Sumoylation and Acetylation Play Opposite Roles in the Transactivation of PLAG1 and PLAGL2*

PLAG1 (pleomorphic adenoma gene 1) and PLAGL2 (PLAG-like 2) are oncogenes involved in various malignancies. Thus the study of their regulatory mechanisms may lead to identification of novel therapeutic targets. In this study, we provide supporting evidence that sumoylation and acetylation regulate functions of PLAG1 and PLAGL2. A conserved transcriptional repression domain exists in both PLAG1 and PLAGL2, whose activity depends on the presence of three sumoylation motifs and an intact sumoylation pathway. In vivo sumoylation assays confirmed that lysines 244, 263, and 353 of PLAG1 and lysines 250, 269, and 356 of PLAGL2 are indeed sumoylation sites. Further study showed that sumoylation inhibits PLAG1-induced IGF-II expression in reporter assays. The repression mediated by sumoylation may be partially explained by its effect on the cellular localization of PLAG1 and PLAGL2, because sumoylation-deficient but not wild-type PLAG1 and PLAGL2 concentrate in the nucleolus. PLAG1 and PLAGL2 are also regulated by acetylation. They are acetylated and activated by p300 and deacetylated and repressed by HDAC7. Interestingly, the sumoylation-deficient mutant of PLAGL2 is acetylated at a lower level than its wild-type counterpart, suggesting that some of the lysine residues may be targets for both modifications. Finally, mutation of three lysine residues in sumoylation motifs significantly impairs the transformation ability of PLAG1 and PLAGL2, suggesting the essential roles of these sites in the oncogenic potential of PLAG proteins. Taken together, the activities of PLAG1 and PLAGL2 are tightly modulated by both sumoylation and acetylation, which have opposite effects on their transactivation. To our knowledge, this is the first demonstration that oncoproteins can be regulated by both sumoylation and acetylation.

PLAG1 (pleomorphic adenoma gene 1) and PLAGL2 (PLAG-like 2) are oncogenes involved in various malignancies. Thus the study of their regulatory mechanisms may lead to identification of novel therapeutic targets. In this study, we provide supporting evidence that sumoylation and acetylation regulate functions of PLAG1 and PLAGL2. A conserved transcriptional repression domain exists in both PLAG1 and PLAGL2, whose activity depends on the presence of three sumoylation motifs and an intact sumoylation pathway. In vivo sumoylation assays confirmed that lysines 244, 263, and 353 of PLAG1 and lysines 250, 269, and 356 of PLAGL2 are indeed sumoylation sites. Further study showed that sumoylation inhibits PLAG1-induced IGF-II expression in reporter assays. The repression mediated by sumoylation may be partially explained by its effect on the cellular localization of PLAG1 and PLAGL2, because sumoylation-deficient but not wild-type PLAG1 and PLAGL2 concentrate in the nucleolus. PLAG1 and PLAGL2 are also regulated by acetylation. They are acetylated and activated by p300 and deacetylated and repressed by HDAC7. Interestingly, the sumoylation-deficient mutant of PLAGL2 is acetylated at a lower level than its wildtype counterpart, suggesting that some of the lysine residues may be targets for both modifications. Finally, mutation of three lysine residues in sumoylation motifs significantly impairs the transformation ability of PLAG1 and PLAGL2, suggesting the essential roles of these sites in the oncogenic potential of PLAG proteins. Taken together, the activities of PLAG1 and PLAGL2 are tightly modulated by both sumoylation and acetylation, which have opposite effects on their transactivation. To our knowledge, this is the first demonstration that oncoproteins can be regulated by both sumoylation and acetylation.
The pleomorphic adenoma gene 1 (PLAG1) 2 is a developmentally regulated (1) transcription factor that plays an important role in tumorigenesis. Dysregulated PLAG1 expression, which results from chromosomal translocation, is crucial in the formation of pleomorphic adenomas of the salivary glands (1) and lipoblastomas (2)(3)(4). PLAG1 overexpression is also detected in tumors without chromosomal translocation, such as uterine leiomyomas, leiomyosarcomas, and smooth muscle tumors (5). It was shown recently that both PLAG1 and its related molecule, PLAGL2, play important roles in the pathogenesis of acute myeloid leukemia in cooperation with Cbfb-MYH11 (6,7).
