Extracellular Signal-Regulated Kinase Induces the Megakaryocyte GPIIb/CD41 Gene through MafB/Kreisler

Abstract

Extracellular signal-regulated kinase (ERK) facilitates cell cycle progression in most mammalian cells, but in certain cell types prolonged signaling through this pathway promotes differentiation and lineage-specific gene expression through mechanisms that are poorly understood. Here, we characterize the transcriptional regulation of platelet GPIIb integrin (CD41) by ERK during megakaryocyte differentiation. ERK-dependent transactivation involves the proximal promoter of GPIIb within 114 bp upstream of the transcriptional start site. GATA, Ets, and Sp1 consensus sequences within this region are each necessary and function combinatorially in ERK-activated transcription. MafB/Kreisler is induced in response to ERK and synergizes with GATA and Ets to enhance transcription from the proximal promoter. The requirement for MafB in promoter regulation is demonstrated by inhibition of transactivation following dominant-negative or antisense suppression of MafB function. Thus, ERK promotes megakaryocyte differentiation by coordinate regulation of nuclear factors that synergize in GPIIb promoter regulation. These results establish a novel role for MafB as a regulator of ERK-induced gene expression.

Mitogen-activated protein kinase (MAPK) pathways are central regulators of mammalian cell proliferation and transformation. Activation of the Raf/MEK/extracellular signal-regulated kinase (ERK) pathway in mammalian cells induces activity of several transcription factors which facilitate cell proliferation in part through induction of cyclin D1 and S-phase entry (15, 26). Conversely, inhibiting ERK with antisense or dominant-negative mutants suppresses cell cycle progression at the G₁/S boundary (42). In consistency with its prominent role in promoting mammalian cell cycle progression, the ERK pathway generally suppresses differentiation across a broad spectrum of cellular systems, including models of myogenesis (3), adipogenesis (21), and epithelialization (48).

Paradoxically, MAPK pathway activation in some systems inhibits cell division and induces cell differentiation. This was first demonstrated with PC12 cells, in which sustained activation of the ERK pathway in response to nerve growth factor causes neurite outgrowth and expression of neuronal markers (7). Studies of PC12 cells have established a paradigm for ERK signaling in which proliferative versus differentiative outcomes are controlled by the kinetics of ERK signaling: transient signaling leads to proliferation, whereas prolonged signaling over several hours promotes differentiation (34). Recent studies show that sustained ERK signaling induces neurite outgrowth by driving expression of the p35 regulatory subunit of cdk5 via induction of the immediate-early Egr1/Krox24 gene (16).

In addition to neuronal differentiation, ERK also promotes hematopoietic cell differentiation and commitment. For example, ERK regulates T-cell development in mice by elevating CD4⁺ cell numbers and reducing CD8⁺ populations, thus favoring commitment to helper T-cell lineages (1, 49). In addition, ERK promotes megakaryocyte differentiation in several blood cell models, including K562, CMK, CHRF228, F36P-mpl, and UT7 (18, 22, 37, 38, 46, 47, 54, 57). In these cell lines, activation of ERK pathway induces many features of megakaryocyte differentiation, including induction of platelet markers, increased cell adhesion and spreading, increased cell size, and polyploidy. ERK also suppresses hemoglobin mRNA in these cell lines, suggesting that the ERK pathway also regulates commitment between megakaryocyte and erythroid cell fates (54, 57).

In UT7 and F36P cells, ERK-dependent megakaryocyte differentiation is responsive to thrombopoietin, which promotes megakaryocyte survival and differentiation (47, 54). Activation of ERK by Tpo/c-Mpl involves stimulation of Ras/Raf1 and Rap1/B-Raf, the latter of which appears to contribute to sustained ERK signaling (14, 20). Prolonged ERK signaling appears to be necessary for megakaryocyte marker induction as well as cell growth arrest (14, 22, 37, 57). Thus, cell systems for neuronal and megakaryocyte differentiation share characteristic kinetic requirements for lineage-specific gene expression. However, little is known about the mechanism underlying ERK control of megakaryocyte gene expression.

GPIIb(CD41) is a megakaryocyte marker whose central role in platelet aggregation makes it a promising target for treatment of thrombosis and cardiovascular disease (5, 27). GPIIb functions as a heterodimer with GPIIIa(CD61) and regulates cell adhesion and platelet aggregation by binding fibrinogen and von Willebrand factor (44). The promoter regulating GPIIb transcription has been localized to 600 bp upstream of the transcriptional start site (28, 45, 55). Within this region are proximal (−114 to −1) and distal (−515 to −478) promoter regions that direct GPIIb expression in megakaryocytes and a bipartite silencer (−198 to −114) that represses GPIIb expression in nonmegakaryocyte cells (13, 28, 45). Transcriptional activation of the promoter in response to the presence of phorbol ester is suppressed by cycloheximide (59), indicating that differentiation-dependent transcription of GPIIb requires protein synthesis.

In this study, we investigated the mechanistic basis for ERK-regulated megakaryocyte differentiation in K562 erythroleukemia cells. We showed that ERK regulates the GPIIb promoter through combinatorial interactions between GATA and Ets family transcription factors, which synergize with MafB/Kreisler, a transcription factor known to repress erythroid differentiation (51). This establishes a novel transcriptional role for MafB/Kreisler as a coregulator of ERK-dependent gene expression during blood cell differentiation.

MATERIALS AND METHODS

Cell culture.

Human erythroleukemia K562 cells were obtained from the American Type Culture Collection. Cells were maintained in spinner culture or T175 flasks at 0.5 × 10⁶ to 1 × 10⁶ cells/ml in RPMI 1640 medium supplemented with 10% (vol/vol) fetal bovine serum (FBS) (Invitrogen), 100 U of penicillin/ml, and 100 U of streptomycin/ml.

Vector construction.

Region −597 to +32 of the human GPIIb promoter was amplified by PCR from genomic DNA extracted from K562 cells with primers 5′-AGCTTGGCTCAAGACGGAGGAGGAGTGAGGAA-3′ (GPIIb sense at −597) and 5′-CTTCCTTCTTCCACAACCTCCCAGGCAGGAATGGGC-3′ (GPIIb antisense at +32), and the 629-bp fragment was subcloned into pT7Blue T-Vector (Novagen) by T/A cloning and then subcloned into pGL3-Basic (Promega) to yield pGL3-GPIIb(Full Length Promoter).

