Differential Regulation of c-Myb-induced Transcription Activation by a Phosphorylation Site in the Negative Regulatory Domain*

The c- myb protooncogene encodes a highly conserved 75–89-kDa transcription factor that contains three functional domains, an amino-terminal DNA binding domain (DBD), a central acidic transactivation domain, and a carboxyl-terminal negative regulatory domain (NRD). Two acute transforming retroviruses, avian myeloblastosis virus and the E26 leukemia virus, transduced por- tions of c- myb and encode Myb proteins that are truncated in both the DBD and the NRD. Several conserved potential sites for phosphorylation by proline-directed serine/threonine protein kinases reside in or near the NRD, suggesting that phosphorylation might play a role in regulating c-Myb. We have previously demonstrated that serine 528, located in the NRD, is a target for p42 mapk in vitro . Serine 528 is phosphorylated in vivo in several cell lines, and substitution of serine 528 to ala- nine (S528A) resulted in an increased ability of Myb to transactivate a synthetic promoter containing five cop- ies of the mim -1A Myb-responsive element and a minimal herpes tk promoter. We have tested the ability of S528A Myb to transactivate a series of cellular target promoters and report that the serine to alanine substi- tution increased the ability of Myb to activate transcription from the CD34 promoter but not the c- myc or mim -1 promoters. This suggests that phosphorylation of serine 528 may differentially regulate c-Myb activity at differ- ent promoters. The DNA binding and multimerization activities of c-Myb appear to be unaffected by the S528A

The c-myb protooncogene encodes a highly conserved 75-89-kDa transcription factor that contains three functional domains, an amino-terminal DNA binding domain (DBD), a central acidic transactivation domain, and a carboxyl-terminal negative regulatory domain (NRD). Two acute transforming retroviruses, avian myeloblastosis virus and the E26 leukemia virus, transduced portions of c-myb and encode Myb proteins that are truncated in both the DBD and the NRD. Several conserved potential sites for phosphorylation by proline-directed serine/threonine protein kinases reside in or near the NRD, suggesting that phosphorylation might play a role in regulating c-Myb. We have previously demonstrated that serine 528, located in the NRD, is a target for p42 mapk in vitro. Serine 528 is phosphorylated in vivo in several cell lines, and substitution of serine 528 to alanine (S528A) resulted in an increased ability of Myb to transactivate a synthetic promoter containing five copies of the mim-1A Myb-responsive element and a minimal herpes tk promoter. We have tested the ability of S528A Myb to transactivate a series of cellular target promoters and report that the serine to alanine substitution increased the ability of Myb to activate transcription from the CD34 promoter but not the c-myc or mim-1 promoters. This suggests that phosphorylation of serine 528 may differentially regulate c-Myb activity at different promoters. The DNA binding and multimerization activities of c-Myb appear to be unaffected by the S528A substitution, suggesting that phosphorylation of serine 528 may mediate its effect on the transcription transactivating activity of c-Myb by regulating interactions with other proteins.
The c-myb proto-oncogene encodes a nuclear DNA binding protein that functions as both a transcriptional activator and repressor (1,2). Expression of c-myb is detected primarily in hematopoietic tissue although c-myb mRNA has been reported in primary chicken embryo fibroblasts (3), smooth muscle cells (4), and several non-hematopoietic human tumors including neuroblastoma (5), colon carcinoma (6), small cell lung carcinoma (7), and breast carcinoma (8). The down-regulation of c-myb expression is associated with hematopoietic maturation, and in each hematopoietic lineage examined the expression of c-myb mRNA and protein is highest in immature normal tissue and tumor cell lines (1). This pattern of expression led to the hypothesis that the c-myb gene product would play a role in regulating hematopoiesis, and this has been supported experimentally in several systems. First, c-myb antisense oligodeoxynucleotides inhibit both erythroid and myeloid colony formation in vitro (9). Second, murine erythroleukemia cells and leukemic myeloid cells stably transfected with either constitutively expressed or inducible c-myb expression vectors were blocked in their ability to terminally differentiate in response to chemical inducing agents in vitro (10,11). Third, transgenic mice lacking a functional c-myb gene developed normally to day 14 and after that they died with severely disrupted patterns of erythroid and myeloid development (12). Finally, Badiani et al. (13) demonstrated that transgenic mice carrying a dominant interfering c-myb allele under the control of a CD2 promoter had disrupted patterns of T-lymphogenesis.
