A synergy control motif within the attenuator domain of CCAAT/enhancer-binding protein alpha inhibits transcriptional synergy through its PIASy-enhanced modification by SUMO-1 or SUMO-3.

One of the most common forms of functional interaction among transcription factors is the more than additive effect at promoters harboring multiple copies of a response element. The mechanisms that enable or control synergy at such compound response elements are poorly understood. We recently defined a common motif within the negative regulatory regions of multiple factors that operates by regulating their transcriptional synergy. We have identified such a synergy control (SC) motif embedded within the "attenuator domain" of CCAAT/enhancer-binding protein alpha (C/EBPalpha), a key regulator of energy homeostasis and cellular differentiation. A Lys(159) --> Arg substitution within the SC motif does not alter C/EBPalpha activity from a single site but leads to enhanced transactivation from synthetic or natural compound response elements. The sequence of SC motifs overlaps with the recently defined consensus SUMO modification site, and we find that the SC motif is the major site of both SUMO-1 and SUMO-3 modification in C/EBPalpha. Furthermore, the disruption of SC motif function is accompanied by loss of SUMO but not ubiquitin modification. C/EBPalpha interacts directly with the E2 SUMO-conjugating enzyme Ubc9 and can be SUMOylated in vitro using purified recombinant components. Notably, we find that PIASy has E3-like activity and enhances both SUMO-1 and SUMO-3 modification of C/EBPalpha in vivo and in vitro. Our results indicate that SUMO modification of SC motifs provides a means to rapidly control higher order interactions among transcription factors and suggests that SUMOylation may be a general mechanism to limit transcriptional synergy.

One of the most pervasive forms of interaction between transcription factors is the synergistic response resulting from the recruitment of an activator to multiple copies of a recognition site (1,2). Transcriptional synergy from such compound response elements provides a means to control both the level and specificity of gene expression (3), yet the mechanisms by which synergy is controlled are still relatively poorly understood.
Through a genetic approach, our laboratory has elucidated the function of a novel protein motif, which limits the transcriptional synergy of DNA binding regulators, including multiple steroid hormone receptors and ETS1 (4). Disruption of these conserved synergy control (SC) 1 motifs dramatically enhances synergistic activation from compound response elements without altering the intrinsic activity from an individual binding site. SC motifs are devoid of activation or repression properties, yet they are both necessary and sufficient to restrict the synergy mediated by a heterologous activation domain (4). We have proposed that SC motifs serve to recruit synergy control factor(s) that directly limit transcriptional synergy (4).
Our functional characterization of eight examples of SC motifs in different regulators revealed that the critical features for SC motif function include a branched aliphatic residue at the first position followed by invariant Lys and Glu residues at positions 2 and 4 (see Fig. 1 and Ref. 4). The core of the motif is preceded and/or followed by Pro residues in a region that often varies in size in different species, suggesting that the motif may lie between secondary structure elements or within a loop that can tolerate insertions (4).
In addition to the cases we have examined, matches to our definition of SC motifs occur frequently within documented negative regulatory regions of numerous unrelated transcription factors (4). These seemingly disparate regions may therefore operate via a common mechanism (i.e. synergy control). A striking example is in the CCAAT/enhancer-binding proteins (C/EBPs), where C/EBP␣ and -⑀ harbor highly conserved SC motifs in previously defined "attenuator" regions (5,6). At least six members of the C/EBP family have been isolated and characterized, C/EBP␣ to C/EBP. They all contain a highly conserved, basic leucine zipper dimerization and DNA binding domain at the C terminus (7). Their divergent N-terminal regions contain the transcriptional regulatory domains and specify their diverse activities. Three conserved regions have been identified in C/EBP␣, -␤, and -⑀ that are involved in transcriptional activation (8).
The ␣-isoform of C/EBP is a central regulator of energy homeostasis (9) as it directly activates the transcription of many metabolically important genes (10,11) and also plays pivotal roles in growth and differentiation (12)(13)(14)(15)(16)(17)(18). Genes regulated by C/EBP␣ usually harbor multiple binding sites like in the case of the peroxisome proliferator-activated receptor ␥ promoter (19) or the myeloperoxidase enhancer (20). C/EBP␣ often functions synergistically with other transcription factors like peroxisome proliferator-activated receptor ␥ or PU.1 (21,22). For example, cross-regulation between peroxisome proliferator-activated receptor ␥ and C/EBP␣ is a key component of the transcriptional control during adipogenesis (23,24). Regulatory mechanisms that affect C/EBP␣ synergy are therefore likely to have a profound impact on its function.
We have hypothesized that the function of SC motifs may be regulated through post-translational modification of the critical Lys residue (4), especially since Arg is not functional at this position. Interestingly, soon after our description of SC motifs, the consensus site for modification by the ubiquitin-like protein SUMO (Fig. 1) came into sharper focus, and it became apparent that our current definition of SC motifs could be viewed as a special case of the more general SUMOylation consensus.
