Sterol regulatory element-binding proteins are negatively regulated through SUMO-1 modification independent of the ubiquitin/26 S proteasome pathway.

Sterol regulatory element-binding proteins (SREBPs) are major transcription factors that activate the genes involved in cholesterol and fatty acid biosynthesis. We here report that the nuclear forms of SREBPs are modified by the small ubiquitin-related modifier (SUMO)-1. Mutational analyses identified two major sumoylation sites (Lys(123) and Lys(418)) in SREBP-1a and a single site (Lys(464)) in SREBP-2. Mutant SREBPs lacking one or two sumoylation sites exhibited increased transactivation capacity on an SREBP-responsive promoter. Overexpression of SUMO-1 reduced whereas its dominant negative form increased mRNA levels of SREBP-responsive genes. Nuclear SREBPs interacted with the SUMO-1-conjugating enzyme Ubc9, and overexpression of a dominant negative form of Ubc9 increased the mRNA levels of SREBP-responsive genes. Pulse-chase experiments revealed that sumoylation did not affect the degradation of SREBPs through the ubiquitin-proteasome pathway. In vitro ubiquitylation assay showed no competition between ubiquitin and SUMO-1 for the same lysine. Considered together, our results indicate that SUMO-1 modification suppresses the transactivation capacity of nuclear SREBPs in a manner different from the negative regulatory mechanism mediated by proteolysis.

SREBPs 1 control the transcription of a number of genes encoding enzymes and proteins involved in cholesterol and fatty acid metabolism (1). These transcription factors belong to a large class of transcription factors containing a basic helixloop-helix leucine zipper (bHLH-Zip) motif. The SREBP family comprises three subtypes: SREBP-1a and SREBP-1c, which are generated by alternative splicing, mainly regulating lipogenic gene expression, and SREBP-2 governing cholesterol metabolism. Unlike other members of the bHLH-Zip transcription factors, the SREBPs are synthesized as membrane-bound precursors on the endoplasmic reticulum (ER) and activated by a two-step proteolytic process (2)(3)(4). The precursor proteins con-tain an N-terminal transcriptional activation domain with a bHLH-Zip motif and a C-terminal regulatory domain separated by two transmembrane regions. The C-terminal regulatory domain associates with SREBP cleavage-activating protein (SCAP), an ER membrane protein with eight membranespanning segments, which contains a sterol-sensing domain (5). An SREBP⅐SCAP complex remains on the ER membrane as long as intracellular cholesterol levels are high, whereas in cells depleted of cholesterol ER-derived membrane vesicles containing this complex moves to the Golgi, where a sequential cleavage of the SREBPs by site 1 and site 2 protease occurs, releasing the active nuclear forms (6). Once the nuclear form of SREBPs is released into the cytoplasm, it is actively transported into the nucleus in an importin ␤-dependent manner (7). In the nucleus, the SREBPs are modified by polyubiquitin chains and rapidly degraded by the 26 S proteasome (8). In the presence of proteasome inhibitors, ALLN and lactacystin, the stabilized nuclear SREBPs are capable of enhancing their responsive gene expression. Thus, ubiquitylation of the nuclear SREBPs and the subsequent turnover play important roles in regulation of lipid metabolism.
Posttranslational modification of a variety of cellular proteins has been variably linked to protein phosphorylation and acetylation other than ubiquitylation. SUMO-1, a 101-amino acid protein bearing 18% identity with ubiquitin but with a remarkably similar secondary structure, has been recently identified. SUMO-1 differs from ubiquitin in its surface-charge distribution, eliciting its specificity (9), and does not have a consensus sumoylation motif, (I/V/L)KX(E/D), in its molecule, explaining why SUMO-1 does not make multichain forms (10,11).
Sumoylation requires a multiple-step reaction similar to that of ubiquitin, but the specific enzymes are distinct from those involved in ubiquitylation (12). SUMO-1 is synthesized as a precursor with the C-terminal extension of several amino acids, which needs to be processed to expose the C-terminal Gly 97 residue that is essential for conjugation to target proteins (13). Then the processed SUMO-1 is recognized as a substrate by SUMO-activating enzyme (E1), which is a heterodimer consisting of SAE1 (also called Uba2) and SAE2 (also called Aos1) subunits (14). Ubc9 is a SUMO-conjugating enzyme (E2), receiving SUMO-1 from the E1 enzyme and transferring it to target proteins (15). Most sumoylated proteins directly interact with Ubc9, which catalyzes the sumoylation of such proteins (16). A recent report showed that Ubc9 recognizes the consensus sequence that surrounds the acceptor lysine residue in sumoylation substrates (17), implying that Ubc9 itself might play to a certain extent a ubiquitin E3-like role in determining the substrate specificity (18). However, recent studies have identified ubiquitin ligase (E3)-like ligases for sumoylation that enhance SUMO-1 conjugation to target proteins in yeasts and mammals (19).