Both PLAG1 and PLAGL2 consist of an N-terminal zinc finger DNA binding domain and a C-terminal transactivation domain. The consensus DNA binding site comprises a core sequence (GRGGC) and a G-cluster (RGGK) (8). Although several potential target genes of PLAG1 have been identified (9), the regulatory mechanisms of transcriptional activation mediated by PLAG1 and PLAGL2 remain unknown. One of the mechanisms of regulation of the activity of transcription factors is post-translational modification, such as phosphorylation (10), acetylation (11), methylation (12), ubiquitination (11), isgylation (13), neddylation (14), and sumoylation (11). Sumoylation is a three-step enzymatic pathway analogous to that of ubiquitin conjugation, which results in the transfer of SUMO from Ubc9 to target proteins (11). The functional consequences of sumoylation are distinct from ubiquitination. Instead of being marked for degradation by ubiquitination, sumoylation has diverse substrate-specific functions. Several transcription factors, including androgen receptor (15), Sp3 (16), c-Myb (17), and Elk-1 (18), are sumoylation targets, and sumoylation represses their transcriptional activities. The exact mechanism of how sumoylation represses transactivation remains unclear, although SUMOdependent recruitment of histone deacetylases (HDACs) has been implicated in the transcriptional repression of p300 (19) and Elk-1 (20).
In addition to ubiquitination and sumoylation, lysine residues of transcription factors can be covalently modified by acetylation, a process known to enhance DNA binding (21), change protein-protein interaction (22), and regulate transactivation (23,24). Numerous nuclear histone acetyltransferases (HATs) have been identified. Among them, p300 and the closely related CREB-binding protein (CBP) are the most potent and versatile of all the acetyltransferases. Consistent with its role as a global co-activator, p300 acetylates and regulates various non-histone transcription factors, such as GATA-1 (25), MyoD (26), E2F-1 (27), and p53 (23). On the other hand, HDACs reverse the acetylation process to maintain the balance between the acetylated and deacetylated states of chromatin and other non-histone proteins. HDACs are divided into four classes: class I HDACs (HDACs 1, 2, 3, and 8) localize to the nucleus; class II HDACs (HDACs 4, 5, 6, 7, 9, and 10) are found in both the nucleus and cytoplasm; class III HDACs are NAD ϩ -dependent enzymes that are similar to the yeast SIR2 proteins; and class IV (HDAC11) is the smallest of the HDAC members (28).
In this study, we have identified regulatory mechanisms for PLAG1 and PLAGL2. Sumoylation occurs in a conserved repression domain and inhibits transactivation of both PLAG1 and PLAGL2, which may be explained by changes of subnuclear localization. Moreover, PLAG1 and PLAGL2 are also regulated by acetylation. They are acetylated and activated by p300, while deacetylated and repressed by HDAC7, suggesting opposite roles of acetylation and sumoylation in the transactivation of PLAG1 and PLAGL2. Interestingly, sumoylation-deficient PLAGL2 * This study was supported by National Institutes of Health Grants DK50570, CA78433, and HL48819 (to Y.-C. Y.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. □ S The on-line version of this article (available at http://www.jbc.org) contains supplementary Figs. S1-S10. 1  shows decreased acetylation, suggesting that some lysine residues could be targets for both modifications. Importantly, mutation of the sumoylation sites greatly reduced transformation abilities of both PLAG1 and PLAGL2, suggesting the importance of these modifications for PLAG1 and PLAGL2 as oncoproteins.
Transient Transfection, Immunoprecipitation, and Western Blot Analysis-HEK-293 cells were transfected by the calcium phosphate precipitation method with various plasmid combinations as indicated. Forty-eight hours later, cells were washed with phosphate-buffered saline (PBS) and 1 ml of ice-cold lysis buffer (radioimmune precipitation assay buffer) (50 mM Tris-HCl pH 7.4, 150 mM NaCl, 1% Nonidet P-40, 1 mM EGTA, 2 mM Na 3 VO 4 , 15 g/ml aprotinin, 10 g/ml leupeptin, 1 mM phenylmethylsulfonyl fluoride) was added. Cells were lysed for 30 min at 4°C with occasional vortexing. The lysates were collected into 1.5-ml tubes and cleared of nuclei by centrifugation for 10 min at 14,000 rpm. The supernatants (whole cell extracts) were incubated with different antibodies for 16 h at 4°C and protein A-agarose beads were added for the last hour. The beads were washed five times in TNEN buffer (20 mM Tris-HCl, pH 8.0, 100 mM NaCl, 1 mM EDTA, 0.5% Nonidet P-40, 2 mM Na 3 VO 4 , 1 mM phenylmethylsulfonyl fluoride, 1 mM NaF). Bound proteins were extracted with SDS-PAGE sample buffer, and analyzed on SDS-PAGE followed by Western blot analysis with the ECL detection system. For sumoylation assays, 0.5 g/well of expression plasmid for a potential substrate was cotransfected with 0.5 g/well of FLAG-SUMO-1 plasmid in 6-well plates. Forty-eight hours after transfection, cells were lysed in a denaturing buffer (2% SDS, 10 mM Tris-HCl pH 8.0, 150 mM NaCl), and analyzed by SDS-PAGE and Western blotting.
