Stress-induced Inactivation of the c-Myb Transcription Factor through Conjugation of SUMO-2/3 Proteins*

Post-translational modifications, such as phosphorylation, acetylation, ubiquitination, and SUMOylation, play an important role in regulation of the stability and the transcriptional activity of c-Myb. Conjugation of small ubiquitin-like modifier type 1 (SUMO-1) to lysines in the negative regulatory domain strongly suppresses its transcriptional activity. Here we report conjugation of two other members of the SUMO protein family, SUMO-2 and SUMO-3, and provide evidence that this post-translational modification negatively affects transcriptional activity of c-Myb. Conjugation of SUMO-2/3 proteins is strongly enhanced by several different cellular stresses and occurs primarily on two lysines, Lys523 and Lys499. These lysines are in the negative regulatory domain of c-Myb and also serve as acceptor sites for SUMO-1. Stress-induced SUMO-2/3 conjugation is very rapid and independent of activation of stress-activated protein kinases of the SAPK and JNK families. PIAS-3 protein was identified as a new c-Myb-specific SUMO-E3 ligase that both catalyzes conjugation of SUMO-2/3 proteins to c-Myb and exerts a negative effect on c-Myb-induced reporter gene activation. Interestingly, co-expression of a SPRING finger mutant of PIAS-3 significantly suppresses SUMOylation of c-Myb under stress. These results argue that PIAS-3 SUMO-E3 ligase plays a critical role in stress-induced conjugation of SUMO-2/3 to c-Myb. We also detected stress-induced conjugation of SUMO-2/3 to c-Myb in hematopoietic cells at the levels of endogenously expressed proteins. Furthermore, according to the negative role of SUMO conjugation on c-Myb capacity, we have observed rapid stress-induced down-regulation of the targets genes c-myc and bcl-2 of c-Myb. Our findings demonstrate that SUMO-2/3 proteins conjugate to c-Myb and negatively regulate its activity in cells under stress.

c-Myb is a DNA-binding transcription factor that plays a major role in the development of erythroid, myeloid, and lymphoid lineages of definitive hematopoiesis. The c-myb gene is abundantly expressed in immature proliferating hematopoi-etic progenitor cells but not in mature non-proliferating cells (1,2). The function of c-Myb as a regulator of hematopoiesis is achieved through transcriptional regulation of genes intimately involved in cellular processes such as proliferation, differentiation, and apoptosis (3,4). Its critical role in establishment of the definitive hematopoietic system has been demonstrated in experiments with targeted disruption of the c-myb gene, where c-myb null mutant mouse embryos die around day 15 in utero due to severe anemia (5). Originally, c-myb was identified as a transforming gene of two avian viruses, avian myeloblastosis virus and avian erythroleukemia virus E26 (6). c-myb can also be oncogenically activated following integration of replicationcompetent retroviruses into the c-myb locus in animal models (4). Oncogenic forms of c-Myb protein are almost invariably accompanied with truncations of the amino and carboxyl termini resulting in removal of structural regions that negatively regulate c-Myb activity (4).
The c-Myb protein is composed of three functional domains: an amino-terminal DNA-binding domain, a central transactivation domain, and a carboxyl-terminal negative regulatory domain. The DNA-binding domain in the NH 2 -terminal region of c-Myb consists of three imperfect tandem repeats and binds to DNA through the consensus sequence 5Ј-PyAAC(G/T)G-3Ј (7)(8)(9). c-Myb-dependent transcriptional activation depends on a centrally located acidic transactivation domain (9) and is stimulated through interaction with histone acetyltransferases p300/CREB 2 -binding protein (10,11). The COOH-terminal negative regulatory domain (NRD) plays an important role in regulation of the transactivation and transforming activities of Myb (12). In this domain the putative leucine zipper (13), the PEST/EVES motif (14), and other conserved sequences have been identified to exert negative regulation through binding of different cellular proteins (15,16). In addition, post-translational modifications of NRD such as phosphorylation (17), acetylation (18,19), ubiquitination (20 -23), and conjugation of SUMO-1 (small ubiquitin-like modifier type 1) (24,25) are crucial for modulation of the transactivation activity of c-Myb.