In recent years, a growing number of SUMO-1 target proteins including several transcription factors have been reported (20). In contrast to ubiquitylation, which usually marks proteins for rapid degradation, sumoylation is involved in the regulation of protein functions through changes in proteinprotein interactions (21,22), subcellular localization (23), and antagonism to ubiquitylation. Thus, sumoylation serves to enhance the stabilization of target proteins (24) and inhibit DNA repair (25).
The effects of SUMO-1 modification of transcription factors are very diverse, depending on the nature of transcription factors. SUMO-1 modification induces nuclear relocalization of p53 and enhances the DNA-binding ability of heat shock transcription factor 1 and 2, resulting in increased transcriptional activities (26 -28). In contrast, sumoylation to p73␣ does not affect its transcriptional activity but rather its subcellular localization (29), whereas SUMO-1 modification attenuates the transcriptional activities of c-Myb, c-Jun, and androgen receptor (30 -32). Thus, SUMO-1 modification might be a common mechanism that regulates specific activities of transcriptional factors.
Here we report a novel posttranslational modification of SREBP-1 and SREBP-2 by the covalent attachment of two molecules and a single molecule of the SUMO-1 protein, respectively. We show that sumoylation does not affect ubiquitylation and the stability but does modulate transcriptional activities of the SREBPs. Our results point to another mechanism through which the SREBPs activities are attenuated in conjunction with degradation by the ubiquitin/26 S proteasome pathway.
The pEGFP-SUMO-1 was a kind gift from Dr. Minoru Yoshida (RIKEN). The pHA-SUMO-1 was kindly provided by Dr. Chiba (Tokyo Metropolitan Institute of Medical Science). The expression plasmid pHisSUMO was constructed by inserting a fragment encoding human SUMO-1 cloned into the pME-His (8). Expression plasmids pHis-SUMO(GG) and pHisSUMO⌬GG were generated by ligating PCR-generated fragments. To construct expression plasmids pHisUbc9 and pG-FPUbc9, a fragment of human Ubc9 was amplified by reverse transcriptase-PCR and inserted into the pME-His and the pME-GFP (8). The expression plasmid pHisUbc9(C93S) was generated using the site-directed mutagenesis kit. To generate the pG5Luc reporter plasmid, five copies of Gal4 binding sites in the pG5CAT (Clontech) were transferred to the pGL3-Basic (Promega, Madison, WI).
Cell Lines and Culture Conditions-HeLa, COS-1, HEK293 and M19 cells and site 2 protease null mutant Chinese hamster ovary cells, which were kindly provided by Dr. Chang (Dartmouth College, Hanover, NH), were cultured as described previously (8,36).
Immunoprecipitation and Immunoblotting-For detection of sumoylated endogenous SREBPs, HeLa cells (20 100-mm dishes) were set up on day 0 in medium A (Dulbecco's modified Eagle's medium; Sigma) supplemented with 10% fetal bovine serum supplemented with 10% fetal bovine serum. On day 1, the cells were transfected with 3 g of pHA-SUMO-1 using X-tremeGENE Q2 Transfection Reagent (Roche Applied Science). After transfection, the cells were refed with medium A containing 5% lipoprotein-deficient serum, 50 M pravastatin, and sodium mevalonate. After incubation for 48 h, the cells were harvested with Buffer C* containing 20 mM HEPES/KOH (pH 7.9), 20% glycerol, 1.5 mM MgCl 2 , 300 mM NaCl, 0.5% Nonidet P-40, and 0.2 mM EDTA supplemented with a mixture of protease inhibitors, 10 M MG-132, 20 mM N-ethylmaleimide, and 10 mM iodoacetamide. After centrifugation at 13,000 ϫ g for 10 min at 4°C, the supernatant was immunoprecipitated by the Seize Classis X Protein A immunoprecipitation kit (Pierce) following the instructions provided by the supplier and resolved on SDS-PAGE and immunoblotted as described previously (8).
