Properties of Ets-1 Binding to Chromatin and Its Effect on Platelet Factor 4 Gene Expression

Abstract

Ets-1 is important for transcriptional regulation in several hematopoietic lineages, including megakaryocytes. Some transcription factors bind to naked DNA and chromatin with different affinities, while others do not. In the present study we used the megakaryocyte-specific promoters platelet factor 4 (PF4), and glycoprotein IIb (GPIIb) as model systems to explore the properties of Ets-1 binding to chromatin. Chromatin immunoprecipitation assays indicated that Ets-1 binds to proximal regions in the PF4 and GPIIb promoters in vivo. In vitro and in vivo experiments showed that Ets-1 binding to chromatin on lineage-specific promoters does not require lineage-specific factors. Moreover, this binding shows the same order of affinity as the binding to naked DNA and does not require ATP-dependent or Sarkosyl-sensitive factors. The effect of Ets-1 binding on promoter activity was examined using the PF4 promoter as a model. We identified a novel Ets-1 site (at −50), and a novel Sarkosyl-sensitive DNase I-hypersensitive site generated by Ets-1 binding to chromatin, which significantly affect PF4 promoter activity. Taken together, our results suggest a model by which Ets-1 binds to chromatin without the need for lineage-specific accessory factors, and Ets-1 binding induces changes in chromatin and affects transactivation, which are essential for PF4 promoter activation.

Ets-1 is the founding member of the Ets family of transcription factors, containing a DNA binding domain, a pointed domain, and a transactivation domain (reviewed in references 39 and 54). The DNA binding domain is conserved among the members of the family and is referred to as the ETS domain. As a transcription factor, Ets-1 binds to sequences on the DNA and participates in transcriptional regulation. The consensus sequences for Ets-1 binding contain a common core trinucleotide, GGA. However, the binding specificities for different Ets family members are not well understood (17) and the binding sequences for Ets proteins are often collectively referred to as Ets sites or Ets elements. Although Ets-1 was first discovered as part of a fused oncogene in an avian retrovirus, multiple roles have been discovered for this widely expressed transcription factor, including tumorigenesis, vertebrate embryogenesis, and hematopoiesis (39). In the hematopoietic system, the importance of Ets-1 in the transcriptional regulation of lineage-specific genes has been reported for lymphocytes, eosinophils, and erythrocytes (28, 30, 40).

Megakaryocytes develop from hematopoietic pluripotent stem cells, giving rise to circulating platelets. After lineage commitment, megakaryocytes enter a unique program of differentiation that includes lineage-specific gene expression and an endomitotic cell cycle, resulting in a mature cell with a large volume and a polyploid nucleus (38, 59). Platelets are generated by fragmentation of mature megakaryocytes. As an indispensable part of the differentiation program, megakaryocytes express a group of genes that are unique to the lineage. Most of the lineage-specific gene products are found in platelets and are important for proper platelet function (46). Two examples are platelet factor 4 (PF4), encoding a chemokine, and glycoprotein IIb (GPIIb), a subunit of the fibrinogen receptor. These genes are expressed from the early stages of lineage development (10, 23, 36). The rat PF4 promoter has been cloned and shown to drive lineage-specific expression in a transgenic mouse model (36), indicating that the elements defining lineage specificity are within this DNA segment.

Megakaryocyte-specific promoters have been shown to be positively controlled by Ets elements (5, 12, 22, 24, 29, 37), like genes from other hematopoietic lineages (3, 16, 30, 31, 49). In a search for potential transcription regulators for megakaryocyte differentiation, it was found that Ets-1 is upregulated during megakaryopoiesis but not during erythropoiesis (45). Furthermore, overexpression of Ets-1 upregulates the megakaryocyte-specific promoter of the PF4 gene in HEL cells (29). In the GPIIb promoter, both Ets-1 and Fli-1 (an Ets family member) are involved in gene expression (22, 52). This suggested that Ets family members control lineage-specific gene expression in megakaryocytes and that Ets-1 may control the expression of some of these genes.

Ets-1 is not unique to megakaryocytes; it is expressed in most of the hematopoietic lineages. It is unclear how a relatively ubiquitous factor regulates lineage-specific function. This problem is a recurring theme in mammalian development. In eukaryotic cells, DNA is present in the form of chromatin, which limits the access of many transcription factors and provides an extra level of control for gene transcription. This important aspect has not been fully explored for Ets-1. Hence, while identification of transcription factors that are coactivators of megakaryocyte-specific genes is important, it is also essential to study the binding properties of Ets-1 to chromatin and its relevance to gene expression. In this paper we focus on this latter aspect.

Ets-1 binding to chromatin might require lineage-specific cofactors and/or chromatin-remodeling enzymes. Ets-1, like many transcription factors, is known to interact with other transcription factors and chromatin-modifying enzymes, such as MafB, USF-1, PAX5, and CBP/p300 (4, 13, 16, 25, 40, 41, 57). It is also possible that the binding of Ets-1 depends on the intrinsic affinity of Ets-1 for binding sites in DNA. Under such a model, the concentration of Ets-1 might determine which binding sites are occupied and then which genes are expressed. A similar mechanism has been found for the Ets family member PU.1 (11).

In the present study we show for the first time that Ets-1 binds to endogenous megakaryocyte-specific promoters in cultured cells. Using a cell-free system, we found that chromatin does not inhibit Ets-1 binding to the PF4 and GPIIb promoters. Ets-1 binding is independent of lineage-specific cofactors and of energy-dependent chromatin remodeling. Ectopic expression of Ets-1 in cultured cells is sufficient for binding to lineage-restricted endogenous genes, in the apparent absence of lineage-specific cofactors. Ets-1 binding is not sufficient to activate lineage-specific genes in cultured cells. Our study is the first to demonstrate that the binding of Ets-1 to lineage-specific promoters does not require lineage-specific factors, and that the affinity of Ets-1 to naked DNA and chromatin is of the same order of magnitude. Via these investigations, we also identified in the PF4 promoter a new Ets-1 binding site that is important for PF4 expression and a DNase I-hypersensitive site which is programmed by Ets-1 binding to chromatin and which enhances promoter activity.

