Regulation of Myeloid Zinc Finger Protein 2A Transactivation Activity through Phosphorylation by Mitogen-activated Protein Kinases*

The myeloid zinc finger protein (MZF)-2 is a C2H2 zinc finger transcription factor that is expressed in myeloid cells and involved in the growth, differentiation, and tumorigenesis of myeloid progenitors. Here we describe a novel isoform of MZF-2, designated MZF-2A, and show that it is phosphorylated by the mitogen-activated protein (MAP) kinases. Anin vitro phosphorylation experiment revealed that the transactivation domain (TAD) of MZF-2A was phosphorylated strongly by extracellular signal-regulated kinase (ERK) and phosphorylated weakly by p38 MAP kinase but not by Jun N-terminal kinase. Experiments using “add-back” mutants showed that three serine residues (Ser257, Ser275, and Ser295) in the TAD were phosphorylated in vitro by ERK. In myeloid LGM-1 cells, various extracellular stimuli induced the phosphorylation of these serine residues, which was differentially inhibited by the protein kinase inhibitors U0126 and SB203580. Substitution of these phosphorylation sites with alanines resulted in a strong enhancement of the ability of MZF-2A to activate transcription in a luciferase reporter assay. Taken together, these results indicate that MZF-2A is a novel target for the ERK and p38 MAP kinase signaling pathways, and its transactivation activity is negatively regulated by MAP kinase-mediated phosphorylation of the TAD.

The development of mature blood cells occurs through the proliferation and differentiation of hematopoietic progenitor cells and is controlled by the combined effects of hematopoietic cytokines and transcription factors (1)(2)(3)(4). A set of zinc finger transcription factors, including GATA, EKLF, and Ikaros, plays important roles in the hematopoietic commitment to specific lineages (1,4). Myeloid zinc finger protein (MZF) 1 is one of the candidates in this family for controlling the development of the myeloid cell lineage. MZF is a Krü ppel-like C 2 H 2 zinc finger protein expressed predominantly in myeloid progenitor cells, particularly at the myeloblast and metamyelocyte stage (5)(6)(7)(8)(9). Blocking MZF expression with antisense oligonucleotides inhibits granulocyte colony-stimulating factor-induced neutrophilic colony formation in vitro, suggesting that MZF is involved in the proliferation of myeloid progenitors and/or their differentiation to neutrophilic granulocytes (6). Furthermore, disruption of the MZF gene in mice results in an accumulation of Mac-1ϩ myeloid cells in bone marrow and the development of neoplasias composed of highly proliferating myeloid blasts in the liver and spleen, indicating that MZF plays a role in negative regulation of proliferation and tumorigenesis of hematopoietic progenitors (10).
We and others have shown that several isoforms of MZF proteins are produced by alternative transcriptional initiation sites and splicing from a single gene in both human and mouse cells (5,7,11). Human MZF-1, originally isolated from myeloid leukemia cells, encodes a 485-amino acid protein consisting of 13 zinc finger motifs that bind specific DNA sequences (5,12). By the extensive screening of mouse and human cDNA libraries, we identified an alternative MZF product, termed MZF-2, which shares the sequence of its C terminus, including the zinc finger domains, with MZF-1 but also contains a long N-terminal extension consisting of an acidic domain, a leucine-rich (LeR; also called SCAN box) domain, and a transactivation domain (TAD) (7). Functional analysis of the MZF-2 subdomains revealed that the TAD works specifically in myeloid cells and that the far N-terminal region spanning the acidic and LeR domains inhibits the neighboring TAD (13). Recently, Peterson and Morris (11) reported another form of the gene product, MZF1B/C, whose entire structure is almost identical to that of MZF-2 but differs in the N-terminal region. The consensus MZF-binding sites are found in the promoters of several hematopoietic cell-specific genes, such as CD34, c-myb, lactoferin, and myeloperoxidase (12,14). The MZF-binding motif is also found in the promoter of the telomerase reverse transcriptase gene, where MZF is likely to function as a negative regulator for gene expression (15). The composite structure of negative and positive regulatory domains for transcription implies that the function of MZF could be regulated by posttranslational modification and/or processing.