COS-1 and M19 cells were set up on day 0. On day 1, the COS-1 cells were transfected using the DEAE-dextran methods and then refed with medium A supplemented with 10% fetal bovine serum. M19 cells were transfected using LipofectAMINE (Invitrogen) and then refed with medium B (a 1:1 mixture of Ham's F-12 medium and Dulbecco's modified Eagle's medium) containing 5% lipoprotein-deficient serum supplemented either with 1 g/ml 25-hydroxycholesterol plus 10 g/ml cholesterol (the sterol-loaded condition) or 50 M pravastatin plus 50 M sodium mevalonate (the sterol-depleted condition). After incubation for 48 h, the cells were harvested with a radioimmune precipitation buffer containing 50 mM Tris/HCl (pH 7.8), 150 mM NaCl, 5 mM EDTA, 15 mM MgCl 2 , 1% Nonidet P-40, 0.75% sodium deoxycholate, 1 mM dithiothreitol for binding assays or Buffer C* for sumoylation assays supplemented with a mixture of protease inhibitors, 10 M MG-132, 20 mM N-ethylmaleimide, and 10 mM iodoacetamide. After centrifugation at 13,000 ϫ g for 10 min at 4°C, the supernatant was incubated with 50 l of a 50% slurry of glutathione-Sepharose 4B or immunoprecipitated with the indicated antibodies and 50 l of a 50% slurry of Protein A-Sepharose CL-4B. All resins were resolved on SDS-PAGE and immunoblotted as described previously (8).
Reporter Assays-Reporter Assays were performed as described previously (36). HEK293 cells were transfected with 0.2 g of the pLDLR (38) or the pG5Luc, 0.01 g of the pRL-CMV, an expression plasmid encoding Renilla luciferase (Promega), and the indicated amounts of various expression plasmids. After incubation for 48 h, the Dual-Luciferase TM Reporter System (Promega) was used to determine luciferase activities.
Pulse-Chase Experiments-On day 0, monolayers of COS-1 cells were set up and transfected on day 1 with wild-type or mutant SREBPs. One day after transfection, the cells were trypsinized and reseeded to normalize the transfection efficiency. On day 3, the cells were preincubated for 1 h with methionine/cysteine-free medium A (Invitrogen) supplemented with 10% fetal bovine serum and then pulsed with 200 Ci of L-35 S cell labeling mix for 1 h. After the pulse period, the cells were incubated for 1 h and then washed with phosphate-buffered saline and refed with a prewarmed complete medium. After each chase period, the cells were harvested and lysed with a cold lysis buffer containing 10 mM Tris/HCl (pH 7.4), 0.1% Triton X-100, 0.1% SDS, and 2 mM EDTA. The immunoprecipitates with anti-FLAG antibodies were visualized by au-toradiography. Exposed filters were quantitatively analyzed on Fluor-Image Analyzer with Image Gauge (Fuji Film).
In Vitro Ubiquitylation Assay-The ubiquitylation assay was previously described (39). Substrate GST-SREBPs were prepared from COS-1 cells transfected with pGSTSBPs. The cells were treated with 20 M MG-132 for the last 12 h of the culture and then lysed. The GST-SREBPs immobilized on the glutathione-Sepharose 4B were incubated with 40 ng of E1, 0.4 g of E2, and 8.0 g of GST-Ub in 40 l of the ubiquitylation buffer containing 50 mM Tris/HCl (pH 7.5), 2 mM MgCl 2 , 1 mM dithiothreitol, and 4 mM ATP for 3 h at 37°C. The reactions were terminated by the addition of the SDS sample buffer.