MATERIALS AND METHODS

Plasmids.

pPF4GH contains 1.1 kb of rat PF4 promoter sequences linked to the human growth hormone (hGH) reporter gene (37). Mutations of specific Ets sites at −379 (relative to the transcriptional start site), −372, −75, or −50 or combinations of the above were introduced by site-directed mutagenesis through PCR and verified by DNA sequencing as outlined previously (53). PCR products were cloned into the original plasmid through the HindIII and HincII sites. All mutational primers used to generate the mutated constructs were tested for their ability to abrogate protein binding by using an electrophoretic mobility shift assay (EMSA) with Y10/L8057 nuclear extract and Ets-1 recombinant protein (see below). The following primers were used (mutated bases are capitalized, and the original bases follow in parentheses): for pPF4GH-m379,372, gtcgggcaaccgTG(ga)agtcgTT(gg)aaggcaacaa; for pPF4GH-m414, caaatcaacagC(g)aaagcacggc; for pPF4GH-m75, gtggatcacttAT(cc)tcatccccta; for pPF4GH-m50, atcccgggtttcAA(cg)gactgggctggca; and for pPF4GH-m50,75, tggatcacttAT(cc)tcatcccctatcccgggtttcAA(cg)gactgggctg. The pGPIIbGH plasmid contains 912 bp of the rat GPIIb promoter sequence upstream of the transcriptional start site, linked to the hGH reporter gene. The p0GH plasmid has the same backbone as the pPF4GH plasmid but contains only a minimal promoter before the hGH gene. The pGPIIbGH plasmid was kindly provided by Mortimer Poncz (5). The above Ets sites in both the rat PF4 promoter and the rat GPIIb promoter are conserved in murine sequences (based on the GenBank database). The pEVRF-Ets-1 plasmid, containing the murine Ets-1 cDNA, and pET14b-Ets-1, a His₆ Ets-1 recombinant protein expression vector, were kindly provided by Barbara Nikolajczyk (31). pcDNA3, a mammalian expression vector, was purchased from Invitrogen (Carlsbad, Calif.). pEGFP-N1, a green fluorescent protein expression vector, was purchased from Clontech (Palo Alto, Calif.).

Cell culture, transfection, and reporter gene assays. (i) Cell culture.

Y10/L8057, a murine megakaryoblastic cell line (58) derived from L8057 cells (18), was maintained in F12 medium (GIBCO BRL, Rockville, Md.) as previously described (53). For Mpl ligand treatment, Y10/L8057 cells were shifted to Iscove's modified Dulbecco's medium (IMDM) (GIBCO BRL) with 10% fetal bovine serum FBS, 5 U of penicillin per ml, 5 μg of streptomycin per ml, and 2 mM l-glutamine. The medium was supplemented, where indicated, with 25 ng of polyethylene glycol (PEG)-rhuMGDF (a gift from Kirin, Tokyo, Japan, referred to throughout the text as MGDF) per ml 2 h after the medium change. The murine NIH 3T3 cell line was obtained from Stephen Farmer, and cultured as reported previously (14).

(ii) Transfection.

Transient transfections were performed using the FuGENE 6 transfection reagent (Roche, Indianapolis, Ind.) as specified by the manufacturer. Y10/L8057 cells were transferred to IMDM and plated at a density of 4 × 10⁵ cells/ml 1 h before the transfection; 10 μg of plasmid (or mixture of plasmids in cotransfection experiments) and 25 μl of FuGENE6 reagent were used per 10 ml of culture medium. Reporter constructs were cotransfected with pCMV-βGal for transfection efficiency normalization (based on β-galactosidase activity). For the NIH 3T3 cell line, cells were plated the night before and transfections were carried out when the cells were 50% confluent.

Stable transfection of Y10/L8057 cells was carried out as described previously, using electroporation (37). The reporter constructs (pPF4GH and mutants) were linearized with EcoRI and purified by phenol-chloroform extraction and ethanol precipitation. A 50-μg portion of reporter plasmid was cotransfected with 5 μg of pcDNA3, which contains the neomycin resistance gene for selection. Cells were selected with 400 μg of Geneticin (G-418; Invitrogen) per ml until the control cells were all dead.

(iii) Reporter gene assays.

Y10/L8057 cells were harvested 2 days after transfection as described previously (53). Reporter gene production was assayed in the culture medium collected, using an hGH detection kit (Nichols Institute Diagnostics, San Juan Capistrano, Calif.) as specified by the manufacturer. Data were normalized for transfection efficiency based on β-galactosidase activity (53).

For stable transfection experiments, the pool of transfected cells was plated at a density of 2 × 10⁵ cells/ml in IMDM in triplicate and incubated for 24 h before being assayed for hGH production. Data were normalized for equal genetic incorporation by Southern blot analyses (see below).

Western blot analysis.

Western blot analyses were performed as described previously (53). Briefly, cells were collected by centrifugation (for Y10/L8057 cells) or by scraping on ice (for NIH 3T3 cells) after being washed in cold 1× phosphate-buffered saline (PBS, GIBCO BRI). Samples were loaded onto a 10% polyacrylamide gel, including 30 μg of whole-cell extract or the indicated amount of recombinant protein or nuclear extract per lane. After sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), the gel was transferred onto a nitrocellulose filter and detected with Ets-1 antibody (C-20; Santa Cruz Biotech, Santa Cruz, Calif.) using the ECL detection kit (Amersham Pharmacia Biotech, Piscataway, N.J.) and quantified using Digital Science 1D software (Eastman Kodak, New Haven, Conn.). The C-20 anti-Ets-1 antibody is documented to be specific for Ets-1 and does not cross-react with other Ets-1 family members (3, 26).

Chromatin immunoprecipitation assay.

Y10/L8057 cells were cultured in IMDM at a density of 5 × 10⁵ cells/ml. Chromatin immunoprecipitation assays were performed as described previously (55) with modifications. Briefly, cells were fixed with 1% formaldehyde at 4°C for 45 min, and the reaction was stopped by adding glycine to 0.125 M. The fixed cells were washed twice in cold 1 × PBS and resuspended for lysis at room temperature for 10 min. Nuclei were collected by centrifugation at 1,000 × g for 5 min and washed once in 10 mM Tris-HCl (pH 8.0)-0.2 M NaCl-1 mM EDTA-0.5 mM EGTA-1× protease inhibitor cocktail. The nuclei were centrifuged and resuspended in ice-cold sonication buffer (10 mM Tris-HCl [pH 8.0], 1 mM EDTA, 0.5 mM EGTA, 1× protease inhibitor cocktail) and sonicated on a model 550 Fisher Sonic Dismembrator to give genomic fragments with an average size of 500 bp. After sonication, the samples were stored at −80°C in aliquots.

NIH 3T3 cells were transfected with 0, 1, 3.3, or 10 μg of the Ets-1 expression plasmid pEVRF-Ets-1 by using FuGENE 6 transfection reagent (pcDNA3 was added as needed to adjust the DNA mass to 10 μg per transfection). Transfection efficiencies were controlled with a green fluorescent protein expression vector at the same dilutions. The estimated transfection efficiency was about 30% for all the dilutions. Cells were fixed for the ChIP assay 2 days after the transfection. After fixation, they were scraped off the plates in cold 1× PBS. The rest of the procedure was as described for Y10/L8057 cells.