Luciferase Assay-Cells were plated in 12-well plates and grown overnight before transfection. The total amount of DNA transfected was adjusted with pcDNA3. Luciferase assay was performed according to the manufacturer's instructions (Promega). Renilla luciferase internal control plasmid was cotransfected with the plasmids as indicated. The relative luciferase units were corrected based on Renilla luciferase activity. For GAL4 fusion-driven luciferase reporter gene assays, 0.1 g/well of reporter (pG5-luc) was cotransfected with 0.1 g of GAL4 fusion expression plasmid in 12-well plates. 0.2 g/well of pcDNA3-DNUbc9 or pCMV6-SSP3 construct was used if indicated. For IGF-II-luciferase reporter assays, 0.05 g/well of reporter plasmid was cotransfected with effector plasmids of indicated amounts in 12-well plates.
Plasmid Construction-The PLAG constructs used in mammalian cells were generated by polymerase chain reactions with primers containing restriction sites for cloning. Every construct was sequenced fully to verify the fidelity of the polymerase chain reaction. PLAG cDNAs were fused in-frame to the DNA binding domain of GAL4 using the pM vector. PcDNA3-PLAG1-Myc and subfragments of PLAG1 were constructed using XhoI and EcoRI sites. PcDNA3-PLAGL2-Myc and subfragments of PLAGL2 were constructed using EcoRI and BamHI sites. PCMV2-FLAG-PLAG1 was constructed using EcoRI and XbaI sites. PCMV2-FLAG-PLAGL2 was constructed using EcoRI and BamHI sites. GFP-PLAG1 or its mutants were constructed by inserting fulllength or mutant PLAG1 into XhoI and EcoRI sites of pEGFP-C1 (Clontech). GFP-PLAGL2 or its mutants were constructed by inserting wildtype or mutant PLAGL2 into the EcoRI and BamHI sites of pEGFP-C1.
Immunostaining-HEK293 cells were seeded in chamber slides (0.5 ϫ 10 5 cells/ml) and transfected 24 h later with 2 g of respective plasmids by the calcium phosphate method. 24 -48 h later, cells were washed in cold PBS, fixed with 3.7% formaldehyde for 30 min and permeabilized with 0.5% Triton X-100 for 5 min. Cells were then blocked for 30 min in blocking buffer (PBS, 5% bovine serum albumin, 0.3% Triton X-100), incubated with primary antibody for 1 h and washed three times with PBS. After incubation with secondary antibody and washing three times with PBS, cells were examined under confocal immunofluorescence microscope.
Focus Formation Assay-NIH-3T3 cells (in 6-well plates) were transfected with 2 g of empty vector, expression plasmid for PLAG1, PLAG1 mutant, PLAGL2, PLAGL2 mutants, or Ha-RAS. The day after transfection, cells were split at 3 ϫ 10 4 /plate in 10-cm plates and selected in 300 g/ml G418 and 10% serum in Dulbecco's modified Eagle's medium for 3 days. Then cells were fed once every 4 days with DMEM plus 1% serum and 300 g/ml G418. After 3 weeks, cells were fixed and stained with methylene blue, and the number of foci was determined.