The activity of many transcription factors is regulated in a rapid and reversible manner by specific phosphorylation. Mitogen-activated protein (MAP) kinases are proline-directed serine/threonine protein kinases activated by various extracellular stimuli, including growth factors, cytokines, and environmental stresses (16 -18). Currently, four distinct MAP kinase pathways are known in mammals. The extracellular signal-regulated kinase (ERK) and ERK5/BMK1 pathways mainly convey signals from mitogenic and differentiation stim-uli, whereas the Jun N-terminal kinase (JNK) and p38 MAP kinase pathways seem to be involved in transducing various stress-and cytokine-triggered signals. Activated MAP kinases in turn phosphorylate a set of cytoplasmic substrates such as MAP kinase-activated protein kinases (18,19) and various nuclear targets including transcriptional activators, c-Myc, Elk-1, c-Jun, ATF2, and HSF1 (20 -29), as well as transcriptional repressors such as Bcl-6 and the ETS-type repressors (30 -32).
Here we first showed that a novel isoform of MZF-2, designated MZF-2A, is produced by an alternative use of transcription start sites and subsequent splicing. We then studied the regulatory mechanism of MZF-2A transcriptional activity and found that three serine residues in its transactivation domain are phosphorylated by ERK and p38 MAP kinases in vitro and in vivo. Mutation of the phosphorylation sites to alanines markedly increased the transactivating potential of MZF-2A, suggesting that MZF-2A activity is negatively regulated by MAP kinase-mediated phosphorylation.
Plasmid Construction-The expression plasmid for FLAG epitopetagged MZF-2A, pFLAG-MZF-2A, was constructed by inserting MZF-2A cDNA downstream of the FLAG epitope sequence (MDYKD-DDDKAG) in a pEF-BOS-EX expression vector (13). To construct the pGL3-4xF113B-G-Luc reporter plasmid, the SV40 promoter sequence of pGL3-Promoter vector (Promega) was replaced by a DNA fragment carrying the tetrameric MZF-binding site (AGTGGGGAgaaaggatctggct-GGTGAGGGGGAAtcggatct) (12) and the granulocyte colony-stimulating factor promoter sequence, derived from the pBPA-4xF113B-G-Luc plasmid (13). The pRL-SV40 plasmid was purchased from Promega. The pGAL4-MZF-TAD plasmid was generated by inserting a fusion construct encoding the DNA-binding domain of yeast GAL4 (amino acids (aa) 1-146) (35), the FLAG epitope sequence, and a cDNA fragment corresponding to the MZF-TAD (aa 155-336) in-frame into pEF-BOS-EX. The deletion mutants of pGAL4-MZF-TAD were constructed by a combination of restriction enzyme digestion and PCR. The pGAL4-Pax6 TAD plasmid was made by replacing the MZF-2A sequence of pGAL4-MZF-TAD with the cDNA fragment corresponding to the transactivation domain (aa 271-422) of mouse Pax6 (36). pMZF-TAD was made by inserting the DNA fragment for MZF-TAD into pEF-BOS-EX. The reporter plasmid pUAS G -Luc containing the interferon-␤ minimal promoter and three GAL4-binding sites was described previously (35). The pGEX-MZF-TAD plasmid was constructed by inserting the DNA fragment for the MZF-TAD (aa 155-336) into the pGEX-5X1 vector (Amersham Biosciences). A series of mutants with disrupted phosphorylation sites was generated by recombinant PCR.