SUMOylation is a reversible process that regulates the function of target proteins in a manner akin to phosphorylation. The functional consequences of SUMOylation are poorly understood but do not directly involve targeting for proteasomal degradation (25). Three different isoforms of SUMO are present in mammals, but whether they subserve different roles is unknown. As in the case of ubiquitination, preparation of SUMO for modification of proteins involves two steps carried by specific E1 activating (SAE1/SAE2) and E2 transfer (UBC9) enzymes. During ubiquitination, a third Ub-ligase or E3 component conveys substrate recognition, often in a signal-regulated manner (26,27). Although SUMOylation can be achieved without an E3 activity in vitro (28), recent studies indicate that proteins like RanBP2 (29) and members of the PIAS family (30 -32) can have E3-like activity for SUMO conjugation.
In an effort to examine the generality of SC motif function and to explore its mode of action, we have probed the functional significance of the SC motif in C/EBP␣ and the role of SUMO modification in its function.

EXPERIMENTAL PROCEDURES
Mammalian Expression Plasmids-The pCDNA3-based expression plasmid for the p42 form of mouse C/EBP␣ was provided by Dr. Ormond MacDougald and is described in Ref. 33. This plasmid (pCDNA3 p42) contains engineered silent restriction sites for ease of manipulation. The K159R substitution was generated by PCR and then transferred to p42 as a 538-bp XhoI/KpnI fragment (pCDNA3 p42 K159R). The C-terminal region of p42 was amplified with primers 5Ј-CGCAACAGAAGGTGCTCGAGTTGACCAGTGACAAT-3Ј and 5Ј-CTA-GAAGCTTCTAA TGATGATGGTGGTGATGGTCGACCGCGCAGTTG-CCCATGGCCTTG ACC-3Ј to add a C-terminal hexahistidine tag and transferred into the XhoI and Hind III sites of the WT and K159R p42 plasmids (pCDNA3 p42 His and pCDNA3 p42 K159R His). The pCDNA3 HA SUMO-1, pCDNA3 HA SUMO-3, and pCDNA3 HA ubiquitin were kind gifts of Dr. Kim Orth (University of Texas Southwestern), and the pCMVFLAG PIASy plasmid was a kind gift of Dr. Tae-Hwa Chun (University of Michigan).
Reporter Plasmids-The p⌬ODLO 02 parental vector contains a polylinker upstream of a minimal Drosophila alcohol dehydrogenase promoter (adh Ϫ33 to ϩ53) and the luciferase gene. The oligonucleotides 5Ј-GATCCTGATTGCGCAATCGA-3Ј and 5Ј-GATCTCGATTGCG-CAATCAG-3Ј, containing a single consensus C/EBP site, were annealed and ligated into the BamHI and BglII sites of p⌬ODLO 02 to yield p⌬(CAAT)1-Luc. Ligation of BseRI/BglII and BamHI/BseRI fragments of the same vector yielded p⌬(CAAT)2-Luc. The same procedure using p⌬(CAAT)2-Luc yielded p⌬(CAAT)4-Luc. The p⌬TAT glucocorticoid response units reporter consists of a fusion of 668-and 300-bp fragments corresponding to the Ϫ5.5 and Ϫ2.5 kb glucocorticoid response units of the rat tyrosine aminotransferase gene (34) inserted at the BamHI and BglII sites of p⌬ODLO 02.
Bacterial Expression Plasmids-The human Ubc9 coding sequence was amplified with primers 5Ј-GCTACGGATCCATGAGTGAGATCGC-CCTCAGCAGACTCGCCCAG-3Ј and 5Ј-GGAGTGCCTTGGCCCCAAG TCCGGTGGTGGTGGTGGAATTCAAAGATC-3Ј, digested with BamHI and EcoRI, and transferred to the same sites of the pGEX-KG vector (35) to yield pGEX-hUbc9. The expression vectors for His SUMO-1 GG , in which the last 4 residues of SUMO-1 have been deleted, and GST-Ulp1 were kind gifts of Dr. Kim Orth. The expression vector for bicistronic expression of GST SAE2 and SAE1 (28) was a kind gift of Dr. R. T. Hay.
The sequences of all of the constructed plasmids were confirmed by sequencing.