The murine c-Myb protein is a 636-amino acid peptide of approximately 75 kDa (1). Chicken c-myb was identified as the cellular homologue of two viral myb genes carried by the avian myeloblastosis virus (AMV) 1 and the E26 virus (14,15). These viruses independently transduced portions of the c-myb gene, deleting both amino-and carboxyl-terminal coding sequences. Three major functional domains have been defined on the c-Myb protein: 1) a DNA binding domain, 2) an acidic transactivation domain, and 3) a negative regulatory domain. The DNA binding domain (DBD) is located near the amino-terminal end of the protein and is a highly conserved region consisting of three imperfect repeats (R1-3) of 50 -52 amino acids each. The DNA binding domain defines a family of Myb-related proteins that have been identified in humans, mice, chickens, Drosophila, yeast, slime molds, and plants (1). Biedenkapp et al. (16) demonstrated that Myb bound DNA specifically to the consensus sequence PyAAC(G/T)G. This led to the finding that Myb could transactivate transcription when this element, referred to as the Myb-responsive element (MRE), was ligated to various test promoters (17,18). Although the DBD provides the only sequences required for DNA binding, it does not appear to activate transcription by itself. Deletion analysis and linker scanning of both c-and AMV v-Myb led to the identification of an approximately 50-amino acid acidic region, carboxyl-terminal to the DBD, that is required for transactivation by Myb (17)(18)(19). The third major functional region is the broadly de-fined negative regulatory domain (NRD). Deletion of this region results in up to a 10-fold increase in Myb-induced transcription activation when assayed by transient co-transfection with several test promoters (18 -20). This region contains a leucine zipper-like motif (16), and point mutations of key residues in this motif increase both the transcription activating and transforming activity of murine Myb (21). Thus, it is of particular interest that potential protein-protein interactions have been identified in blotting studies using a peptide containing this motif to probe HeLa cell extracts (21). Several other features of the NRD are also consistent with it playing a role in regulating c-Myb activity. First, it is a site of proviral integration as well as modification by alternative splicing (15). Second, disruption of the carboxyl-terminal portion of AMV v-Myb (which retains a portion of the NRD) by linker insertion results in an increase in the transactivating activity of AMV v-Myb (22). Third, both amino-and carboxyl-terminal deletions have been reported to independently activate the transforming potential of c-Myb. Finally, cells transformed by differentially truncated forms of c-Myb differ phenotypically (23,24). This finding suggests that changes in the structure of c-Myb may be important in determining cell lineage and stage of development in hematopoietic cells presumably by altering the array of genes regulated by c-Myb.
The sequence-specific DNA binding by Myb proteins led to the demonstration that both c-and v-myb encoded transcription factors (17,18), and a number of candidate target promoters have been identified, including mim-1 (25), c-myc (26), CD34 (27), CD4 (28), T cell receptor-␦ (29), and a murine thymic locus control region (30). Each of these promoters requires the DBD for transactivation by Myb and contains at least one Myb binding site. The sequence of these Myb binding sites is surprisingly variable, and Myb proteins bind these sites with varying affinity (31). In the case of the mim-1 promoter there are three MRE's that bind Myb with different affinities in vitro (25,31). Mutation of the highest affinity site (mim-1A) abrogates activation by Myb proteins while mutation of the lower affinity sites has less of an effect. In some cases, c-Myb has been demonstrated to activate transcription in cooperation with Ets-2, core binding factor and NF-M (25,29,32,33). In contrast, the DBD is not required for the activation of transcription from the human hsp 70 promoter or the avian MD1 promoter, and it has been suggested that c-Myb may bind and inactivate a negative acting transcription factor (34,35). Thus, c-Myb may regulate gene expression via at least two distinct mechanisms. However, little is known about how c-Myb activates transcription or the mechanisms that regulate c-Myb activity.
Both v-and c-Myb are phosphorylated on serine and threonine at multiple sites in vivo (36 -39). Lü scher et al. (36) demonstrated that serines 11 and 12 are phosphorylated in vivo and are targets for phosphorylation in vitro by casein kinase II. Phosphorylation of serines 11 and 12 results in decreased sequence-specific DNA binding in vitro, and substitution of these sites by alanine results in decreased cooperativity with NF-M (40). In addition, c-Myb becomes hyperphosphorylated at several unidentified sites during mitosis, and mitotic c-Myb binds DNA less efficiently than interphase c-Myb (41). We have previously demonstrated that murine c-Myb is phosphorylated on serine 528 (which lies within the NRD) in vivo and that it is a target for phosphorylation by p42 mapk in vitro. Substitution of serine 528 by alanine results in 3-7-fold increase in the ability of c-Myb to activate transcription from an artificial promoter/ reporter construct consisting of five copies of the mim-1A MRE and a minimal herpesvirus tk promoter (38,42). We now demonstrate that substitution of serine 528 by alanine modulates the transcription activating properties of c-Myb on some target promoters, but not others, suggesting the phosphorylation of serine 528 provides a mechanism to differentially regulate c-Myb activity. Interestingly, this substitution does not affect the ability of c-Myb to bind DNA or to form multimerized complexes. We suggest that phosphorylation of serine 528 may serve to regulate the interaction between c-Myb and other proteins that modulate c-Myb activity.

MATERIALS AND METHODS
Cell Lines and Culture Conditions-The African green monkey kidney cell lines, CV-1 and CMT3COS, were obtained from Dr. David Rekosh (University of Virginia) and maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum (Hyclone) and 2 mM glutamine (Life Technologies, Inc.) in a humidified incubator at 37°C in 10% CO 2 . The CMT3COS cell line is a derivative of CV-1 and is transformed by SV40 large T antigen driven by the mouse metallothionein I promoter (43).
Plasmids-The murine c-myb expression vector pRMb3SV.wt (44) contains the entire murine c-myb coding sequence driven by the Rous sarcoma virus long terminal repeat and sequences for polyadenylation from SV40. The empty control vector pR3SV that lacks c-myb coding sequences was generated by NcoI digestion, which removes the c-myb coding sequences, and religation. Oligodeoxynucleotide mutagenesis was used to create alanine (pRMb3SV.S528A) and aspartate (pRMb3SV.9) substitution mutations at serine 528 of murine c-Myb. Briefly, the 2-kilobase pair NcoI fragment, containing the murine c-myb coding sequences, was isolated from pRMb3SV.wt (44,45) and subcloned into pALTER (Promega). Oligodeoxynucleotide-directed mutagenesis was carried out using the Altered Sites II System (Promega) per the manufacturer's instructions. Mutant cDNAs were then subcloned into pR3SV via their NcoI sites, creating pRMb3SV.S528A and pRMb3SV.9. Insert orientation was confirmed by restriction endonuclease digestion, and the presence of appropriate mutations was confirmed by double-stranded dideoxynucleotide chain termination DNA sequencing using a Sequenase version 2.0 kit per the manufacturer's instructions (U. S. Biochemical Corp.).