The distal enhancer region (−597 to −406) was excised by restriction digestion with KpnI and PvuII, yielding pGL3-GPIIb Δ(−597 to −406). Excision of a silencer region defined by Fong and Santoro (13) was accomplished by restriction digestion of pGL3-GPIIb(Full Length Promoter) with NlaIII, yielding pGL3-GPIIb Δ(−223 to −121). The proximal promoter fragment from −114 to +32 was generated by PCR and ligated into pGL3-Basic, yielding pGL3-GPIIb Δ(−597 to −115) [(pGL3-GPIIb Proximal Promoter)]. Point mutations in pGL3-GPIIb(Full Length Promoter) and pGL3-GPIIb(Proximal Promoter) were created using a QuickChange Kit (Stratagene) and mutagenesis primers 5′-GGGGTTCCAGTTTCCGTAGAAAAGACTTCCTG-3′ (−55 GATA site mutation), 5′-GAAAAGACTGTCTGTGGTGCTATGTGAAGGGAAGGAGG-3′ (−41 Ets site mutation), and 5′-GAAGGGATGGTCGAGGACCTGGCCCATTCCTG-3′ (−14 Sp1 site mutation). pRLTK and pRL-null constructs for expression of Renilla luciferase (Promega) were used to normalize transfection efficiencies.

Plasmids for mammalian expression of wild-type MKK1 (WT-MKK1), WT-MKK2, constitutively active MKK1 (CA-MKK1) (R4F), and CA-MKK2 (KW71A) have been described previously (29, 30, 57). Human GATA1 and Fli1 were cloned by reverse transcription-PCR (RT-PCR) from K562 cells stimulated for 48 h with phorbol 12-myristate 13-acetate (PMA; Sigma). Human MafB was cloned by PCR from a genomic clone, pBeloBACII 94m17 (Incyte Genomics). The PCR products were digested and cloned into pCMV-Tag2 (N-terminal FLAG tag) and pCMV-Tag3 (N-terminal MYC tag) (Stratagene). pBS KS+ c-Ets1 and pBS KS+ Ets2 for expression of murine Ets factors were gifts of James Hagman (National Jewish Medical Center, Denver, Colo.). After digesting pBS KS+ c-Ets1 with EagI/BamHI and pBS KS+ Ets2 with BamHI, they were ligated into pCMV-Tag2 and pCMV-Tag3. Dominant-negative Ets1 (amino acids 331 to 440) and dominant-negative Ets2 (amino acids 358 to 468) were created by PCR and cloned into pCMV-Tag2. Human pCMV-Sp1 was a gift of James Goodrich (University of Colorado, Boulder, Colo.). Dominant-negative MafB (amino acids 204 to 337 containing the DNA binding domain) was cloned by digesting pCMV-Tag2-MafB with SrfI/EcoRI and religating the 406-bp product (corresponding to residues 204 to 337) into pCMV-Tag2. Engrailed repressor MafB was created by PCR amplification of the mouse engrailed-2 repressor domain (amino acids 2 to 207) (gift of Ilona Skerjanc, University of Western Ontario, London, Ontario) and then cloning the fragment into BamHI-digested pCMV-Tag2-MafB. Antisense MafB was created by reversing the orientation of the coding sequence of pCMV-Tag2-MafB.

Cell stimulation and transfection.

K562 cells were transfected by electroporation (Bio-Rad Gene Pulser) (960 μF, 210 V). Cells (5 × 10⁶) were washed once with Hank's buffered saline solution, resuspended in 0.4 ml of OptiMEM (Invitrogen) supplemented with 1% (vol/vol) FBS without antibiotics, and transferred to a 4-mm-gap electroporation cuvette (BTX). Plasmid DNA was added to cells; after 5 min at room temperature, cells were subjected to a single electric pulse followed by 5 min of incubation at room temperature and dilution into 10 ml of RPMI 1640-12% (vol/vol) FBS plus antibiotics. After recovery for 12 to 16 h, cells were centrifuged at 250 × g for 10 min, resuspended into 20 ml of fresh RPMI 1640-10% FBS plus antibiotics, and further incubated for 48 h. Transfection efficiencies under these conditions ranged between 80 to 90%, as assessed by expression of green fluorescent protein (pEGFP; Clontech). Cells were stimulated with 10 nM PMA for 24 to 48 h with 1 h of pretreatment with 20 μM U0126 (Promega) or were left untreated.

Luciferase assays.

Transfected cells were plated in 12-well plates. Cells were processed and assayed using the Dual-Luciferase Reporter assay system (Promega). Briefly, each well was extracted in 200 μl of passive lysis buffer, 20 μl was mixed with 100 μl of firefly luciferase substrate, and luminescence was quantified on a MicroLumat LB96P luminometer (EG&G Berthold). Renilla luciferase activity was determined by adding 100 μl of Stop and Glow solution and measurement of luminescence.

RNA isolation and RT-PCR.

Total RNA was isolated from K562 cells using an RNeasy Mini kit (Qiagen) and treated with DNaseFree (Ambion) to clear RNA of possible genomic DNA contamination. RT was performed using random primers and Superscript II RNase H-reverse transcriptase (Invitrogen). PCR amplification of selected genes used the following primers: 5′-CAGGTACTCAGTGCACCAAC-3′ (human GATA1 sense), 5′-TGGTAGAGATGGGCAGTACC-3′ (human GATA1 antisense), 5′-CTATGGTATTGAGCATGCCCAGTG-3′ (human Ets1 sense), 5′-AAGGTGTCTGTCTGGAGAGGGTCT-3′ (human Ets1 antisense), 5′-ATTCACACCTCACCTCCGTTCCTC-3′ (human Ets2 sense), 5′-GGAAGTCCTGACTGACAGAGCAGT-3′ (human Ets2 antisense), 5′-CCAGAACATGGATGGCAAGGA-3′ (human Fli1 sense), 5′-CCCAGGATCTGATACGGATCTGGC-3′ (human Fli1 antisense), 5′-CCTGACGCCCGAGGACG-3′ (human MafB sense), 5′-CGCACGGACATGGACACG-3′ (human MafB antisense), 5′-AAGAGATGGCCACGGCTGCT-3′ (human β-actin sense), 5′-TCGCTCCAACCGACTGCTGT-3′ (human β-actin antisense), 5′-GGGATGGGAGGCATGATC-3′ (human GPIIb sense), and 5′-GTCTGGGTATCCGTTGTCA-3′ (human GPIIb antisense). PCR was performed using a Fail Safe PCR kit (Epicentre). [α-³²P]dATP (ICN) was added to the PCRs, and radiolabeled product was quantified on a phosphorimager (Molecular Dynamics).