Cell Culture, Transfections, and Immunoblotting-Human embryonic kidney 293T cells were maintained in Dulbecco's modified Eagle's medium (Invitrogen) supplemented with 10% fetal bovine serum. Cells were transfected by liposome-mediated transfection using Lipo-fectAMINE and Plus reagent (Invitrogen). In all cases, cells received equimolar amounts of each type of expression plasmid to control for promoter effects. For functional assays, 5 ϫ 10 3 cells were seeded into 96-well plates and transfected with the indicated amounts of expression plasmids, 30 ng of the indicated reporter plasmid, and 10 ng of the control pRSV␤gal plasmid (36). The total amount of DNA was supplemented to 70 ng/well with pBSKS (Ϫ). Cells were lysed 36 h after transfection, and luciferase and ␤-galactosidase activities were determined as described previously (37). For SUMOylation/ubiquitination experiments, 2 ϫ 10 6 cells were seeded in 10-cm plates and transfected with the indicated amounts of expression plasmids. Cells were harvested 36 h post-transfection in 0.7 ml of urea lysis buffer (8 M urea, 0.5 M NaCl, 45 mM Na 2 HPO 4 , 5 mM NaH 2 PO 4 , 10 mM imidazole, Complete miniprotease inhibitor mixture tablets (1 tablet/10 ml) (pH 8.0)) and sonicated. For ubiquitination experiments, the cells were treated with 10 M lactacystin for 1 h before harvest with urea lysis buffer. Lysates were incubated with 0.1 ml of Ni 2ϩ -NTA-agarose (Qiagen) for 1 h at room temperature in a rotator. The resin was washed three times with 10 bed volumes of wash buffer 1 (8 M urea, 0.4 M NaCl, 17.6 mM Na 2 HPO 4 , 32.4 mM NaH 2 PO 4 , 10 mM imidazole (pH 6.75)) and three times with 10 bed volumes of wash buffer 2 (buffer 1 with 150 mM NaCl and no urea). Examination of the supernatants revealed that the binding is quantitative under these conditions. For Ulp1 treatment, beads were incubated with 3.5 g of purified GST or GST-Ulp1 for 60 min at 30°C. Proteins were eluted by incubating at 90°C in elution buffer (100 mM Tris-HCl, pH 6.8, 2% SDS, 10% glycerol, 500 mM imidazole, 0.015% bromphenol blue, 10 mM dithiothreitol), resolved by SDS-PAGE, and processed for immunoblotting. Membranes (Immobilon) were incubated with goat polyclonal anti-C/EBP␣ IgG (Santa Cruz Biotechnology, Inc., Santa Cruz, CA) or monoclonal HA 11 anti IgG (Covance) or monoclonal anti-FLAG IgG (Sigma). Anti-goat IgG peroxidase conjugate (Santa Cruz Biotechnology) or anti-mouse IgG peroxidase conjugate (Bio-Rad) were used as secondary antibodies, and visualization was with Super Signal West Femto substrates (Pierce). Images were captured with an Eastman Kodak Co. Image Station 440 CF. All of the experiments were performed at least twice with similar results.
Protein Expression and Purification-BL21 DE3-CodonPlus cells harboring the pGEX-hUbc9 or pGEX-SAE2/SAE1 expression vector and BLR (DE3) pLysS cells (Novagen) containing pGEX-Ulp1 or pT7His SUMO-1 GG were grown at 37°C in LB medium containing carbenicillin, chloramphenicol, and tetracycline (50 g/ml, 25 g/ml, and 12 g/ml). Cultures (1 liter; A 600 ϭ 0.8) were induced with 1 mM isopropyl-1-thio-␤-D-galactopyranoside for 2 h at 37°C. Cells were centrifuged at 8,000 ϫ g for 15 min at 4°C. For GST fusion proteins, the pellet was resuspended in buffer A (10 mM Tris-HCl (pH 8.0), 150 mM NaCl, 1 mM EDTA, 5 mM dithiothreitol, 10% glycerol, and Complete miniprotease inhibitor mixture tablets (1 tablet/10 ml)). After lysozyme treatment (40 g/ml for 30 min) and sonication at 4°C, the suspension was centrifuged at 35,000 rpm at 4°C for 30 min. The supernatant was incubated with 2 ml of glutathione-agarose (Sigma) for 60 min at 24°C. The matrix was washed (4°C) with 10 bed volumes of buffer A without protease inhibitors and with 10 bed volumes of buffer B (buffer A with 400 mM NaCl). Proteins were eluted in buffer B (24°C) supplemented with 20 mM reduced glutathione. For His SUMO-1 GG , cells were resuspended and lysed in buffer C (50 mM sodium phosphate buffer, pH 8.0, 300 mM NaCl, 10% glycerol 10 mM imidazole, 5 mM ␤-mercaptoethanol Complete miniprotease inhibitor mixture tablets (1 tablet/10 ml)). Incubation of the extract was with 2 ml of Ni 2ϩ -NTA resin (Qiagen) for 1 h at 4°C. The resin was washed with 10 bed volumes of buffer C, followed by 2 bed volumes of buffer C containing 20 mM imidazole. Protein was eluted in buffer C containing 250 mM imidazole. All proteins were exchanged into buffer D (10 mM Tris-HCl (pH 8.0), 150 mM NaCl, 1 mM EDTA, 5 mM dithiothreitol, 10% glycerol) via gel filtration and stored at Ϫ80°C. For immunopurification of PIASy, COS-7 cells were maintained and transiently transfected in 10-cm plates as described above for 293T cells with 10 g of expression plasmid, pCMVFLAG-PIASy, or the empty pCMV FLAG. Cells were harvested after 36 h in Buffer E (20 mM Hepes, pH 7.5, 5 mM EDTA, 1 mM EGTA, 5% glycerol, 400 mM NaCl, Complete miniprotease inhibitor mixture tablets (1 tablet/10 ml)). The extracts were incubated with 50 g of anti-FLAG antibody (Sigma) for 1 h at 4°C. Complexes were recovered using 50 l of protein A-Sepharose (Sigma) and washed three times in buffer E with 200 mM NaCl and twice in 50 mM Tris (pH 7.5), 5 mM MgCl 2 .
