Multiple ETS Family Proteins Regulate PF4 Gene Expression by Binding to the Same ETS Binding Site

In previous studies on the mechanism underlying megakaryocyte-specific gene expression, several ETS motifs were found in each megakaryocyte-specific gene promoter. Although these studies suggested that several ETS family proteins regulate megakaryocyte-specific gene expression, only a few ETS family proteins have been identified. Platelet factor 4 (PF4) is a megakaryocyte-specific gene and its promoter includes multiple ETS motifs. We had previously shown that ETS-1 binds to an ETS motif in the PF4 promoter. However, the functions of the other ETS motifs are still unclear. The goal of this study was to investigate a novel functional ETS motif in the PF4 promoter and identify proteins binding to the motif. In electrophoretic mobility shift assays and a chromatin immunoprecipitation assay, FLI-1, ELF-1, and GABP bound to the −51 ETS site. Expression of FLI-1, ELF-1, and GABP activated the PF4 promoter in HepG2 cells. Mutation of a −51 ETS site attenuated FLI-1-, ELF-1-, and GABP-mediated transactivation of the promoter. siRNA analysis demonstrated that FLI-1, ELF-1, and GABP regulate PF4 gene expression in HEL cells. Among these three proteins, only FLI-1 synergistically activated the promoter with GATA-1. In addition, only FLI-1 expression was increased during megakaryocytic differentiation. Finally, the importance of the −51 ETS site for the activation of the PF4 promoter during physiological megakaryocytic differentiation was confirmed by a novel reporter gene assay using in vitro ES cell differentiation system. Together, these data suggest that FLI-1, ELF-1, and GABP regulate PF4 gene expression through the −51 ETS site in megakaryocytes and implicate the differentiation stage-specific regulation of PF4 gene expression by multiple ETS factors.


Introduction
Megakaryocytic differentiation is accompanied by drastic morphological changes that are induced by endomitosis and proplatelet formation. To understand the molecular mechanism of megakaryocyte-specific gene regulation during this unique differentiation process, several megakaryocyte-specific gene promoters, including the promoters of platelet factor 4 (PF4), c-Mpl, Glycoprotein (GP) IIb, GPV, GPIX, GPVI, GPIb and platelet basic protein (PBP), have been studied (reviewed in [1]). Each of these promoters includes multiple GATA and ETS motifs.
GATA-1 is shown to bind to the GATA motifs in most megakaryocyte-specific gene promoters. GATA-1 is one of the zinc finger transcription factors; it recognizes the T/AGATAA/G motif and promotes megakaryocytic and erythroid development. GATA-1 binds to GATA motifs in the megakaryocytic gene promoters and activates the gene expression (reviewed in [1]). On the other hand, multiple ETS family transcription factors, such as FLI-1 and PU.1, are known to bind to ETS motifs in each megakaryocyte-specific gene promoter. The members of the ETS family of transcription factors share an evolutionarily conserved DNA-binding domain of 85 amino acids with a winged-helix-turnhelix configuration [2]. ETS factors bind to GGAA/T core sequences. FLI-1 is known to be a positive regulator of megakaryocyte-specific gene expression. FLI-1 regulates several megakaryocyte-specific genes, including GPIIb, GPVI, GPIX and c-Mpl. FLI-1 interacts with GATA-1 and enhances its binding to DNA [3]. Homozygous FLI-1 knockout is embryonic lethal in mice and shows severe dysmegakaryopoiesis. PU.1 is known to be a transcriptional activator of the GPIIb and PBP genes [4,5]. However, PU.1 is also reported to interact with GATA-1 and inhibit GATA-1 function through several possible mechanisms [6,7,8,9,10,11]. This indicates that PU.1 may function as a negative regulator of megakaryocyte-specific gene expression.
To investigate the mechanism underlying megakaryocytespecific gene expression, we have been studying the regulatory mechanism of the PF4 gene expression by using the rat PF4 promoter. We demonstrated that several transcription factors (e.g. GATA-1, ETS-1, MEIS1, PBX1/2, PREP1 and USF1/2) bind to the proximal promoter region and activate the PF4 gene expression [12,13,14,15]. ETS-1 has been shown to activate the PF4 promoter through the 273 ETS site. However, the functions of several other ETS motifs in the PF4 promoter are still unknown. In the present study, we identified the 251 ETS site as a novel functional site in the promoter. The importance of the 251 ETS site for the activation of PF4 gene expression in physiological megakaryocytic differentiation was demonstrated by a novel ES cell differentiation system. Based on EMSA and coexpression assays, 3 ETS family proteins, FLI-1, ELF-1, and GABP were shown to bind to the 251 ETS site and regulate the PF4 promoter. Analysis of expression patterns of FLI-1, ELF-1, and GABP during megakaryocytic differentiation suggested the differentiation stage-specific function of ETS factors. Thus, we succeeded in identifying multiple ETS family proteins that regulate megakaryocyte-specific PF4 gene expression through the novel ETS site.