The SV40 origin of replication was introduced into pR3SV, pRMb3SV.wt, pRMB3SV.S528A, and pRMb3SV.9 for autonomous replication in the SV40 large T antigen-transformed CMT3COS cell line. Briefly, pSV4Ori, which carries a 376-bp EcoRI/HindIII fragment containing the SV40 origin of replication ligated into pUC18, was obtained from Dr. David Rekosh (University of Virginia). pSV4Ori was digested with EcoRI and HindIII and made blunt-ended by filling using Klenow fragment, and the 376-bp fragment was isolated by agarose gel electrophoresis. pR3SV, pRMb3SV.wt, pRMb3SV.S528A, and pRMb3SV.9 were linearized with BamHI, made blunt-ended by filling using Klenow fragment, treated with calf intestinal phosphatase, and ligated with the 376-bp fragment to yield pR3SVori, pRMb3SVori.wt, pRMb3SVori.S528A, and pRMb3SVori.9. In pRMb3SV-based plasmids, the unique BamHI site lies immediately 3Ј to the polyadenylation sequences.
To generate pOriGST.FLwt and pOriGST.FLS528A, pR3SVori was linearized by digestion with HindIII and BglII and treated with calf intestinal phosphatase. Polymerase chain reaction was used to amplify sequences encoding GST-fusion proteins from pG2T.wt, pG2T.S528A, pGEX-2T, and pGEX2TK.E1A12S. The pG2T.wt and pG2T.S528A plasmids encode GST fused amino-terminally to full-length wild type or S528A murine c-Myb, respectively (42). pGEX2TK.E1A12S was obtained from Dr. Daniel Engel (University of Virginia) and encodes GST fused amino-terminally to the full-length adenovirus Type 2 E1A 12S coding sequence. Forward (forgex2: 5Ј-GACAAGCTTGCCATG TCCCCTATACTAGGTTATTGG-3Ј) and reverse (revgex: 5Ј-CA-GAGATCTTCAGTCAGTCACGATGAATTCC-3Ј) oligodeoxynucleotide primers for polymerase chain reaction amplification were commercially synthesized (Oligos Etc.). Forgex2 includes sequences beginning at the translation initiation codon of GST (nucleotide position 270 in pGEX-2T, underlined above) and ending within the GST coding region at nucleotide position 294. Forgex2 also contains a HindIII site near the 5Ј-end and a minimal Kozak sequence (46) (corresponding to nucleotides Ϫ3 through Ϫ1 in murine c-myb) immediately 5Ј to the GST translation initiation codon. Revgex contains sequences corresponding to nucleotides 961 through 939 in the polylinker of pGEX-2T plus a BglII site at the 5Ј-end. Each 100-l reaction included 500 ng each of forgex2 and revgex primers, 20 ng of DNA template, 200 M dNTPs, 10 l of 10 ϫ reaction buffer (100 mM KCl, 100 mM (NH 4 )SO 4 , 200 mM Tris-HCl (pH 8.8 at 25°C), 20 mM MgSO 4 , 0.1% Triton X-100) and 0.5 l (1 unit) Deep Vent DNA polymerase (New England Biolabs). Additional MgSO 4 was added to bring the final concentration to 6 mM. Amplification was carried out in a Perkin-Elmer DNA thermocycler under the following conditions: 30 cycles of 94°C for 1 min, 50°C for 2 min, and 72°C for 4 min, followed by one extension cycle at 72°C for 15 min and a 4°C soak cycle. Polymerase chain reaction products were extracted with phenol/chloroform, precipitated with ethanol, and digested with HindIII and BglII. The digested DNA products were purified by agarose gel electrophoresis and ligated into the pR3SVori vector. The presence of appropriate mutations was confirmed by DNA sequencing as described above. The pOri.GSTFLwt and pOri.GSTFLS528A plasmids consist of pR3SVori with the GST wild type c-myb or the GST-S528A c-myb fusion insert, respectively. pOri.GST contains only the GST coding sequences, and pOri.GSTE1A12S contains the GST-E1A12S fusion gene insert.
The pCD34[-3.7k/ϩ302]d.luc plasmid was obtained from Dr. Brian Davis (Medical Research Institute of San Francisco, CA) and contains the human CD34 promoter (nucleotides Ϫ3700 through ϩ302) cloned upstream of the firefly luciferase reporter gene (47). To construct pMy-cBg.luc, a 1.7-kilobase pair BglII fragment from the murine c-myc promoter containing both the P1 and P2 c-myc promoters was prepared from pLSBgCAT (a gift from Dr. Jenny Ting, University of North Carolina at Chapel Hill) and ligated into the BglII site of pGL2-BASIC (Promega). The pdE.luc plasmid was obtained from Dr. Scott Ness (Northwestern University, Evanston, IL) and contains a 242-bp chicken mim-1 promoter fragment cloned upstream of firefly luciferase (25). Diagrams of the Myb-responsive promoters are shown in Fig. 1.