Posttranslational Modification of Endogenous SREBPs by
Covalent Attachment of SUMO-1-Analysis of the amino acid sequences of the nuclear forms of human SREBPs revealed that SREBP-1a contains four matches to the consensus sumoylation sequence centered around Lys 123 , Lys 381 , Lys 418 , and Lys 470 , and that SREBP-2 contains two matches, centered around Lys 420 and Lys 464 (Fig. 1A). To examine whether the endogenous SREBPs are modified by SUMO-1, HeLa cells were transiently transfected with an expression plasmid for HA-SUMO-1 and cultured under sterol-depleted conditions for 48 h to increase in the amount of the nuclear SREBPs. The nuclear extracts treated with isopeptidase inhibitors, N-ethylmaleimide and iodoacetamide, were subjected to immunoprecipitation and immunoblot analysis. In the case of SREBP-1a, anti-HA antibodies recognized several slower migrating bands in the immunoprecipitates with anti-SREBP-1 antibodies (Fig. 1B,  left top). Anti-SREBP-1 antibodies also detected weaker bands for SUMO-1-modified forms above the parental form of SREBP-1a (left bottom). In the case of SREBP-2, anti-HA antibodies recognized a single band in the immunoprecipitates with anti-SREBP-2 antibodies (right top), and anti-SREBP-2 antibodies detected a faint SUMO-1-modified form above the parental form of SREBP-2 (right bottom). These results demonstrated that endogenous SREBPs are modified by SUMO-1. Based on the fact that SUMO-1 can be covalently attached to a lysine residue only in a monomeric form (40), our findings suggest that more than two residues of lysine among four potential sites in SREBP-1a and a single residue of lysine between two potential sites in SREBP-2 are modified.
A Dominant Negative Form of SUMO-1 Induces the Expression of SREBP-responsive Genes-To determine whether the transcriptional activities of endogenous SREBPs are regulated by sumoylation, either SUMO-1(GG), a processed mature form of SUMO-1 (13), or SUMO-1⌬GG, a dominant-negative form of SUMO-1 that is unable to attach target proteins (42), was overexpressed in HeLa cells. Northern blot analyses for SREBP-responsive genes shown in Fig. 2 indicate that overexpression of SUMO-1(GG) decreased the amounts of HMG-CoA synthase and LDL receptor mRNA in a dose-dependent manner, suggesting that sumoylated endogenous SREBPs (Fig. 1B) down-regulate transcription of their target genes ( Fig. 2A). In contrast, overexpression of SUMO-1⌬GG increased the amounts of HMG-CoA synthase and LDL receptor mRNA in a dose-dependent manner, indicating that block of sumoylation accelerated transcription of SREBP target genes (Fig. 2B). These results indicate that sumoylation of endogenous SREBPs can affect regulation of the SREBP-responsive gene expression.
Two Lys Residues (Lys 123 and Lys 418 ) in SREBP-1a Are Sumoylated-To determine which Lys residues in SREBP-1a are sumoylated, we cotransfected COS-1 cells with expression plasmids for His-SUMO-1 and either wild-type or mutant versions of GST-SREBP-1a. GST-SREBP-1a bound to glutathione-Sepharose resins were subjected to immunoblotting with anti-SUMO-1 antibodies (Fig. 3, A and B, top panels). Three sumoylated bands were detected in cells transfected with an expression plasmid for wild-type SREBP-1a, whereas a single sumoylated band was detected in cells expressing either SREBP-1a K381R/K418R/K470R or K123R/K381R/K470R (Fig. 3A, top, lanes 1, 2, and 4). No bands were observed after the removal of four lysine residues in potential sumoylation sites (lane 6). Another immunoblotting analysis with anti-SREBP-1a antibodies confirmed that all sumoylated bands observed in the top panel were derived from sumoylated SREBP-1a (bottom). It is likely that the most slowly migrating sumoylated band in GST-SREBP-1a (top, lane 1) represents doubly sumoylated proteins at Lys 123 and Lys 418 . Fig. 3B also shows that mutation at both Lys 123 and Lys 418 completely abolished sumoylation of SREBP-1a (lane 10). Mutation of one of two possible sumoylation sites, Lys1 23 and Lys 418 , resulted in a single sumoylated band (lanes 8 and 9), confirming that these residues are major sumoylation sites. Taken together, these results clearly demonstrate that the molecule of SREBP-1a contains two possible sumoylation sites, Lys 123 and Lys 418 .
SREBP-2 Is Sumoylated at Lys 464 -To determine which Lys residues of two putative sumoylation sites in SREBP-2 are modified, COS-1 cells were transiently transfected with expression plasmids for His-SUMO-1 and either wild-type or mutant versions of SREBP-2. Fig. 4 shows that a slowly migrating SUMO-1-conjugated band (ϳ83 kDa) was detected when either wild-type or K420R SREBP-2 was expressed together with His-SUMO-1. These results clearly indicate that Lys 464 serves as the sumoylation site in the nuclear form of SREBP-2.