Preparation of plasmids and targeting vectors
The Rat PF4 promoter (PF4-luc) and human PF4 (hPF4-luc) reporter constructs were previously described [12,14]. To generate the PF4 promoter deletion constructs (Del-500, Del-300, and Del-100), rat PF4 promoter upstream fragments were amplified by polymerase chain reaction (PCR) using 3 different primer sets (DNA sequences are shown in Table S1). The resulting fragments were phosphorylated by T4 DNA kinase and cloned into the SmaI site of PGV-B (TOYO B-Net, Tokyo, Japan). All DNA sequences were verified by automated DNA sequencing.
To generate the two PF4 promoter mutants containing a single mutation in 273 or 251 ETS site (273 ETSmut and 251 ETSmut), an intermediate plasmid containing mutations in both sites (273/251 ETSmut) was prepared by two-step PCR amplification. In the first step, the rat PF4 promoter fragment was amplified by PCR using PF4-luc as a template and primers for the first PCR. In the second step, PCR was performed using primers (the first PCR product and a new primer) and PF4-luc as a template. The resulting 0.2-kb promoter fragment containing the 273 and 251 ETS mutations was digested with KpnI and ligated with the 6.4-kb fragment of PvuII-KpnI digested PF4-luc. To generate the 251 ETSmut, 273/251 ETSmut was digested with SmaI and the 0.7-kb fragment containing the 251 ETS mutation was ligated with the 5.9-kb fragment of SmaI-digested PF4-luc. To generate the 273 ETSmut, PF4-luc was digested with SmaI and the 0.7-kb fragment was ligated with the 5.9-kb fragment of SmaIdigested 273/251 ETSmut. The DNA sequences of all the constructs were verified by automated DNA sequencing. All the primer sequences are shown in Table S1.
To generate FLI-1, PU.1, and GATA-1 expression vectors, cDNAs for FLI-1, PU.1, and GATA-1 were PCR amplified using HEL cDNA and gene-specific primers (shown in Table S1). The resulting PCR fragments were phosphorylated with T4 DNA kinase and cloned into the EcoRV site of pcDNA3 (Invitrogen, Carlsbad, CA). DNA sequences were verified by automated DNA sequencing. To generate the ELF-1 expression vector (pcDNA3-ELF-1), pCI-ELF-1 (a generous gift from Peter Oettgen) was digested with KpnI and XbaI, and the resulting ELF-1 fragment was cloned into pcDNA3. Preparation of ETS-1 and GABP expression vectors was described previously [12,16].
To generate the Hprt-targeting vector with the rat PF4 promoter (pMP8II-PF4-AcGFP), a 1.1-kb region of the PF4 promoter was amplified by PCR using PF4-luc as a template. The resulting promoter fragment was digested with XhoI and SacII, and cloned into the XhoI-SacII site of pAcGFP1-1 (Takara, Siga, Japan) to generate PF4-AcGFP. The promoter sequence was verified by automated DNA sequencing. PF4-AcGFP was then digested with AflII and blunt-ended with T4 DNA polymerase. The resulting DNA was digested with MluI. A 2.1-kb fragment containing PF4 promoter coupled to GFP was purified and cloned between the MluI and PmeI sites of pMP8II [17]. To generate the targeting vector with a mutation of the 251 ETS site (pMP8II-PF4 (251 ETSmut)-AcGFP), a 170 bp fragments spanning the mutation was PCR amplified using the 251 ETSmut plasmid as a template. The resulting promoter fragment was digested with ScaI and SacII, and cloned between the ScaI and SacII sites of PF4-AcGFP. The promoter sequence in the obtained plasmid (PF4(251ETSmut)-AcGFP) was verified by automated DNA sequencing. PF4(251ETSmut)-AcGFP was then digested with AflII and blunt-ended with T4 DNA polymerase. The resulting DNA was digested with MluI and 2.1-kb fragment was cloned between the MluI and PmeI sites of pMP8II.

