N-terminal A/B region, which usually contains the ligand-independent transactivation AF-1 domain. Amino acids 180 -265 in the hinge region of SF-1 may function as AF-1. In this domain, the single serine residue (Ser 203 ) is phosphorylated and required for activation by mitogen-activated protein kinase (9). However, phosphorylation of SF-1 at Ser 203 does not respond to adrenocorticotropin and gonadotropin signals. The functional significance of this putative AF-1 domain remains unclear. Another activation domain, AF-2, is located at the C terminus of SF-1; this domain interacts with transcription co-activators such as steroid receptor co-activator 1 (10), glucocorticoid receptor-interacting protein 1 (9), and p300/CBP/co-integrator-associated protein (11).
Recently, SUMO-1 (also known as Pic1, Ubl1, hSmt3, and sentrin) conjugation (sumoylation) has been reported to play an important role in many cellular processes (12). Sumoylation resembles ubiquitination, but the enzymes involved in these two processes are distinct. During sumoylation, four C-terminal amino acids of the precursor of SUMO-1 are removed to expose the conjugation site to the E1 activating enzyme. Subsequently, the activated SUMO-1 is transferred to the E2conjugating enzyme UBC9. Although E1 and E2 enzymes are sufficient to trigger sumoylation in vitro, SUMO-1 is conjugated to target proteins by E3 ligase in vivo. Analysis of sumoylated proteins has revealed that SUMO-1 is conjugated to substrate at the consensus sequence KXE ( is any hydrophobic amino acid, and X is any amino acid).
In the present report, we show that SF-1 is sumoylated at Lys 119 and Lys 194 . Mutations of sumoylation sites enhance the transcriptional activity of SF-1 but do not affect the DNA binding affinity of SF-1. The sumoylation sites of SF-1 function as intrinsic repression domains. Moreover, overexpression of SUMO-1 directs wild-type (WT) SF-1, but not sumoylationdeficient SF-1 mutant, to nuclear speckles. These results indicate that sumoylation at repression domains is an important posttranslational modification for the transcriptional activity of SF-1.
Transcriptional Activity Assay-For transfection into NIH-3T3 or Y1 cells in 24-well culture plates, 100 ng of SF-1 expression plasmid was co-transfected with 100 ng of pS2.3H-Luc or 5xGAL4-E1B-Luc, respectively. Luciferase activities were assayed 24 h after transfection. The firefly luciferase activities were normalized with Renilla luciferase activities from phRLuc, used as an internal control. The results are the average of three independent transfection experiments, with error bars representing standard deviations.
In Vitro Protein Binding Assay-In vitro GST pull-down assays to measure protein binding were performed as described previously (8), with some modifications. Briefly, in vitro transcribed/translated SF-1-FH(S) fragment or full-length SF-1 was incubated with 1 g of purified GST, GST-UBC9, or GST-TFIIB in NENT buffer (20 mM Tris-HCl, pH 8.0, 0.5% Nonidet P-40, 5 mM EDTA, and 150 mM NaCl). The protein complexes were then incubated with glutathione-coated beads. After extensive washing, the samples were separated by 12% polyacrylamide gel and detected by autoradiography.
In Vitro and in Vivo Sumoylation Assay-The substrate proteins, luciferase and SF-1-HA, were in vitro transcribed/translated to generate radiolabeled proteins. Two microliters of in vitro translation products were added to sumoylation buffer (4 mM MgCl 2 , 1 mM dithiothreitol, and 2 mM ATP) containing 100 ng of SAE1/SAE2 and 1 g of His-SUMO-1 in the presence or absence of 1 g of His-UBC9 at 30°C for 30 min. The reaction samples were separated by 10% polyacrylamide gel followed by autoradiography for 24 h. To examine SF-1 sumoylation in cells, SF-1-HA was co-expressed with SUMO-1 in 293T cells as indicated. Twenty-four hours after transfection, cells were washed twice with ice-cold phosphatebuffered saline containing 20 mM N-ethylmaleimide. Whole cell extracts were prepared in IPH buffer (50 mM Tris-HCl, pH 8.0, 150 mM NaCl, 5 mM EDTA, 0.5% Nonidet P-40, and 1ϫ complete protease inhibitor mixture) containing 20 mM N-ethylmaleimide and subjected to immunoprecipitation with anti-HA antibody. The anti-HA immunoprecipitates were fractionated by 10% polyacrylamide gel electrophoresis and transferred to a polyvinylidene difluoride membrane. The membrane was analyzed by anti-HA immunoblotting, stripped, and then re-blotted with anti-SUMO-1 antibody (Chemicon, Temecula, CA) and visualized by chemiluminescence detection.