Sumoylation Decreases the Transactivation Activities of SREBPs-To assess the potential consequences of sumoylation of SREBPs, we examined whether sumoylation influences the transcriptional activities of SREBPs. HEK293 cells were cotransfected with a reporter plasmid, pLDLR, containing the promoter region of human low density lipoprotein receptor gene, and expression plasmids encoding either wild-type or mutant versions of SREBPs. The cells were cultured under the sterol-loaded condition to suppress the processing of endogenous SREBPs, and luciferase assays were carried out. As shown in Fig. 5A, both SREBP-1aK123R and -K418R signifi-cantly increased luciferase activities compared with wild-type SREBP-1a. Double mutation markedly activated the transcription of the reporter gene. Fig. 5B also shows that SREBP-2K464R markedly activated transcription of the reporter gene compared with wild-type SREBP-2. Similar results were obtained using a reporter plasmid containing the promoter region of the human HMG-CoA synthase gene (data not shown). These results indicate that sumoylation of both SREBP-1a and SREBP-2 negatively regulates their transcriptional activities and that sumoylation at both Lys 123 and Lys 418 of SREBP-1a coordinately attenuates the transcription.
SREBPs require co-regulatory transcription factors, such as Sp1 and CBF/NF-Y, that bind DNA sequences adjacent to the SREBP binding sites and enhance the transcriptional activities of SREBPs by forming complexes with SREBPs (33,41). Based on these early findings, we speculated that sumoylation might influence the interaction between SREBPs and these co-regulatory factors. Alternatively, it is possible that modification by SUMO-1 may weaken the affinity between SREBPs and their responsive DNA sequences. Instead of focusing on these possibilities, we evaluated whether sumoylation of SREBPs directly resulted in inactivation of these proteins. Accordingly, we used the heterologous Gal4 system with a reporter plasmid, pG5Luc. In this assay the luciferase gene transcription is driven by a promoter that contains five consensus Gal4-binding sites, without any co-regulatory transcription factors. Sumoylated SREBPs were shown in the insets in Fig. 5, C and D. The mutations at Lys 123 and Lys 418 resulted in a 1.9-and 2.1-fold increase in luciferase activities compared with Gal4-SREBP-1a, respectively (Fig. 5C). Furthermore, Gal4-SREBP-1aK123R/K418R, which was no longer sumoylated (Fig. 5C, inset), was more potent than Gal4-SREBP-1aK123R and -K418R. Fig. 5D also  shows that Gal4-SREBP-2K464R significantly enhanced transcription of the reporter gene to an extent elicited by double mutation in SREBP-1a. These results indicate that SUMO-1 modification of SREBPs directly reduces their own transcriptional activities. On the other hand, we cannot rule out the possibility that sumoylation of SREBPs might influence their interaction with co-regulatory factors or their DNA affinity. Effect of Sumoylation of SREBPs on Their Stability-We previously reported that nuclear SREBPs are rapidly degraded via the ubiquitin/26 S proteasome pathway (8). Since both sumoylation and ubiquitylation require a lysine residue of target proteins, we examined whether sumoylation competes with ubiquitylation, thereby affecting the ubiquitin-dependent degradation of nuclear SREBPs. Pulse-chase experiments using COS-1 cells transfected with expression plasmids encoding either FLAG-tagged wild-type or K464R SREBP-2 were performed. Fig. 6A shows that the calculated half-life of wild-type and K464R SREBP-2 was ϳ2 h and that there were no obvious differences in their stabilities. Similar results were obtained when cells were transfected with expression plasmids encoding either wild-type or mutant versions of FLAG-SREBP-1a (data not shown). Taken together, these results indicate that sumoylation of SREBPs does not influence their degradation through the ubiquitin/26 S proteasome pathway.
To directly determine whether wild-type and mutant versions of SREBPs are similarly modified by ubiquitin, we devised an in vitro ubiquitylation assay for SREBPs. In this system, the GST-SREBPs expressed in COS-1 cells were conjugated with glutathione-Sepharose resins and then incubated with recombinant Ub-activating enzyme (E1) and Ub-conjugating enzyme (E2) in the presence of GST-Ub and ATP. We confirmed that Ubc4, but not UbcH7, collaborates as E2 for the in vitro ubiquitylation assay of SREBPs (data not shown). As shown in the top panels of Fig. 6B, in the presence of Ubc4, the amounts of parental forms of SREBPs were decreased (arrowheads), and high molecular mass smear bands (closed bars) were observed by immunoblotting with anti-SREBP antibodies. To confirm that these multiple smear bands are indeed ubiquitin-modified forms of SREBPs, we performed immunoblotting with anti-multiubiquitin antibodies. High molecular mass smear bands were also detected with anti-multiubiquitin antibodies (lower panels). Weak ubiquitylated smear bands were observed when UbcH7 was used as E2, which probably reflected endogenously ubiquitylated SREBPs. These results allow us to conclude that wild-type and mutant versions of SREBPs are similarly modified by ubiquitin and that sumoylation does not compete with ubiquitylation in the SREBPs.