In Vitro Phosphorylation of GST-MZF-TAD-Active human ERK1 was produced in insect Sf9 cells as described previously (37). To produce active human p38 and JNK1, HeLa cells were transiently transfected with expression plasmids for FLAG-tagged p38 (38) or hemagglutinintagged JNK1 (39) and stimulated with 0.7 M NaCl for 30 min or 10 g/ml anisomycin for 30 min, respectively. Activated kinase was immunoprecipitated using specific antibodies (anti-FLAG M2 Affinity Gel (Sigma) for FLAG-p38 and anti-hemagglutinin 16B12 monoclonal antibodies (BAbCO) for hemagglutinin-tagged JNK1), and the immune complexes were used for in vitro phosphorylation of GST fusion proteins.
Luciferase Assay-LGM-1 cells were suspended at a density of 1 ϫ 10 7 cells/ml in 500 l of serum-free RPMI 1640 medium supplemented with 10 mM Hepes (pH 7.2) and mixed with 20 g of plasmid DNA. After electroporation at 360 V with a capacitance of 960 microfarads in a 0.4-cm gap cuvette, the cells were transferred to 2 ml of growth medium and cultured for 18 h. Firefly and Renilla luciferase activities were assayed using the Dual-Luciferase Reporter Assay System (Promega). The firefly luciferase activity of each sample was normalized to the Renilla luciferase activity used as an internal standard.

Identification of a Novel Form of the Mouse MZF-2 Gene
Product-Previously, we isolated a murine cDNA for MZF-2 that encoded a polypeptide of 814 amino acids ( Fig. 1A) (7). In the course of characterizing the MZF-2 transcripts by Northern blotting, however, we found that the major MZF-2 mRNA species of about 3 kb encoded an alternative protein product lacking the N-terminal region of the reported MZF-2 sequence. As shown in Fig. 1B, when poly(A) ϩ RNA extracted from mouse myeloid cell lines was analyzed by Northern blotting, the NP2 probe spanning the acidic and LeR domains detected a major 3-kb transcript and two minor transcripts of 4 and 8 kb in the RNAs from LG and NFS-60 cells, whereas the NP1 probe derived from the 5Ј end of the cDNA hybridized only to the 4-kb transcript, indicating that the NP1 sequence was not present in the 3-kb transcript. A BLAST search of the expressed sequence tag data base using the NP2 sequence as a query identified an expressed sequence tag clone (GenBank accession number AA499285) that contained not only the NP2 region but also an additional 5Ј sequence that was different from the NP1 sequence. Sequencing analysis of the 5Ј-flanking region of the MZF genomic clones as well as reverse-transcription-PCR analysis of the RNA from the myeloid cell lines revealed that the 5Ј sequence of AA499285 was derived from an upstream exon (designated exon 0) and represented the 5Ј-untranslated region of the 3-kb mRNA species (Fig. 1A). Consequently, this mRNA would encode a 735-amino acid polypeptide that lacked the N-terminal 79 amino acids of the previously reported MZF-2 protein. We conclude that the short form of MZF-2, designated hereafter as MZF-2A (MZF-2 Alternative form), is the major product of the MZF gene, whereas the long form is very minor.
In our previous study, we showed that the N-terminal region of MZF-2 negatively regulates its transcriptional activity (13). To examine whether the short form MZF-2A still retains the inhibitory domain, mouse myeloid LGM-1 cells, which do not express endogenous MZF (Fig. 1B), were co-transfected with a luciferase reporter plasmid carrying the MZF-binding sites (pGL3-4xF113B-G-Luc) and an effector plasmid expressing MZF-2, MZF-2A, or an N-terminal deletion mutant, KM3 ( Fig.  1A) (13). The level of reporter expression was then compared. Although MZF-2A showed significant transactivating activity when compared with MZF-2 or the vector control, its activation level was much lower than that of the KM3 mutant, which lacked a 155-aa N-terminal sequence that MZF-2A contained, indicating that the N-terminal domain of MZF-2A functions as a negative regulatory domain (Fig. 1C).