Electrophoretic Mobility Shift Assay-Electrophoretic mobility shift assay was performed as described previously (24), with some modifications. SF-1-containing samples were prepared from in vitro synthesis/ sumoylated assay without [ 35 S]methionine and incubated with 32 Plabeled oligonucleotide containing the SF-1 binding sequence on ice for 30 min. Subsequently, the samples were separated by 5% native acrylamide gel followed by autoradiography. The intensity of each band was quantified using Image Gauge Version 3.2 software with a FujiFilm LAS-1000plus image reader.
Immunostaining and Confocal Microscopy-Y1 cells were cultured on coverslips at a density of 1ϫ 10 4 cells/well in 6-well plates. Expression plasmids were transfected into cells as indicated. Twenty-four hours after transfection, cells were fixed in 3.7% formaldehyde for 10 min, permeabilized with 0.2% Triton X-100/phosphate-buffered saline (PBST), and blocked in 2% blocking reagent (Roche Applied Science).
Subsequently, the cells were immunostained with an anti-SF-1 antiserum (1:100) in blocking buffer at 4°C for 16 h. The cells were washed with PBST and stained with Cy3-conjugated goat anti-rabbit IgG in blocking buffer for 1 h at room temperature. The coverslips were washed three times with PBST and stained with 10 mg/ml Hoechst 33258 for 15 min. After three extensive washes, the coverslips were mounted on glass slides in 50% glycerol/phosphate-buffered saline. Fluorescent cells were examined with a Zeiss LSM 510 confocal microscope.

SF-1 Is Sumoylated in Vitro and in Vivo-
To understand the components that regulate SF-1 action, we searched for SF-1interacting proteins by yeast two-hybrid screening using the FH(S) fragment (aa 78 -213) of SF-1 as bait. During this screening, UBC9 was identified as one of the many proteins that interacted with SF-1 (data not shown). An in vitro protein binding assay further confirmed this interaction (Fig. 1). Both the SF-1-FH(S) fragment and full-length SF-1 interacted with UBC9. SF-1-FH(S) and SF-1 also interacted with TFIIB ( Fig. 1,  lanes 3 and 5), as demonstrated previously (8). They did not interact with GST (Fig. 1, lane 2), which was used as a negative control. The addition of 25-hydroxycholesterol, a proposed ligand for SF-1, did not affect the interaction between SF-1 and UBC9 (Fig. 1, lane 4). Because UBC9 is the single E2-conjugating enzyme for SUMO-1, and many sumoylation substrates interact with UBC9 (25), this result led us to investigate SF-1 conjugation by SUMO-1.
To determine whether SF-1 can be conjugated by SUMO-1, in vitro translated, 35 S-labeled SF-1-HA was incubated with SUMO-1 and the E1-activating enzyme in the presence or absence of UBC9. Luciferase ( Fig. 2A, Luc), a negative control, was not modified, regardless of matter whether UBC9 was present or absent (lanes 1 and 2). Whereas SF-1 remained unmodified in the absence of UBC9 (Fig. 2A, lane 3), in the presence of UBC9 two slow-migrating bands of 73 and 90 kDa appeared in addition to SF-1-HA (lane 4). The slow-migrating bands indicated the presence of SUMO-1-conjugated SF-1. This result indicated that SF-1 was modified by SUMO-1 in vitro.
To examine SUMO-1 conjugation of SF-1 in cells, SF-1-HA was co-expressed with GFP-or YFP-tagged SUMO-1 in 293T cells. After immunoprecipitation with anti-HA antibody, the presence of SF-1-HA was detected by immunoblotting with an anti-HA antibody (Fig. 2B, top panel). In the absence of SF-1-HA transfection, no band was detected, indicating the specificity of the immunoprecipitation and immunoblotting (Fig. 2B,  lanes 1 and 2). The SF-1-HA band appeared when it was co-transfected with GFP alone (Fig. 2B, lane 3), whereas an additional slow-migrating band besides SF-1-HA was detected

SF-1 Sumoylation
in the presence of GFP-or YFP-tagged wild-type SUMO-1 (lanes 4 and 6). These slow-migrating bands were further confirmed as SUMO-SF-1 by anti-SUMO-1 immunoblotting (Fig.  2B, bottom panel). To verify the involvement of SUMO-1, Gly to Ala mutation was performed at the 97th Gly residue of SUMO-1, which is essential for the ability of SUMO-1 to conjugate to substrates (25). When the resulting GFP-SUMO-1 GA mutant was co-transfected instead of wild-type GFP-SUMO-1, no additional band was detected (Fig. 2B, lane 5). This indicated the specificity of SUMO-1 conjugation. Unlike the in vitro assay, only one slow-migrating band of SF-1 was detected in the in vivo analysis. These results indicated that although SF-1 can be conjugated by two SUMO-1 molecules in vitro, it is probably sumoylated at only one site in vivo.