Transient Transfections and Luciferase Assays-Transient transfections into CV-1 and CMT3COS cells were performed using the CaPO 4 precipitation method (48) as described previously (38). Briefly, cells were plated at 1 ϫ 10 5 cells per 60-mm dish 16 -20 h prior to transfection. Cells were incubated overnight with the CaPO 4 -DNA precipitate in a 37°C humidified incubator containing 10% CO 2 , washed twice with PBS (137 mM NaCl, 2.7 mM KCl, 4.3 mM Na 2 HPO 4 ⅐H 2 O, 1.4 mM KH 2 PO 4 ), and further incubated in growth media described above until harvest at 48 h post-transfection. All cotransfection assays included 1 g of individual reporter plasmid and the indicated amounts of c-Myb expression plasmids per dish. The total amount of DNA per transfection was made equal by the addition of an appropriate amount of the empty pR3SV expression vector. Transfections for luciferase assays were performed in triplicate.
Cell lysates were prepared and luciferase assays performed using the luciferase assay kit (Promega) per the manufacturer's instructions. Transfected cells were washed once with PBS and incubated in 400 l of 1 ϫ Reporter Lysis Buffer (Promega) for 15 min at 27°C. Cells were harvested by scraping with a rubber policeman, transferred to microcentrifuge tubes, and frozen in a dry ice/ethanol bath. The lysates were thawed and clarified by centrifugation in an Eppendorf microcentrifuge at the maximum setting for 30 s. The supernatants were transferred to fresh microcentrifuge tubes and stored at Ϫ80°C. Lysates were thawed and kept on ice throughout the assay. Twenty microliters of lysate was added to a microcentrifuge tube containing 100 l of luciferase assay reagent (Promega), and the mixture was agitated gently by hand for 5 s. Luciferase activity was assayed in a LKB RackBeta scintillation counter. Events registering on either photomultiplier tube (open channels) over a 15-s period were recorded. Data are reported as the mean counts over background plus or minus the standard error. Background was defined for each assay as the counts generated in 15 s by Luciferase Assay Reagent without the addition of cell lysate.

Detection of Myb Multimers in Transfected Cell
Lysates-Detection of Myb-Myb interactions was performed essentially as described (49) with several modifications. CMT3COS cells were seeded at 5 ϫ 10 5 cells per 100-mm dish 20 h prior to transfection by the CaPO 4 method (48). For each dish, 15 g of either pOri.GSTFLwt, pOriGSTFLS528A, pOri.GSTE1A12S, or pOriGST was cotransfected with 1 g of either pRMb3SVori.wt or pRMb3SVori.S528A. At 48 h post-transfection, the cells were harvested by gentle scraping in PBS and transferred to a 50-ml conical tube (Sarstedt). Cells were pelleted by centrifugation at 500 ϫ g for 10 min at 4°C, resuspended in 1 ml of ice-cold PBS, and transferred to a microcentrifuge tube. Cells were then pelleted by brief microcentrifugation, resuspended in 300 l of HNNE ϩ (10 mM HEPES, pH 7.9, 1% Nonidet P-40, 150 mM NaCl, 1 mM EDTA, 1% aprotinin, 1 mM phenylmethylsulfonyl fluoride, 10 g/ml leupeptin, 0.01 unit/ml ␣2-macroglobulin, 50 mM ␤-glycerophosphate (Calbiochem), 100 nM microcystin (Calbiochem)), and mixed by inversion for 2 h at 4°C. Lysates were clarified by centrifugation in a refrigerated Eppendorf microcentrifuge at the maximum setting for 10 min at 4°C and the supernatants then transferred to fresh microcentrifuge tubes and kept on ice. Lysates were precleared with 100 l of a 1:1 slurry of protein A-Sepharose 4B beads (Pharmacia Biotech Inc.) in 150 mM NaCl, 10 mM Tris-HCl, pH 8, for 1 h at 4°C on a rotating wheel. Protein A-Sepharose 4B beads were removed by brief microcentrifugation, and the supernatants were transferred to microcentrifuge tubes. Thirty-microliter aliquots were removed from precleared lysates as controls for protein expression from transfected plasmids. Precleared lysates were incubated with 50 l of a 1:1 slurry of preequilibrated glutathione-Sepharose 4B (Pharmacia) for 2 h at 4°C. Glutathione-Sepharose 4B beads were preequilibrated in TNE (20 mM Tris-HCl, pH 8, 1% Nonidet P-40, 1 mM EDTA), blocked in TNE plus 4% nonfat dry milk for 1 h, washed once with TNE, and resuspended as a 1:1 slurry in TNE. The beads were harvested by brief centrifugation and washed three times with 1 ml of TNE per wash. After the final wash, the beads were resuspended in 100 l of 2 ϫ SB (20% glycerol, 124 mM Tris-HCl, pH 6.8, 4% SDS, 10% ␤-mercaptoethanol, 0.1% saturated bromphenol blue), boiled for 5 min, and briefly microcentrifuged. The supernatants were fractionated by 7.5% SDS-PAGE, electrophoretically transferred to nitrocellulose filters (Schleicher & Schuell), and immunoblotted with anti-Myb or anti-GST monoclonal antibodies as described below.
Protein Immunoblotting-Nitrocellulose filters were blocked in TBS (50 mM Tris-HCl, pH 8, 150 mM NaCl) plus 5% nonfat dry milk for 1 h at room temperature or overnight at 4°C. After several rinses with TBS, filters were incubated with an anti-Myb mAb at 1 g/ml (Type I anti-Myb mAb, UBI) for 1 h at room temperature. Filters were washed at room temperature for 30 min with several changes of TBS and incubated with horseradish peroxidase-conjugated sheep anti-mouse IgG (Amersham Corp., diluted 1:1333) for 1 h at room temperature. Filters were washed with TBS as described above, and Myb proteins were visualized using enhanced chemiluminescence per the manufacturer's instructions (Amersham Corp.). To detect GST and GST fusion proteins, filters were stripped for 30 min at 70°C in strip buffer (2% SDS, 0.7% ␤-mercaptoethanol, 62.5 mM Tris-HCl, pH 6.8), rinsed several times in TBS, incubated in blocking buffer for 1 h, rinsed three times in TBS, and incubated for 2 h at 27°C with 9D9 mouse anti-GST hybridoma supernatant, provided by Dr. J. Thomas Parsons (Univer-sity of Virginia), diluted 1:20 in TBS with 0.1% Tween 20. Nitrocellulose filters were washed and probed with horseradish peroxidase-conjugated sheep anti-mouse IgG as described above except that 0.1% Tween 20 was included in all washes. GST and GST fusion proteins were visualized by enhanced chemiluminescence.