Cell culture and transient transfection assays
HEL cells were maintained in RPMI 1640 medium supplemented with 10% fetal bovine serum, 100 IU/ml penicillin and 100 mg/ml streptomycin. HepG2 cells were maintained under the same conditions except that Dulbecco's modified Eagle's medium (DMEM) was used. In transient transfection assays with HepG2 cells, 0.5 mg of PF4-luc or hPF4-luc were transfected into 2610 5 cells with or without 0.5 mg of each expression vector, using the Lipofectamine 2000 reagent (GIBCO BRL, Gaithersburg, MD). To control for transfection efficiency, the cells were also transfected with 0.5 mg of pbactin-lacZ. In the assay with HEL cells, 7 mg each of the reporter plasmids and pbactin-lacZ were transfected into 1610 7 cells by electroporation as described previously [18]. Transfected cells were cultured for 48 hr and assayed for luciferase and b-galactosidase activity. Each assay was performed in duplicate more than three times.

Chromatin immunoprecipitation (ChIP) assay
ChIP assays were performed using a ChIP-IT Express kit (Active Motif, Carlsbad, CA). Briefly, crosslinked genomic DNA was prepared from HEL cells and sheared by sonication using the Digital sonifer model 250 (Branson, Danbury, CT). Resulting DNA-protein complexes were immunoprecipitated using 3 mg of antibodies against FLI-1, ELF-1, GABPa, or control IgG. The precipitated DNA fragments were analyzed by real-time PCR using primers (sequences are shown in Table S1) to amplify the promoter region including the 251 ETS sites or GAPDH locus as a control.
siRNA -mediated knockdown of FLI-1, ELF-1, and GABPa HEL cells (1610 7 cells) were electroporated with 150 to 300 pmol of siRNAs for FLI-1, ELF-1, GABPa or control (sequences are shown in Table S1) as described previously [19]. Cells were incubated for 2 days and then processed for total RNA isolation using RNeasy plus mini kit (Qiagen, Germantown, MD). Each cDNA was synthesized from 0.5 mg of total RNA using SuperscriptIII first-strand synthesis system (Invitrogen) and used for real-time RT-PCR analyses. Expression levels of PF4, FLI-1, ELF-1, and GABPa were normalized by a GAPDH expression level. siRNA target and primer sequences are shown in Table S1.

RT-PCR using cDNA from human megakaryocytes or cell lines
Buffy coat peripheral blood (PB) cells were obtained from volunteer blood donors. PB was collected after written consent that was obtained from the volunteer blood donors. All the protocols were approved by the ethic committees in Japanese Red Cross Osaka Blood Center and Osaka University. AC133 + cells were isolated from PB mononuclear cells by a MACS AC133 Cell Isolation Kit (Miltenyi Biotech, Auburn, CA), as described previously [20]. Purified AC133 + cells were cultured in IMDM containing 20% human serum, 50 IU/ml penicillin, 50 mg/ml streptomycin, and 10 ng/ml thrombopoietin (TPO) (PeproTech Inc., Rocky Hill, NJ). The same amount of TPO was added to the medium every two days, and half of the medium was replaced with new medium after 6 days of incubation. Cells were incubated for 0, 4, 6, or 10 days and then processed for total RNA using ISOGEN (Nippon gene, Tokyo, Japan). Each cDNA was synthesized from 0.5 to 1 mg of total RNA using ReveTra Ace (Toyobo, Osaka, Japan) and used for real-time RT-PCR analyses. The same procedure was used for the preparation of cDNA from HepG2 and HEL cells. All the primer sequences are shown in Table S1.