SUMOylation is a reversible process that regulates the function of target proteins in a manner akin to phosphorylation (26,27). The functional consequences of SUMOylation are not completely understood but this modification (unlike ubiquitination) does not directly involve targeting for proteasomal degradation. Rather it causes changes in protein-protein interactions, subnuclear localization, and conformational changes (26,27). Three different isoforms of SUMO (SUMO-1, -2, -3) are present in mammals. SUMO-2 and SUMO-3 are 97% identical to each other and about 66% homologous to SUMO-1 (28). At present, it is not entirely clear whether SUMO-1 and SUMO-2/3 play similar roles within cells. It was shown that oxidative stress, mild heat stress, or genotoxic stress (UV irradiation) cause a dramatic increase in the amount of SUMO-2/3 incorporated into high molecular weight complexes within cells (28). However, the identity of these target proteins that are modified by conjugation of SUMO-2/3 in response to different stresses largely remains to be identified. Whereas there is a growing list of proteins modified with SUMO-1, only a few substrates for SUMO-2/3 modifications are validated in vivo.
Here we show that Lys 523 and Lys 499 , located in NRD of c-Myb, are modified with SUMO-2/3 in vivo. This modification is greatly enhanced by several different stresses such as heat stress, osmotic stress, and metabolic stress and to a lesser extent by genotoxic stress. Two SUMO-E3 ligases, PIASy and PIAS3, enhance conjugation of SUMO-2/3 proteins to c-Myb in a qualitatively different way. In hematopoietic cells under stress, SUMO-2/3 conjugation to c-Myb at the endogenous protein levels demonstrates the physiological relevance of this posttranslational modification in regulation of c-Myb activity.
Transient Transfections-For transient transfections, COS-7 cells were plated at a density of 3 ϫ 10 5 cells/100-mm tissue culture dish 1 day prior to transfection. Transfections were carried out with 2 g of plasmid DNA/dish using the Effectene TM transfection reagent (Qiagen) according to the manufacturer's instructions. Expression of the transfected gene was analyzed 36 h post-transfection by Western immunoblotting.
Western Immunoblotting-Cells were disrupted in ice-cold lysis buffer (20 mM Tris-HCl (pH 7.5), 50 mM NaCl, 0.5% Nonidet P-40, 0.5% SDS, and 0.5% sodium deoxycholate) supplemented with N-ethylmaleimide (10 mM, Sigma) and a mixture of protease inhibitors (Complete TM , Roche Molecular Biochemicals). Cell lysates were sonicated for 20 s and clarified by centrifugation at 13,000 ϫ g for 15 min at 4°C. The c-Myb protein was immunoprecipitated with rabbit antiserum (24) and SUMO-2/3 proteins were immunoprecipitated with rabbit polyclonal anti-SUMO-2/3 antibody (33), kindly provided by Y. Azuma (NICHD, National Institutes of Health). Immunoprecipitated proteins or whole cell lysates were fractionated by SDS-PAGE on an 8% gel and electrophoretically transferred to a nitrocellulose membrane. Immunoblots were incubated with primary antibodies and visualized with SuperSignal West Pico chemiluminescent substrate (Pierce) in accordance with the manufacturer's instructions. Primary antibodies were anti-c-Myb monoclonal antibody (a kind gift of Eric Westin), anti-FLAG monoclonal antibody M2 (Sigma), and anti-HA monoclonal antibody (Covance), and anti-T7 monoclonal antibodies (Novagen).
Confocal Laser Microscopy-COS-7 cells were seeded onto glass coverslips and cotransfected with constructs encoding green (pEGFP-SUMO-2 or pEGFP-PML) and red (pDsRed2-cMyb) fluorescent fusion proteins. Thirty-six hours after transfection coverslips were fixed in 2% paraformaldehyde in phosphate-buffered saline for 10 min, washed two times in phosphate-buffered saline, and mounted on microscopic slides using Fluoromount G (Southern Biotechnology). Confocal images were acquired with a Leica DMIRBE inverted microscope (Leica) equipped with a digital scanning head (Leica SP2 Confocal Microscope) using excitation wavelengths of 488 nm for GFP and 594 nm for Red2. The channels were recorded independently and pseudocolor images were generated and superimposed using Imaris version 5.0.3 software (Bitplane). The acquired digital images were processed using Adobe Photoshop 6.0 software.