For immunoprecipitation, equal volumes of sonicated chromatin (1 ml) were precleared with 60 μl of protein A/G-agarose beads (Santa Cruz Biotech) on a 4°C rotator for 2 h. The precleared chromatin samples were divided into two tubes and incubated at 4°C overnight with 1 μg of Ets-1 antibody (C-20) or 1 μg of normal rabbit immunoglobulin G (IgG) (Santa Cruz Biotech). Protein A/G-agarose beads (30 μl) were then added and further incubated at 4°C for 2 h with rotation. The beads were washed sequentially for 5 min each at 4°C with low-salt wash buffer (20 mM Tris-HCl [pH 7.6], 150 mM NaCl, 1 mM EDTA, 0.5 mM EGTA, 0.1% Triton X-100, 100 μg of sonicated herring sperm DNA [GIBCO BRL] per ml), medium-salt wash buffer (low-salt wash buffer with 300 mM NaCl), and high-salt wash buffer (low-salt wash buffer with 500 mM NaCl, 1% Triton X-100, and 0.1% SDS). This was followed by three washes with TE pH 7.6 buffer (10 mM Tris-HCl [pH 7.6], 1 mM EDTA). Immunoprecipitation samples and samples from the sonication step (input) were then digested with proteinase K and incubated at 65°C overnight to reverse the cross-linking. DNA fragments were recovered by phenol-chloroform extraction and ethanol precipitation. Equal volumes of immunoprecipitation samples and input samples (diluted 1:500) were analyzed by PCR. Primers that specifically amplify the PF4, GPIIb, and CD5 promoters and immunoglobulin μ heavy-chain enhancer (Ig-μE) region were used in the PCR amplifications, as listed (5′ to 3′, sense and antisense): for the PF4 gene, agatgtggatcctgctgccagctacta and tcccaagcacgttttatcttcactgcc; for the GPIIb gene, gggcttgtggctgagaaacagagacaa and cttcccaaacgtcctaaacaggaatgg; for the GPIIb downstream region, ggtcactaggagtgatagtaag and ccaattaggtggtaacagcttg; for the CD5 gene, ttctgcccagccatgtgaatgg and cagtgtcttactgtcttggagg; and for Ig-μE, gagcaatgttgagttgagtcaa and ggtgttttgctcagcctggact.

The number of cycles for each set of primers was chosen based on the optimal amplification and linear range of the PCR reactions, as determined by a serial dilution of the immunoprecipitation inputs. A total of 28 cycles were used for the PF4 and GPIIb primers; 23 cycles were used for the GPIIb downstream sequence primers; and 28 cycles were used for the CD5 and Ig-μE primers. PCR products were analyzed on a 5% acrylamide gel run with 0.5× TBE buffer and stained with GelStar nucleic acid stain (BioWhittaker Molecular Applications, Baltimore, Md.).

Total RNA preparation and RT-PCR.

Total RNA from either NIH 3T3 cells or Y10/L8057 cells was prepared with Trizol as described previously (6). For reverse transcription (RT), 2 μg of RNA was used in the reaction mixture as specified by the manufacturer for random priming (Superscript II; GIBCO BRL). The product of the reverse transcription reaction was then used in subsequent PCR amplifications with primers corresponding to the PF4 or glyceraldehyde-3-phosphate dehydrogenase (GAPDH) cDNA. For the PF4 gene, the gaaagcgatggagatcttag and gacatttaggcagctgatac mouse primers were used (expected product, 228 bp); for the GAPDH gene, the tcaccatcttccaggag and gcttcaccaccttcttg primers were used (expected product, 568 bp).

Recombinant protein preparation.

Recombinant His-Ets-1-expressing plasmid pET14b-Ets-1 was transformed into HMS174-DE3 competent cells (Novagen, Madison, Wis). A colony was grown until the bacterial concentration reached an optical density of 0.5 at 600 nm. Recombinant protein expression was induced by adding 1 mM isopropyl-β-d-thiogalactopyranoside (IPTG; GIBCO BRL), and bacteria were collected 2 h after the induction. The recombinant His-Ets-1 was purified with His-Bind resin (Novagen) as specified by the manufacturer, with the following modifications. To obtain a concentrated product, elution buffer from the His-Bind column was collected in 0.3-ml fractions. The rEts-1-containing fractions were pooled and dialyzed overnight with three changes of a dialysis buffer containing 50 mM Tris-HCl (pH 7.9), 0.1 M KCl, 12.5 mM MgCl₂, 1 mM EDTA, 10% glycerol, 1 mM benzamidine, and 0.1 mM phenylmethylsulfonyl fluoride. The recombinant protein was stored at −80°C in aliquots. The concentration of the recombinant protein was determined by comparison with a dilution series of bovine serum albumin on a Coomassie blue-stained SDS-PAGE gel.

In vitro chromatin assembly and purification.

The cell-free chromatin assembly was carried out as described previously (34) with some modifications. In a 200-μl assembly reaction mixture, 50 μl of S-190 Drosophila extract was added to 120 μl of buffer R (10 mM potassium HEPES [pH 7.5], 10 mM KCl, 1.5 mM MgCl₂, 0.5 mM EGTA, 10% glycerol, 10 mM β-glycerophosphate, 1 mM dithiothreitol, 1× protease inhibitor cocktail) and incubated for 30 min at room temperature. The reaction mixture was supplemented with 29 μl of Mg-ATP mix (0.21 M creatine phosphate, 21 mM ATP, 0.04 μg of creatine kinase per μl 30 mM MgCl₂), and 250 ng of plasmid DNA. No additional histone proteins were added because sufficient histones are present in the S-190 extract for proper assembly with the amount of plasmid used. The mixture was incubated at 27°C for 3.5 to 4 h. Apyrase (Sigma, St Louis, Mo.) was added to a final concentration of 0.025 U/μl, together with the Mg-ATP mix, before the assembly where indicated. Two batches of apyrase were used, and they yielded similar results. The assembled chromatin was then used in the micrococcal nuclease digestion assay or the DNase I primer extension footprinting assay. The chromatin was subjected to Sarkosyl treatment and purification where indicated. Sarkosyl (Sigma) was added to a final concentration of 0.1% (wt/vol) from a 5% stock and incubated at room temperature for 10 min before being loaded onto a Sephacryl S-500 HR column (Amersham Pharmacia Biotech). The column was washed with three 200-μl steps of the column buffer and eluted with six 200-μl steps of column buffer (17 ml of buffer R as described above, 3 ml of H₂O, 24 μl of 1 M MgCl₂). Fractions (100 μl each) were collected during the elution steps, and the chromatin-containing fractions were identified by running 5 μl of each fraction on a 1% agarose gel. These chromatin-containing fractions were pooled and used in subsequent experiments or stored at 4°C for up to 1 week without any apparent structural change.

Micrococcal nuclease assay and DNase I primer extension footprinting.

The micrococcal nuclease assay and DNase I primer extension footprinting assay were performed by published methods (34), with a few changes. Each chromatin sample was divided into three tubes of 25 μl each for digestion with 2.5 μl of diluted micrococcal nuclease (Sigma). The enzyme dilutions used with nonpurified chromatin were 0.44, 1.3, and 4.0 U/ml, and those used with purified chromatin template were 0.009, 0.027, and 0.083 U/ml. In the experiment investigating the effect of apyrase on chromatin assembly, enzyme dilutions were titrated and modified to ensure that the resulting DNA fragments from the micrococcal nuclease digestions were within the same range as that of the standard assembly reaction. After the digestion was stopped, DNA was recovered by proteinase K digestion, phenol-chloroform extraction, and ethanol precipitation. Samples were run on a 1% agarose gel and visualized by staining with ethidium bromide. For increased sensitivity when needed, the gel was transferred to a nylon filter and probed with an end-labeled oligonucleotide corresponding to the region of interest.