Generation and differentiation of HPRT-targeted ES cells
Undifferentiated Hprt-deficient BK4 ES cells (a generous gift from Sarah Bronson) were maintained and propagated on mouse embryonic fibroblast cells in Knockout DMEM (Invitrogen) supplemented with 15% FBS, 1% Glutamax-I (Invitrogen), 100 IU/ml penicillin, 100 mg/ml streptomycin, 0.1 mM MEM non-essential amino acids (Invitrogen), 0.1 mM 2-mercaptoethanol, and 1000 U/ml ESGRO (LIF) (Millipore, Billerica, MA). To generate homologous recombinants, ES cells were electroporated with 25 mg of SalI-linearized targeting vectors (pMP8II-PF4-AcGFP and pMP8II-PF4(251 ETSmut)-AcGFP). Homologous recombinants were selected in the medium containing HAT (SIGMA, St. Louis, MO) for approximately 10 days, at which time individual colonies were picked for expansion and verification of the desired recombination. At least 2 independent ES cell lines were established for each construct. Established ES cell lines were differentiated into megakaryocytes on OP9 stroma cells by culturing for 8 to 10 days in the medium containing 10 ng/ml TPO according to the method as described previously [21].

Flow cytometry
Two ES cell lines containing the PF4 promoter, with or without a mutation in the 251 ETS site, and a non-recombined control ES cell line were differentiated into megakaryocytic cells for 8 days. The resulting cells were treated with 0.25% Trypsine and collected. The cells (4610 5 cells) were then incubated with phycoerythrin-conjugated rat anti-mouse CD41 antibody (BD pharmingen, Franklin Lakes, NJ) or phycoerythrin-conjugated control rat IgG (BD pharmingen) in phosphate-buffered saline containing 0.5% BSA at 4uC for 30 min. The cells were washed 3 times and analyzed by a flow cytometer.

DNA sequences for the PF4 promoter
The DNA sequences of the rat, mouse, and human PF4 promoters are obtained from Genbank. Accession numbers are M83070.1, NW_001030791.1, and NT_022778.16 for the rat, mouse, and human promoters, respectively.

Statistical analyses
The statistical significance of differences of the means was determined by Student t test.
To determine the temporal expression of these genes, we employed a human megakaryocyte differentiation assay. Hematopoietic progenitor (AC133 + ) cells from human peripheral blood cells were differentiated into megakaryocytes by adding thrombopoietin (TPO). The cells were processed 0, 4, 6, and 10 days later for RNA isolation and assayed by real-time PCR for expression of PF4 and the relevant transcription factors ( Figure 1). PF4 demonstrated a time-dependent increase in expression, indicating that AC133 + cells were differentiating into the megakaryocytic lineage. A similar pattern was observed for NF-E2p45, a transcription factor that has been shown to play an important role during the late stages of megakaryocytic differentiation (reviewed in [22]). The expressions of ETS-1, PU.1, and FLI-1 preceded that of PF4. Detectable expression of the megakaryocytic transcription factor, GATA-1, coincided with that of PF4. These results are consistent with a role for FLI-1 and PU.1 in regulating PF4 gene expression.
To determine whether PU.1 or FLI-1 mediate PF4 promoter activity, we employed co-expression assays using HepG2 cells. We chose these cells because it lacks detectable expression of ETS-1, PU.1, FLI-1, and GATA-1 (Figure 2A). Overexpression of ETS-1 or PU.1 resulted in a small induction of the rat PF4 promoter activity, while FLI-1 strongly activated the PF4 promoter ( Figure 2B). Overexpression of GATA-1 with FLI-1 but not with ETS-1 or PU.1 resulted in synergistic activation of the PF4 promoter ( Figure 2B; Figure S1). A similar pattern was observed with the human PF4 promoter ( Figure 2C). These results suggest a positive role for FLI-1 in regulating PF4 gene expression.