Transactivation Assay-COS-7 cells were plated at a density of 1 ϫ 10 4 cells/well in 12-well tissue culture plates and grown overnight prior to transfections. Transfections were carried out using the FuGENE 6 Transfection Reagent (Roche). Cells were cotransfected with 100 ng of the c-Myb-responsive reporter plasmid p5xMRE-A-luc (containing five copies of the mim-1A, Myb-responsive element (MRE) upstream of minimal herpes simplex virus thymidine kinase promoter and the firefly luciferase gene), 50 ng of pRL-tk-Renilla luciferase vector (Promega), and expression vectors encoding wild-type (cMybWT) or a mutant form of cMyb (cMyb2K/R) (100 ng of each). To assess the effect of PIAS3 expression on the transcriptional activity of c-Myb, increasing concentrations (1, 10, and 50 ng) of plasmid encoding murine PIAS3 was also transfected as indicated. The total amount of transfected DNA was adjusted to 500 ng with empty vector pcDNA3.1 (Invitrogen). Forty-eight hours post-transfection, cells were processed for both firefly and Renilla luciferase activities using the Dual Luciferase reporter assay system (Promega) and Turner TD-20e luminometer (Turner Designs). Each transfection experiment was performed in triplicate and repeated two times.
RNA Isolation and Northern Blot Analysis-Total RNA was prepared using the TRIzol reagent (Invitrogen). Samples containing 10 g of total RNA were resolved on a 1.2% agarose gel containing 0.25 M formaldehyde in MOPS buffer system. Separated RNAs were capillary transferred onto nylon membrane (Nytran SuperCharge, Schleicher & Schuell) and UV crosslinked using a Stratalinker 1800 (Stratagene). Blots were hybridized with cDNA probes labeled by the random priming method using Ready-to-Go DNA Labeling Beads (Amersham Biosciences). Quantitative analysis was performed with a Storm 840 phosphorimager using ImageQuant 5.1 software (GE Healthcare).

Post-translational Modification of c-Myb by Covalent
Attachment of SUMO-2/3-It has been demonstrated that the negative regulatory domain of c-Myb can be modified by SUMO-1 (24, 25) within its so-called PEST/EVES motif, and this modification is important for the inhibitory function of this domain. Whereas many targets have no preferences for SUMO isoforms, others are preferentially modified by one isoform, for example, RanGAP1 with SUMO-1 (28) or topoisomerase-II by SUMO-2/3 (34). To determine whether, in addition to SUMO-1, c-Myb is modified in vivo by SUMO-2/3, we analyzed transiently transfected COS-7 cells with plasmids encoding either wild-type c-Myb (c-MybWT), HA-tagged SUMO-2 (HA-SUMO-2), or both. Thirty-six hours post-transfection the cells were lysed, c-Myb was immunoprecipitated and analyzed by Western immunoblotting with anti-HA and anti-c-Myb monoclonal antibodies. As shown in Fig. 1A, anti-HA antibody recognized two HA-immunoreactive bands in the immunoprecipitates from cells transfected with both c-MybWT and HA-SUMO-2-encoding constructs, but not in cells transfected with either construct alone. Anti-c-Myb monoclonal antibodies were used to confirm that similar levels of c-Myb were expressed and immunoprecipitated from cells transfected with c-MybWT. Anti-c-Myb monoclonal antibodies also detected a SUMO-2-modified form migrating ϳ20 kDa above the nonmodified form of c-MybWT (Fig. 1A). Transfection efficiency was evaluated by Western immunoblotting of total cellular proteins. Anti-HA antibody detected a ladder of bands representing the HA-SUMO-2-conjugated cellular proteins only in cells transfected with HA-SUMO-2 (Fig. 1B). Anti-c-Myb monoclonal antibodies detected the c-MybWT protein (75-kDa form) only in cells transfected with the c-MybWT construct (Fig. 1B). Identical results were obtained with HA-tagged SUMO-3 constructs (data not shown). Both proteins SUMO-2 and-3 are virtually