Immunochemistry.

Transfected K562 cells were washed once with Hank's buffered saline solution and then lysed in radioimmunoprecipitation assay buffer (9.1 mM Na₂HPO₄, 1.7 mM NaH₂PO₄, 150 mM NaCl, 1% NP-40, 0.5% deoxycholic acid, 0.1% sodium dodecyl sulfate [SDS]) with protease inhibitors (1 μg of pepstatin A/ml, 10 μg of aprotinin/ml, 10 μg of leupeptin/ml) and clarified by centrifugation at 17,000 × g for 5 min. Clarified lysate was incubated with 20 μl of mouse anti-c-MYC (9E10) (sc-40; Santa Cruz Biotechnology) for 1 h at 4°C, and 40 μl of protein A/G PLUS-Agarose (sc-2003; Santa Cruz Biotechnology) was added and incubated for 30 min at 4°C. After removing supernatant, agarose beads were washed four times with phosphate-buffered saline (9.1 mM Na₂HPO₄, 1.7 mM NaH₂PO₄, 150 mM NaCl). Immunoprecipitated proteins were eluted with 1× SDS-polyacrylamide gel electrophoresis (PAGE) sample buffer (50 mM Tris [pH 6.8], 100 mM dithiothreitol, 2% [wt/vol] SDS, 0.1% bromophenol blue, 10% [vol/vol] glycerol). Both eluates and whole-cell extracts in radioimmunoprecipitation assay buffer were analyzed using SDS-PAGE.

For Western blotting, proteins were transferred to Immobilon-P (Millipore), blocked for 1 h in 3% nonfat milk-2% bovine serum albumin or 5% nonfat milk, and incubated with primary antibodies (1:100 to 1:200 dilution) followed by 1 h of incubation with goat anti-rabbit or goat anti-rat (Jackson Laboratories) or donkey anti-goat (Santa Cruz Biotechnology) immunoglobulin G coupled to horseradish peroxidase (1:10,000 dilution). After extensive washing, blots were visualized by enhanced chemiluminescence (Amersham). After visualization, the blots were stripped and reprobed per the manufacturer's protocol (Amersham). Primary antibodies used were rat anti-GATA1 (N6) (sc-265; Santa Cruz) (1:200), rabbit anti-Ets1 (C-20) (sc-350; Santa Cruz) (1:100), and goat anti-MafB (P-20) (sc-10022; Santa Cruz Biotechnology) (1:200).

Nuclear extract preparation.

K562 cells (3 × 10⁹) were treated with 10 nM PMA for 48 h or were left untreated. Nuclear extracts were prepared as previously described (10) with the following modifications. Buffers for low-salt lysis (buffer A) and high-salt extraction (buffer C) contained protease inhibitors (1 μg of pepstatin A/ml, 10 μg of aprotinin/ml, 10 μg of leupeptin/ml) and phosphatase inhibitors (0.1 mM sodium orthovanadate, 1 μM microcystin). High-salt crude nuclear extracts were desalted into electrophoretic mobility shift assay (EMSA)-pulldown buffer (20 mM HEPES [pH 7.9], 50 mM KCl, 0.05% NP-40, 1 mM dithiothreitol, 1 mM EDTA, 5% glycerol, protease inhibitor and phosphatase inhibitors) with PD-10 desalting columns (Amersham), immediately frozen in liquid nitrogen, and stored at −80°C. Small-scale nuclear extraction procedures were performed with transfected K562 cells as described previously (6) except that buffers A and C contained protease and phosphatase inhibitors and high-salt crude nuclear extracts were desalted into EMSA-pulldown buffer with G-25 Macro SpinColumns (Harvard Apparatus), immediately frozen in liquid nitrogen, and stored at −80°C.

Oligonucleotide pulldown assays.

The GPIIb proximal promoter (both WT and mutant) was PCR amplified and gel purified using primers 5′-biotin-TTGCTCTGCCCGTTGCTCA-3′ and 5′-CTTCCTTCTTCCACAACCTC-3′. GATA and Ets double-stranded DNA (dsDNA) oligonucleotides were annealed using 5′-biotin-CACTTGATAACAGAAAGTGATAACTCT-3′ and 5′-AGAGTTATCACTTTCTGTTATCAAGTG-3′ for GATA consensus, 5′-biotin-CACTTCTTAACAGAAAGTCTTAACTCT-3′ and 5′-AGAGTTAAGACTTTCTGTTAAGAAGTG-3′ for GATA mutant, 5′-biotin-GATCTCGAGCAGGAAGTTCGA-3′ and 5′-TCGAACTTCCTGCTCGAGATC-3′ for Ets, and 5′-biotin-GATCTCGAGCAAGAAGTTCGA-3′ and 5′-TCGAACTTCTTGCTCGAGATC-3′ for Ets mutant. A total of 300 μg of nuclear extract in EMSA-pulldown buffer was incubated with 10 μg of dI-dC/ml, 200 μg of insulin/ml, and 2 μg of dsDNA oligonucleotide in 500 μl and incubated for 1 h at 4°C. A total of 50 μl of blocked streptavidin Dynabeads (M-280; Dynal Biotech) blocked in 10 μg of dI-dC/ml, 100 μg of sheared calf thymus DNA/ml, and 200 μg of insulin/ml was added, incubated for 30 min at 4°C, washed three times in 500 μl of EMSA-pulldown buffer-10 μg of dI-dC/ml-200 μg of insulin/ml, and eluted in 1× SDS-PAGE sample buffer.