FLI-1 activates the PF4 promoter activity through the 251 ETS site
In order to identify the FLI-1 binding site that regulates PF4 promoter activity, we cotransfected HepG2 cells with progressively shortened 59-deletion fragments of the PF4 promoter. As shown in Figure 3A, deletion constructs containing 500, 300, and 100 bp of the upstream promoter region (Del-500, Del-300, and Del-100, respectively) were all activated by FLI-1. These findings suggest that the FLI-1-responsive DNA element is contained within 100 bp of the transcriptional start site. A comparative analysis of this region revealed 2 highly conserved ETS consensus elements at   Figure 3B). To determine whether 1 or both sites were responsible for mediating the positive effect of FLI-1 on PF4 promoter activity, we co-transfected HepG2 cells with a FLI-1 expression vector and a wild type or a mutant PF4 promoterreporter plasmid containing a mutation in the 273 or 251 ETS sites (273 ETSmut and 251 ETSmut, respectively). Mutation of the 251 ETS site resulted in a significant reduction in FLI-1mediated promoter activation, while mutation of the 273 ETS had little effect ( Figure 3C). Consistent with this result, the mutation of the 251 ETS site also resulted in a significant reduction in synergistic promoter activation mediated by FLI-1 and GATA-1, while the mutation of the 273 ETS had no effect ( Figure 3C). A similar result was obtained with an assay using Del-100, with or without a mutation in the 273 or 251 ETS site ( Figure S2). These results indicated that the 251 ETS site is a FLI-1 responsive element and important for synergistic promoter activation by FLI-1 and GATA-1. A mutation of the 251 ETS site resulted in a significant decrease of the promoter activity in the PF4-expressing megakaryocytic HEL cell ( Figure 3D). To determine whether FLI-1 plays a role in mediating the expression of the endogenous PF4 gene, siRNA against FLI-1 was transfected into HEL cells. FLI-1 siRNA resulted in 67% and 56% reduction of FLI-1 and PF4 mRNA expression, respectively ( Figure 3E). Together, these data suggest that FLI-1 regulates PF4 gene expression through the 251 ETS site.

ETS transcription factors bind to the 251 ETS site
To determine which ETS factors bind to the 251 ETS site in megakaryocytic cells, electrophoretic mobility shift assay (EMSA) was performed using a probe spanning the 251 ETS site and nuclear extract from HEL cells. Four shifted bands were observed ( Figure 4A; Band I, II, IIIa, and IIIb, respectively). Band IIIa and IIIb were detected almost at the same position. Band I, II, and IIIa were inhibited by the addition of a cold wild type competitor probe, but not a mutant competitor probe ( Figure 4A; lanes 3-6). Band IIIb was inhibited by the addition of both wild type and mutant competitor probes, suggesting that band IIIb is not derived from a protein which specifically binds to the 251 ETS site. Incubation of the reaction mixture with anti-FLI-1 antibody, but not control antibody resulted in a supershifted band ( Figure 4A; lanes 7 and 8). Consistent with these results, EMSA using in vitro translated FLI-1 indicated that FLI-1 binds to the 251 ETS site and that a FLI-1-DNA complex is observed to have a size similar to that of band IIIa ( Figure 4B).
The above results suggested that ETS factors other than FLI-1 were involved in the DNA-protein complexes corresponding to band I and II. To identify these factors, we performed supershift assays with antibodies against other ETS proteins. Band I was inhibited by addition of anti-ELF-1 antibody, while band II was inhibited by anti-GABP antibody ( Figure 4C; lanes 3-5). Control IgG had no effect on the DNA-protein complexes ( Figure 4C; lane 7). EMSA was performed with in vitro translated ELF-1 and GABP. GABP binds as a complex consisting of heterodimers of GABPa and GABPb or c (reviewed in [23]). The recombinant ELF-1 resulted in a DNA-protein complex that migrated at the same position as band I ( Figure 4D; lane 4). Anti-ELF-1 antibody inhibited the complex and resulted in a weak super-shift band ( Figure 4D; lane 5). The recombinant GABPa together with GABPb and/or GABPc resulted in DNA-protein complexes that corresponded in size to band II ( Figure 4E). FLI-1, ELF-1, and GABP (containing recombinant GABPa, GABPb and GABPc) did not bind to a probe containing a mutation of the 251 ETS site, and this multiple protein binding of 3 ETS family proteins was not observed with the 273 ETS site ( Figure 4F; Figure S3). Consistent with the EMSA, a chromatin immunoprecipitation (ChIP) assay using HEL cells also demonstrated the bindings of FLI-1, ELF-1, and GABP to the 251 ETS site ( Figure 4G). Together, these findings suggest that FLI-1, ELF-1, and GABP bind to the 251 ETS site in the PF4 promoter.