Identification of Repression Domains in PLAG1
and PLAGL2-In this study, PLAG1 and PLAGL2 were found to share similar regulatory mechanisms. To avoid redundancy, the data for PLAGL2 is under "Results" wherever indicated; the rest of the data is under supplemental materials. The structures of PLAG1, PLAGL2, and their GAL4 fusion derivatives are shown in Fig. 1A. Both PLAG proteins have a zinc finger domain in the N terminus, and a transactivation domain in the C terminus. A previous study by Kas et al. (30) identified a region in the middle of PLAG1 and PLAGL2 with a transcriptional repression function. As shown in Fig. 1B, when PLAGL2-(387-496), which includes only the transactivation domain, and PLAGL2-(238 -496), which includes both the transactivation domain and the middle region, were fused with the GAL4 DNA binding domain, there was a 10-fold difference in the transactivation capacity between these two constructs. Similarly, PLAG1-(361-500) exhibited about 10 times more transactivation potential compared with PLAG1-(232-500) (supplemental Fig. S1). These results suggest that PLAG1-(232-361) and PLAGL2-(238 -387) have repressive activity. The repression domains of some transcription factors such as Elk-1 have been shown to act in trans (18); that is, when bound to a promoter, the repression domain itself is able to repress the activity of another adjacent transcription factor. To test whether the repression domains of PLAG proteins also repress transcription in trans, we performed transrepression assays in which GAL4 fusion proteins were used to repress the activity of a Lex-VP16 activated reporter (18). We found that both PLAG1-(232-361) (data not shown) and PLAGL2-(238 -387) (supplemental Fig. S2) have no significant repression activity in the assay, suggesting that the repression domains of PLAG1 and PLAGL2 function differently from that of Elk-1.
Potential SUMO Modification Sites Play a Repressive Role in PLAG1 and PLAGL2-Because repression activities are conserved in the middle regions of PLAG1 and PLAGL2, we aligned their sequences to find the possible conserved sites that may be crucial for the repressive activ-

Regulation of PLAG1/PLAGL2 by Acetylation and Sumoylation
40776 JOURNAL OF BIOLOGICAL CHEMISTRY ity. We found 3 sumoylation motifs, whose consensus sequence is KXE ( is a hydrophobic amino acid, X is a random amino acid), well conserved in both PLAG1 and PLAGL2 (supplemental Fig. S3). We tested whether they are crucial for the repression domain by introducing triple mutations K244R/K263R/K353R into PLAG1 and K250R/K269R/ K356R into PLAGL2. In the context of a GAL4 fusion protein tethered to an artificial promoter, both GAL4-PLAG1-(232-500)(K244R/ K263R/K353R) (supplemental Fig. S4A) and GAL4-PLAGL2-(238 -496)(K250R/K269R/K356R) ( Fig. 2A) exhibited a 9 -10-fold higher transcriptional activity than their wild-type counterparts. Moreover, these mutants became equally active to the transactivation domains alone, suggesting that the activities of the repression domains were abolished by the mutations of the sumoylation motifs. Western blot analysis showed that the effect of the mutations is not caused by higher expression of the mutants (Fig. 2B and supplemental Fig. S4B). These results suggest that these sumoylation motifs are required for the repression domains of PLAG1 and PLAGL2.
Next we investigated whether the sumoylation pathway is involved in the activity of the repression domains of PLAG1 and PLAGL2. We used two approaches to block the sumoylation pathway: a catalytically inactive form of the SUMO-conjugating enzyme, Ubc9 C93S (DNUbc9), which acts as a dominant-negative mutant; and SSP3, which is a SUMOspecific protease. As shown in Fig. 2C, transfection of pcDNA3-DNUbc9 completely rescued the difference between GAL4-PLAGL2-(238 -496) and GAL4-PLAGL2-(387-496), suggesting that the repressive activity of PLAGL2-(238 -387) was abolished. A similar observation was made using SSP3 (Fig. 2D). When we converted the luciferase activity to fold induction (the ratio between transfected and untransfected with pcDNA3-DNUbc9 or pCMV6-SSP3), as shown in supplemental Fig. S4C and S4D, both sumoylation pathway inhibitors significantly enhanced the transcriptional activity of PLAG1-(232-500), but had much less effect on PLAG1-(361-500) and PLAG1-(232-500)(K244R/K263R/K353R). Moreover, the repression domain of either PLAG1 or PLAGL2 alone, but not vector control or a control protein ZNF76K411R, was activated by DNUbc9 (supplemental Fig. S5). These results suggest that both the sumoylation motifs and the intact sumoy-lation pathway are essential for the repression domains of PLAG1 and PLAGL2.