In Vitro Protein-Protein Interaction and SUMOylation Assays-Proteins were translated in vitro using the T 7 -TNT Quick Coupled Transcription-Translation system (Promega) in the presence of [ 35 S]methionine using the pCDNA3 p42, pCDNA3 p42 K159R, or control T7 Luc (Promega) plasmids as templates. Binding reactions (50 l) were carried out at 4°C for 1 h and contained 1.2 nmol of purified GST or GST-hUBC9 fusion proteins bound to 20 l of glutathione-Sepharose 4B (Amersham Biosciences) and 10 l of 35 S-labeled proteins in a binding buffer containing 50 mM NaCl and 1 mg/ml bovine serum albumin. The resin was washed four times with 1 ml of 0.1% Nonidet P-40 in phosphate-buffered saline. The beads and 2 l of the corresponding load were boiled in a final volume of 40 l of SDS-PAGE sample buffer, and 25% of the samples were resolved by SDS-PAGE. The gels were fixed in 45% methanol, 10% acetic acid and dried, and radioactive proteins were visualized using a PhosphorImager (Amersham Biosciences). In vitro SUMOylation reactions (20 l) were assembled on ice in 50 mM Tris, 5 mM MgCl 2 (pH 7.5) and contained 1 g of GST SAE2/SAE1, 5 or 0.5 g of GST hUBC9, 5 g of His-SUMO-1 GG , 5 l of in vitro translated [ 35 S]methionine-labeled p42 C/EBP␣ WT or K159R, and 10 l of control or PIASy-containing beads as indicated. Reactions were initiated by the addition of an ATP regeneration system (10 units/ml creatine kinase, 25 mM phosphocreatine, 5 mM ATP final concentrations) and pyrophosphatase (0.6 units/ml final concentration) and incubated at 30°C for 2 h. Disruption buffer (50 mM Tris-HCl, pH 6.8, 1.67% SDS, 10% glycerol, 0.24 M ␤-mercaptoethanol, 0.015% bromphenol blue) was added to terminate the reaction. The samples were heated at 95°C for 5 min, resolved by SDS-PAGE, and dried, and the radioactive proteins were visualized in a PhosphorImager. All of the results were confirmed in at least two independent experiments.

Identification of a Synergy Control Motif in C/EBP␣-Min-
ing of the Swiss-Prot protein data base for the occurrence of SC motifs allowed us to identify a number of transcription factors that harbor conserved SC motifs. For several of them, like the progesterone receptor, Sp3, C/EBP␣, C/EBP⑀ and c-Myb, the motifs reside in demonstrated negative regulatory regions (4). In the case of C/EBP␣, the SC motif lies within an "attenuator" domain (5). By comparing multiple sequences from distantly related vertebrates, we find that the SC motif in C/EBP␣ constitutes a highly conserved small stretch surrounded by regions of lower conservation (see Fig. 1A). This suggests an evolutionary pressure for the preservation of this sequence and the function it subserves.
To assess whether the SC motif in C/EBP␣ functions by inhibiting transcriptional synergy, we replaced the predicted critical lysine at position 159 with arginine. This substitution inactivates SC motif function in all of the cases we have examined (4). We then compared the activity of the WT and mutant proteins at promoters harboring zero, one, two, or four C/EBP␣ sites. We chose 293T cells, since they do not contain endogenous C/EBP␣ (38). As can be seen in Fig. 1B, WT C/EBP␣ activates the reporter with a single binding site by 10-fold. Adding a second site enhances the activity only 2.8-fold, and adding two more sites does not significantly increase transcription. These results indicate that the ability of C/EBP␣ to engage in synergistic interactions is limited. The K159R mutant is indistinguishable from the WT at a single site. However, the activity of the mutant at the reporters with two and four sites is ϳ6and 14-fold higher than that at a single site. This translates to a 2.5-and 5-fold higher activity of the mutant versus the WT protein at these reporters. Thus, disruption of the SC motif leads to enhanced transcriptional synergy, indicating that its normal function is to restrict the potential of C/EBP␣ to synergize but without affecting its intrinsic transactivation potential. Similar results were obtained in CV-1 cells (not shown). Importantly, we also observed a 5-fold higher activity of the mutant C/EBP␣ from a reporter driven by the natural enhancer regions of the rat tyrosine aminotransferase gene, which harbor multiple C/EBP␣ sites (see Fig. 1B, right) (34). This indicates that the effect is not restricted to synthetic promoters.