EMSAs.

dsDNA oligonucleotides were annealed and labeled with T4 kinase (Invitrogen) and [γ-³²P]ATP (ICN) following the manufacturers' instructions. Oligonucleotides were 5′-AGTTTGATAAGAAAAGACTTCCTGTGGAGGAATCTGA-3′ and 5′-TCAGATTCCTCCACAGGAAGTCTTTTCTTATCAAACT-3′ for the GPIIb probe and 5′-GGAATTGCTGACTCAGCATTACTC-3′ and 5′-GAGTAATGCTGAGTCAGCAATTCC-3′ for the MafB probe. GATA and Ets probes (GATA consensus gel shift oligonucleotides [sc-2531] and Ets1/PEA3 consensus gel shift oligonucleotides [sc-2555]; Santa Cruz Biotechnology) were purchased. For gel shift assays, 40,000 cpm of dsDNA (approximately 0.1 ng) was incubated with 2 μg of nuclear extract in EMSA-pulldown buffer containing 50 ng of dI-dC in a final volume of 20 μl. Binding competition was performed using excess unlabeled oligonucleotide. Supershift assays were performed with 1 μl (2 μg) of monoclonal anti-c-MYC (M4439; Sigma). After 30 min at room temperature, 2 μl of loading dye (25% Ficoll, 0.01% bromophenol blue, 0.25× Tris-borate-EDTA [22.5 mM Tris-borate, 0.5 M EDTA]) was added, and samples were separated in preelectrophoresed nondenaturing 6% polyacrylamide gels in 0.25× Tris-borate-EDTA.

RESULTS

ERK regulates GPIIb transcription through the proximal promoter.

Because of its unique expression in megakaryocytes and platelets in humans, GPIIb integrin was chosen as a model for ERK-dependent regulation of lineage-specific markers. The human GPIIb promoter was cloned by PCR and included bases −597 to +32 of the GenBank sequence AF160252. Figure 1A shows the full-length GPIIb promoter and consensus binding sites for known transcription factors.

FIG. 1.

The proximal promoter of GPIIb is responsive to ERK signaling. (A) The megakaryocyte-specific promoter of GPIIb, showing consensus binding sites for transcription factor families. (B) GPIIb promoter mutants (cloned into luciferase reporter vector pGL3) used in this study. (C) K562 cells were transfected with the GPIIb full-length promoter or the GPIIb proximal promoter vector (5 μg), pRL null vector (5 μg), and carrier pUC119 DNA (60 μg) and then treated with 10 nM PMA and 20 μM U0126 or left untreated. After 48 h of drug treatment, cells were harvested and luciferase activity levels were measured. (D) Cells were transfected with GPIIb promoter mutant constructs (as described for panel B) or empty pGL3 vector (5 μg) in addition to pRLTK (5 μg) and either WT-MKK1 plus WT-MKK2 (35 μg each) or CA-MKK1 plus CA-MKK2 (35 μg each). Transfected cells were maintained for 48 h and harvested, and luciferase activity levels were measured. (E) Cells were transfected with GPIIb promoter mutant constructs (as described for panel B) or empty pGL3 vector (5 μg) in addition to pRLTK (5 μg). Cells were treated with 10 nM PMA for 48 h or left untreated and were harvested, and luciferase activity levels were measured. Firefly luciferase activity levels were normalized to corresponding Renilla luciferase activity levels to correct for transfection efficiency. Values represent averages and standard deviations of duplicate transfections normalized to the highest value (+PMA). Similar results were observed in two or three independent experiments.

Transcription from the full-length GPIIb promoter was strongly enhanced in response to the presence of 10 nM PMA and suppressed by U0126, the MKK1/2 inhibitor (Fig. 1C). Coexpressing the full-length promoter reporter with constitutively active forms of MKK1 and MKK2 (CA-MKK) induced transcription to the same extent observed with PMA (Fig. 1D). This indicates that ERK signaling is both necessary and sufficient for activation of the GPIIb promoter and accounts for the characteristic responsiveness of GPIIb to the presence of PMA.

To identify the region of the GPIIb promoter responsive to ERK signaling, promoter deletion mutants were constructed and tested for regulation by CA-MKK and PMA (Fig. 1D and E). Deletion Δ(−597 to −406) removes the distal enhancer (which regulates GPIIb expression in megakaryocytes) (45), deletion Δ(−223 to −121) removes part of a silencer region which suppresses GPIIb expression in nonmegakaryocyte cells (13), and deletion Δ(−597 to −115) retains only the proximal promoter (which also regulates megakaryocyte expression) (28). K562 cells were transfected with each deletion mutant and cotransfected with CA-MKK versus WT-MKK or treated with PMA for 48 h. The strongest induction by CA-MKK or PMA was observed with the full-length promoter (Fig. 1D and E). Although removal of the distal enhancer region reduced responsiveness to both CA-MKK and PMA, the construct containing the minimal proximal promoter still yielded 50% of maximal activation. PMA-induced transcription from the proximal promoter was strongly inhibited by the presence of U0126 (Fig. 1C), confirming that the promoter response within this region can be primarily attributed to ERK signaling. Partial disruption of the silencer region elevated basal transcription but had little effect on CA-MKK- or PMA-induced transcription (Fig. 1D and E). A small (fourfold) increase in responsiveness to CA-MKK or PMA was observed in the promoterless vector (Fig. 1C to E). Taken together, the results demonstrated that the region of the GPIIb promoter most responsive to ERK is located within 114 bp upstream of the transcriptional start site. Therefore, further studies focused on the region between −114 to +32 containing the proximal promoter.

ERK regulates GPIIb expression through coordinate regulation of multiple promoter response elements.

Consensus response elements for GATA, Ets, and Sp1 DNA binding proteins are within the proximal region of the GPIIb promoter (Fig. 2A). To test the dependence of ERK-regulated transcription on these binding sites, point mutations were introduced into each element individually and in combination. The Ets element consisted of two potential binding sites, which were mutated together. In the proximal promoter, disruption of each transcription factor binding site individually reduced ERK-stimulated transcription by 65 to 85% (Fig. 2B, left panel). Disruption of any two sites reduced transcription by 85 to 90%, and disruption of all three sites together reduced transcription to the level observed with empty vector. In the full-length promoter, disruption of each site individually reduced ERK-stimulated transcription by 45 to 75%. Disruption of any two sites reduced transcription by 75 to 85%, and disruption of all three sites reduced transcription by 90% (Fig. 2B, right panel). Similar results were seen in response to PMA with each reporter construct (data not shown). The results demonstrate that consensus binding sites for GATA, Ets, and Sp1 factors are each necessary and together act combinatorially to confer the ERK sensitivity of transcription from the proximal promoter.