Sumoylation of SREBPs under Sterol-loaded or -depleted Conditions-Our next question was whether intracellular sterol levels alter the rate of sumoylation of SREBPs. It is possible that high sterol levels could accelerate sumoylation of nuclear SREBPs, thus suppressing the transcription of SREBP-responsive genes. To investigate this issue, we used M19 cells, which lacks site 2 protease (43), resulting in the absence of endogenous nuclear SREBPs. In these cells, one can determine sumoylation of exogenously expressed SREBPs under either steroldepleted or sterol-loaded conditions in the absence of endogenous SREBPs. The cells were transfected with expression plasmids encoding His-SUMO-1 and GST-SREBPs and cultured under either sterol-depleted or sterol-loaded conditions. GST-SREBPs bound to glutathione-Sepharose resins were subjected to immunoblotting analysis with anti-SUMO-1 (Fig. 7, top) and anti-SREBP-1/2 (bottom) antibodies. The amounts of sumoylated and unmodified SREBPs under either sterol-loaded (lanes 1 and 3) or sterol-depleted (lanes 2 and 4) conditions were similar. These results demonstrate that sterols do not alter the sumoylation rate of SREBPs.
The SUMO-conjugating Enzyme, Ubc9, Interacts with SREBPs-To investigate the direct interaction of SREBP with endogenous Ubc9, COS-1 cells were transfected with an expression plasmid encoding either FLAG tag or FLAG-SREBP-1a/-2.  6. Effects of sumoylation of nuclear SREBPs on their stability and ubiquitylation. A, COS-1 cells were transfected with an expression plasmid encoding either wild-type or K464R SREBP-2. After a 24-h incubation, the cells were trypsinized and reseeded in 60-mm plates to normalize the transfection efficiency. After further incubation for 24 h, pulse-chase experiments were performed as described under "Experimental Procedures." At the indicated time points, the cell lysates were subjected to immunoprecipitation with anti-FLAG antibodies. All precipitates were fractionated with SDS-PAGE and visualized by autoradiography. The same results were obtained in three separate experiments. B, COS-1 cells were transfected with an expression plasmid encoding either wild-type or mutant versions of SREBPs. The purified GST-SREBPs from the cells were subjected to an in vitro ubiquitylation assay, as described under "Experimental Procedures." Ubiquitin-modified SREBPs were detected by immunoblotting (IB) with anti-SREBP-1/2 (top) and multiubiquitin (bottom) antibodies. The mobilities of the ubiquitylated and unmodified SREBPs are indicated on the right.
To examine whether the dominant negative form of Ubc9 is capable of repressing sumoylation of endogenous SREBPs, thereby enhancing the transcriptional activities, HeLa cells were transfected with an expression plasmid for Ubc9(C93S). Northern blots of SREBP-responsive genes shown in Fig. 8C indicate that overexpression of Ubc9(C93S) increased the amounts of HMG-CoA synthase and LDL receptor mRNA in a dose-dependent manner, indicating that blockade of sumoylation accelerated transcription of SREBP target genes. These results suggest that Ubc9 interacts with SREBPs, thereby enhancing the attachment of SUMO-1 to SREBPs and suppressing their transcriptional activities.

DISCUSSION
The aim of this study was to identify posttranslational modification of the SREBPs other than ubiquitylation and to elucidate the mechanisms that modulate SREBP activities through such modifications. The major findings of the present study were 1) SUMO-1 modifies nuclear SREBPs in a Ubc9-dependent manner, 2) residues Lys 123 and Lys 418 of SREBP-1a and Lys 464 of SREBP-2 can act as potential SUMO-1 acceptor sites, and 3) SUMO-1 can negatively regulate the transactivation function of SREBPs. These sumoylation motifs are all conserved in vertebrate SREBPs whose sequences have been so far reported (U00968 for human, U09103 for Chinese hamster, AF286470 for rat, and AY029224 for chicken SREBP-1; U02031 for human, U12330 for Chinese hamster, and AJ414379 for chicken SREBP-2).