RESULTS
Transcription Transactivating Activities of Wild Type and S528A c-Myb-Considerable evidence indicates that the NRD functions to influence c-Myb activity, yet little is known about how sequences in this region function. We have previously demonstrated that sequences in the negative regulatory domain serve as substrate for phosphorylation by p42 mapk in vitro and that at least one of these sites, serine 528, is phosphorylated in vivo (see Ref. 42 and Fig. 2). Substitution of serine 528 with alanine results in a 3-7-fold increase in the ability of Myb to activate transcription from an artificial promoter containing one or five copies of the mim-1A Myb response element (1 ϫ MRE or 5 ϫ MRE) inserted into a minimal herpesvirus thymidine kinase promoter (38). This suggests that phosphorylation at serine 528 serves to negatively regulate c-Myb transcription transactivating activity. To determine whether serine 528 may be involved in regulating the ability of c-Myb to activate transcription from cellular promoters, we have used a transient transfection assay to compare the ability of wild type and S528A Myb to transactivate several Myb-responsive cellular promoters. The promoters tested in this study (see Fig. 1) include the human CD34 hematopoietic stem cell antigen promoter (pCD34[-3.7k/ϩ302]d.luc), the chicken mim-1 promoter (pdE.luc), and the murine c-myc promoter (pMycBg.luc). As a negative control, the empty expression vector lacking c-myb coding sequences (pR3SV) was included in each assay.
To determine whether wild type, S528A, and S528D c-Myb proteins were expressed at similar levels in transfected CV-1 cells, anti-Myb immunoblots of transfected CV-1 cell lysates were performed. CV-1 cells were transiently transfected with pRMb3SV.wt, pRMb3SV.S528A, or pRMb3SV.9 and harvested 48 h post-transfection. Lysates from 2 ϫ 10 6 cells per transfection were fractionated by 8% SDS-PAGE and immunoblotted to detect c-Myb proteins. As demonstrated in Fig. 3A, c-Myb expression is not detected in lysates from CV-1 cells transfected with the empty control plasmid pR3SV (Fig. 3A, lane 4). In contrast, wild type c-Myb and S528A c-Myb are readily detectable in CV-1 cells transfected with pRMb3SV.wt and pRMb3SV.S528A (Fig. 3A, lanes 1 and 2, respectively). Furthermore, the amounts of wild type and S528A c-Myb expressed in transfected CV-1 cells are equivalent. The S528D c-Myb protein encoded by pRMb3SV.9 was not detected in transfected CV-1 cells (Fig. 3A, lane 3) and was not used in transfections designed to assay transcription transactivating activities. Results from this experiment demonstrated that wild type and S528A c-Myb were expressed at similar levels in transfected CV-1 cells and allowed for a direct comparison of their relative transcription transactivating activities in the cotransfection assays described below.
We first examined the relative abilities of wild type and S528A Myb to activate transcription from the human CD34 promoter. CD34 is a cell surface glycoprotein whose pattern of expression in early hematopoietic precursors in bone marrow is similar to that of c-Myb (50) and transcription from the CD34 promoter is activated by c-Myb (27). As demonstrated in Fig.  3B, both wild type and S528A c-Myb activated transcription from the CD34 promoter at each concentration of transfected expression plasmid tested. However, S528A c-Myb activated transcription from the CD34 promoter approximately 3-4-fold more effectively than did wild type c-Myb. These results are representative of three independent experiments and are similar to our previous results using a synthetic 5 ϫ MRE contain-ing promoter/reporter construct (38).
To extend our analysis to another known Myb-responsive cellular promoter, the relative abilities of wild type and S528A c-Myb to transactivate the chicken mim-1 promoter were examined. Although the mim-1 promoter contains three potential MREs referred to as mim-1A, mim-1B, and mim-1C (Fig. 2), the mim-1A MRE is predominantly responsible for activation by c-Myb (51). The mim-1 promoter also contains two binding sites for the NF-M myeloid-specific transcription factor (see Fig. 2), and c-Myb and NF-M synergistically activate transcription of mim-1 in myeloid cells (32,51). In contrast to the results obtained using the 5 ϫ MRE and CD34 promoters, wild type and S528A c-Myb equivalently activated transcription from the mim-1 promoter at each amount of expression plasmid tested (Fig. 3C). Thus, in transfected CV-1 cells, there is no significant difference between the abilities of wild type and S528A c-Myb to activate transcription from the intact mim-1 promoter. The relative abilities of wild type and S528A c-Myb to transactivate the murine c-myc promoter were also examined. The murine c-myc promoter contains 16 Myb binding sites located in two clusters. The cluster distal to the transcription start site contains 10 and the cluster proximal to the site of transcription contains 6 Myb binding sites (Fig. 2) (52). The results of transient transfection assays obtained using the murine c-myc promoter were similar to those obtained using the chicken mim-1 promoter (Fig. 3D). The relative abilities of wild type and S528A c-Myb to transactivate the murine c-myc promoter were equivalent throughout the range of expression plasmid amounts tested. Both proteins effectively activated transcription of the c-myc promoter. These results indicate that phosphorylation of serine 528 may provide a mechanism to differentially regulate c-Myb transcription transactivating activity at different promoters.