For DNase I primer extension footprinting, chromatin was assembled or purified as described above. In samples requiring apyrase treatment, chromatin was further digested with 0.025 U of apyrase per μl for 10 min. A 15-μl volume of chromatin or 18.75 ng of naked DNA was combined with the indicated amount of recombinant protein or nuclear extract for binding in a final volume of 25 μl. The DNA concentration in each sample is 0.2 nM, which is more than 10-fold lower than the apparent binding affinity (10 to 30 nM) of Ets-1 for the sites studied. Samples were subjected to digestion with 2.5 μl of DNase I for 1 min at room temperature; 0.166 μg/ml was added for the naked DNA control, and 250 μg/ml was added for the chromatin control. The concentrations of DNase I used for other samples were titrated to give the same degree of digestion as the control samples. After digestion, DNA was recovered and dissolved in 5 μl of TE (10 mM Tris-HCl [pH 8.0], 1 mM EDTA) and subjected to primer extension with Vent exo⁻ DNA polymerase (NEB, Beverly, Mass.). The primers used in the primer extension reactions were as follows: for the PF4 −50/−75 region, agctgtccacaggaccctgagtggttcg; for the PF4 −379/−372 and −414 regions, acagtctacagaggtagaaacagttgga; for the GPIIb −35 region, tcatcagcacctatccccgggacagacc. Samples from the primer extension reactions were precipitated with ethanol and analyzed on an 8% sequencing gel.

Nucleosome ladder disruption assay, nucleosome positioning assay, and DNase I hypersensitivity assay.

The nucleosome ladder disruption assay, nucleosome positioning assay, and DNase I hypersensitivity assay were performed as described previously (34). For the nucleosome ladder disruption assay, chromatin was incubated with or without recombinant Ets-1 for 30 min. Naked DNA or chromatin samples were digested with serial dilutions of micrococcal nuclease (as described for the micrococcal nuclease assay [see above]). DNA fragments were recovered by phenol-chloroform extraction and ethanol precipitation and resolved on a 1% agarose gel. The gel was transferred in 10× SSC (1× SSC in 0.15 M NaCl plus 0.015 M sodium citrate) onto a nylon filter, which was probed with oligonucleotides corresponding to the PF4 −50 region, the PF4 −379 region, or a region in the vector backbone (−1092). The primers used were atcccgggtttccggactgggctggcagt (−50 region), ggtcgggcaaccggaagtcgggaaggca (−379 region), and ctgcgaatgctgaaatctttggtgaagg (−1092 region).

Nucleosome positioning was determined similarly to the nucleosome ladder disruption assay, except that recovered DNA fragments from micrococcal nuclease digestion were further digested completely with NdeI, which recognizes a single site 1.1 kb upstream of the transcription start site. After the DNA had been transferred to nylon, the filter was probed with an oligonucleotide adjacent to the NdeI site (−1092 [the sequence is given above]). In experiments analyzing nucleosome positioning by sequencing resolution, DNA fragments generated after micrococcal nuclease digestion were subjected to primer extension analysis as described for DNase I primer extension footprinting (see above).

The DNase I hypersensitivity (HS) assay was performed by a procedure similar to the nucleosome positioning assay, except that chromatin samples were digested with serial dilutions of DNase I instead of micrococcal nuclease. The DNase I dilutions used for chromatin samples were 150, 250, and 420 μg/ml, and those used for naked DNA samples were 0.056, 0.166, and 0.5 μg/ml.

Southern blot analysis.

Southern blot analyses were performed as described previously (60). Briefly, genomic DNA was prepared from stably transfected Y10/L8057 cells in duplicate and digested with PstI. Digested DNA was resolved on a 1% agarose gel with ethidium bromide staining and transferred to a nylon filter. Plasmid pPF4GH was digested with PstI to generate the hybridization probe. Signals from the hybridized blot was analyzed with Digital Science 1D software (Eastman Kodak) and normalized to the ethidium bromide staining intensity from the gel.

RESULTS

Ets-1 binds to megakaryocytic promoters in vivo.

Due to the rarity of megakaryocytes in the bone marrow, megakaryocytic cell lines that display lineage-specific gene regulation are often used. Y10/L8057 is a murine megakaryocytic cell line that was subcloned from L8057 megakaryocytic cells (18, 58). Y10/L8057 cells express most megakaryocyte-specific genes examined, including PF4 and GPIIb. It was shown recently that these cells regulate the GPIIb gene similarly to primary megakaryocytes, further validating the use of Y10/L8057 cells for studying megakaryocyte transcriptional regulation (52).

Overexpression of Ets-1 positively regulates megakaryocyte-specific genes (12, 29). However, it has never been examined whether this is associated with direct binding of Ets-1 to the endogenous promoters of these genes in megakaryocytes. Using chromatin immunoprecipitation, we found that Ets-1 was bound to proximal regions of both the PF4 (primers spanning −210 to −5 bp) and the GPIIb (primers spanning −237 to +24 bp) promoters in Y10/L8057 cells (Fig. 1A). Ets-1 binding was specific, since negligible binding was detected with a genomic sequence downstream of the GPIIb gene or with control antisera (Fig. 1A). We used conditions that allowed linear PCR amplification (Fig. 1B). This result suggested that Ets-1 is a physiological regulator interacting with megakaryocyte-specific promoters.

FIG. 1.

Ets-1 binds to megakaryocytic promoters in vivo. The binding of endogenous Ets-1 to genomic sites in living Y10/L8057 cells was analyzed using the chromatin immunoprecipitation assay. (A) Y10/L8057 cells were fixed and sonicated. Genomic fragments were either saved as input material or subjected to immunoprecipitation with 1 μg of anti-Ets-1 antibody (Ab) or 1 μg of normal IgG. The genomic fragments recovered from input material or enriched by immunoprecipitation were subjected to semiquantitative PCR amplification with specific primers corresponding to the PF4 (spanning bp −210 to −5) and GPIIb (spanning bp −237 to +24) promoters, as well as to a genomic sequence downstream of the GPIIb gene (GD). (B) Chromatin immunoprecipitation (IP) samples from Y10/L8057 cells immunoprecipitated with either anti-Ets-1 antibody or normal IgG were used in semi-quantitative PCR analyses with primers for the PF4 promoter (as above). Serial dilutions of input material in twofold steps were analyzed to demonstrate that the analyses were within the linear range of PCR amplification. Data shown are representative of two experiments.

Ets-1 binds to consensus Ets sites on the PF4 and GPIIb promoters.