Consistent with their comparable activities at a single site, Western blot analysis indicated that the WT and mutant proteins are expressed at equivalent levels (not shown). Furthermore, similar results were observed at both higher and lower amounts of plasmid, ruling out preferential squelching effects (not shown). Taken together, these results confirm our assignment of this region of C/EBP␣ as a functional synergy control motif and argue that the SC motif is responsible for the de- scribed "attenuator" property of this region (5).
The SC Motif in C/EBP␣ Is Modified in Vivo by SUMO-1 and SUMO-3-Our sequence definition of SC motifs, which is based purely on functional effects, can be viewed as a subset of the more general SUMO modification consensus (Fig. 1A). Therefore, we tested whether the SC motif in C/EBP␣ can be modified by SUMO isoforms in vivo. To this end, we co-transfected 293T cells with expression vectors for His-tagged WT and mutant C/EBP␣ forms with vectors for HA-tagged SUMO-1 or SUMO-3. Cells were lysed under denaturing conditions to protect SUMOylated proteins from isopeptidases. His-tagged C/EBP␣ forms were purified via metal chelate chromatography, resolved by SDS-PAGE, and immunoblotted. As can be seen in Fig. 2, when WT C/EBP␣ is coexpressed with HA-SUMO-1 or HA-SUMO-3, we can detect ϳ91-kDa HA immunoreactive bands corresponding to HA SUMO-1-and HA SUMO-3-modified C/EBP␣. The SUMO-3-modified form migrates slightly faster than the SUMO-1 counterpart, presumably due to the smaller size of SUMO-3 versus SUMO-1. As is the case for other SUMO-modified proteins, the migration of modified forms is slower than that expected for their molecular sizes. The slower migrating species can also be detected as a minor band in the anti C/EBP blot for both SUMO-1 and SUMO-3. As in the case of other targets (25), relative quantitation revealed that less than 5% of the total C/EBP␣ is modified by either SUMO isoform. This may reflect the transient and reversible nature of this modification. Notably, these slower migrating forms were completely absent in the case of the K159R mutant, although its expression is indistinguishable from that of the WT protein. Comparable results were obtained in COS-7 cells. These results show that C/EBP␣ is modified in vivo by SUMO-1 or SUMO-3 and argue strongly that the SC motif is the main target for SUMO modification in C/EBP␣. Moreover, the fact that the K159R mutation disrupts both SC motif function and SUMOylation implies that this modification is key for SC motif function.

Ulp1 Can Remove SUMO-1 and SUMO-3 from C/EBP␣-
The yeast ubiquitin-like protein specific protease 1 (Ulp1) deconjugates SUMO from the lysine ⑀-amino group of modified proteins (39). This deconjugase activity is specific for SUMO versus ubiquitin. To confirm that the higher order species of C/EBP␣ that we observe is a SUMO-modified form, we treated the purified C/EBP␣ preparations with GST alone or with GST-Ulp1 (Fig. 3). In contrast to the GST treatment, we did not observe HA immunoreactive bands at ϳ91 kDa in SUMO-1 or SUMO-3 samples treated with GST-Ulp1. Instead, we saw a species of ϳ24 kDa corresponding to free SUMO. Ulp1 did not display nonspecific protease activity, since the unmodified C/EBP␣ protein was not affected by the treatment. These results suggest that C/EBP␣ can be modified by SUMO-1 and SUMO-3 and that Ulp1 can remove either SUMO isoform from C/EBP␣. Although the cleavage of SUMO-1 from substrates by Ulp1 is well established, to our knowledge, this is the first demonstration that SUMO-3-modified proteins can also be deconjugated by this yeast enzyme.

Disruption of the SC Motif Does Not Prevent Ubiquitination-Both ubiquitin and SUMO are linked to proteins through
Lys residues, and in the case of IB, both modifications appear to occur at the same site (40). Desterro et al. (40) proposed that SUMO modification of IB prevents its ubiquitination and therefore contributes to the stabilization of this protein. We therefore explored whether C/EBP␣ is ubiquitinated and, if so, whether disruption of the SC motif alters this modification. We used the same experimental paradigm as for SUMO modification and treated the cells with the proteasome inhibitor lactacystin to allow the accumulation of ubiquitinated proteins. As can be seen in the HA immunoblot in Fig. 4, we can detect mono-and polyubiquitinated forms of C/EBP␣. To our knowledge, this is the first demonstration that C/EBP␣ is ubiquitinated. Notably, we observed an identical pattern using the K159R mutant. These results clearly indicate that although ubiquitination is likely to be important for the degradation of C/EBP␣, the mechanism of action of the SC motif does not involve preventing ubiquitination.