Substitution of Serine 528 Does Not Affect DNA Binding-To determine whether the different transactivation abilities observed for wild type and S528A c-Myb was reflected by a differential ability to bind DNA, EMSAs were performed. For these experiments, wild type c-Myb and the Myb mutants were transiently expressed in CMT3COS cells. Ser-528 is phosphorylated in CMT3COS cells, and the differential transcription activation properties of wild type and S528A Myb are similar in CV-1 and CMT3COS cells (38). As demonstrated in Fig. 4C, all three proteins, including S528D, were expressed in CMT3COS cells. The S528D c-Myb substitution mutant was also tested in these experiments since it was stably expressed in CMT3COS cells. A single major complex with retarded mobility was detected in extracts from CMT3COS transfected with each expression vector (Fig. 4, A and B). This complex was not detected in cells transfected with the empty pR3SV control vector. The presence of c-Myb in the major protein-DNA complex was demonstrated by a supershift of the protein-DNA complex when the lysate was preincubated with a c-Myb monoclonal antibody. The specificity of this interaction was demonstrated by cold competition assay, and the binding of DNA to each Myb protein was effectively competed by preincubation with unlabeled MRE oligodeoxynucleotide at a molar excess of 100 -500-fold (Fig.  4B). In contrast, unlabeled cyclic AMP response element oligodeoxynucleotide, at the same molar excess, did not compete for DNA binding (Fig. 4B). Scatchard analysis, using a more extensive range of competitor concentrations, revealed no differences between the affinities of these three proteins for several MREs. 2 In addition, no difference in the on or off rates of these proteins was detected. Thus, the difference in the abilities of wild type and S528A c-Myb to transactivate the CD34 and synthetic 5 ϫ MRE containing promoters does not appear to be due to differences in DNA binding due to substitution of serine 528 with alanine, suggesting that phosphorylation of serine 528 does not regulate c-Myb sequence-specific DNA binding activity.

Substitution of Serine 528 Does Not Affect the Ability of c-Myb to Form Multimerized
Complexes-It has been reported that c-Myb can homodimerize in vitro via a leucine zipper-like motif (amino acids 375-403) in the NRD (53). Those authors reported that c-Myb homodimers cannot bind DNA in a sequence-specific manner and are therefore prevented from activating transcription from Myb-responsive promoters. Since the leucine zipper of c-Myb resides in close proximity to serine 528, we tested whether substitution of serine 528 by alanine affected the ability of c-Myb to multimerize in transfected CMT3COS cell lysates. A similar approach to that reported by Nomura et al. (53) was used, with two modifications. First, CMT3COS cells were used instead of NIH-3T3 cells for high level expression of transfected gene products since it is not known whether serine 528 of murine c-Myb would be a target for phosphorylation in NIH-3T3 cells. Second, our experiments utilized full-length rather than truncated Myb proteins to address the issue of Myb multimerization.
CMT3COS cells were transiently cotransfected with plasmids encoding wild type or S528A c-Myb and individual plasmids encoding either GST wild type c-Myb, GST S528A c-Myb, GST E1A, or GST. Whole cell lysates were prepared and incubated with glutathione-agarose to harvest the GST-containing proteins. After a series of washes, proteins bound to the beads were released by boiling in sample buffer, fractionated by 8% SDS-PAGE, and transferred to nitrocellulose for detection by either anti-c-Myb or anti-GST monoclonal antibodies. GST c-Myb fusion proteins can be distinguished from native c-Myb proteins by their slower mobility during SDS-PAGE (GST c-Myb migrates at approximately 100 kDa and c-Myb migrates at approximately 75 kDa). Whereas GST E1A and GST did not form complexes with wild type or S528A c-Myb (Fig. 5A, lanes  4 -7), both GST wild type c-Myb and GST S528A c-Myb associated with c-Myb (Fig. 5A, lanes 1 and 2). Thus, the S528A substitution did not have an apparent effect on the ability of c-Myb to multimerize with GST wild type c-Myb. To determine whether S528A substitutions on both partners would affect multimerization, we also tested the ability of GST S528A c-Myb to form a complex with S528A c-Myb. As shown in lane 3 of Fig.  5A, GST S528A c-Myb also associated with S528A c-Myb to a similar extent as observed using one or two wild type partners (compare Fig. 5A, lanes 1, 2, and 3).  2. Schematic representation of murine c-Myb. The c-Myb protein is depicted schematically as a box with regions transduced and expressed by the E26 and AMV avian acute transforming retroviruses represented by lines above c-Myb. The three tandem repeats comprising the DNA binding domain (DNAB) are represented as black triangles. The transactivation (TA) and negative regulatory (NR) domains are also shown. The two nuclear localization (NL) motifs are shown as stippled rectangles and the leucine zipper (LZ) is depicted as a striped rectangle. Five previously reported potential sites for phosphorylation by proline-directed serine/threonine protein kinases (42) are labeled with the single-letter amino acid code followed by their position in murine c-Myb (e.g. S528 represents serine at position 528). T448 lies within a TP motif (small stippled box), S459 and S528 lie within SP motifs (small black circles) and T462 and T486 lie within PXTP motifs (small stippled triangles). Amino acid positions delineating the amino terminus, DNAB domain, TA domain, NR domain and carboxyl terminus are indicated on the scale below c-Myb (19,45,(72)(73)(74)(75).