Since DNA is in the form of chromatin in vivo, the binding of Ets-1 to the PF4 and GPIIb promoters suggests that Ets-1 binds to chromatin. However, in the complex cellular environment, Ets-1 binding could be assisted by other factors. Furthermore, the low resolution of chromatin immunoprecipitation hinders further analysis of the binding. We found that the consensus Ets binding sites (starting from 5′ to 3′) at −379, −372, −75, and −50 in the PF4 promoter (relative to the transcriptional start site) display specific binding to Y10/L8057 megakaryocytic nuclear extracts in EMSA (unpublished data), as previously reported for the sites −75 and −379/−372 by using different megakaryocytic cell lines (22, 29). Based on supershift analyses with Ets-1 antibody, it was concluded that Ets-1 binds to these sites (29). As to the GPIIb promoter, Fli-1 was reported to bind to the proximal Ets site (at bp −39) in the mouse GPIIb promoter (52) and recombinant Ets-1 was reported to bind to a proximal Ets site (at −40 bp) in the human GPIIb promoter, as revealed by EMSA (22). To confirm the binding of Ets-1 to the above sites, we used a cell-free chromatin assembly system and analyzed the PF4 and GPIIb promoters for binding to recombinant Ets-1 by DNase I footprinting. We selected this assay also because we wished to determine whether Ets-1 binding to chromatin containing lineage-specific promoters requires help from lineage-specific factors.

Chromatin was assembled in vitro on DNA containing the PF4 promoter, using Drosophila extract (34). The integrity of the assembled chromatin was assayed by micrococcal nuclease digestion. A regular ladder of many bands with a periodicity of about 170 bp indicated the formation of a regular nucleosomal array (Fig. 2A). Moreover, the chromatin is unlikely to have nucleosome-free gaps, since the ladder extends to the size of the linearized plasmid. While this assay reveals that the nucleosome spacing is highly regular within each molecule of plasmid, typically the exact locations of each nucleosome vary on different DNA molecules within a pool of the same plasmid.

FIG. 2.

Sequence-specific binding of recombinant Ets-1 to chromatin containing megakaryocytic promoters. (A) Chromatin was assembled in vitro on the pPF4GH plasmid containing the 1.1-kb PF4 promoter by using Drosophila S-190 embryo extract. Chromatin was digested with serial dilutions of micrococcal nuclease (MNase), analyzed on a 1% agarose gel in comparison to a molecular weight (MW) ladder, and detected with ethidium bromide. (B) Recombinant Ets-1 (rEts-1) was analyzed by SDS-PAGE and detected with Coomassie blue. The sizes of molecular mass markers (MW) are indicated at the side. (C) DNase I- footprinting assays were performed on chromatin containing either the PF4 or the GPIIb promoter. Recombinant Ets-1 was used at a final concentration of 400 nM. The protected areas are indicated with bars and labeled in relation to the transcriptional start site. Data shown are representative of three experiments. (D) Sequences of the PF4 and GPIIb promoters, numbered relative to the transcriptional start sites. The DNase I-protected Ets-1 binding sites are underlined.

To study Ets-1 binding to chromatin, we generated recombinant Ets-1 protein. The preparation was assayed by SDS-PAGE and Coomassie blue staining and was shown to be more than 95% pure (Fig. 2B). Ets-1 protected three regions of PF4 chromatin from DNase I digestion (Fig. 2C). The first region spanned from −78 to −43, containing two consensus Ets sites at −75 and −50, respectively (referred to as the −75/−50 region). The second region covered −387 to −368, containing two consensus Ets sites at −379 and −372 (referred to as the −379/−372 region). The third region spanned from −418 to −405, containing a consensus Ets site at −414. On the GPIIb promoter, Ets-1 protected a region (−33 to −46) in which a functional site was previously reported (Fig. 2C) (5). The sequences of these sites are presented in Fig. 2D. These results confirmed binding of Ets-1 to specific sites in the PF4 and GPIIb promoters. Moreover, it is unlikely that Ets-1 binding to chromatin containing lineage-specific promoters required lineage-specific factors, because only recombinant Ets-1 and Drosophila embryo extract were present in the binding-reaction mixtures. Our data, however, do not exclude the possibility that widely expressed factors assist the binding of Ets-1 to chromatin.

Ets-1 binding to endogenous genes in cultured cells does not require lineage-specific cofactors and is not sufficient to activate transcription.

We reasoned that Ets-1 binding in cells could also be independent of lineage-restricted factors. We examined Ets-1 binding to the PF4 gene in NIH 3T3 cells, a murine fibroblast cell line, to eliminate the influence of megakaryocytic factors. Ets-1 cDNA was used to transfect NIH 3T3 cells; note that our NIH 3T3 line does not express significant amounts of endogenous Ets-1 (Fig. 3A). The NIH 3T3 cells transiently expressed Ets-1, with a range covering the physiological Ets-1 concentration in megakaryocytes (Fig. 3A).

FIG. 3.

Ets-1 can bind to endogenous, lineage-specific genes without lineage-specific cofactors. (A) NIH 3T3 cells were transiently transfected with 0 μg (−), 1 μg (+), 3.3 μg (++), or 10 μg (+++) of Ets-1 cDNA. The cells were harvested 2 days after transfection. Ets-1 levels in transfected cells were analyzed by Western blotting (30 μg/lane) with anti-Ets-1 antibody. Extracts from MGDF-treated Y10/L8057 cells (30 μg/lane) or 1 ng of recombinant Ets-1 (rEts-1) were also analyzed. Recombinant Ets-1 migrated slower due to the presence of a fused multihistidine tag. The blot was stained with Ponceau S after being transferred, and the intensity of a major band was used as the loading control. The transient transfection efficiency was determined to be 30% to 40% by parallel transfections with cytomegalovirus-driving GFP (not shown). (B) Transfected NIH 3T3 cells were subjected to the chromatin immunoprecipitation assay. After immunoprecipitation with either normal IgG or anti-Ets-1 antibody (Ab), enriched genomic fragments were PCR amplified with primers specific for the PF4 promoter (as in Fig. 1) or a downstream region of the GPIIb gene (GD). Samples before immunoprecipitation were used to control the amount of starting material (input). (C) PF4 and GAPDH mRNA levels in the transfected NIH 3T3 cells and in Y10/L8057 cells were analyzed by RT-PCR, using primers specific for the PF4 or GAPDH cDNA. (D) Binding of endogenous Ets-1 to endogenous, lymphocyte-restricted genes in Y10/L8057 megakaryocytes was assayed using chromatin immunoprecipitation. After immunoprecipitation with either normal IgG or anti-Ets-1 antibody, enriched genomic fragments were amplified with specific primers for the CD5 promoter or for the Ig-μE region. Samples before immunoprecipitation were used as input. Data shown are representative of two experiments.