To map the site of SUMO conjugation, we analyzed SF-1 for the consensus sumoylation KXE sequence and found Lys 119 in the sequence of FKLE and Lys 194 in the sequence of IKSE of SF-1 to be potential sumoylation sites (Fig. 2C). To determine whether these two sites are indeed SUMO-conjugated, Lys to Arg mutagenesis was carried out at both sites to generate the K119R, K194R, and K119R/K194R mutants. The mutants were analyzed using both in vitro and in vivo sumoylation assays. In vitro sumoylation reactions revealed that whereas two slowmigrating SUMO-SF-1 bands were detected in the presence of wild-type SF-1 (Fig. 2D, lane 2 4 and 5). The absence of modification in the K194R mutant indicated that Lys 194 of SF-1 is a major conjugation site. In the absence of SUMO-conjugated Lys 194 , Lys 119 can not be conjugated. Therefore, Lys 119 is a minor conjugation site, which may depend on prior conjugation at Lys 194 in vitro. Analysis of these mutants in 293T cells also revealed similar results, except that Lys 119 was not used for SUMO conjugation in vivo (Fig. 2E). Collectively, these results showed that SF-1 is sumoylated predominantly at Lys 194 in vitro and in vivo.
Sumoylation Sites of SF-1 Are Intrinsic Repression Domains-To determine the effect of sumoylation on SF-1, we first checked whether sumoylation affects SF-1 DNA binding activity. WT and mutated SF-1-HA (K119R, K194R, and K119R/K194R) were synthesized and SUMO-conjugated in vitro before being subjected to DNA binding and electrophoretic mobility shift assays. As calculated from the anti-HA immunoblot, about 70% of wild-type SF-1-HA and the K119R protein were modified in vitro (Fig. 3, top panel, lanes 4 and 6). However, the intensities of the DNA-protein complexes of sumoy-lated and non-sumoylated SF-1 were similar (Fig. 3, bottom  panel). Therefore, sumoylation does not appear to affect the DNA binding activity of SF-1.
To examine the effect of SUMO conjugation on SF-1 transcriptional activity, wild-type or sumoylation-deficient SF-1 mutants were co-transfected with a SF-1-dependent reporter gene in NIH-3T3 cells. As shown in Fig. 4A, wild-type SF-1 activated the expression of the reporter; however, all of the sumoylation-deficient mutants (K119R, K194R, and K119R/ K194R) were even more active. Because SF-1 is not endogenous to NIH-3T3 cells, we examined the activity of mutant SF-1 in a physiologically relevant cell line, Y1. Because Y1 expresses abundant SF-1, to avoid interference of endogenous SF-1, we examined the ability of exogenous GAL4-SF-1 fusion proteins (GAL4-DBD fused to aa 110 -462 from wild-type or sumoylation-deficient SF-1) to activate a GAL4-dependent reporter gene (Fig. 4B). The activity of wild-type GAL4-SF-1 (Fig. 4B, WT) was very low in Y1 cells. The activities of all the GAL4-SF-1 mutants were enhanced, and the K119R, K194R, and K119R/K194R mutants had activities 2-, 39-, and 12-fold greater than the WT activity, respectively. These results indi- FIG. 4. Sumoylation sites of SF-1 are repression domains. A, mutation of sumoylation sites increases the activity of SF-1. The luciferase activity of SF-1-dependent reporter pS2.3H-Luc was determined after co-transfection with expression plasmids for wild-type or mutated SF-1 in NIH-3T3 cells. B, mutation of sumoylation sites increases the activity of Gal4-SF-1. The activities of wild-type and mutated Gal4-SF-1 fusion proteins are shown after they were co-transfected with 5xGAL4-E1B-Luc and luciferase activities were measured in Y1 cells. C, sumoylation sites of SF-1 are repression domains. GAL4-VP16 fusion protein constructs are shown in the left panel. Wild-type or mutated sequences (12 amino acids) around the sumoylation sites of SF-1 were inserted between the GAL4-DBD and VP-16 activation domain of all constructs; residue numbers in SF-1 are indicated. The transcriptional activities of GAL4-fusion proteins were assayed with the 5xGAL4-E1B-Luc reporter in Y1 cells. The relative luciferase activities are shown in the right panel. ‫,ء‬ p Ͻ 0.05; ‫,ءء‬ p Ͻ 0.01. cated that the regions surrounding Lys 119 and Lys 194 function as repression domains for SF-1.