PLAG1 and PLAGL2, but Not PLAGL1, Are Modified by Sumoylation-To verify that sumoylation of PLAGL2 occurs in cells, we cotransfected HEK293 cells with expression plasmids for PLAGL2-Myc and FLAG-SUMO-1 and analyzed sumoylation of PLAGL2 by immunoblotting with anti-Myc antibody. A slower migrating band reacting with anti-Myc antibody was clearly detected in cells ectopically expressing FLAG-SUMO-1 (Fig. 3A). Because PIAS1 is a well established E3 ligase in sumoylation, we tested whether PIAS1 promotes sumoylation of PLAGL2. HEK293 cells were transfected with PLAGL2-Myc, FLAG-SUMO-1, and FLAG-PIAS1 or FLAG-PIAS1 C335A , a construct encoding an inactive form of PIAS1. As shown in Fig. 3B, cotransfection of FLAG-PIAS1, but not FLAG-PIAS1 C335A , strongly enhanced the sumoylation of PLAGL2. Moreover, DNUbc9 abolished the appearance of the slower migrating bands (Fig. 3C), suggesting that the higher molecular weight bands correspond to sumoylated PLAGL2. To test whether lysine residues in the consensus sumoylation motifs are indeed sumoylation targets, wild-type and mutant PLAGL2 were cotransfected with FLAG-SUMO-1 and FLAG-PIAS1. PLAGL2 K250R , partially, and the triple mutant PLAGL2 K250,269,356R , completely, lost their abilities to be sumoylated (Fig. 3D), which correlates with the loss of repression activity ( Fig. 2A). Similar results were obtained for PLAG1 (supplemental Fig. S6). We also tested single mutants of PLAGL2. As shown in Fig.  3E, each PLAGL2 single mutant still retained the ability to be sumoylated, suggesting that all three lysine residues are sumoylation targets. PLAGL1 is another member in the PLAG family. Unlike PLAG1 and PLAGL2, PLAGL1 is a tumor suppressor gene rather than an oncogene. As shown in Fig. 3F, no sumoylation was detected for PLAGL1 even when FLAG-SUMO-1 was overexpressed. These results suggest that both PLAG1 and PLAGL2, but not PLAGL1, can be sumoylated, and lysine residues inside the sumoylation motifs are indeed sumoylation targets.
Sumoylation Pathway Represses PLAG1/PLAGL2-mediated Transcription-Because both the intact sumoylation motifs and the sumoylation pathway are required for the function of the repression  DECEMBER 9, 2005 • VOLUME 280 • NUMBER 49 domain (Fig. 2), and PLAGL2 can be physically sumoylated at those sites (Fig. 3), we next tested in the full-length context, whether sumoylation represses PLAG1/PLAGL2-mediated transcription. We fused fulllength PLAGL2 with the GAL4 DNA binding domain, and tested their activity in the presence or absence of DNUbc9. Though DNUbc9 did not change the basal reporter activity, it activated PLAGL2 about 15-fold (Fig. 4A). A similar observation was made for PLAG1 (supplemental Fig. S7). We also tested whether sumoylation regulates PLAG1induced expression of IGF-II, a well established target gene of PLAG1 (8). DNUbc9 did not have any effects on the basal reporter activity, but transfection of as low as 0.02 g of pcDNA3-DNUbc9 enhanced PLAG1-induced IGF-II promoter activity, and the increase was dosedependent (Fig. 4B). These results suggest that sumoylation negatively regulates PLAG1/PLAGL2-mediated transcription.

Regulation of PLAG1/PLAGL2 by Acetylation and Sumoylation
Sumoylation-deficient PLAG1 and PLAGL2 Localize to Nucleoli-To further understand sumoylation-mediated functions, we compared the cellular localization of wild-type PLAG1 and PLAGL2 with that of the sumoylation-deficient mutants. Both GFP-PLAG1 (supplemental Fig.  S8) and GFP-PLAGL2 (Fig. 5A) showed a diffused nuclear pattern except for the nucleolus. However, after lysine residues responsible for sumoylation were mutated, 80% of GFP-PLAG1 K244,263,353R (supplemental Fig. S8) and 60% of GFP-PLAGL2 K250,269,356R (Fig. 5A) exhibited strong and distinct nucleolar localization. PIAS1, which promotes sumoylation of PLAG1 and PLAGL2, colocalized with wild-type but not the mutant PLAGL2 (Fig. 5B). These results suggest that sumoylation regulates the subnuclear localization of PLAG1 and PLAGL2, which may contribute to its transcriptional repression effect.