We detected Ets-1 binding on the PF4 promoter in the chromatin immunoprecipitation assays in cells expressing Ets-1 (Fig. 3B). By contrast, control antibodies did not immunoprecipitate the PF4 promoter. Moreover, Ets-1 did not bind to the PF4 promoter in the absence of Ets-1 expression and did not bind to a nonpromoter region downstream of the GPIIb gene, confirming the specificity of the assay. Hence, when Ets-1 is ectopically expressed at physiologic levels in a nonmegakaryocytic cell line, it is able to bind to a megakaryocyte-specific promoter. Thus, Ets-1 could bind endogenous, megakaryocyte-specific genes in the absence of lineage-specific factors.

We asked whether Ets-1 can induce low-level megakaryocyte-specific gene expression without lineage-restricted factors. We measured PF4-specific transcripts by RT-PCR. Control NIH 3T3 cells did not express PF4 (Fig. 3C). Ectopic Ets-1 did not induce PF4 expression, although Ets-1 expression was sufficient for binding to the PF4 gene. Similarly, Ets-1 overexpression did not increase GPIIb mRNA levels in these cells (data not shown). By contrast, megakaryocytes expressing similar amounts of Ets-1 expressed large amounts of PF4, relative to the GAPDH control. This showed that Ets-1 was not sufficient to drive detectable levels of megakaryocytic gene expression.

We asked whether our observations of Ets-1 binding were specific to megakaryocyte genes by measuring Ets-1 binding to lymphocyte-specific genes in the Y10/L8057 megakaryocyte cell line. The activity of the CD5 promoter in T lymphocytes and the Ig-μE in B lymphocytes require Ets-1 (30, 49). We found that endogenous Ets-1 in megakaryocytes was bound to the regulatory regions of these two lymphocyte-specific genes (Fig. 3D). Thus, binding of Ets-1 to chromatin does not require lineage-specific cofactors for B- and T-lymphocytic genes.

Ets-1 binds to chromatin and naked DNA with similar affinity.

Binding of Ets-1 to chromatin may be a result of two mechanisms. It could be an intrinsic property of the Ets-1 protein, which enables it to bind to chromatin by itself, similar to nuclear hormone receptors and HNF-3 (8, 35). Alternatively, Ets-1 binding to megakaryocytic promoters may be facilitated by chromatin-remodeling enzymes or other transcription factors specific to the megakaryocyte lineage or within the Drosophila extract used.

We first measured the binding affinity of the Ets-1 protein to naked DNA and to chromatin through DNase I primer extension footprinting. Compared to naked DNA, chromatin inhibits digestion by DNase I; consequently, chromatin templates are digested with 1,000-fold more DNase I than are naked DNA templates (see Materials and Methods). Thus, the difference in signal between protected and unprotected regions is much smaller with chromatin than with naked DNA.

Transcription factors that need facilitating factors typically display a large difference (50- to 1,000-fold) between chromatin binding affinity and naked DNA binding affinity (see, for example, references 1, 43, and 56). We determined the concentration of Ets-1 that would give half the maximum protection in DNase I footprint reactions. Since the DNA concentration is over 10-fold lower than the binding affinity (see Materials and Methods), such a half-maximum protection concentration gives a good estimation of the apparent dissociation constant (K_D). The affinity of Ets-1 to the PF4 −50, −75 and −380 sites and the GPIIb −35 site are all in the range of 10 to 30 nM on naked DNA templates (Fig. 4A and 5A). The affinity we measure here on natural Ets-1 binding sites with DNase I footprinting is somewhat higher than a previous determination with a synthetic site and EMSA measurements (20). The highest concentrations of Ets-1 indicated in Fig. 4A and 5A were predetermined to produce near-maximum protection (data not shown). The presence of chromatin did not noticeably alter the affinity of Ets-1 to any particular site compared to the affinity on naked DNA templates (Fig. 4A and 5A). While small changes (less than 3-fold) in binding affinity would not be detected by our method, it is important to note we did not find large (10- to 1,000-fold) changes in affinity. Therefore, we concluded that chromatin does not significantly inhibit Ets-1 binding.

FIG. 4.

Binding of Ets-1 to chromatin is ATP independent. (A) The indicated amounts of recombinant Ets-1 (rEts-1) were incubated with naked DNA, chromatin after complete assembly, or chromatin treated with 0.025 U of apyrase per μl after assembly. Ets-1 binding was then assayed by DNase I footprinting. The example shown here is for the PF4 promoter −75/−50 region. Arrows point to bands that showed dose-dependent protection with increasing amount of rEts-1 (bp −83 to −41). Similar results were obtained with the PF4 −379/−372 region and the GPIIb −35 region (not shown). (B) Different conditions for chromatin assembly were tested to demonstrate that apyrase was active under the experimental conditions used. Chromatin assembly reactions were carried out in the presence or absence of ATP or apyrase (0.025 U/μl), as indicated, and DNA was added after chromatin assembly (lane A) or during assembly (lanes D), also as indicated. After assembly, the reactions were tested by the micrococcal nuclease assay for regularly spaced nucleosomes. Molecular markers are shown on the left. Data shown are representative of two experiments.

FIG. 5.

Binding of Ets-1 to chromatin is independent of Sarkosyl-sensitive factors. (A) The affinity of Ets-1 binding to Ets sites was tested on naked DNA, chromatin, or chromatin treated with Sarkosyl and purified by gel filtration after assembly. The indicated amounts of recombinant Ets-1 (rEts-1) were incubated with the indicated templates and assayed by DNase I footprinting. The example shown here is for the GPIIb −35 region. Arrows point to bands that showed dose-dependent protection with increasing amounts of rEts-1 (bp −46 to −33). Similar results were obtained with the PF4 −75/−50 region and the PF4 −379/−372 region (not shown). (B) To demonstrate that chromatin integrity was not affected by the Sarkosyl treatment and subsequent purification, treated chromatin was subjected to micrococcal nuclease digestion, analyzed on a 1% agarose gel, transferred to a nylon filter, and probed with a radiolabeled oligonucleotide corresponding to the GPIIb promoter −35 region. Data shown are representative of two experiments.

Ets-1 binding to chromatin does not require ATP-dependent or Sarkosyl-sensitive chromatin remodeling.

The Drosophila S-190 extract used in the chromatin assembly reactions is known to contain several chromatin-remodeling enzymes (19, 47, 51). We used two methods to examine whether they facilitate Ets-1 binding.

First, we inhibited ATP-dependent remodeling enzymes by depleting ATP with the enzyme apyrase (33). Chromatin was assembled to completion, and then apyrase was incubated with the indicated samples, followed by the addition of Ets-1. Ets-1 affinity for the PF4 −75/−50 region was similar, within our detection limit, for naked DNA and chromatin treated with apyrase (Fig. 4A). Similar results were obtained for the PF4 −379/−372 region and the GPIIb −35 region (data not shown). To verify that apyrase was active within the experimental system, we measured chromatin assembly, an ATP-dependent process, in the presence of apyrase. As expected, apyrase abrogated the regular nucleosomal spacing seen in the presence of ATP. Instead, addition of apyrase before assembly resulted in a smeared pattern similar to that for the digestion reactions without ATP or to that for the reactions with DNA added after the assembly reaction (Fig. 4B).