Structural Similarity between the Transactivation Domains of MZF-2A and Pax6 -We previously showed that the TAD of MZF-2A resided in the middle region of the molecule (Fig. 1A) (13). To characterize the TAD function independent of the zinc finger domain, we generated several chimeric proteins in which various regions of the TAD were fused to the DNA-binding domain of yeast GAL4 protein and examined their ability to activate transcription of the reporter gene carrying a GAL4binding promoter. As shown in Fig. 2A, the GAL4-MZF-TAD fusion protein activated the transcription of the reporter gene in LGM-1 cells. Deletions from the C terminus severely impaired transcriptional activation (D1 and D2 mutants), whereas N-terminal deletions up to aa 239 did not affect it significantly (D3 and D4). Further combinations of deletions revealed that the D5 and D6 mutants were able to activate transcription, but D7 was not, indicating that the 50-aa region from aa 239 -288 is the minimal TAD of MZF-2A. This finding was further supported by the inability of the D8 mutant, which lacked only the D6 region, to activate transcription.
To explore the mechanism of transcriptional activation by LGM-1 cells were co-transfected with 1 g of pGAL4-Pax6-TAD, 3 g of pUAS G -Luc, and increasing amounts (0, 1, 3, and 6 g) of pMZF-TAD. The total amount of transfected DNA was kept at 10 g by adding the empty expression vector. The luciferase activities are expressed as percentages of the values obtained in the absence of pMZF-TAD plasmid.
MZF-2A, we searched databases for proteins showing any homology with the MZF-TAD and found that the D6 region showed marked homology with the transactivation domain of mouse Pax6 (Fig. 2B) (42), which is a transcription factor that plays a pivotal role in the development of the central nervous system, eyes, nose, and pancreas (43,44). This similarity implies that MZF-2A and Pax6 might utilize common transcriptional machinery to regulate transcription. In support of this hypothesis, we found that the expression of the MZF-TAD protein inhibited GAL4-Pax6-TAD-mediated transactivation in a dose-dependent manner, presumably through competitive binding to a shared transcriptional cofactor (Fig. 2C).
The Transactivation Domain of MZF-2A Is Phosphorylated by MAP Kinases in Vitro and in Vivo-Mikkola et al. (45) showed that the transactivation domain of zebrafish Pax6 is phosphorylated at serine 413 by the ERK and p38 MAP kinases. The conservation of this serine residue in the MZF-TAD (Fig. 2B) prompted us to investigate whether MZF-2A was also phosphorylated and regulated by MAP kinases. For this purpose, we first examined the phosphorylation of recombinant MZF-TAD by purified ERK, p38, and JNK MAP kinases in vitro. As shown in Fig. 3A, ERK1 phosphorylated GST-MZF-TAD more efficiently than GST-ATF2 (left panel), and p38 MAP kinase weakly phosphorylated MZF-TAD (middle panel). In contrast, JNK1 did not phosphorylate MZF-TAD at all under the same conditions in which it phosphorylated ATF2, Elk-1, and c-Jun very well (right panel). Similarly, JNK2 did not phosphorylate MZF-TAD (data not shown).
Because MAP kinases are proline-directed protein kinases, we focused our next set of studies on five serine/threonine residues, Thr 167 , Thr 170 , Ser 257 , Ser 275 , and Ser 295 , all of which are followed by a proline residue and conserved in both human and mouse MZF-TAD sequences (Fig. 3B, left panel). To determine which residue(s) is actually phosphorylated, a series of MZF-TAD mutants in which these serine or threonine residues were replaced with alanine residues were generated and tested for their ability to be phosphorylated by ERK1 in vitro. When all five of the sites were replaced with alanines (A5 mutant), no phosphorylation was observed (Fig. 3B, right panel). Starting with this A5 mutant, we constructed "add-back" mutants in which each of the mutated residues was changed back to the wild-type residue, and we examined them for in vitro phosphorylation. The S257, S275, and S295 add-back mutants were phosphorylated efficiently by ERK1, whereas T163 and T170 were not. These results indicate that ERK1 phosphorylates the MZF-TAD in vitro at three serine residues, Ser 257 , Ser 275 , and Ser 295 , among the five putative phosphorylation sites.