PLAG1 and PLAGL2 Are Regulated by Acetylation-Surprisingly, when we compared wild-type and sumoylation-deficient PLAG1 for their ability to activate IGF-II expression in reporter assays, we found that PLAG1 K244,263,353R has lower transactivation activity than wild-type PLAG1 (Fig. 6A). We speculated that besides being the sumoylation sites, these lysine residues may play other critical roles in transactivation. Because lysine residues can also be acetylated, and acetylation has been shown to regulate transactivation of transcription factors, we considered the possibility that one or more of the three lysine residues of PLAG1 and PLAGL2 are also targets for acetylation. To determine whether PLAG1 and PLAGL2 can be acetylated in cells, GFP-PLAG1 or FLAG-PLAGL2 were transiently transfected into HEK293 cells with or without HA-p300, a well studied lysine acetyltransferase. Cell lysates were immunoprecipitated with anti-FLAG or anti-GFP antibody, and acetylated forms of PLAGL2 and PLAG1 were detected by Western blotting with anti-acetyl lysine antibody ( Fig. 6B and supplemental Fig.  S9). P300 significantly enhanced acetylation of both PLAG1 and PLAGL2. TSA, which is a histone deacetylase inhibitor, further increased the acetylation level of PLAGL2 (Fig. 6C), suggesting that HDACs may regulate the acetylation of PLAG1 and PLAGL2. Indeed, HDAC7, a histone deacetylase, significantly decreased acetylation of PLAGL2 (lane 4 compared with lane 3, Fig. 6D). Interestingly, the triple mutant PLAGL2 K250,269,356R , was less acetylated than the wild-type (lane 3 compared with lane 6, Fig. 6D), suggesting that these lysine residues may also be the targets for acetylation, which could explain why the sumoylation-deficient mutant PLAG1 has weaker activity than the wild-type PLAG1 (Fig. 6A). To investigate the functional consequences of acetylation, we tested whether p300 affects the transactivation of PLAG1 and PLAGL2. When GAL4-fused full-length PLAG1 and PLAGL2 were transfected alone or with p300, p300 significantly enhanced the transactivation potential of both PLAG proteins (Fig. 6E and data not shown), Moreover, transfection of p300 also enhanced PLAG1-induced IGF-II expression in luciferase reporter assays (supplemental Fig. S10). Because acetylation is a reversible process and HDAC7 decreased p300-induced PLAGL2 acetylation (Fig. 6D), we also tested the effect of HDAC7 on PLAG1-mediated IGF-II expression in reporter assays. As shown in Fig. 6F, ectopic expression of HDAC7 repressed the PLAG1-induced IGF-II expression in a dose-dependent manner. Taken together, these results suggest that acetylation activates, whereas deacetylation represses PLAG1 and PLAGL2.
Lysine Residues Responsible for Sumoylation/Acetylation Are Important for the Transforming Activity of PLAG1 and PLAGL2-Both PLAG1 and PLAGL2 are oncogenes involved in the pathogenesis of different malignancies. Overexpression of PLAG1 or PLAGL2 transforms NIH-3T3 cells to low serum growth (31). To evaluate the significance of sumoylation/acetylation in their transforming activity, expression plasmids for PLAG1, PLAGL2, or their lysine mutants were transfected into NIH-3T3 cells. One day after transfection, cells were split at the density of 3 ϫ 10 4 cells/10-cm plate, and grown as a monolayer in the medium containing 10% serum and G418 (300 g/ml). After 3 days the medium was changed to 1% serum plus 300 g/ml G418. The ability of transfected cells to form foci was analyzed 3 weeks following selection. As shown in Fig. 7, both activated Ha-RAS oncogene, PLAG1 and PLAGL2 promoted focus formation in NIH-3T3 cells in low serum condition. It was noticed that although RAS group and PLAG1/ PLAGL2 groups had similar numbers of focus formation, RAS transformed foci were bigger in size than the ones transformed by PLAGs (data not shown), which suggests that PLAG1 and PLAGL2 are weaker oncogenes than RAS. Importantly, whereas all the single mutants of PLAGL2 still retained partial transforming ability, the abilities of PLAG1/PLAGL2 sumoylation-deficient mutants to transform NIH-3T3 cells were greatly reduced (Fig. 7). These results suggest that lysine residues that are sumoylation/acetylation targets are important for the transforming ability of PLAG1 and PLAGL2.