transfected with plasmids encoding GST c-Mybs alone. 3 Neither wild type nor S528A c-Myb bound to glutathione agarose in the absence of any GST-fusion protein. 3 To ensure that the negative control GST fusion proteins were appropriately expressed in this assay, the nitrocellulose filter was stripped and reprobed with the 9D9 anti-GST mAb that detects each GST fusion protein used in this assay. Fig. 5B demonstrates that all of the GST fusion proteins were appropriately expressed in this experiment. In fact, GST E1A and GST were more effectively harvested by incubation with glutathione-agarose than were GST wild type c-Myb and GST S528A c-Myb (Fig. 5B, lanes 1-7). Together, these data indicate that full-length c-Myb multimerized with GST c-Myb and that the S528A substitution does not affect multimerization. Thus, the difference in the abilities of wild type and S528A c-Myb to transactivate the 5 ϫ MRE and CD34 promoters was not reflected by an apparent difference in their ability to participate in Myb-Myb interactions. DISCUSSION We have examined the potential role of serine 528 in regulating c-Myb transcription transactivating activity on a series of cellular promoters that are known to be activated by c-Myb. Substitution of serine 528 for alanine (S528A c-Myb) resulted in a substantial increase in c-Myb-activated transcription from the human CD34 promoter (see Fig. 3B) as well as from the previously reported synthetic Myb-responsive promoters containing one or five copies of a mim-1A-based MRE (38). These results suggest that phosphorylation of serine 528 may act to suppress c-Myb transcription transactivating activity at these promoters. In contrast to the results obtained using the CD34 promoter, S528A c-Myb and wild type c-Myb were equally effective at stimulating transcription from the chicken mim-1 and murine c-myc promoters (see Fig. 3, C and D) indicating that phosphorylation of serine 528 does not play a role in regulating c-Myb-activated transcription from either the mim-1 or c-myc promoters. These results strongly support the notion that phosphorylation of serine 528 provides a mechanism to differentially regulate c-Myb activity by modifying the structure of the NRD.
The observation that wild type and S528A c-Myb equivalently transactivated the murine c-myc promoter is consistent with previously published data demonstrating that truncation of the NRD does not potentiate c-Myb-activated c-myc transcription (54). Thus, it appears that the NRD does not act as a negative regulator in the context of Myb-activated c-myc transcription, further supporting our hypothesis that the c-Myb NRD is differentially active at different promoters. The finding that S528A c-Myb and wild type c-Myb were equally effective at stimulating transcription from the chicken mim-1 promoter was unexpected since the c-Myb NRD has previously been demonstrated to negatively regulate the synergistic activation of mim-1 transcription by c-Myb and CEBP/␤ (also referred to as NF-M) (32). Although the S528A substitution in c-Myb had no measurable effect on transactivation of the mim-1 promoter in the current study, CV-1 cells do not express CEBP/␤. It should be noted that promoter context contributed to the differential activity of wild type and S528A c-Myb, since the dominant mim-1A MRE from the mim-1 promoter was differentially responsive to these proteins within the context of a heterologous minimal promoter containing one or five copies of the mim-1A MRE in the absence of CEBP/␤.
It is not surprising that c-Myb activity might be regulated by post-translational modification under some circumstances. In mature hematopoietic and some non-hematopoietic cell lines, the activity of c-Myb appears be regulated primarily by changes in c-Myb expression during the cell cycle (55,56). However, c-Myb expression does not appear to be regulated during the cell cycle in immature hematopoietic cells (3,57). If c-Myb plays a similar role in regulating gene expression during the cell cycle in both immature and mature hematopoietic cells, then a mechanism must be available to regulate c-Myb activity during the cell cycle in immature hematopoietic cells. Phosphorylation provides a rapid and efficient mechanism by which c-Myb activity may be regulated. Indeed, Lü scher and colleagues (37) have recently reported that hyperphosphorylation of c-Myb during mitosis, at unidentified sites, in a pre-B cell lymphoma cell line correlates with decreased DNA binding activity. These observations, along with the data presented in this article, suggest that phosphorylation may serve to negatively regulate c-Myb activity. However, it should be noted that these observations have been made using a relatively restricted set of Myb target promoters and that phosphorylation at these sites, or other unidentified sites, may serve to increase the ability of c-Myb to activate transcription at other promoters in a lineage or differentiation stage-specific fashion.
How phosphorylation of serine 528 may regulate c-Myb transcription transactivating activity is not understood. Serine 528 resides within a region (residues 495-554) that was previously demonstrated to have a negative effect on c-Myb transcription transactivating activity (58). Phosphorylation of c-Myb serine 528 may mediate the direct physical interaction between the NRD and the transactivation domain, thus masking the transactivation domain and suppressing the transcription transactivating activity of c-Myb. Release of phosphate from serine 528 would then allow c-Myb to assume a more active conformation. However, we do not favor this model because it fails to account for the differential activity of wild type and S528A c-Myb at different promoters in the same cell type.