To investigate whether the serine residues identified above could be phosphorylated by MAP kinases in vivo, we estab-  stimulated or unstimulated cells and treated with calf intestine alkaline phosphatase (Fig. 4B, CIAP), the upper band of wildtype MZF-2A was converted into the lower band, whereas the mobility of the S257/275/295A mutant was not affected (Fig.  4B). These results indicated that the mobility shift of MZF-2A was caused by phosphorylation and that one or more of the three serine residues (Ser 257 , Ser 275 , and Ser 295 ) were phosphorylated by the extracellular stimuli. Western blot analysis using antibodies specific for the active form of MAP kinases revealed that a distinct set of MAP kinases was activated by the individual stimuli (Fig. 4A, bottom panels). To explore which MAP kinase is responsible for MZF-2A phosphorylation, we used two specific inhibitors for MAP kinase pathways, U0126 and SB203580. U0126 is a potent inhibitor of MEK, an ERK-activating kinase, and consequently blocks the ERK pathway in vivo (46). SB203580 inhibits some of the p38 members such as p38␣/SAPK2a and p38␤/SAPK2b, but not p38␥/SAPK3 or p38␦/SAPK4 (47)(48)(49)(50). As shown in Fig. 4C, pretreating cells with U0126 strongly blocked TPA-induced MZF-2A phosphorylation and partially inhibited FCS-or IL-3-induced phosphorylation. This result, together with the observation that ERKs were strongly activated by TPA but very weakly activated by FCS or IL-3 (Fig. 4A), indicates that ERK MAP kinases phosphorylate MZF-2A in vivo. On the other hand, SB203580 partially blocked the anisomycin-induced mobility shift, indicating that p38 MAP kinase members, which are activated by anisomycin (Fig. 4A), also contributed to the phosphorylation of MZF-2A. Interestingly, combined treatment with both U0126 and SB203580 did not completely inhibit the MZF-2A phosphorylation induced by anisomycin or IL-3, suggesting that other protein kinases insensitive to these inhibitors may also be involved. These results indicate that, depending on the types of stimulation, the ERK and p38 MAP kinase pathways are differentially involved in MZF-2A phosphorylation.
The Transactivation Activity of MZF-2A Is Negatively Regulated by Phosphorylation-To test whether MAP kinase-mediated phosphorylation regulates the transactivating activity of MZF-2A, we constructed a series of FLAG-MZF-2A mutants in which one or two of the three serine residues were replaced with alanine(s) and assessed their transactivation potentials by a luciferase reporter assay using LGM-1 cells. Western blot analysis of the FLAG-MZF-2A proteins expressed in LGM-1 cells demonstrated that all the wild-type and mutant MZF-2A proteins except for the triple mutant (S257/275/295A) migrated to the position of phosphorylated MZF-2A (Fig. 5, bottom panel). This result indicates that in the growing LGM-1 cells, the wild-type and mutated MZF-2A proteins were constitutively phosphorylated at the unmutated site(s) among the three serine residues. As shown in the top panel of Fig. 5, when compared with wild-type MZF-2A, each mutant with a single substitution (S257A, S275A, and S295A) exhibited a 1.4 -2.6-fold increase in reporter activity. Introduction of an additional alanine substitution resulted in a further increase in reporter expression (S275/295A, S257/295A, and S257/275A). The greatest augmentation (ϳ6-fold) in transcriptional activity was observed with the S257/275/295A mutant, in which all three serine residues were replaced with alanines. These results suggest that phosphorylation of these serine residues cooperatively reduces the transactivation activity of MZF-2A. Among the three serine residues, the substitution of Ser 295 showed a greater enhancement in reporter expression than the others, suggesting that phosphorylation of Ser 295 is the most effective in repressing the transactivation activity of MZF-2A. The enhancement of reporter expression was not a result of an increase in the protein level of the mutants because comparable amounts of MZF-2A protein were expressed in all of the con-structs (Fig. 5, bottom panel). In an attempt to confirm the negative regulation of MZF-2A