FIG. 2.

Promoter response elements mediate ERK-dependent transcription combinatorially. (A) Sequence of the proximal promoter. GATA, Ets, and Sp1 consensus binding sites are indicated in boxes, and base changes that disrupt these sites are indicated in boldface characters. (B) Cells were transfected with WT- or CA-MKK1/2, pRLTK, and various mutant GPIIb promoter constructs (indicated at the left side of the panel). After 48 h, cells were harvested and firefly luciferase activity levels were measured and normalized to Renilla luciferase activity levels. Averages and standard deviations of duplicate transfections are shown.

Next, representative members of GATA, Ets, and Sp1 families were tested for their ability to activate the GPIIb proximal promoter. Epitope-tagged forms of human GATA1, Ets1, Ets2, and Sp1 were cotransfected with the GPIIb proximal promoter reporter into K562 cells. GATA1, Ets1, and Ets2 enhanced transcription by 5- to 10-fold when expressed individually (Fig. 3A). A small synergistic effect was observed upon coexpression of GATA1 plus Ets1, although the effect of GATA1 plus Ets2 appeared additive. Sp1 had little effect on transcription and in fact partially repressed transcription by GATA plus Ets. Overall, promoter activation observed upon transcription factor expression reached less than 25% of the level observed with PMA. Fli1, an Ets family member previously reported to induce GPIIb mRNA when virally delivered into K562 cells (2), was ineffective in enhancing proximal promoter activation (either alone or in combination with GATA1 and Sp1) (Fig. 3B). These results indicate that GATA, Ets, and Sp1 factors cooperate to facilitate transactivation from the GPIIb promoter, although they do not fully account for the extent of ERK- or PMA-dependent regulation.

FIG. 3.

GPIIb promoter activation by expression of GATA1 and Ets1. (A) Cells were transfected with the GPIIb proximal promoter (5 μg), pRL null (5 μg), and indicated combinations of FLAG-GATA1, FLAG-Ets1, FLAG-Ets2, and Sp1 (20 μg each). (B) The experiment was conducted as described for panel A except that cells were transfected with combinations of FLAG-GATA1, FLAG-Fli1, and Sp1 (20 μg each). The amount of plasmid DNA in each transfection was held constant at 70 μg by addition of pUC119. Cells were maintained for 48 h and harvested, and firefly luciferase activity levels were measured and normalized to cell numbers. Values represent averages and standard deviations of duplicate transfections.

MafB/Kreisler regulates the GPIIb promoter in synergy with GATA and Ets.

The potential involvement of other transcription factors in regulating the GPIIb promoter was tested. In particular, effects were observed with MafB/Kreisler, a factor reported to directly bind Ets1 and repress markers of erythroid differentiation (51, 52) but never previously linked to ERK signaling. Semiquantitative RT-PCR (performed with RNA isolated from cells transfected with WT-MKK or CA-MKK) showed that steady-state levels of MafB mRNA increased significantly in response to ERK activation (Fig. 4A). Likewise, ERK signaling led to coordinate induction of Ets1 and Fli1, although steady-state levels of GATA1 and Ets2 remained unchanged. Similar increases in MafB, Ets1, and Fli1 (but not GATA1 or Ets2) levels were obtained in K562 cells treated with PMA (data not shown). Protein levels of MafB were also elevated in response to both PMA stimulation and transfection by CA-MKK (Fig. 4B).

FIG. 4.

ERK signaling induces MafB and Ets1. (A) Cells transfected with WT-MKK1/2 or CA-MKK1/2 were maintained for 48 h followed by the harvesting of total RNA for RT-PCR. ³²P-labeled PCR products were separated by PAGE and visualized by phosphorimager analysis. Each PCR sampled three different cycles that were identical between WT- and CA-MKK experiments and were chosen by the range of linear amplification of each product as follows: for MafB, cycles 32, 34, and 36; for Ets1, cycles 32, 34, and 36; for Ets2, cycles 34, 36, and 38; for Fli1, cycles 28, 30, and 32; for GATA1, cycles 26, 28, and 30; and for β-actin, cycles 18, 20, and 22. Similar results were obtained in three independent experiments. (B) Western blotting of MafB protein. Whole-cell extracts (50 μg per lane) from cells left untreated, treated with 10 nM PMA, transfected with WT-MKK1/2, or transfected with CA-MKK1/2 for 48 h were used.

Because MafB mRNA and protein were induced in response to ERK, we asked whether this factor contributed to activation of the GPIIb promoter. Coexpressing MafB with either GATA1 plus Ets1 or GATA1 plus Ets2 markedly enhanced GPIIb promoter activation compared to that observed upon expressing GATA1 plus Ets1/2 alone, reaching 60% of the activity level achieved with PMA (Fig. 5A). In contrast, no enhancement was observed upon coexpressing MafB with GATA1 plus Fli1 (Fig. 5B), suggesting that Fli1 is not as effective as Ets1/2 in promoter regulation. Western blot analyses confirmed that the increased responses observed in the MafB cotransfection experiments were not caused by increased nuclear factor expression (Fig. 5C). The results demonstrate that MafB synergizes with GATA1 and Ets1/2 but not with Fli1, indicating selectivity in combinatorial regulation of the GPIIb promoter.

FIG. 5.

MafB synergizes with GATA and Ets to activate transcription from the GPIIb promoter. Cells were transfected with GPIIb proximal promoter (5 μg), pRL null (5 μg), and the indicated combinations of FLAG-GATA1, FLAG-Ets1, FLAG-Ets2, and MafB (20 μg each) (A) or FLAG-GATA1, FLAG-Fli1, and FLAG-MafB (20 μg each) (B). The amount of plasmid DNA in each transfection was held constant at 70 μg by addition of pUC119. Cells were maintained for 48 h and were harvested, and firefly luciferase activity levels were measured and normalized to cell numbers. Averages of the results of two independent experiments (with error bars representing the standard deviations) are shown. (C) GATA1, Ets1, and MafB were immunoprecipitated from extracts of transfected cells with anti-FLAG antibody and reprobed by Western blotting. (D) Cells were transfected with GPIIb promoter mutant constructs, pRL null, and FLAG-GATA1, FLAG-Ets1, and FLAG-MafB as described for panel A and were harvested at 48 h, and firefly luciferase activity levels were measured and normalized to Renilla luciferase activity levels. Values represent averages and standard deviations of duplicate transfections.