Mutation analysis revealed that residues both Lys 123 and Lys 418 are required for a full repression of the SREBP-1a transcriptional activity, whereas Lys 464 is sufficient for a full repression of SREBP-2 to an extent comparable with that elicited by double mutation in SREBP-1a (Fig. 5). It remains unknown why SREBP-1a needs two sumoylation sites to control its transactivation function. Interestingly, the amino acid sequence surrounding Lys 123 in SREBP-1a, PGIK 123 EESVP, perfectly conforms to the consensus sequence, PX0 -4(I/V)K(Q/T/ S/L/E/P)EX0 -3P, identified recently in the negative regulatory domain of several transcription factors and named the synergy control (SC) motif (45). The putative SC/sumoylation sites in androgen receptor, glucocorticoid receptor, and c-Myb were found to be functional (31,32,46). It is hypothesized that an as yet unidentified SC factor might be recruited to the SC motifs, thereby limiting the transcriptional activities of certain transcription factors. The current finding that covalent attachment of SUMO-1 at Lys 123 in the SC motif attenuates the transactivation function of SREBP-1a implies involvement of such SC factor in a sumoylation-dependent manner. However, whether the hypothetical SC factor recognizes unmodified SC motifs or sumoylated motifs remains obscure, because sumoylation of the SC motifs in SREBP-1a as well as androgen receptor and c-Myb attenuates their transactivation capacity, whereas sumoylation of glucocorticoid receptor elicits the opposite effect.
We found that the C-terminal domain containing ϳ90 amino acids of the nuclear SREBPs exerts a transcriptional repression function when fused to the Gal4 DNA binding domain plus the SREBP-1a activation domain containing the N-terminal 50 amino acids (data not shown). In a previous study, we indeed noticed that truncation of the domain in a nuclear form of SREBP-1 somehow potentiated the transcriptional activities (2). The residues Lys 418 of SREBP-1a and Lys 464 of SREBP-2 are located in this negative regulatory domain and are likely to be involved in such repression. The amino acid sequence surrounding Lys 464 in SREBP-2, VK 464 DEP, matches the consensus sumoylation site found in the negative regulatory domain of four CCAAT/enhancer-binding protein family members, which is evolutionally conserved in a variety of vertebrate species (47). However, SUMO-1 attachment augmented the inhibitory effect of the SREBP-2-negative regulatory domain but decreased that of the CCAAT/enhancer-binding protein regulatory domain. In the case of SREBPs, covalent conjugation of SUMO-1 to Lys 418 or Lys 464 could potentially influence the flexibility of the C terminus as observed in sumoylated RanGAP1 (21). In support of this notion, the different migration patterns of the sumoylated bands modified at Lys 123 and Lys 418 in SREBP-1a (Fig. 3) suggest that sumoylation of SREBPs might induce conformational changes in these proteins.
Sumoylation has been shown to be responsible for the translocation of some proteins into discrete subnuclear matrix-associated structures, called the promonocytic leukemia nuclear bodies (23,48). Several proteins, including Sp100, Daxx, CCAAT/enhancer-binding protein, and ISG20, have been reported to be targeted to the nuclear bodies after SUMO-1 modification (19); however, this does not necessarily mean that all SUMO-1-conjugated proteins are destined to be localized in the nuclear body (49). Interestingly, although the biological function of the nuclear body is unclear, it has been recently shown that sumoylation directly or indirectly promotes promonocytic leukemia degradation (50). Although we examined whether mutations in the SUMO-1 sites affect the intracellular localization of the GFP-tagged SREBPs, sumoylation did not induce significant changes in the distribution of them in the nucleus (data not shown). Further studies are required to investigate the role of SUMO-1 modification of several target proteins including SREBPs on their subnuclear localization in order to understand the biological functions of sumoylation. Figs. 1, 3, and 4 show that the sumoylated SREBPs were barely detected compared with unmodified SREBPs in these  1 and 3) or FLAG-SREBPs (lanes 2 and 4). The cell lysates were subjected to immunoprecipitation (IP) with anti-FLAG antibodies and immunoblotting (IB) with anti-SREBP-1/2 (top) and Ubc9 (middle) antibodies. Expression of endogenous Ubc9 was evaluated by immunoblotting in normalized whole cell lysates (10 g/lane) using anti-Ubc9 antibodies (bottom). B, COS-1 cells were cotransfected with expression plasmids for GST-SREBPs, His-SUMO-1, and either GFP (lanes 5 and 7) or GFP-Ubc9 (C93S) (lanes 6 and 8). GST-SREBPs bound to