Ubc9 Interacts with C/EBP␣ Independently of SC Motif Function-Ubc9 (E2 SUMO-conjugating enzyme) is an essential component of the SUMOylation pathway and can directly mediate the conjugation of SUMO to target proteins. We therefore performed an in vitro binding assay to determine whether C/EBP␣ can interact with Ubc9. Full-length human Ubc9 was expressed in bacteria as a GST fusion protein, coupled to glutathione-Sepharose beads, and incubated with in vitro translated 35 S-labeled C/EBP␣ in its WT and K159R mutant forms as well as with luciferase as a control. As shown in Fig. 5, both the WT and mutant forms of C/EBP␣ interacted with GST-Ubc9 but not with GST alone. The control protein luciferase did not bind to either matrix. These results show that C/EBP␣ interacts with Ubc9 and that the K159R mutation does not affect the interaction although it disrupts both SC motif function and SUMOylation.
PIASy Enhances SUMO-1 and SUMO-3 Modification of C/EBP␣ in Vivo and in Vitro-Recent reports suggest that members of the PIAS family of proteins can enhance SUMO conjugation in vivo and in vitro (31,32). In order to initiate our analysis of the role of PIAS proteins in C/EBP␣ SUMOylation and SC motif function, we examined the effect of expressing PIASy on the extent of C/EBP␣ modification by either SUMO-1 or SUMO-3 in vivo. The results in Fig. 6 show that expression of PIASy enhances the extent of SUMOylation of p42C/EBP␣ WT, both with SUMO-1 and SUMO-3 (Fig. 6, A and B, respectively) without altering the expression or recovery of C/EBP␣. The slower migrating band in the anti-C/EBP␣ blot was also enhanced to the same extent, consistent with our assignment of this band as SUMO-modified C/EBP␣. Notably, we did not detect significant SUMO-1 or SUMO-3 modification of the K159R mutant either in the absence or presence of PIASy. The enhancement of SUMO modification by PIASy was not limited to C/EBP␣, however, because immunoblot analysis of cell extracts with anti-HA antibodies revealed that PIASy substantially enhanced general SUMO-1 and SUMO-3 modification of cellular proteins (data not shown). Unfortunately, the broad changes in SUMO modification levels upon PIASy expression lead to global effects on transcription, presumably because multiple transcription factors and coregulators (41,42) are subject to SUMO modification. This has prevented us from specifically assessing the functional consequences of enhancing C/EBP␣ SUMOylation on its activity (see "Discussion"). Taken together, our results indicate that PIASy enhances the in vivo SUMO-1 and SUMO-3 modification of Lys 159 in C/EBP␣ and reveal the novel finding that the ability of PIASy to enhance SUMO modification is not limited to SUMO-1 but also extends to the conjugation of SUMO-3. To investigate the mechanism of C/EBP␣ SUMOylation, we established an in vitro SUMOylation system using purified recombinant SUMO-1, SUMO-activating (SAE1/SAE2) and -conjugating (Ubc9) proteins, and in vitro transcribed, 35 Slabeled p42 C/EBP␣ WT and K159R proteins as substrates. In the case of the WT form, a 2-h incubation at 37°C produced a well defined higher molecular mass C/EBP␣ band (ϳ91 kDa) and a faint higher order slow migrating species at ϳ101 kDa corresponding to mono-and probably di-SUMOylated C/EBP␣ (Fig. 7A). Under these conditions, more than 55% of the C/EBP␣ was in the mono-SUMOylated form. We did not detect these species in the case of the K159R mutant. The formation of these bands was dependent on the presence of SUMO-1 and the E1 (SAE2/SAE1) and E2 (Ubc9) activities. To examine the effect of PIASy, we immunopurified the protein and carried out 30-min reactions in the presence of either control beads derived from vector-transfected cells or PIASy-loaded beads (Fig. 7B). At 30 min, the extent of SUMO modification was only 30% in the control reaction, whereas in the presence of PIASy, ϳ70% of the protein was modified. As in the in vivo experiments, the K159R mutant was not modified. The presence of PIASy in the beads was confirmed by immunoblotting (Fig. 7B, lower panel). This indicates that PIASy accelerates the rate of SUMO modification even in the presence of excess Ubc9. When similar reactions were carried out for 1 h but with an amount of Ubc9 (0.5 g) that supports modification of only a small fraction of C/EBP␣, PIASy enhanced the stochiometry of SUMOylation to ϳ35% (Fig. 7C). Thus, our results indicate that PIASy enhances the rate of SUMO modification of the SC motif in C/EBP␣ both in vivo and in vitro and therefore has the properties of an E3 for conjugation of both SUMO-1 and SUMO-3.