We also treated chromatin with the ionic detergent Sarkosyl and purified the chromatin by gel filtration to remove or inactivate remodeling enzymes (48). There was little or no change in the affinity of Ets-1 binding to the GPIIb −35 region on Sarkosyl treatment, compared to naked DNA or untreated chromatin (Fig. 5A). Similar results were obtained for the PF4 −75/−50 region and the PF4 −379/−372 region (data not shown). To verify that the Sarkosyl treatment and subsequent purification did not alter the chromatin structure, a micrococcal nuclease assay was performed, yielding a regular DNA ladder (Fig. 5B). We conclude that the binding of Ets-1 to chromatin containing these megakaryocyte-specific promoters is independent of Sarkosyl-sensitive and ATP-dependent remodeling enzymes.

Ets-1 induces chromatin remodeling in the presence of a Sarkosyl-sensitive factor.

The previous experiments showed that Ets-1 binding to chromatin did not require energy-dependent remodeling enzymes. However, for some transcription factors, remodeling is an independent event that occurs after transcription factor binding (33). We asked whether Ets-1 could program DNase I HS site formation, which is one indicator of a change in chromatin structure that would allow increased accessibility of DNA to protein. We found that chromatin of the PF4 promoter template with Ets-1 bound had two DNase I HS; these were not present in chromatin without Ets-1 or in naked DNA (Fig. 6A). The stronger DNase I HS site on PF4 was located around −390, between the Ets-1 binding sites at −368 to −387 (binding region −379/−372) and −414. A weaker HS site was located at around −560. This indicated that Ets-1 changed the local chromatin structure in this region to make it more accessible to the nuclease DNase I.

FIG. 6.

Sarkosyl-sensitive DNase I HS site formation is induced by Ets-1 binding to the PF4 promoter. (A) Samples containing the PF4 promoter in the forms of naked DNA (lane N), chromatin (lanes C), or chromatin treated with Sarkosyl (lanes S), were incubated with (+) or without (−) 200 nM recombinant Ets-1 (rEts-1). Samples were digested with serial dilutions of DNase I and assayed by indirect end labeling. Arrows point to regions (around bp −390 and −560) with increased DNase I digestion. A scheme of the PF4 promoter is presented on the left, based on molecular weight markers run on the gel. The arrow and “+1” represent the transcriptional start site. The two thick horizontal bars represent the −75/−50 and −379/−372 Ets-1 sites. The thin horizontal bars are 200 bp apart, starting from the transcriptional start site. (B) Chromatin was assembled on wild-type (Wt) PF4 promoter or PF4 promoter carrying mutations at the −379/−372 Ets-1 binding sites (Mut). DNase I footprinting assays were performed on naked DNA and chromatin with or without incubation with 200 nM of rEts-1. Numbers correspond to positions relative to the transcription start site, as determined from the sequencing ladder run on the same gel. The thick vertical bar indicates areas with increased DNase I digestion. The two ovals represent the −414 (top) and −379/−372 (bottom) Ets-1 binding sites. Data shown are representative of two experiments.

Transcription factors can program changes in chromatin structure; in some cases this occurs through energy-dependent remodeling enzymes, and sometimes it is an innate property of the transcription factor (7). We inactivated Sarkosyl-sensitive remodeling enzymes after assembly and then allowed recombinant Ets-1 to bind. Sarkosyl treatment blocked Ets-1-mediated DNase I HS formation (Fig. 6A). Since Ets-1 binding to chromatin was not inhibited by Sarkosyl treatment (Fig. 5A), this result suggested that one or more chromatin-remodeling enzymes were recruited by Ets-1 binding and changed chromatin structure; moreover, such a factor was not specific to megakaryocytes.

To exactly map the DNase I HS sites, fragments obtained after DNase I digestion and primer extension were resolved on a sequencing gel (similarly performed to the DNase I footprinting experiments). As shown in Fig. 6B, increased digestion with DNase I was detected at the region from −388 to −398 on the chromatin template in the presence of Ets-1. The HS site occurred only in the presence of chromatin, suggesting that it indicated a change in chromatin structure. This location corresponds to the first HS site and is adjacent to the PF4 −379/−372 Ets-1 binding sites. For the weaker HS site located at around −560, however, we were not able to exactly map it by the sequencing gel method and no binding of Ets-1 to putative Ets sites around that region was detected (data not shown). The −390 and −560 HS may form the boundaries of a nucleosome positioned in the presence of Ets-1. The appearance of the strong HS site adjacent to the −379/−372 region suggested that the HS site formation was the result of the binding of Ets-1 to the PF4 −379/−372 region. Indeed, when Ets-1 sites in −379/−372 region were mutated, strong DNase I digestion disappeared from the −390 HS site (Fig. 6B). In contrast, mutation of the −379/−372 Ets-1 region did not abolish Ets-1 binding to the −414 site.

Ets-1 binding results in changes in nucleosome positioning at the DNase I HS regions.

We asked whether binding of Ets-1 to chromatin might result in changes in the regular nucleosomal array or nucleosome positioning on the PF4 promoter. Nucleosomal ladder disruption assay was used to determine whether any changes have occurred to nucleosomes at or adjacent to a particular site of interest. Some chromatin-remodeling events are reflected by disruption of the regular DNA ladder (34), revealed through micrococcal nuclease digestion and probing with oligonucleotides specific to the regions of interest. As shown in Fig. 7A, Ets-1 binding to chromatin did not result in detectable disruption of the local nucleosomal array structure at the PF4 −379 or −50 region, compared to that at a distal region (−1092). This result is consistent with positioned nucleosomes that are regularly spaced (32) or the absence of positioned nucleosomes. This result also showed that the spacing between nucleosomes at the Ets-1 sites of interest is not detectably different from that at a distal region.

FIG. 7.

Ets-1 binding results in changes in nucleosome positioning at the DNase I HS regions. Chromatin containing the PF4 promoter was incubated with or without 200 nM recombinant Ets-1 (rEts-1). The chromatin was digested with serial dilutions of micrococcal nuclease (MNase) in threefold steps. (A) A nucleosome array disruption assay was performed on the digested samples. Probes corresponding to the PF4 −379 and −50 regions, as well a distal region (−1092), were used to probe the transferred blot. (B) Chromatin (lanes C) and naked DNA (lane N) samples were subjected to nucleosome positioning assays, using indirect end labeling. A scheme of the PF4 promoter is presented on the left, based on molecular weight markers run on the gel. The arrow and “+1” represent the transcriptional start site. The two thick horizontal bars represent the −75/−50 and −379/−372 Ets-1 sites. The thin horizontal bars are 200 bp apart, starting from the transcriptional start site. (C) Nucleosome positioning of chromatin in the absence or presence of rEts-1 was assayed using primer extension after micrococcal nuclease digestion. Arrows point to bands with increased digestion in the presence of rEts-1. Numbers correspond to positions relative to the transcription start site, as determined from the sequencing ladder run on the same gel. The thick vertical bar indicates the −390 DNase I HS region. The two ovals represent the −414 (top) and −379/−372 (bottom) Ets-1 binding sites. Data shown are representative of two experiments.