activity by phosphorylation, we examined the effect of the U0126 and SB203580 inhibitors on the luciferase reporter assay. However, we found that the luciferase assay with these inhibitors could was not a reliable experiment because treatment of cells with the inhibitors resulted in strong reduction of gene expression in general (data not shown). We finally examined the effect of the phosphorylation site mutation on the activity of the N-terminally deleted KM3 mutant, which showed an enhanced transcriptional activity (Fig. 1). Interestingly, the transactivation activity of the S257/275/295A mutant was almost the same as that of KM3, and disruption of the three phosphorylation sites of the KM3 mutant (KM3-S257/275/295A) resulted in only a slight additional enhancement of reporter activation (Fig. 5, top panel), indicating that the two different mutations, the N-terminal deletion and the alanine substitution of the phosphorylated sites, did not work additively to augment reporter expression. Together, these results demonstrate that the transactivation activity of MZF-2A is negatively regulated by MAP kinasemediated phosphorylation at the three serine residues in the transactivation domain. DISCUSSION In this report, we identified and described a novel isoform of mouse MZF-2, designated MZF-2A, which is generated by the use of an alternative transcription start site and subsequent splicing. Compared with MZF-2, MZF-2A lacks 79 N-terminal amino acids and shows a basal level of transactivation activity. Recently, Peterson and Morris (11) identified two novel transcripts of the human MZF gene, MZF1B and MZF1C. These mRNAs differ in their 5Ј-untranslated region but encode the same polypeptide of 734 amino acids, designated MZF1B/C (11). Mouse MZF-2A and human MZF1B/C are highly homologous along their entire sequences (81.7% amino acid identity) and are likely to represent the major MZF gene product in myeloid cells.
We showed that the previously identified transactivation domain of MZF-2A was functional when fused to the heterologous DNA-binding domain of yeast GAL4 and defined the minimal TAD as a 50-amino acid region (aa 239 -288), which is necessary and sufficient to activate transcription. The amino acid composition of this minimal TAD is very characteristic of TADs; i.e. it is rich in acidic amino acid and proline residues. There are five aspartic acid and four glutamic acid residues but no basic (arginine or lysine) residues, resulting in a low isoelectric point of 3.8. Similar acidic and proline-rich features are observed in the TAD of many other transcription factors (51) and are likely to be important for their interaction with other transcription factors.
Our studies demonstrate that MZF-2A is phosphorylated by ERK and p38 MAP kinases. In vitro phosphorylation experiments showed that MZF-TAD is phosphorylated strongly by ERK, weakly by p38 kinase, and not at all by JNK1, and further analysis using "add-back" mutants revealed that three serine residues (Ser 257 , Ser 275 , and Ser 295 ) in the TAD could be phosphorylated by ERK1 in vitro. The mobility shift analysis of MZF-2A expressed in LGM-1 cells (Fig. 5) demonstrated that any of the single (S257A, S275A, and S295A) and double (S275/ 295A, S257/295A, and S257/275A) serine residue mutants showed the same mobility as wild-type MZF-2A, indicating that these mutants were still phosphorylated at the remaining unmutated sites. This result indicates that all three serine residues are susceptible to phosphorylation by MAP kinases and that the phosphorylation of any one of them is sufficient to cause the mobility shift. Therefore, the phosphorylation of any of the three serine residues is likely to cause a similar, repressive effect on the transactivation potential of MZF-2A. Consistent with this, additional alanine substitutions for any of the serines cumulatively enhanced the transactivation activity of MZF-2A, and the highest activation was observed when all the phosphorylation sites were disrupted (Fig. 5). These results indicate that the phosphorylation of each serine residue functions cooperatively to repress the ability of MZF-2A to activate transcription.