ELF-1 and GABP regulate the PF4 gene expression through the 251 ETS site
To determine the role of ELF-1 and GABP in mediating PF4 promoter activity, coexpression assays were performed. Overexpression of ELF-1 or GABP (GABPa, b, and c) resulted in strong activation of the wild-type promoter, but not the 251 ETS mutant promoter in HepG2 cells ( Figure 5A). The effect of GABP on promoter activation was recapitulated by combined expression of either GABPa/b or GABPa/c ( Figure 5B). To determine whether ELF-1 and/or GABP functionally interact with GATA-1 to mediate promoter activity, co-transfection assays in HepG2 cells were performed. Unlike FLI-1, ELF-1 and GABP did not synergistically activate the PF4 promoter with GATA-1 ( Figure 5C). These results demonstrated that ELF-1 and GABP activate the PF4 promoter through the 251 ETS site while these factors cannot synergistically activate the promoter with GATA-1.
To investigate whether ELF-1 and/or GABP regulate the endogenous PF4 gene expression in megakaryocytic cells, siRNA against ELF-1 or GABP were transfected into HEL cells. ELF-1 siRNA resulted in 62% and 40% reduction in ELF-1 and PF4 mRNA expression ( Figure 5D). On the other hand, GABP siRNA resulted in 32% reduction of GABP and 66% increase of PF4 mRNA expression ( Figure 5E). These results indicated that both ELF-1 and GABP regulates the PF4 gene expression.  lanes 3 and 6). The supershift assay was performed with the antibody against FLI-1 and control antibody (lanes 5 and 6). (C) The supershift assay was performed with antibodies against ELF-1 and GABP or control antibody (lanes 3-7). GABPb/c antibody recognizes both GABPb and its splicing isoform GABPc. (D) EMSA was performed with the in vitro translated ELF-1 and control protein (lanes 3-6). The supershift assay was performed with the antibody against ELF-1 and control antibody (lanes 5 and 6). (E) EMSA was performed with the in vitro translated GABP (GABP a, b, and c) and control protein (lanes 3-13). The supershift assay was performed with the antibody against GABP and control antibody. Shifted bands derived from GABPa, GABPa/b and GABPb/c complexes are labeled a, a/b, and a/c, respectively. (F) EMSA was performed with the 251 ETSmut probe and in vitro translated FLI-1, ELF-1, GABP, and control protein. (G) ChIP assay was performed with HEL cells by using antibodies against ETS factors or control antibody. Immunoprecipitated DNA was measured by real-time PCR with primers to amplify the human PF4 promoter region, including the 251 ETS site, or the GAPDH locus as a negative control region. A representative of 3 independent experiments is shown. doi:10.1371/journal.pone.0024837.g004 Although the result of co-expression assay showed that both ELF-1 and GABP are positive regulators for the PF4 gene expression, siRNA knockdown of GABP led to increased expression of PF4 in HEL cells. We speculated that this controversial effect was observed because the decrease of GABP promoted FLI-1 binding to the 251 ETS sites. In fact, EMSA using recombinant proteins showed competitive binding of FLI-1, ELF-1, or GABP at the 251 ETS site ( Figure S4). Thus, these data suggest that FLI-1, ELF-1, and GABP competitively bind to the 251 ETS sites and regulate PF4 gene expression.

Expression patterns of FLI-1, ELF-1, and GABP in human megakaryocytic differentiation
To investigate and compare the temporal expression patterns of the ETS factors FLI-1, ELF-1, and GABP during megakaryocytic differentiation, mRNA levels of ELF-1 and GABPa were measured by real-time PCR (using the same cDNA prepared in Figure 1). As shown in Figure 1, both ELF-1 and GABPa were expressed during megakaryocytic differentiation. However, unlike the expression pattern of FLI-1 expressions of ELF-1 and GABPa had decreased by day 4. This finding suggests that while all 3 ETS factors may play a role in mediating PF4 expression, they likely function in temporally distinct ways.