Phosphorylation has been shown to negatively regulate the DNA binding activities of several transcription factors, including c-Myb, c-Jun, Oct-1, and Ets-1 (59 -62). However, our results indicate that serine 528 does not play a significant role in regulating the DNA binding activity of c-Myb. This is consistent with the findings of Dubendorff et al. (58) who demonstrated that amino acids 495-554 of avian c-Myb (serine 528 of murine c-Myb corresponds to serine 533 of avian c-Myb) suppress c-Myb transactivating activity in-cis and in-trans without affecting DNA binding activity. However, since the MRE consensus sequence is degenerate, phosphorylation of serine 528 may affect c-Myb DNA binding activity at only a subset of MREs that have not yet been examined. In addition, it should be noted that in the EMSA experiments presented in this study, c-Myb was expressed at very high levels in transfected CMT3COS cells. Thus, it is possible that high level expression of c-Myb effectively titrated out the protein kinase(s) responsible for phosphorylating serine 528. However, this is unlikely as c-Myb is phosphorylated on Ser-528 in CMT3COS cells, and the differences between the transcription activating properties of wild type and S528A Myb are similar in CV-1 and CMT3COS cells (38).
Mutations in the c-Myb leucine zipper that abolish the ability to homodimerize in vitro also increase the transcription transactivating, transforming, and DNA binding activities of c-Myb (53). Since serine 528 lies in close proximity to the leucine zipper within the NRD (residues 374 -403), phosphorylation of serine 528 might regulate the ability of the leucine zipper to mediate homodimerization. However, data presented in this article do not support a role for phosphorylation of serine 528 in regulating c-Myb homodimerization. Although the ability of c-Myb to homodimerize per se was not tested in vitro using purified proteins, the S528A substitution in c-Myb had no measurable effect on the detection of steady state complexes containing two or more c-Myb proteins in transfected CMT3COS cells.
We favor a model in which phosphorylation of serine 528 regulates the interaction between c-Myb and other transcription activators or repressors. Phosphorylation has been demonstrated to regulate the interaction of several transcription factors, including NF-B, E2F, and PU.1, with heterologous binding partners (63)(64)(65). For example, the PU.1 transcription factor binds to its cognate sequence on the immunoglobulin 3Ј enhancer where it can be phosphorylated on serine 148. This functions to recruit NFEM-5 to its cognate DNA element directly 3Ј to the PU.1 site and stimulate transcription (64). Studies performed in our laboratory 2 and by others (66) indicate that the c-Myb NRD can associate with a number of as yet uncharacterized cellular proteins. While our data suggest that phosphorylation at serine 528 does not affect the interaction between c-Myb and a general inhibitor of transcription, it may limit the ability of c-Myb to interact with specific transcription repressors, thus differentially regulating c-Myb activity.
The c-Myb NRD is commonly truncated in oncogenic Myb proteins generated by either retroviral transduction or integration, thus deleting serine 528 (see Fig. 2). This suggests that the inability of these proteins to be phosphorylated on serine 528 might contribute to their oncogenic activation. However, substitution of chicken c-Myb serine 533 with alanine (chicken c-Myb serine 533 corresponds to murine c-Myb serine 528) is not sufficient to transform chicken yolk sac cells in vitro. 4 One potential explanation for this observation is that phosphorylation of serine 528 may not regulate c-Myb-activated transcription of genes directly involved in transformation. Alternatively, deletion or substitution of serine 528 may contribute to Myb oncogenicity only in the context of other lesions in the carboxylterminal region or amino-terminal truncation. Indeed, carboxyl-terminal truncated c-Myb is weakly transforming compared with amino-terminal truncated c-Myb, whereas truncation of both termini synergize to oncogenically activate Myb, indicating that intramolecular interactions may play a key role in the regulation of c-Myb function.
While the data presented in this article support a role for phosphorylation of serine 528 in regulating c-Myb transcription transactivating activity, other phosphorylation sites will also likely be important for the regulation of c-Myb activity. Indeed, phosphorylation of serines 11 and 12 by CK II in vitro inhibits the sequence-specific DNA binding activity of c-Myb, and mutation of these sites to alanine decreases the ability of c-Myb to activate transcription (36). These sites are also phosphorylated in vivo in at least two cell lines, and their substitution by alanine results in increased DNA binding and transcription transactivating activity (36,40). In addition, common and lineage-specific phosphopeptides have been identified by our laboratory 5 and others (36 -38). This raises the possibility that lineage-specific phosphorylation may regulate c-Myb function in a cell type-specific manner. It will be of considerable interest to identify these sites of phosphorylation and assess their roles in the regulation of c-Myb activity in different cell types.
p42 mapk is generally viewed as a cytoplasmic protein kinase that is activated during entry into the cell cycle (67). However, it has been reported to translocate to the nucleus upon activa-tion (68,69). Although the expression and activity of p42 mapk in developing hematopoietic cells has yet to be well characterized, the activation of p42 mapk in response to differentiation factors such as GM-CSF has been reported (70). Thus, during the differentiation of hematopoietic cells, c-Myb activity could be modulated by phosphorylation on serine 528 by p42 mapk in response to growth and differentiation factors. In addition, activated p42 mapk is constitutively detected in the C19.396 murine erythroleukemia cells, suggesting that p42 mapk might phosphorylate c-Myb in this cell line. 6 Although murine c-Myb serine 528 is a substrate for p42 mapk in vitro, it is not clear whether p42 mapk is responsible for phosphorylation of serine 528 in vivo. Other MAPKs, such as p44 mapk , or other prolinedirected serine/threonine protein kinases may phosphorylate c-Myb serine 528. One such kinase, p34 cdc2 , may be a good candidate for being a regulator of c-Myb, since it is active during the G 1 /S transition of the cell cycle when c-Myb is expressed at its highest levels in many cell types (71). Interestingly, the c-Myb carboxyl terminus contains consensus sequences for a number of serine/threonine protein kinases, including protein kinase A, GSK III, CK II, and protein kinase C. It will be of interest to determine whether other sites in the NRD are targets for protein kinase activity and whether they serve to integrate information from multiple signal transduction pathways.