glutathione-Sepharose resins were subjected to immunoblotting (IB) with anti-SUMO-1 (top) and SREBP-1/2 (bottom). The asterisks mark the nonspecific bands observed in all lanes. C, HeLa cells in a 60-mm dish were transfected with the indicated amounts of expression plasmid for His-Ubc9(C93S). After transfection, the cells were cultured under sterol-depleted conditions for 48 h. Total RNA (8 g/lane) was subjected to Northern blot analysis as described in the legend to Fig. 2 experiments. This does not appear to be in line with the results of transactivation experiments shown in Fig. 5, in which mutations in the SUMO-1 acceptor sites resulted in marked increases of the transcriptional activities of SREBPs, and the findings that expression of SREBP-responsive genes was diminished by overexpression of SUMO-1 but increased by overexpression of dominant negative SUMO-1 (Fig. 2). Sumoylated proteins are considered as very transient structures resulting from a dynamic equilibrium between SUMO-1-conjugated and -deconjugated forms (51). Thus, sumoylation is likely to represent a mechanism for rapid and reversible control, and target proteins are conjugated during a short period of time, which is nevertheless sufficient to affect their biological activities. So far, a number of mammalian desumoylating enzymes, which remove SUMO-1 from its protein conjugates, have been identified (19). The different patterns of subcellular or tissue distribution of these enzymes may reflect the specific function of each enzyme at its location and suggest that desumoylation, like deubiquitylation, plays an important role in regulation of SUMO-1-mediated cellular processes.
The biological consequences of sumoylation and ubiquitylation are quite different despite the facts that both SUMO-1 and ubiquitin are conjugated to a lysine residue in target proteins and that their modification machineries are mechanistically very similar. The addition of a polyubiquitin chain to a lysine residue marks modified proteins for a rapid degradation by the 26 S proteasome. In contrast, sumoylation has been shown to increase the stability of IB␣ by antagonizing ubiquitylation (24). We previously reported that nuclear SREBPs are degraded through the ubiquitin/26 S proteasome pathway (8). In the present study, we provided experimental evidence that SUMO-1 modification of the SREBPs could not alter their stability (Fig. 6A) and that SUMO-1 conjugation did not compete with ubiquitin modification (Fig. 6B). We devised a unique in vitro ubiquitylation assay, in which ubiquitin conjugation occurred without the addition of a specific E3 ligase for SREBPs in the presence of recombinant E1 and E2. It is likely that the GST-SREBPs prepared from cells treated with a proteasome inhibitor trapped an unidentified functional E3 ligase, which was capable of attaching ubiquitin to the SREBPs. Based on these results, we conclude that lysine residues of sumoylation sites in the SREBPs are at least not the major potential ubiquitylation sites. Taken together, the independence of these two processes implies that inactivation of SREBPs is doubly regulated by sumoylation and by their destruction through the ubiquitin/26 S proteasome pathway. These results also support the notion that the two independent processes might be required for sufficient elimination of unexpected increases in the expression of the SREBPresponsive genes to maintain homeostasis of lipid metabolism.
SREBPs are synthesized as membrane-bound precursors and activated by a proteolytic process regulated by intracellular sterol levels. In addition to the proteolytic processing, a rise in intracellular sterol levels, probably cholesterol levels in nuclear membranes, might accelerate the rate of SUMO-1 conjugation to the nuclear SREBPs, which in turn negatively regulates their transactivation function to suppress the expression of genes encoding enzymes for cholesterol biosynthesis. However, the results shown in Fig. 7 clearly rule out this hypothesis. In a previous study, we also demonstrated that intracellular sterol levels do not alter the rapid degradation of nuclear SREBPs through the ubiquitin/26 S proteasome pathway (8). Taken together, these studies suggest that two independent processes, ubiquitin-dependent degradation and SUMO-1-mediated inactivation, are not regulated in re-sponse to changes in lipid metabolism but act constitutively to control the transcriptional activities of the nuclear SREBPs.
In conclusion, the importance of SUMO-1 modification for modulating the transactivation function of nuclear SREBPs is now emerging. Our work reports a new mechanism through which sumoylation can exert its negative effect on SREBPs function and govern lipid metabolism.