Taken as a whole, the data presented indicate that the synergy control motif we have identified in C/EBP␣ is functional and is responsible for the negative or attenuator functions attributed to this region. Moreover, the SC motif constitutes a post-translational modification site for SUMO-1 and -3. Since mutation of the critical lysine within the SC motif disrupts both SUMO modification and function, our data provide a direct link between SUMOylation and synergy control. DISCUSSION A Synergy Control Motif in C/EBP␣ Inhibits Its Transcriptional Synergy-We recently characterized the properties of a novel functional region defined by one or more copies of a short amino acid motif that operates by restraining the ability of regulators to engage in synergy (4). In the case of C/EBP␣, an SC motif is present within a region (amino acids 107-170) that Pei and Shih (5) identified as an "attenuator" domain. As we anticipated (4), this study shows that the ability of C/EBP␣ to engage in synergy is indeed regulated through this SC motif. Disruption of the SC motif does not affect the activity of C/EBP␣ at a single site, but it leads to a substantial enhancement of activity at natural or synthetic compound response elements. This brings the number of functionally confirmed examples of SC motifs to nine, spanning three very different classes of transcriptional regulators (nuclear receptors, ETS, and C/EBP) (4). Thus, a common synergy control mechanism may be responsible for the function of negative regulatory regions that harbor SC motifs.
Gene activation by C/EBP␣ relies on cooperation between three separate trans-activation elements (TE-I through TE-III) that operate by distinct mechanisms (43). It will be interesting to examine whether one or more of these functions is affected by the SC motif. Notably, the p30 isoform of C/EBP␣, which antagonizes the effects of the full-length p42 form, retains the SC motif as well as TE-III (44). We are currently characterizing the effects of the SC motif in the context of p30 and examining whether the activity of the SC motif extends to synergy with other transcription factors. The recent characterization of a region with SC motif properties in the Drosophila activator Dorsal suggests that heterotypic effects are possible (45).
SC Motifs as SUMO Acceptor Sites-The convergence between the SUMO modification and SC motif consensus sequences led us to test the role of this modification in SC motif function. Our results, using a transfection paradigm indicate that C/EBP␣ is modified by SUMO and that the critical lysine within the SC motif is the major site for this modification. Mutation of this site to Arg leads to concomitant loss of SC motif function and SUMO modification, arguing for a critical role of SUMO in synergy control. This post-translational modification therefore provides a means to rapidly regulate SC motif function. Although SUMO modification of endogenously expressed C/EBP␣ remains to be demonstrated, a key role for SUMO in the function of SC motifs is supported by the fact that the site of SUMO modification in other transcription factors maps to functional SC motifs like in the case of the androgen and glucocorticoid receptors (46). In the case of c-Myb, we proposed that the SC motif embedded in its C-terminal negative region was involved in its function (4). Recent evidence by Bies et al. supports this idea (47). Other factors such as p53 and c-Jun are SUMO-modified (48), and disruption of the main SUMOylation sites leads to an enhancement of activity. Interestingly, in both cases, the modification site matches our definition of SC motifs except for the first position, which is occupied by Phe and Leu in p53 and c-Jun, respectively. If these regions do function as synergy control motifs, our SC motif definition will have to relax the identity at the first position to accommodate these hydrophobic residues.
Interestingly, although SUMO-1 and SUMO-3 are substantially divergent in sequence, both isoforms can modify the SC motif of C/EBP␣. Except for the preferential SUMO-1 modification of Ran GAP1 (49), little is known regarding the subtype specificity for modification of proteins. In contrast to SUMO-1, the closely related SUMO-2 and -3 each harbor a SUMO modification site, and SUMO-2 chains have been observed in vivo (50). Whether these or other differences result in distinct regulatory outcomes remains to be explored.
Many SUMO-1 modified proteins interact directly with Ubc9, and mutagenesis studies of Ran GAP1 indicate that the residues surrounding the modification site are both necessary and sufficient for Ubc9 binding (51). This correlates with our observations that SC motifs are self-contained modules (4). Interestingly, although the K159R mutation abolishes SUMOylation, it does not perturb Ubc9 binding, presumably due to the structurally conservative nature of this substitution. A similar result was observed for Ran Gap1 (51).