Inspection of the proximal promoter sequence showed no obvious consensus binding site for MafB, suggesting that its effects might be mediated indirectly through GATA, Ets, or Sp1 binding. This was tested by examining the synergistic effect of MafB on proximal promoter mutants disrupted at the GATA, Ets, and Sp1 binding sites. GATA1, Ets1, and MafB were cotransfected with the various promoter mutants, and transcription rates were measured (Fig. 5D). Mutation of any site individually reduced transcriptional activation by 75 to 80%, mutation of two sites reduced activity by >95%, and mutation of all three sites reduced transcription to the background levels seen with promoterless vector. This behavior paralleled the responses to the presence of CA-MKK observed upon promoter mutation (Fig. 2B). Thus, the enhancement of activity introduced by MafB depends on GATA, Ets, and Sp1 binding sites without necessarily requiring separate MafB-DNA binding interactions. This is consistent with a model in which ERK facilitates promoter activation by upregulating MafB, which then promotes transcription directly or indirectly through interactions with GATA or Ets.

The necessity of MafB for GPIIb promoter regulation was tested using three approaches to the disruption of MafB function. First, the N terminus of MafB was removed, leaving only the bZIP domain (residues 204 to 337) (Fig. 6A). Previous reports demonstrated the dominant-negative activity of this truncation mutant (23). Second, full-length MafB was fused to the repressor domain of mouse engrailed-2 (residues 2 to 207), which has been shown to confer dominant repression of transcriptional activators (43). Third, the full-length MafB open reading frame was subcloned in reverse orientation for antisense expression. Each construct was effective in inhibiting PMA-stimulated activation of the GPIIb proximal promoter (Fig. 6B). Engr-MafB reduced PMA-stimulated transcription by 70%, whereas dominant-negative and antisense MafB each reduced transcription by 85%. In addition, dominant-negative mutants of Ets1 and Ets2 (corresponding to the DNA binding domains of each protein) (24) strongly suppressed PMA-induced promoter activation (Fig. 6B). Because the PMA response can be primarily attributed to ERK signaling, these results demonstrate the importance of MafB as a regulator of ERK-dependent promoter activation.

FIG. 6.

MafB and Ets factors are necessary for activation of the GPIIb promoter. (A) Nuclear factor constructs and dominant-negative mutations. DN-Ets1/2 and DN-MafB consist of the DNA binding domain for each factor, and Engr-MafB fuses the engrailed repressor domain with full-length MafB. (B) Cells were transfected with GPIIb proximal promoter (5 μg), pRL null (5 μg), and pUC119 (pUC control) or DN-Ets1, DN-Ets2, DN-MafB, Engr-MafB, or antisense MafB (60 μg each). Transfected cells were then treated with 10 nM PMA for 48 h or left untreated and were harvested, and firefly luciferase activity levels were measured and normalized to Renilla luciferase levels. Values represent averages and standard deviations of duplicate transfections, with similar results obtained in two independent experiments. (C) Cells were transfected with pUC119 (pUC control), DN-Ets1, or DN-MafB and were then treated with 10 nM PMA for 48 h or left untreated and were harvested. Intensities of ³²P-labeled RT-PCR products derived from endogenous GPIIb mRNA were normalized to levels of β-actin RT-PCR products measured in parallel. Averages and standard deviations of the results of two independent experiments are shown.

We also tested the ability of dominant-negative mutants to suppress PMA-dependent transcription of endogenous GPIIb. Cells transfected with DN-Ets1 or DN-MafB and then challenged with PMA were analyzed by RT-PCR with primers corresponding to sequences within exons 11 and 14 of GPIIb, which predicts a 404-bp product. Controls transfected with pUC DNA showed 2.5-fold induction of GPIIb mRNA in response to the presence of PMA (Fig. 6C). DN-Ets1 and DN-MafB suppressed the PMA response by 98 and 60%, respectively, revealing the importance of MafB and Ets factors in regulating expression of the endogenous GPIIb gene during megakaryocyte differentiation.

GATA and Ets factors interact with the endogenous promoter.

Conceivably, overexpressed GATA, Ets, and Maf factors might activate the GPIIb promoter indirectly by inducing other factors that bind the promoter directly. Therefore, direct interaction of each factor with the GPIIb promoter was tested using oligonucleotide pulldown experiments and EMSAs. Endogenous GATA1 and GATA2 interacted with double-stranded oligonucleotides corresponding to both the GPIIb proximal promoter and a consensus GATA recognition sequence (Fig. 7A).

FIG. 7.

GATA and Ets proteins bind the endogenous GPIIb promoter. (A) The GPIIb promoter (−98 to +32) was PCR amplified with biotinylated primers and gel purified. Nuclear extract (300 μg) from K562 cells treated with 10 nM PMA or left untreated was incubated with 2 μg of biotinylated GPIIb promoter, consensus GATA oligonucleotides, or consensus Ets oligonucleotides for 1 h at 4°C with insulin (200 μg/ml) and dI-dC (10 μg/ml). Protein/DNA complexes were pulled down with 50 μl of blocked streptavidin Dynabeads for 30 min at 4°C. Dynabeads were washed, and complexes were eluted in sample buffer and analyzed by SDS-PAGE. Western blot analyses were performed to detect the presence of GATA, Ets, and MafB family members. Oligonucleotides used were either WT (W) or mutant (M) oligonucleotides in which the DNA binding sequence was mutated in the GATA, Ets, and Sp1 elements in GPIIb (Fig. 2) or in the GATA or Ets sequence in the consensus oligonucleotide. GATA1/2 and Ets1/2 proteins bound to GPIIb or consensus oligonucleotides were visualized by Western blotting. Western blot analyses showed no MafB bound to GPIIb oligonucleotides. (B) EMSA analysis of protein complexes with the GATA/Ets motif in GPIIb (−60 to −24). K562 cells were transfected with the indicated tagged constructs of GATA1, Ets1, and MafB (20 μg each) and treated with 10 nM PMA for 48 h or left untreated. Nuclear extracts were prepared from these cells, mixed with oligonucleotide probes, and separated by PAGE. Competition experiments used consensus GATA (lanes 5 and 6), Ets (lanes 11 and 12), or MafB (MARE) (lanes 17 and 18) oligonucleotide sequences. Anti-c-MYC antibody was used in supershift experiments (lanes 3, 4, 9, 10, 15, and 16). The low-molecular-weight complex was most likely an overexpressed Ets1 degradation product and was only present in experiments with overexpressed Ets1 and PMA treatment. 100X Comp, 100-fold molar excess of unlabeled oligonucleotide.