identical (in the processed form they differ only in 3 amino acids) and presently no antibodies are available that would distinguish between these two isoforms. All the figures show representative results that were obtained with the  DECEMBER 29, 2006 • VOLUME 281 • NUMBER 52 HA-SUMO-2 construct, even though all experiments in COS-7 cells were performed with both SUMO-2 and SUMO-3 with virtually identical results. These experiments clearly show that SUMO-2/3 proteins efficiently modify c-Myb. We also observed that, at least in COS-7 cells, SUMO-2/3 have a higher affinity toward c-Myb than SUMO-1. Transfection of c-MybWT and equimolar amounts of either HA-SUMO-2 or HA-SUMO-3 leads to conjugation of SUMO-2/3, where 20 -40% of c-Myb is detected in the SUMOylated form, whereas transfected c-MybWT and SUMO-1 leads to only 5-10% of SUMOylated c-Myb (Refs. 24 and 25 and data not shown). Previously we have shown that SUMO-1 protein conjugates to two lysines, Lys 499 and Lys 523 , located in the negative regulatory domain of c-Myb. Sequences around both Lys 499 and Lys 523 conform perfectly to the minimal SUMO modification consensus sequence ⌿KXE (where ⌿ is a hydrophobic amino acid and X any amino acid). Interestingly, we also observed that the single mutation K523R completely abolished modification of c-Myb with SUMO-1, suggesting that SUMOylation of c-Myb is an interdependent process, where modification of Lys 523 is required for modification of the second Lys 499 to occur. To determine whether the same lysines are sites of SUMO-2/3 conjugation, we cotransfected COS-7 cells with HA-SUMO-2/3 and wild-type c-Myb, as well as c-Myb constructs in which Lys 499 (cMybK499R), Lys 523 (cMybK523R), or both (cMyb2K/R) were changed to arginines. Lysates from transfected cells were immunoprecipitated with anti-c-Myb polyclonal antiserum and analyzed by immunoblotting using anti-HA and anti-c-Myb antibodies (Fig. 1C). These results demonstrate that Lys 499 and Lys 523 are major modification sites for SUMO-2/3 proteins, with an obvious preference for modification of Lys 523 with SUMO-2/3. The single mutation K523R strongly inhibited modification of c-Myb with SUMO-2/3 at both sites. Thus, conjugation of SUMO-2/3 is ordered, where the first modification of c-Myb with SUMO-2/3 takes place almost exclusively at Lys 523 , and is necessary for the attachment of the second molecule to Lys 499 . As expected, mutation of both sites completely abolished modification of c-Myb with SUMO-2/3. Thus, Lys 499 and Lys 523 are the two major sites in the NRD of c-Myb modified with SUMO-2/3.

Stress-induced SUMOylation of c-Myb
Subnuclear Co-localization of c-Myb with SUMO-2-Next, we examined the co-localizations of c-Myb with SUMO-2/3 in COS-7 cells. COS-7 cells were transfected with constructs encoding the fusion fluorescent proteins EGFP-SUMO2 and DsRed2-cMyb, and analyzed the cells by confocal microscopy. DsRed2-cMyb was detected in the nucleus in the form of bright nuclear speckles, as well as in a fine granular form. GFP-SUMO-2 protein had a more homogenous distribution in the nucleus than c-Myb, but overall there was a strong co-localization of DsRed2-cMyb and EGFP-SUMO-2 ( Fig. 2A; superimposition of confocal images). It was shown previously that c-Myb associates with PML in vivo in very bright dot-like structures that represent PML (ND10) nuclear bodies (35). We also observed co-localization of c-Myb and PML in COS-7 cells when these cells were cotransfected with HA-SUMO-2, and fluorescent fusion proteins encoding DsRed2-cMyb and EGFP-PML (Fig. 2B). However, the majority of DsRed2-cMyb protein was localized outside of GFP-PML-positive nuclear bodies.
Thus, modification of c-Myb with SUMO-2/3 proteins seems to occur in subnuclear compartments that may differ from PML (ND10) nuclear bodies.