DISCUSSION
In our study, we found that PLAG1 and PLAGL2 are modulated by both sumoylation and acetylation. Sumoylation occurs in a conserved repression domain and is required for the repression activity. In contrast, acetylation by p300 activates the transcriptional activity of PLAG1 and PLAGL2. Mutation of the sumoylated lysine residues severely impairs the transforming ability of PLAG1 and PLAGL2, suggesting the importance of these modifications on the transforming potential of these proteins.
It was shown that sumoylation pathway is activated under stress situations, such as hypoxia or genotoxic condition (32,33,34), and involved in DNA repair and maintenance of genome stability (35). Several important enzymes involved in DNA repair, such as thymine DNA glycosylase (36), DNA topoisomerase I (37) and II (38), are regulated by sumoylation. At the same time, sumoylation prevents further cell FIGURE 6. PLAG1 and PLAGL2 are regulated by acetylation. A, sumoylation-deficient PLAG1 has impaired ability to activate IGF-II expression. HEK293 cells in 12-well plates were transfected with 50 ng of Hup3-luc reporter and different amounts of PLAG1-Myc or PLAG1 K244,263,353R -Myc as indicated. All data are representative of at least three independent experiments. B, PLAGL2 can be acetylated by p300. Lysates from HEK-293 cells (in 6-well plates) transfected with 0.5 g of FLAG-PLAGL2 and 0.5 g of HA-p300 either alone or in combination were immunoprecipitated (IP) with anti-FLAG. Immunoprecipitates or whole cell lysates were analyzed by immunoblot with indicated antibodies. C, TSA augments PLAGL2 acetylation. Lysates from HEK-293 cells (in 6-well plates) transfected with 0.5 g of FLAG-tagged PLAGL2 and 0.5 g of HA-p300 with or without TSA treatment (330 nM TSA for 12 h before harvesting) were immunoprecipitated (IP) with anti-FLAG. Immunoprecipitates or whole cell lysates were analyzed by immunoblot with indicated antibodies. D, acetylation of the sumoylation-deficient mutant of PLAGL2 is impaired. HEK293 cells in 10-cm plates were transfected with 3 g of PLAGL2-Myc or PLAGL2 K250,269,356R -Myc, 3 g of HA-p300, and 3 g of HA-HDAC7 either alone or in combination as indicated. Whole cell lysates or its anti-Myc immunoprecipitates were analyzed by immunoblot with antibodies indicated. E, P300 activates PLAGL2 transcriptional activity. HEK293 cells in 12-well plates were transfected with 0.1 g of pG5-luc reporter and 0.1 g of PM-PLAGL2, in the presence or absence of 0.2 g of expression plasmid for HA-p300 as indicated. Luciferase assays are representatives of at least three independent experiments. F, HDAC7 represses PLAG1induced IGF-II expression. HEK293 cells in 12-well plates were transfected with 50 ng of Hup3-luc reporter, 0.1 g of pCMV-FLAG-PLAG1, and different amounts of HA-HDAC7 expression plasmid either alone or in combination as indicated. Luciferase assays are representatives of at least three independent experiments. Total cell lysates were blotted with anti-HA antibody to detect the expression of HDAC7 (bottom panel).
growth through the repression of transcription factors with mitogenic potential, such as c-Jun (39), Elk-1 (18), androgen receptor (15), Sp3 (16), and PLAG1/PLAGL2 in the current study. It is tempting to speculate that sumoylation pathway may be one of the mechanisms to facilitate cellular adaptation to various stress conditions by preventing cell growth. Consistent with the model that transcription factors that promote cell growth tend to be repressed by sumoylation, another PLAG family member, PLAGL1, which is a tumor suppressor rather than an oncogene, is not regulated by sumoylation (Fig. 3F).