Binding was not dependent on PMA but was specific for intact GATA sites, because it was reduced upon disruption of GATA, Ets, and Sp1 binding sites in the GPIIb promoter and the GATA binding site in the GATA consensus oligonucleotide. Likewise, endogenous Ets1 and Ets2 interacted with oligonucleotides corresponding to both the GPIIb proximal promoter and consensus Ets recognition sequence (Fig. 7A). Ets1 binding to both the GPIIb promoter and Ets consensus was PMA dependent, while Ets2 binding was not PMA dependent. Under the conditions used, MafB binding to the GPIIb promoter was undetectable in the oligonucleotide pulldown experiments.

An alternative approach monitored protein-DNA complex formation by EMSA. K562 cells were cotransfected with MYC-tagged GATA1, Ets1, and MafB, and binding of each protein to a radiolabeled double-stranded oligonucleotide containing the GPIIb GATA/Ets motif (−60 to −24) was examined. MYC-GATA1 clearly bound the GATA/Ets motif (demonstrated by supershifting of GATA1/GPIIb complexes with anti-c-MYC antibodies) (Fig. 7B, lanes 3 and 4). Competition with a double-stranded oligonucleotide containing the GATA consensus sequence reduced the signal from the GATA/GPIIb complex (Fig. 7B, lanes 5 and 6). Likewise, the presence of a MYC-Ets1/GPIIb complex was also indicated by EMSA. Although it could not be selectively supershifted using anti-c-MYC antibody (Fig. 7B, lanes 9 and 10), its abundance was reduced by competition with an Ets1/2 consensus sequence (Fig. 7B, lanes 11 and 12). In contrast, no evidence was found for selective electrophoretic mobility shifting of a MYC-MafB/GPIIb complex (Fig. 7B, lanes 15 and 16) and a MafB (MARE) consensus probe did not selectively compete away observable protein/GPIIb complexes. Interestingly, the electrophoretic mobility of the GATA1/GPIIb complex showed a slower-migrating form in two extracts from PMA-treated cells (Fig. 7B; compare lanes 1 and 2), both of which were almost completely competed away with the GATA consensus probe. This suggests a PMA-dependent modification of components in the complex that might reflect regulated transcription.

Regulation of cell adhesion by GATA, Ets, and MafB.

Increased cell adhesion and spreading is a morphological response to megakaryocyte differentiation. To test whether the nuclear factors that regulated GPIIb expression could also alter cell morphology, transfections were performed under conditions shown to synergistically regulate GPIIb and were examined by phase-contrast microscopy (Fig. 8). As previously observed (57), cells transfected with CA-MKK showed increased cell adhesion and spreading compared to the results seen with WT-MKK controls (Figs. 8A and B). Little adherence was observed upon transfection with GATA1, Ets1, or MafB alone (Fig. 8C, D, and F). But the number of adherent cells increased after cotransfecting GATA1 plus Ets1 and increased further with GATA1 plus Ets1 plus MafB (Fig. 8G and H). Interestingly, adherence with cells transfected with Fli1 alone was higher than the results seen with Ets1 and increased further in coexpression with GATA1, although the combination of Fli1 plus GATA1 did not increase GPIIb promoter activation compared to the results seen with GATA1 alone (Fig. 8E and I). MafB showed a small enhancement increase over the GATA1-plus-Fli1 combination (Fig. 8J). Quantitation of adherent cells highlighted the synergy between MafB and GATA1/Ets1 with respect to cell adherence (Fig. 8K). Thus, although cell adhesion and spreading are regulated by mechanisms distinct from GPIIb induction they nevertheless can be controlled by MafB coordinately with GATA and Ets.

FIG. 8.

MafB enhances cell adhesion and spreading induced by GATA and Ets. Phase-contrast images of cells transfected for 48 h with WT-MKK1/2 (A), CA-MKK1/2 (B), GATA1 (C), Ets1 (D), Fli1 (E), MafB (F), GATA1 plus Ets1 (G), GATA1 plus Ets1 plus Fli1 (H), GATA1 plus Fli1 (I), and GATA1 plus Fli1 plus MafB (J). Transfections were carried out with 30 μg of each MKK plasmid and 20 μg of each nuclear factor. Similar results were obtained in three independent experiments. (K) Cells were plated in 96-well plates, and all adherent cells were counted. Values represent averages and standard deviations of triplicate platings (normalized to CA-MKK levels). Cells transfected with WT-MKK, GATA1, and MafB showed no adherent cells.

DISCUSSION

Our investigation shows that ERK regulates the GPIIb promoter by combinatorial signaling through multiple response elements and transcription factors. We show that GPIIb transcription is primarily regulated by ERK in the proximal promoter region via combinatorial protein interactions between GATA and Ets. Significantly, ERK induces MafB/Kreisler, a transcription factor known to repress erythroid differentiation (51). MafB in turn synergizes with GATA plus Ets to enhance GPIIb transcription. The requirement for MafB was confirmed using dominant-negative and antisense suppression. Such behavior establishes a novel transcriptional role for MafB/Kreisler and a novel mechanism for ERK-dependent gene expression.