Regulation of SUMO Modification-SUMO modification is a dynamic and reversible process regulated by both positive (E3like) and negative (Ulp-like) activities. Our results show that PIASy can act as an E3 to enhance the SUMO modification of C/EBP␣. Notably, the effect applies to both SUMO-1 and SUMO-3 conjugation. This novel finding implies that caution should be taken in assigning the effects of PIASy proteins to SUMO-1 modification exclusively. PIASy expression enhances the modification of many proteins in addition to C/EBP␣ (not shown). This has prevented us from assessing the effects of selectively enhancing C/EBP␣ SUMOylation on its function. In our hands, PIASy expression leads to a strong dose-dependent inhibition of certain constitutive promoters as well as C/EBP␣dependent transcription. However, both the WT and mutant forms of C/EBP␣ are affected (not shown). Similarly, PIASy inhibits both WT and non-SUMOylatable forms of Lef-1 (31), and PIAS1 and PIASx␤ inhibit p53 even when the major SUMO modification site is mutated (30). In other cases, expression of PIAS family members alters transactivation by a given activator in complex ways (32). Given that multiple transcriptional regulators including co-activators (41) and co-repressors (42) are also regulated by SUMO modification, the net effect of broad spectrum E3-like activities such as PIASy probably reflects the balance between positive and negative effects on both the activator and its cofactors. A similar argument can be made for the variable and noncongruent effects of overexpressing SUMO-1 on transcriptional activation by several SUMO-modified factors (30,52). To circumvent these problems, we are developing novel approaches to selectively affect the SUMOylation status of C/EBP␣.
A Model for the Function of SC Motifs in C/EBP␣ and Other Activators-On the basis of the properties of SC motifs, we proposed that synergy control is unlikely to be due to a direct intramolecular interaction and suggested that a synergy control factor (SCF) is recruited to an assemblage of activators by multivalent recognition of SC motifs (4). Our observations lead us to propose that SC motifs need to be SUMOylated to be functional. In this modified model (Fig. 8), recruitment of the machinery responsible for synergy control is likely to depend on its interaction with SUMO itself. The selective effect of SC motifs at compound response elements could then be the result of SUMO modification only when the transcription factor is bound at a compound response element. Conversely, SC motifs could be modified irrespective of the DNA binding status, but recruitment of SCF may depend on multivalent contacts and therefore only occur at compound response elements. We are currently examining these nonexclusive possibilities experimentally.
Although the nature of SCF remains to be determined, our observations suggest that the factor(s) responsible for synergy control may comprise a SUMO-binding protein. In this regard, the ability of Ubc9 to interact with the K159R mutant makes it unlikely that Ubc9 is involved in this capacity. The mechanisms involved in synergy, especially those involving steps subsequent to DNA binding, are not well understood, but effects beyond DNA binding imply the alteration of the transcription complex. An SCF could thus interfere with or favor the disassembly of an active transcription complex. This may be similar to the role of SUMO modification in septin disassembly in yeast. Conversely, SCF function could lead to the sequestration of factors in specific subnuclear domains. Sachdev et al. (31) proposed that PIASy inhibits Lef-1 by altering its subnuclear localization. Our preliminary experiments, however, suggest that at concentrations that fully affect function, PIASy does not significantly alter the distribution of C/EBP␣ (not shown).
Role of SC Motif and SUMO Modification in C/EBP␣-regulated Processes-C/EBP␣ plays key roles in adipocyte and myeloid differentiation by regulating the expression of differenti- ation regulators and markers. Through its SUMO modification, the SC motif in C/EBP␣ may permit the selective deployment or utilization of functional synergy surfaces at the appropriate promoter context or developmental stage. For example, selective loss of SUMO modification when preadipocytes approach a terminally differentiated state would allow the characteristic high level expression of the appropriate set of adipocyte-specific genes. We are currently examining the SUMO modification of endogenous C/EBP␣ in an in vitro model of adipocyte differentiation. This approach will also allow us to extend our observations beyond the transfection paradigm we have used here. Recently, Wang et al. have provided evidence that the antimitotic activity of C/EBP␣ may be mediated by direct inhibition of Cdk2 and Cdk4 (53). The critical region for this function lies immediately downstream of the SC motif, which raises the interesting possibility that SUMO modification of C/EBP␣ may modify its antimitotic activity. Whether SUMO modification alters the pattern or the effects of GSK-mediated phosphorylation of C/EBP␣ (33) remains to be determined. Moreover, regulation by SUMO modification is likely to extend to other members of the C/EBP family; in fact, during the preparation of this manuscript, Kim et al. (54) showed that in addition to C/EBP␣, C/EBP⑀ can also be modified by SUMO with important regulatory consequences.
The finding that SC motifs function as SUMO modification sites provides a versatile regulatory mechanism to control their function. Since the extent of SUMO modification can be modulated via factors that enhance conjugation, such as PIASy, or reduce it, such as Ulp-like activities, SUMO modification of SC motifs has all of the hallmarks of a cellular regulatory device designed to control higher order interactions among transcription factors.