There are several possible mechanisms that could explain the repressive effects exerted by sumoylation. The first one is that sumoylation regulates subcellular localization of the substrates. Early studies have suggested that sumoylation is involved in regulating nucleocytoplasmic transport. For example, RanGAP1 is targeted to the nuclear pore complex after being sumoylated (40). Recent studies indicated that sumoylation may play a more important role in regulating subnuclear distribution of proteins. Interestingly, sumoylation is essential for PML (promyelocytic leukemia) to accumulate in the nuclear bodies (41), and for Sp3 to localize in the nuclear periphery and nuclear dots (16). Our data suggest that sumoylation affects the subnuclear localization of PLAG1 and PLAGL2 (Fig. 5), which may contribute to its transcriptional repression effect. Another possible mechanism for sumoylationmediated transcriptional repression is that sumoylation may abolish some of the crucial protein-protein interactions thus rendering the substrates inactive. One of such examples is that sumoylation of ZNF76 affects its interaction with TATA-binding protein (TBP) (42), which is a critical component for transcription initiation. Finally, it is also possible that sumoylation may cause a gain of function such that sumoylated PLAG1/PLAGL2 may recruit novel repressors to repress their transcriptional activity. It was shown that SUMO alone in a promoter is sufficient to inhibit promoter activity (18). Because SUMO itself does not have repression activity, it suggests that other factors are recruited by SUMO to repress transcription. HDACs and HP1 have been shown to be recruited to p300 (19), Elk-1 (18), and histones (43) after sumoylation to affect gene expression. Importantly, these mechanisms are not mutually exclusive. Although sumoylation affects the subnuclear localization of PLAG1/PLAGL2, whether changes in protein-protein inter-actions or recruitment of novel repressors by sumoylation affect PLAGmediated transactivation requires further investigation.
Acetylation has been shown to modulate activities of a broad range of transcription factors including DNA binding, protein-protein interactions, protein stability, and transcriptional potency. In our study, we found that PLAG1 and PLAGL2 can be activated by acetylation and repressed by deacetylation. Interestingly, sumoylation-deficient PLAGL2 is partially acetylation-defective. We propose the functions of acetylation are 2-fold: first, to compete lysine residues for sumoylation, second, to activate transcription factors possibly through enhancing DNA binding or changing protein-protein interactions such as recruitment of co-activators. Thus, a model emerges whereby PLAG1 and PLAGL2 can respond to activating signals by desumoylation and subsequent acetylation at the same lysine residues, which not only eradicates the function of the repression domain, but also enhances transcriptional activation.
During the preparation of the current report, Van Dyck et al. (44) published their study on the biochemical characterization of PLAG1 sumoylation. Our study not only agrees with the published data but also provides additional novel findings: first, our study shows that both PLAG1 and PLAGL2 (7, 45), but not PLAGL1, which is a tumor suppressor rather than an oncogene, are regulated by sumoylation. Thus regulation by sumoylation may be one of the mechanisms to differentiate the functions of these three PLAG family members. Second, using the GAL4 reporter system, our data clearly demonstrate that sumoylation is required for the function of the repression domains of PLAG1 and PLAGL2 described previously (30). Third, our cellular localization studies suggest that sumoylation may play a role in the nuclear localization of PLAG1 and PLAGL2 (Fig. 5). Fourth, we identify acetylation as another post-translational modification for PLAG1 and PLAGL2, which has an opposite function from sumoylation. Finally, in the transformation assay, our data clearly show that acetylation/sumoylation sites of PLAG1 and PLAGL2 are important for their transforming ability.
Notably, both sumoylation and acetylation are reversible processes in which SUMO-specific proteases and deacetylases can desumoylate and deacetylate modified proteins, respectively. Acetylation and sumoylation of PLAG1/PLAGL2 may differentially modulate their affinity for different interacting protein partners, which contributes to different functional consequences of the two modifications: sumoylation causes a switch to a repressed state, whereas acetylation of PLAGL2/PLAG1 causes a switch to an activated state. Thus, the transactivation potential of PLAG1 and PLAGL2 can be regulated by a dynamic interplay of acetylation, deacetylation, sumoylation, and desumoylation in response to various biological signals. In our study, sumoylation and acetylation not only affect the transactivation of PLAG1 and PLAGL2, but also affect their transforming abilities (Fig. 7), suggesting a profound biological effect of these modifications. Given that both PLAG1 and PLAGL2 are oncoproteins involved in the pathogenesis of certain cancers, the enzymes involved in regulating their sumoylation and acetylation could be considered as potential therapeutic targets in associated malignancies.