Stress Induces Conjugation of SUMO-2/3 Proteins to c-Myb-It was shown previously that SUMO-2/3 proteins are more abundant in their free form than SUMO-1 in COS-7 cells. Additionally, Saitoh and Hinchey (28) reported that pools of free SUMO-2/3 decrease rapidly when cells are exposed to heat, ethanol, or hydrogen peroxide. Therefore, SUMO-2/-3 modification of cellular proteins may be involved in the cellular response to environmental stresses. To investigate whether conjugation of SUMO-2/3 to c-Myb is modulated by stress, COS-7 cells transfected with HA-SUMO-2/3 and wild-type c-Myb were subjected to different forms of stress and analyzed by immunoprecipitation and Western immunoblotting. As shown in Fig. 3A, the amount of SUMO-2/3 conjugated to c-Myb dramatically increased after exposure of cells to heat stress (43°C for 30 min), osmotic stress (0.7 M NaCl for 30 min), and also metabolic stress (ethanol 7% for 30 min). In contrast, genotoxic stress (UV irradiation) increased conjugation only mildly (Fig. 3B). With the environmental stress, we observed, not only was there a dramatic increase in modification of two major SUMOylation sites Lys 523 and Lys 499 , but also several slower migrating species that correspond to c-Myb modified with more than two molecules of SUMO-2/3 (Fig. 3A). SUMO-2 and SUMO-3 contain the consensus SUMO modification site (KXE) in their N-terminal regions and it is absent in the sequence of SUMO-1. These sites have been shown to be utilized by SUMO-E1-activating and SUMO-E2-conjugating enzymes to form polymeric chains of SUMO-2/3 on protein substrates in vitro, and SUMO-2/3 chains have also been detected in vivo (32). Thus, although all SUMO species share the same conjugation machinery, modification by SUMO-1 and SUMO-2/-3 may have distinct functional consequences. We decided to explore the possibility that the slower migrating bands detected in cells under stress (Fig. 3A) are actually c-Myb species modified by the conjugation of poly-SUMO-2/3 chains. Mutation K11R in SUMO-2/3 proteins (HA-SUMO-2/3- K11R) destroys the ability of the proteins to create polymeric chains in vitro and in vivo (32). COS-7 cells were transfected with c-Myb and either wild-type or mutated forms (K11R) of SUMO-2/3 proteins. Thirty-six hours post-transfection cells were subjected to heat stress, lysed, and c-Myb protein was immunoprecipitated. Immunoprecipitates were subjected to Western immunoblotting analysis with anti-HA and anti-cMyb antibodies. Heat stress strongly induced SUMOylation of c-Myb and none of the SUMO-2/3-modified c-Myb forms were affected by cotransfection of c-Myb with the HA-SUMO-2/3-K11R mutants (Fig. 3C). Thus, this result suggests that Lys 11 in the NH 2 -terminal part of SUMO-2/3 is not involved in the creation of the poly-SUMO-2/3 chains that could be attached to c-Myb. However, it is still possible that the slower migrating forms of c-Myb contain the endogenously expressed SUMO-2/3 isoforms capable of forming chains.
Further data showed that minor SUMOylation sites, in addition to Lys 523 and Lys 499 , exist in c-Myb and can be used under stress as acceptor sites for conjugation of SUMO-2/3 proteins. As shown on Fig. 3D, heat stress induced conjugation of SUMO-2/3 proteins not only to wild-type but also to mutant forms of c-Myb. SUMOylation of mutants K523R and 2K/R is much weaker than that of wild-type, but we could still clearly detect some conjugation of SUMO-2/3 to the cMyb2K/R mutant that is not observed under normal growth conditions. The SUMOylated forms of both K523R and 2K/R mutants migrated again approximately as mono-, di-and, multiple SUMOylated species. (Fig. 4A). In addition, conjugation of SUMO-2/3 to c-Myb is independent of de novo protein synthesis, as treatment of transfected cells with cycloheximide (an inhibitor of protein synthesis) did not influence the stress-induced SUMOylation (data not shown). This prompt and protein synthesis-independent accumulation of SUMO-2/3-modified c-Myb suggests the existence of a stress-associated signaling pathway that activates SUMO-2/3 conjugation. The JNK and p38 MAPK pathways represent the two major MAPK cascades in mammalian cells that are activated in response to a variety of stress signals (36). We tested whether either of these pathways were activated in COS-7 cells in response to heat stress. COS-7 cells were subjected to heat stress for the indicated times in the presence or absence of an inhibitor of the p38 MAPK pathway, SB 202190. Western immunoblotting with phospho-specific anti-P-p38MAPK antibodies revealed a rapid and strong induction of phosphorylation of Thr 180 /Tyr 182 in p38 MAPK. This phosphorylation was inhibited in cells treated with SB 202190 (Fig. 4B, upper panel). Immunoblotting with anti-p38 MAPK confirmed equal loading and expression of p38 MAPK in the tested cell lysates (Fig. 4B, lower panel). Similar activation of the JNK pathway in response to stress was also observed (data not shown). To determine whether either of these pathways is involved in stress-induced conjugation of SUMO-2/3 proteins to c-Myb, we analyzed SUMOylation of c-Myb in cells subjected to heat stress and treated with inhibitors of the p38 MAPK (SB 202190) and JNK (SP 600125) pathways as indicated (Fig. 4C). To our surprise, inhibition of either p38 MAPK, or JNK, or both pathways did not influence heat stress-increased SUMOylation of c-Myb (Fig. 4C).  DECEMBER 29, 2006 • VOLUME 281 • NUMBER 52