The 251 ETS site is important for PF4 promoter activation in differentiating megakaryocytes
To confirm a role for the 251 ETS site in mediating the PF4 gene expression in megakaryocytes, the wild type or the 251 ETSmut PF4 promoter was coupled to GFP. A single copy of the resulting transgenic cassette was targeted to the Hprt locus in ES cells by homologous recombination (Figure S5A; Figure 6A). Two correctly targeted recombinant clones for each construct were differentiated into megakaryocytic lineage by culturing with OP9 stroma cells and TPO. At day 10, a number of differentiated cells were positive for acetylcholine esterase activity ( Figure 6B upper panel), and 30% of the cells were positive for CD41 (data not shown). In addition, proplatelets were also observed ( Figure 6B lower panel), suggesting differentiation into the megakaryocytic lineage. Megakaryocytes and proplatelets from cells carrying the wild type PF4 promoter had strong GFP expression, while those from 251 ETSmut showed significantly decreased GFP expression ( Figure 6C; Figure S5B). FACS analysis demonstrated that the mutation of the 251 ETS site significantly decreased the number of GFP positive megakaryocytic cells ( Figure 6D). These observations suggest that the 251 ETS site is important for the activation of the PF4 promoter during megakaryocytic differentiation.

Discussion
Multiple ETS binding sites are often found in the megakaryocyte-specific gene promoters. However, it has not been clearly shown how multiple ETS binding sites regulate megakaryocytespecific genes. The rat PF4 promoter contains more than 20 ETS core sequences in it. In this study, we identified a novel, functional ETS binding site (251 ETS site) in the PF4 promoter. To evaluate the physiological function of the 251 ETS site for the PF4 promoter activity in megakaryocytes, we developed and used a novel reporter gene assay using homologously recombined mouse ES cells (Figure 6; Figure S5). It has been difficult to analyze the promoter activity in intact megakaryocytes because of the small number of megakaryocytes in vivo and the technical difficulty of transient transfection of plasmids into megakaryocytes and proplatelets. Our system enables the promoter activity in normal differentiating megakaryocytes and proplatelets to be evaluated. Using this system we evaluated the function of the 251 ETS site using 2 independent ES cell lines containing a wild type or mutant PF4 promoter coupled to the GFP gene. The insertion of a singly copy transgene into the Hprt locus allowed to compare the expression levels of the reporter gene between ES cell lines in the same genomic DNA context [24]. By differentiating these ES cell lines into megakaryocytic lineage, we succeeded in demonstrating that the 251 ETS site is important for the activation of PF4 gene expression in megakaryocytes and proplatelets. This reporter assay system should be a powerful tool for analyzing physiological promoter function in normal megakaryocytes and proplatelets.
Our data demonstrate that multiple ETS family proteins regulate the PF4 gene expression. EMSA and coexpression assays indicate that FLI-1, ELF-1, and GABP bind to the 251 ETS site and activate the PF4 promoter. In addition, the ChIP assay showed that FLI-1, ELF-1, and GABPa bind to the 251 ETS site in vivo. ELF-1 was originally described as a regulator of T-cell specific genes [25]. To our knowledge, this is the first study to demonstrate a role for ELF-1 in mediating the megakaryocytespecific gene expression. GABP consists of 2 subunits, GABPa and GABPb (or its splicing isoform, GABPc). GABPa contains the ETS DNA-binding domain, while GABPb is required for nuclear translocation and transactivation. Mammalian GABP is ubiquitously expressed in all tissues and has been implicated in several critical cellular processes such as cellular differentiation, cell cycle, cell survival and mitochondrial respiration [23]. In addition to regulating the expression of housekeeping genes, GABP has been shown to regulate the expression of cell-type specific genes in several distinct lineages, including myeloid cells, lymphocytes, mast cells and endothelial cells [16,23]. A few reports indicate the importance of GABP in the regulation of megakaryocyte-specific genes (e.g., GPIIb and cMpl) [26,27]. Comparing with FLI-1, ELF-1, and GABP, ETS-1 has a small effect on the PF4 gene expression in the co-expression assay using HepG2 cells, although we have previously demonstrated that ETS-1 activates the PF4 promoter [12]. This difference may be caused by the difference in the cell types used in the co-expression assay. ETS-1 is known to inhibit its DNA binding via the autoinhibitory domain, and its DNA binding is enhanced by interacting with other transcription factors, such as CBFb [28]. We speculated that some interacting protein that is expressed in HEL cells but not HepG2 cells is needed for the strong promoter activation by ETS-1.