The involvement of GATA in ERK-regulated GPIIb induction is consistent with the importance of this transcription factor family in hematopoietic cell development. GATA1 regulates transcription of several megakaryocyte genes (including GPIIb), and mice deficient in GATA1 show decreased megakaryocyte proliferation and platelet numbers (35, 50). Expression of GATA1 in erythroleukemia cell lines induces megakaryocyte differentiation and is needed for Ras-dependent megakaryocyte maturation (36, 56). The selective expression of GATA1 and GATA2 in blood cells may confer cell specificity of promoter activation, because no ERK-dependent induction of the GPIIb proximal promoter was observed in MCF10A breast epithelial cells or IMR90 fibroblasts (J. R. Sevinsky and N. G. Ahn, unpublished data). GATA1 mRNA levels were unchanged in response to the presence of ERK; however, GATA1 and GATA2 are substrates for direct phosphorylation by ERK (8, 53), suggesting that posttranslational regulation might also be relevant to the requirement for sustained signaling.

Ets family members have previously been implicated in megakaryopoiesis. Ets1, Fli1, and PU.1 are expressed in megakaryocytes, and overexpression of Fli1 and PU.1 leads to increased GPIIb expression in UT7-Mpl and K562 cells, respectively (2, 11). Our results demonstrate that both Ets1 and Fli1 are targets of mRNA induction in response to ERK signaling. However, we found that responses to the presence of Fli1 within the proximal promoter were minimal. Thus, although Fli1 is a candidate for GPIIb promoter regulation it appears to regulate expression outside the ERK-responsive promoter region. In addition, EMSAs have shown that the proximal Ets site binds Ets1 but not PU.1 (11). The available evidence indicates selectivity in factor binding to the proximal promoter and suggests that Ets1/2 might be most important in mediating the ERK response.

Common features of several megakaryocyte gene promoters are composite GATA/Ets cis-acting sequences, which confer Tpo-inducible gene expression. Examples include GPIIb, subunits GP1bα, GP1bβ, GPV, and GPIX of the GP1B-V-IX vWF receptor complex, platelet factor 4, integrin α2/CD49b, and c-Mpl (9, 12, 17, 19, 25, 28, 39, 40, 58, 60). Some of these genes, including GPIIb, GPIX, and α₂(CD49b), are also responsive to ERK (12, 57, 60). This suggests that combinatorial interactions of GATA and Ets with composite promoter elements might represent a common mechanism by which ERK regulates several megakaryocyte markers.

An unexpected result from our study was the involvement of MafB/Kreisler in ERK-dependent gene expression. The Maf family of transcription factors includes large Maf proteins (c-Maf, MafB, Nrl, and L-Maf), which contain an N-terminal acidic domain fused to a bZIP DNA binding domain, and small Maf proteins (MafK/p18, MafF, MafG, and hMaf), which contain only the DNA binding domain (reviewed in references 4 and 33). Some members include known regulators of erythroid and megakaryocyte differentiation. MafK/p18 forms a complex with p45/NF-E2, a factor required for erythroid and megakaryocyte development. Both overexpression of MafK and knockout of MafG in mice yielded a megakaryocyte-deficient phenotype, suggesting that MafK/NF-E2 confers both positive and negative regulation of platelet development (41).

MafB was first discovered as a protein that interacted with Ets1 in a yeast one-hybrid screen. It was shown to interact directly with the Ets1 and repress transactivation of erythroid-specific markers, transferrin receptor, α- and β-hemoglobin, and porphobilinogen deaminase (51). In adult organisms, MafB is expressed primarily in myeloid lineages and promotes monocyte differentiation, leading to the proposal that MafB primarily regulates the monocyte-macrophage differentiation program (23). During vertebrate development, MafB regulates hindbrain segmentation by controlling the expression of Hoxa3 and Hoxb3 (31, 32). MafB synergizes with Krox20/Egr-2 in Hoxb3 transcription, and promoter analysis shows that the synergy involves binding interactions of both proteins with overlapping binding sites. A nearby Ets binding site implicates an Ets-related factor in Hoxb3 regulation. Thus, MafB coregulates gene transcription through interactions with Ets1 and Krox20/Egr-2 nuclear factors.

Our results indicate that MafB expression can be upregulated in response to ERK signal transduction and can thus facilitate megakaryocyte differentiation. Importantly, the synergistic interaction between MafB, GATA1, and Ets1 in enhancing transcription (combined with the negative effects of mutating the consensus GATA and Ets binding sites) suggest that MafB does not independently activate the GPIIb promoter but requires GATA and Ets factor recruitment to the proximal promoter. The precedent for MafB/Ets1 binding interactions provides a reasonable hypothesis to explain this functional behavior. As MafB/Ets1 binding involves the DNA binding domains of both factors (51), however, direct DNA interactions with MafB may be precluded. We note that the GPIIb proximal promoter lacks a consensus Maf binding sequence (4) and speculate that MafB interacts with the promoter via Ets family members. Similarly, Maf regulation of Ets1 has also been demonstrated using a multimerized Ets response element in a manner that does not require a Maf consensus sequence (51).

We conclude that ERK regulates GPIIb expression in part by promoting MafB expression. Various mechanisms could be envisioned. So far, the results of attempts to identify a DNA element for direct MafB binding within the GPIIb proximal promoter have been negative; we have also been unable to demonstrate direct binding of MafB to the GPIIb promoter by oligonucleotide pulldown or EMSA analyses. Thus, MafB might indirectly contribute to GATA/Ets-dependent GPIIb transcription by controlling synthesis of other nuclear factors that directly cooperate with GATA/Ets. Alternatively, the synergy between MafB, GATA, and Ets in promoter regulation may involve a more direct mechanism (either through direct binding of MafB to an as-yet-unidentified promoter element or through indirect binding of MafB via higher-order protein complexes on the GPIIb promoter). The latter hypothesis is consistent with previous reports of functional MafB/Ets1 interactions at specific promoter elements.

In any event, the mechanisms by which MafB contributes to K562 differentiation are not limited to transcriptional control of GPIIb (given that characteristic morphological changes induced by GATA1 plus Ets1 are also further elevated by MafB). Cell adhesion and spreading onto tissue culture plates have been shown not to require GPIIb/IIIa and more likely involve upregulation of vitronectin receptor subunits (S. Santoro, personal communication). This finding suggests that multiple genes important for megakaryocyte function and differentiation might be controlled by ERK through MafB upregulation and combinatorial control of GATA/Ets composite elements.