JOURNAL OF BIOLOGICAL CHEMISTRY 40069
Phosphorylation of Ser 528 in the PDSM Motif Only Modestly Affects Stress-induced SUMOylation of c-Myb-Recently, the bipartite motif named PDSM (phosphorylation-dependent sumoylation motif) was identified in several SUMO protein substrates (37). It comprises a SUMO consensus site and a proline-directed phosphorylation site, separated by two amino acids (KXEXXSP). Phosphorylation of the critical serine in the PDSM motif was shown to be required for SUMOylation of the neighbor lysine residue in many unrelated transcriptional regulators, including the erythroid transcription factor GATA-1, myocyte-specific enhancer factors 2A and 2D (MEF2A and MEF2D), and two members of the heat-shock factor family, HSF1 and HSF4b (37,38). The sequence Ile-Lys 523 -Gln-Glu-Val-Glu-Ser, which surrounds the primary SUMOylation site Lys 523 within the PEST/EVES domain of c-Myb, contains a perfect PDSM consensus sequence that is conserved across species. It was shown that Ser 528 serves as a substrate for the p42/44 MAPK signaling pathway. Substitution of S528A increased transactivation of c-Myb on some promoters, suggesting that phosphorylation of this residue negatively regulates the transactivational activity of c-Myb (17). To test whether the phosphorylation of Ser 528 has any influence on the conjugation of SUMO-2/3 to c-Myb under stress, we subjected wild-type c-Myb, as well as two mutant forms (cMybS528A and cMybS528D) to an in vivo SUMOylation assay. The results of this experiment show that both mutant forms of c-Myb, one that cannot be phosphorylated at Ser 528 (cMybS528A) and one (cMybS528D) that through replacement of serine with acidic amino acid mimics phosphorylation of this residue, had an overall similar level of SUMO-2/3 modification (Fig. 5). The only subtle difference we reproducibly observed was a slight increase in conjugation of two and three molecules of SUMO-2/3 to cMybS528A mutant under stress (Fig. 5, asterisks).
Qualitatively Different Enhancement of SUMO-2/3 Conjugation to c-Myb by PIAS3 and PIASy-PIAS family proteins have been identified as either positive or negative regulators of many transcriptional factors or cofactors (39). It was shown previ-  ously that PIASy enhances conjugation of SUMO-1 to c-Myb (25). However, homozygous mutant mice with a targeted Piasy gene have a normal level of c-Myb SUMOylation, suggesting that more than one cofactor may be involved in c-Myb SUMOylation (40). Therefore, we investigated whether PIASy and/or other members of the PIAS family of proteins function as SUMO-2/3 E3 ligases for c-Myb by screening all known mammalian members of the PIAS SUMO E3 ligase family for their capacity to modulate the extent of SUMO-2/3 conjugation to c-Myb. Although cotransfection of PIAS1, PIASx␣, or PIASx␤ did not have a significant effect on the SUMOylation of c-Myb (data not shown), coexpression of PIASy (T7-PIASy, 200 ng) or PIAS3 (FLAG-PIAS3, 200 ng) with c-Myb (300 ng) and HA-SUMO-2/3 (100 ng) strongly induced modification of c-Myb with SUMO-2/3 (Fig. 6A). This experimental setup did not lead to detectable modification of c-Myb by conjugation of SUMO-2/3 in the absence of stress. However, in the presence of stress SUMOylation of c-Myb was easily detected (Fig. 6A, first  two lanes). Coexpression of SUMO E3 ligases, PIAS3 and PIASy, strongly induced conjugation of SUMO-2/3 to c-Myb even in the absence of stress. Stress increased the modification of c-Myb with PIAS3 but not PIASy coexpression. We also observed that the PIAS3 and PIASy SUMOylation profile differed. Whereas PIASy mainly increased conjugation by one or two molecules of SUMO-2/3 (most probably to the major SUMOylation sites Lys 523 and Lys 499 in NRD of Myb), coexpression of PIAS3 enhanced modification of c-Myb that results in slower migrating forms (corresponding to conjugation of two and more molecules of SUMO-2/3 conjugated to c-Myb). Also, we noticed that simple coexpression of PIAS3 in COS-7 resulted in profile of SUMOylated c-Myb that was similar to that observed in cells subjected to stress (Fig. 6A). PIAS proteins are characterized by a central cysteine-rich SPRING (Siz/ PIAS RING) domain, which is exclusively found in SUMO E3 ligases and is required for SUMOylating activity (41). To confirm that the c-Myb-specific SUMO E3 ligase activity of PIAS3 requires the SPRING domain, we prepared a mutant form of PIAS3, where the central Cys 334 of the SPRING domain was replaced with Ser (PIAS3Mut; C334S) and used it in our in vivo SUMOylation assay. As shown in Fig. 6B, conjugation of SUMO-2/3 to c-Myb was severely compromised in the presence of the mutant, both in cells propagated under regular growth conditions and those subjected to heat stress.