To investigate the SF-1 repression activity around its SUMO conjugation sequences, we inserted peptides with either wildtype or mutant sequences centered around Lys 119 or Lys 194 sites of SF-1 into a transcriptional activator composed of a DNA-binding domain from GAL4 and an activation domain from VP16 (Fig. 4C, left panel). The resulting P-119WT and P-194WT plasmids encoded proteins with the wild-type Lys 119 and Lys 194 motifs, respectively, whereas P-119R and P-194R contained the mutated sequences. After transfection, P-119WT and P-194WT repressed reporter gene expression by 15% and 48%, respectively (Fig. 4C, right panel). This repression was statistically significant (p Ͻ 0.05 and p Ͻ 0.01, respectively). In contrast, P-119R and P-194R had the same activity as the control. Thus, the mutants lost the ability to repress transcription. This result indicated that both Lys 194 and Lys 119 lie in intrinsic repression domains, and the repression domain at Lys 194 is stronger than that at Lys 119 .
SF-1 Is Co-localized with SUMO-1 in Nuclear Speckles-SUMO-1 conjugation affects the localization of some substrate proteins (25). To study whether sumoylation regulates the localization of SF-1, GFP-tagged SF-1 and RFP-tagged SUMO-1 were co-expressed in Y1 cells. Consistent with other reports, wild-type RFP-SUMO-1 formed many nuclear speckles in transfected cells (Fig. 5A). In about 60% of transfected cells, wild-type SF-1-GFP was located in nuclear speckles in the presence of wild-type SUMO-1 (Fig. 5A, top panels). In contrast to wild-type SUMO-1, the conjugation-deficient SUMO-1 mutant, RFP-SUMO-1 AA, did not form nuclear speckles, and the co-expressed SF-1 was found homogenously in the nucleus (Fig.  5A, middle panels). To ascertain that localization of SF-1 to the nuclear speckles is due to SUMO-1 conjugation, the SF-1 mutant K194R, which cannot be conjugated, was co-expressed with wild-type RFP-SUMO-1. This SF-1-GFP K194R was homogenously distributed in the nucleus of all examined transfected cells, whereas the co-transfected SUMO-1 still formed nuclear speckles (Fig. 5A, bottom panels). Thus, the lack of SUMO-1 conjugation renders SF-1-GFP unable to go into nuclear speckles. These data indicate that targeting of SF-1-GFP into nuclear speckles by SUMO-1 requires intact conjugation sites in both SUMO-1 and SF-1-GFP.
Because the localization of exogenous SF-1-GFP is affected by SUMO-1 conjugation, we next examined whether the localization of endogenous SF-1 was also affected by SUMO-1. We transfected YFP-SUMO-1 into Y1 cells and then examined the location of SF-1 by indirect immunofluorescence (Fig. 5B). In some cells that expressed YFP-SUMO-1, nuclear speckles formed, and endogenous SF-1 was localized in these speckles (Fig. 5B, arrows). When YFP-SUMO-1 was absent, SF-1 was evenly distributed in the nucleus (Fig. 5B). These data suggest that overexpression of SUMO-1 can direct SF-1 to nuclear speckles formed by SUMO-1.

DISCUSSION
In this report, we demonstrate that SF-1 is modified by SUMO-1 in vitro and in vivo. We detect two SUMO conjugation sites, Lys 119 and Lys 194 , in SF-1. SUMO conjugation does not affect the DNA binding activity of SF-1 but is associated with transcriptional repression. In addition, SUMO-1 conjugation modulates the localization of SF-1 in the nucleus. These results suggest that SUMO modification at the repression domains is an important posttranslational regulation of SF-1 activity.