We have shown that various extracellular stimuli, including TPA, anisomycin, FCS, and IL-3, induce the phosphorylation of MZF-2A. Experiments with the protein kinase inhibitors U0126 and SB203580 revealed that distinct MAP kinase pathways are responsible for MZF-2A phosphorylation, depending on the type of stimulation. TPA activates ERK MAP kinases strongly, and the MEK inhibitor U0126 blocked TPA-induced MZF-2A phosphorylation almost completely, clearly demonstrating that ERKs play a predominant role in phosphorylation of MZF-2A in response to TPA stimulation. ERKs are also partly involved in FCS-or IL-3-induced phosphorylation. On the other hand, anisomycin-induced phosphorylation of MZF-2A was partially inhibited only by SB203580, suggesting that p38 MAP kinases, but not ERKs, are involved. Interestingly, combined treatment with both inhibitors mostly blocked MZF-2A phosphorylation induced by TPA or FCS but still resulted in partial inhibition of anisomycin-or IL-3-induced phosphorylation, suggesting that other protein kinase(s) insensitive to these inhibitors are involved downstream of anisomycin and IL-3. Although the JNK pathway is known to be resistant to these inhibitors (47,50), JNK MAP kinases are unlikely to play this role because neither JNK1 (Fig. 3) nor JNK2 (data not shown) phosphorylated GST-MZF-TAD in vitro. Another possibility is that the SB203580-insensitive p38 isoforms, such as p38␥/SAPK3 and p38␦/SAPK4 (49,50), phosphorylate MZF-2A. These kinases are likely to mediate anisomycin-inducible and SB203580-resistant phosphorylation of eukaryotic elongation factor 2 kinase in KB cells (52), and they may similarly play a prominent role in anisomycin-or IL-3-dependent phosphorylation of MZF-2A in LGM-1 cells. These findings demonstrate that the MZF-2A transcription factor integrates MAP kinase signaling pathways in response to different extracellular stimuli, such as Elk-1, Sap-1a, and ATF2 (26,27,(53)(54)(55).
Specific phosphorylation events can regulate the activity of transcription factors via several different mechanisms involving changes in protein stability, subcellular localization, or protein-protein interactions (19,28,30,32). We showed that protein stability is not significantly affected by alanine substitution of the phosphorylation sites (Fig. 5). We also found by subcellular fractionation experiments that MZF-2A is present exclusively in the nuclear fraction, irrespective of its phosphorylation state. 2 The KM3 mutant shows enhanced transcriptional activity, indicating that the N-terminal domain inhibits the transcriptional activity of the TAD. We previously proposed a model for this negative regulation in which an intramolecular interaction of the N-terminal domain with the adjacent TAD impedes the recruitment of a transcriptional cofactor(s) (13). The findings in this report that disruption of all three phosphorylation sites elevates MZF-2A transactivation activity to a level comparable with that of the KM3 mutant and that alanine substitution of the phosphorylation sites in the KM3 construct enhances its activity only weakly (Fig. 5, top panel) suggest that the alanine substitution and the N-terminal truncation could cause a similar effect on TAD function, which may imply that TAD phosphorylation is required for the N-terminal regulatory domain to repress transactivation activity. Combining this possibility with the previous model, dephosphorylation might release the TAD from the inhibitory domain and allow the TAD to interact with the transcriptional machinery (Fig. 6). Alternatively, because the LeR/SCAN domain is implicated to function as a module for protein-protein interaction (56,57), TAD phosphorylation may cooperate with the LeR/SCAN domain to bind a repressor protein. 2 H. Ogawa, S. Nagata, and R. Fukunaga, unpublished observation. In summary, this work represents the first report of MZF-2A phosphorylation by MAP kinases. A challenge for future studies is to clarify the physiological significance of MZF-2A phosphorylation and the subsequent repression of its transactivating function. Because serum and IL-3 are potent growthpromoting factors, inducible or constitutive phosphorylation of MZF-2A by these factors would contribute to reducing the growth-suppressing function of MZF-2A, and dephosphorylation and the subsequent activation of MZF-2A may be involved in cell cycle arrest and terminal differentiation of myeloid cells. In this regard, it would be intriguing to examine the regulation of MZF-2A phosphorylation in granulocyte colonystimulating factor-driven granulocytic differentiation and maturation.