Suppression of Transactivation Activity of c-Myb through Conjugation of SUMO-2/3-Although post-translational modification by SUMO has diverse effects, in most cases SUMOylation has been found to inhibit transcription (26,27). Mutation of SUMOylation sites in the NRD of c-Myb increased its transactivation capacity, suggesting that conjugation of SUMO negatively regulates its activity (Fig. 7A) (24,25). One consequence of SUMOylation is to support interaction of transcription factors with co-repressors from HDAC families (42,43). To explore whether coexpression of SUMO-2/3 leads to decreased transactivation activity of c-Myb, and whether this effect is sensitive to inhibition of histone deacetylases (HDACs), COS-7 cells were transfected with c-Myb-responsive reporter plasmid 5xMRE-A-luc, wild-type c-Myb, and HA-SUMO-2. Thirty-six hours post-transfection cells were treated with trichostatin A (100 ng/ml) or phosphate-buffered saline for another 12 h and analyzed by luciferase assay. As shown in Fig. 7B, c-Myb was able to activate the reporter gene activity ϳ3-fold and this activity was strongly inhibited by co-expression of HA-SUMO-2. Interestingly, treatment of cells with HDAC inhibitor trichostatin A not only abolished this repression, but resulted in an ϳ2-fold increase in c-Myb transactivation. PIASy was previously shown to enhance SUMO-1 conjugation to c-Myb, however, its effect on the activity of c-Myb could not be evaluated because its expression has a strong, nonspecific effect on reporter constructs (25). Because PIAS3 also increases conjugation of SUMO-2/3 to c-Myb, we asked whether PIAS3 regulates c-Myb-induced transcriptional activation. When FLAGtagged PIAS3 was co-transfected with wild-type c-Myb, a strong and dose-dependent reduction in gene activation was observed (Fig. 7C). The mutant form of c-Myb (cMyb2K/R) activated luciferase gene expression about 2-fold stronger than wild-type protein in this assay. Surprisingly, the cMyb2K/R mutant was also inhibited, but to a lesser extent, by PIAS3 (Fig.  7C). These results demonstrate that the PIAS3 protein is able to inhibit c-Myb-mediated gene activation and that this inhibition can be, at least in part, independent of c-Myb modification by SUMO-2/3. However, we cannot rule out the possibility that PIAS3 induces modification of minor SUMOylation sites such as those that are modified in the c-Myb2K/R mutant in cells subjected to stress (Fig. 3D).