It has been reported that multiple transcription factors can bind to the same site and regulate gene expression. Recent genomewide ChIP-seq analyses have also shown an overlap in the chromatin regions commonly bound by different members of the ETS family [29,30,31]. In some cases, a transcriptional activator competes with a suppressor for binding to the same site. This mechanism enables positive or negative regulation of promoter activity through the same site. However, in our reporter gene assays, individual overexpression of FLI-1, ELF-1, and GABP resulted in similar levels of PF4 promoter activity. This suggests that all these factors are transcriptional activators of the PF4 promoter. The question is: why is the PF4 promoter regulated by multiple transcriptional activators through the same site? In the coexpression assay, FLI-1 but not ELF-1 and GABP synergistically activated the PF4 promoter with GATA-1 ( Figure 5C). One potential mechanism to explain this synergistic activation may be a direct interaction between FLI-1 and GATA-1 on the GATA-1 binding site, because FLI-1 expression enhanced promoter activation by GATA-1 without binding to the 251 ETS site ( Figure 3C; Figure S2). Alternatively, GATA-1 may enhance the FLI-1 binding to the other ETS site, such as the 271 ETS site. In either case, this synergistic activation by FLI-1 and GATA-1 suggests distinct roles for the ETS family proteins in mediating gene expression by interacting with other transcription factors. The siRNA knockdown of GABPa slightly increased the PF4 gene expression although the overexpression of GABP itself activated the promoter activity in the co-expression assay ( Figure 5A, E). This controversial observation also suggests the possibility that FLI-1, ELF-1, and GABP competitively bind to the 251 ETS site, and that the relative expression levels of these 3 factors are important for the regulation of the PF4 gene expression. Our realtime PCR analysis indicated that the expression of FLI-1 increased during megakaryocytic differentiation, whereas the expression of ELF-1 and GABP decreased at day 4 and did not drastically change at days 6 and day 10 ( Figure 1). This suggests the possibility that FLI-1 plays a role in the late stage of megakaryocytic differentiation, whereas ELF-1 and GABP are necessary during the early stage. Consistent with this hypothesis, the importance of the FLI-1 at the late stage of megakaryocytic differentiation and gene expression was demonstrated by inducible FLI-1 knock out and FLI-1 mutant mice [32,33]. Furthermore, Pang et al. reported that GABP regulates megakaryocyte-specific genes that are expressed at the early differentiation stage, while FLI-1 regulates genes expressed at the late stage [27]. In the regulation of the late megakaryocytic marker PF4 either GABP or ELF-1 may bind to the 251 ETS site and moderately activate the promoter at the early stage, whereas FLI-1 may displace them and strongly activate the promoter with GATA-1 at the late stage. Our EMSA data indicate the possibility that FLI-1 displaces ELF-1 and GABP from the 251 ETS site ( Figure S4). Together, these results implicate the differentiation stage-specific regulation of PF4 gene expression by multiple ETS factors.
Inconsistent with our result, Pang's group demonstrated that GABP synergistically activates the GPIIb promoter with GATA-1 under the existence of FOG-1. This suggests the possibilities that FOG-1 is essential for synergistic activation of the PF4 promoter by GABP and GATA-1 and that synergistic activation by GABP and GATA-1 does not occur on the PF4 promoter. More detailed analyses about ETS family proteins and their functionally associated proteins will be important to understand megakaryocyte-specific and differentiation stage-specific PF4 gene expression.  Table S1. (TIF) Figure S4 Competitive binding of FLI-1, and ELF-1 or GABP to the 251 ETS site. (A) EMSA was performed with the 251 ETS probe, and in the protein mixture containing ELF-1 and various amounts of FLI-1. The arrows indicate the shifted bands derived from ELF-1 (E) and FLI-1 (F). (B) EMSA was performed with the 251 ETS probe, and the protein mixture containing GABPa and various amounts of FLI-1. The arrows indicate the shifted bands derived from GABPa (G) and FLI-1 (F). (TIF) Figure S5 Promoter activity of the wild type or mutant PF4 promoter during megakaryocytic differentiation. (A) The transgene was inserted into the Hprt locus by homologous recombination. (B) Two ES cell lines with transgenes containing the PF4 promoter with or without a mutation in the 251 ETS site were differentiated into megakaryocytic lineage. The GFP expression level was compared between the 2 cell lines from days 6 to 12. (TIF)