SF-1 is Modified by SUMO-1-We investigated SF-1 sumoylation because SF-1 interacts with UBC9, and many UBC9interacting proteins are sumoylated (12). In the present study, we demonstrate that SF-1 interacts with E2-conjugating en-zyme UBC9 in vitro and is a substrate for sumoylation in vitro and in vivo.
The potential sumoylation sites in SF-1 are located at Lys 119 and Lys 194 . Our data showed that Lys 194 was efficiently conjugated by SUMO-1 in vitro and that Lys 119 was poorly conjugated by SUMO-1 in vitro (Fig. 2D). In vivo, only Lys 194 was sumoylated (Fig. 2E). This indicates that Lys 119 is inefficiently sumoylated, although its sequence matches sumoylation consensus well. This suggests that regulation of SUMO conjugation depends on many factors in addition to pure substrate recognition. The large difference in the sumoylation efficiency of Lys 194 and Lys 119 could be of relevant importance to physiological condition. Similar results were also observed with c-Myb, which is also unequally sumoylated at two different sites (26).
Sumoylation Sites Are Intrinsic Repression Domains-The single SF-1 sumoylation site of Lys 194 can repress transcription more efficiently than Lys 119 (Fig. 4C). In our experiments, FIG. 5. SF-1 is targeted into nuclear speckles through SUMO-1 conjugation. A, exogenous SF-1-GFP co-localizes with SUMO-1. GFPtagged wild-type or K194R mutant SF-1 and RFP-tagged wild-type or mutant SUMO-1 were co-expressed in Y1 cells as indicated. Cells were fixed after 24 h, and confocal images of fluorescence are shown. B, SUMO-1 can direct endogenous SF-1 into nuclear speckles. After expression of YFP-tagged SUMO-1 in Y1 cells, endogenous SF-1 was visualized by indirect immunofluorescence with an anti-SF-1 antiserum followed by Cy3-conjugated anti-rabbit IgG antibody. DNA was stained by Hoechst 33258. Arrows indicate YFP-SUMO-1-expressing cells. mutation of Lys 194 enhanced the transcriptional activity of full-length SF-1 by 2-fold and of GAL4-SF-1 by 39-fold. Mutation of Lys 119 also enhanced the transcriptional activity of SF-1, although at a much reduced level. One would expect that the K119R/K194R mutant would have an even higher activity than that of the single mutants, K119R and K194R, but our data showed that the activities of the double mutant were between those of the single mutants (Fig. 4, A and B). We do not know how to explain this because the mechanism of this repression is still unclear. The Lys 194 sequence directly interacts with a DEAD box-containing protein, DP103, which represses the activity of SF-1 (27). It will be interesting to find out whether sumoylation of Lys 194 is related to DP103 interaction or to recruitment of other co-repressor proteins. Furthermore, this SUMO conjugation motif also overlaps with the synergistic control motif, which has been suggested to modulate higherorder interactions among transcriptional regulators (28). These observations reveal that Lys 194 could be an important and highly regulated domain involved in interactions among transcriptional regulators.
SF-1 Is Localized to SUMO-1-formed Nuclear Speckles-It has been proposed that SUMO-1 conjugation targets proteins to different cellular localizations. For example, sumoylation directs RanGAP1 from the cytoplasm to the nuclear membrane (14). SUMO-1 modification of promyelocytic leukemia and p53 also directs them into promyelocytic leukemia nuclear bodies (20,29). Besides this, SUMO-1-conjugated homeodomain-interacting protein kinase 2 is concentrated in nuclear speckles, which are distinct from promyelocytic leukemia bodies (30). Like most transcription factors, SF-1 is evenly distributed in the nucleus. Both endogenous and exogenous SF-1 can be directed into nuclear speckles in the presence of SUMO-1, but not in the presence of conjugation-deficient SUMO-1. The SF-1 SUMO conjugation-deficient mutant also cannot be directed into nuclear speckles (Fig. 5A). This re-localization of SF-1 may be associated with transcriptional repression. Therefore, we propose that sequestration of SF-1 from the nucleoplasm might be a mechanism regulating the transcriptional activity of SF-1.
Based on our data, SUMO conjugation sites of SF-1 are transcriptional repression domains. Repression could be achieved by two separate means: (a) directly, probably through recruitment of co-repressor proteins to the transcriptional activation complex and/or disruption of the transcriptional activation complex; and (b) through the sequestration of SF-1 to nuclear speckles, thus removing it from transcriptionally active sites. Both possibilities are not mutually exclusive, and more experiments are needed to test these possibilities.