Stress Regulates the Activity of Endogenous c-Myb in Hematopoietic Cells-We have shown that in COS-7 cells, with transiently expressed c-Myb and SUMO-2/3 proteins,  DECEMBER 29, 2006 • VOLUME 281 • NUMBER 52 stress induces inactivation of c-Myb through rapid conjugation of SUMO-2/3. Next, we looked for evidence of stressinduced inactivation of c-Myb in hematopoietic cells, a physiologically relevant system where c-Myb and SUMO-2/3 proteins are expressed at normal endogenous levels. Erythroleukemic cells DS19 were subjected to heat stress for the indicated time points and modification of c-Myb was detected by immunoprecipitation and Western immunoblotting using anti-c-Myb antibody. Two forms of c-Myb, normal and an alternatively spliced form (Fig. 8A, open triangle) were identified in cells growing without stress. Shortly after heat stress we observed accumulation of slower migrating forms of c-Myb and these increased over the duration of stress (Fig. 8A, closed triangles). The electrophoretic mobil-ity of these forms was similar to SUMO-2/3 modified forms of c-Myb detected in COS-7 cells. We also noted that heat stress increased the steady state level of c-Myb in DS19 cells. This is consistent with a previous finding that c-Myb protein accumulates in heat-stressed cells due to the decreased proteolytic turnover of c-Myb (44). To determine whether the slower migrating forms of c-Myb contain SUMO-2/3, we immunoprecipitated SUMO-2/3-conjugated proteins and analyzed immunoprecipitates by Western immunoblotting to detect c-Myb. In stress-treated cells, anti-c-Myb antibody detected two bands that represent c-Myb conjugated with one and two molecules of SUMO-2/3 (Fig. 8B, closed triangles). Next we evaluated the activity of c-Myb on target gene activity in stress-treated hematopoietic cells by Northern  A, murine erythroleukemic cells DS19 were subjected to heat stress as indicated. After treatments, c-Myb protein was immunoprecipitated (IP) with polyclonal rabbit anti-c-Myb antiserum (␣-cMyb). Immunoprecipitates were separated by SDS-PAGE and analyzed by Western immunoblotting (WB) using anti-c-Myb (␣-cMyb) monoclonal antibody. B, treated and untreated DS19 cells were lysed and SUMO-2/3 proteins were immunoprecipitated with polyclonal rabbit anti-SUMO-2/3 antiserum (␣-SUMO-2/3) and analyzed by Western immunoblotting using anti-c-Myb monoclonal antibody. The open triangle marks the position of an alternatively spliced form of c-Myb, closed triangles mark the positions of SUMO2/3 modified species of c-Myb. C, total RNAs were isolated from heat stress-treated and untreated DS19 cells and analyzed by Northern blot hybridization using radioactively labeled probes encoding murine c-myc, c-myb, bcl-2, and gapdh cDNAs. D, the relative amount of c-myc, c-myb, and bcl-2 mRNAs in DS19 cells treated or untreated with heat stress. Quantitative analysis of Northern blots shown in C was performed on a Storm 840 phosphorimager using ImageQuant 5.1 software. Values were normalized to gapdh mRNA expression levels. Normalized values of mRNAs in untreated cells were assigned to 100%.

Stress-induced SUMOylation of c-Myb
mapping. The region in the NRD between amino acids 460 and 528 is conserved and contains several Thr residues that are potential phosphorylation sites for protein kinases (58). We prepared mutants of all Thr residues from this region and tested them in our SUMOylation assay, but no single Thr/Ala mutation affected the stress-induced SUMOylation of c-Myb (data not shown).
We have identified two c-Myb-specific SUMO E3 ligases from the PIAS family, PIASy and PIAS3. PIASy was described previously as a c-Myb-specific SUMO E3 ligase for constitutive conjugation of SUMO-1 to c-Myb (25). Interestingly, we have observed that PIASy and PIAS3 have slightly different activities regarding conjugation of SUMO-2/3 to c-Myb. PIASy primarily catalyzes conjugation of one or two molecules of SUMO-2/3 to Lys 523 and Lys 499 . PIAS3 enhances modification of c-Myb with two and more molecules of SUMO-2/3 and strongly resembles stress-induced SUMOylation of c-Myb. Conjugation of SUMO-2/3 protein to c-Myb was severely compromised in cells that coexpressed the SPRING domain mutant under stress, arguing that PIAS3 may play an important role in stress-induced conjugation of SUMO-2/3 to c-Myb. Stress may also induce phosphorylation of SUMO E3 ligase that in turn results in the enhanced conjugation of SUMO-2/3 to specific targets. In regard to this, it was reported very recently that stressinduced phosphorylation of SUMO-E3 ligase PIASx␣ causes enhanced SUMOylation of the transcription factor Elk-1 (51).
In summary, data presented here reveals that the c-Myb transcription factor is dynamically regulated by conjugation of SUMO-2/3 proteins. Thus, we uncovered a novel post-translational modification of c-Myb with a strong impact on its activity. Stress-induced modification of c-Myb with SUMO-2/3 was also confirmed in hematopoietic cells at normal endogenous protein levels. This highlights the physiological relevance of this modification for c-Myb and raises the possibility that this novel post-translational modification is part of a regulatory pathway that controls the activity of c-Myb in cellular environments, such as hematopoietic tissues, where it plays critical roles.