We next examined the nucleosome positioning on the PF4 promoter before and after Ets-1 binding. We first analyzed the positioning by agarose gel-based indirect end labeling. Comparison with naked DNA reveals chromatin-specific changes in nuclease sensitivity; these changes, if present, can be used to map nucleosome positions. The digestion pattern on chromatin was very similar to that of naked DNA (Fig. 7B), suggesting that the nucleosome positioning was random before Ets-1 binding. Considering that the nucleosome repeat length is 170 bp (Fig. 2A), a nucleosome core particle contains 146 bp of DNA (21), and Ets-1 contacts ∼8 bp of DNA (16), only 10% of single Ets-1 sites are fully in linker regions between core particles while paired sites such as in the PF4 promoter-proximal region are too large to be fully accommodated between regularly spaced core particles. This excludes the model that nucleosomes are positioned away from most of these sites, allowing Ets-1 to bind to nucleosome-free regions. Indirect end labeling failed to detect any significant change in the micrococcal nuclease digestion pattern (Fig. 7B). However, the resolution on the agarose gel may not reveal subtle changes.

We next analyzed the fragments after micrococcal nuclease digestion on a sequencing gel by using primer extension (Fig. 7C). With higher resolution, we found increases in micrococcal nuclease digestion in the presence of Ets-1 at the −390 HS region and at a region further upstream (close to −560), which is approximately the site of the weak DNase I HS detected in Fig. 6A. The distance between the two regions with increased micrococcal nuclease digestion is 170 to 180 bp, corresponding to the size of a single nucleosome. This result suggested that the DNase I HS sites are generated by increased positioning of a nucleosome between −390 and −560, creating short regions at the two HS sites that are less well protected by the nucleosome core particle.

Ets-1 binding sites are important for the transcriptional activity of the PF4 promoter.

As described above, we have identified sites on the PF4 promoter that bind to Ets-1, including −50, −75, −379/−372, and −414 (Fig. 2C). The PF4 −75 site was previously reported as a functional Ets site, based on a 50% reduction in promoter activity when mutated (29). We found the PF4 −379/−372 region was important for the formation of a strong DNase I HS site (Fig. 7B). To assess the significance of these Ets-1 binding sites for promoter activity, we mutated the core of the Ets-1 binding sites in the above regions in the PF4 promoter and analyzed the transcriptional activity by reporter gene assays. Reporter constructs containing the rat PF4 promoter linked to the hGH gene were transiently transfected into Y10/L8057 cells. Mutation of the reported PF4 −75 site led to a moderate reduction of promoter activity (by 40%), while mutation of the PF4 −50 site reduced the promoter activity fivefold (Fig. 8A). When both the −50 and −75 sites were mutated, the promoter activity was abolished. Mutation of the −414 site had no effect on promoter activity (Fig. 8A). Transient transfection of the PF4 −379/−372 mutation construct, in which both −379 and −372 sites were mutated, showed a 28% reduction of promoter activity (Fig. 8B).

FIG. 8.

Mutations of Ets-1 binding sites downregulate the transcriptional activity of the PF4 promoter. Single or double mutations (with sites separated by a comma) of Ets-1 binding sites were generated in the PF4 promoter reporter construct pPF4GH. (A) Y10/L8057 cells were transiently transfected with control plasmid (p0GH) with no promoter upstream of hGH or reporter constructs carrying wild-type PF4 promoter (PF4), −50 site mutation (m50), −75 site mutation (m75), −414 site mutation (m414), or −75/−50 site double mutations (m50,75). Cells were harvested 2 days after transfection and assayed for hGH production. hGH data were normalized to β-galactosidase activity for transfection efficiency and were expressed as the percentage of wild-type promoter activity. Data presented are the means of three experiments, with error bars representing standard deviations. (B) Reporter constructs carrying wild-type PF4 promoter or the double mutation (m379,372) were transiently transfected into Y10/L8057 cells. hGH production was assayed and normalized to β-galactosidase activity and expressed as the percentage of wild-type promoter activity. Data for m379,372 represent the mean from six determinations and the error bars represent standard deviation. The wild-type PF4 promoter construct and the m379,372 construct were stably transfected into Y10/L8057 cells. hGH production was assayed and normalized to genomic incorporation rates determined by Southern blot analyses, as described in Materials and Methods. Data for m379,372 represent the mean of three determinations, and error bars represent standard deviation for hGH production.

We reexamined the −379/−372 mutation in the context of chromatin by using stable transfections. This was done because Ets-1 binding to these sites was found to cause chromatin changes, which might impact promoter activity, while transiently transfected plasmids do not form physiologic chromatin and may not reflect effects on chromatin (reviewed in reference 42). To avoid cell clone variations, the pool of stable transfectants was used for reporter assays. To correct for copy number differences, Southern blot analyses were used to normalize the data for reporter readings. As shown in Fig. 8B, stable transfection showed that the −379/−372 mutation caused a 70% downregulation of PF4 promoter activity. Our results indicated that Ets-1 binding sites −75/−50 and −379/−372 play a critical role in PF4 gene regulation.

DISCUSSION

The importance of Ets-1, a widely expressed transcription factor, in the transcriptional regulation of specific genes in different tissues and several hematopoietic lineages has long been recognized (28-30, 40). Although Ets-1 binding to chromatin in vivo in some tissues has been shown (44), the mechanism by which Ets-1 interacts with lineage-specific promoters in the context of chromatin is not well understood. Since chromatin can inhibit binding of transcription factors to DNA (35, 56), the presence of Ets-1 on lineage specific promoters may be the result of recruitment by lineage-specific factors, such as transcription factors or chromatin remodeling enzymes. Ets-1 binding might also be determined by lineage-specific combinations of non-lineage-specific factors. In fact, Ets-1 can bind to ubiquitous transcription factors such as USF-1 (41) and ubiquitous chromatin modifying enzymes such as CBP/p300 (57). We tested these possibilities by examining recombinant Ets-1 binding to chromatin in vitro, using two megakaryocyte-specific gene promoters, PF4 and GPIIb, as well as by chromatin immunoprecipitation assays in vivo. We showed that binding of Ets-1 in vitro can be achieved without any lineage-specific factors. Ectopically expressed Ets-1, at physiological concentrations, displayed binding to PF4 promoter in NIH 3T3 cells. Moreover, endogenous Ets-1 in megakaryocytes showed binding to lymphocytic promoters. These results indicated that Ets-1 binding to lineage-specific regulatory sequences in the context of chromatin does not require lineage-specific factors or lineage-specific combinations of factors.

The in vivo experiments, interestingly, also showed that transcription factor binding and transcriptional activation of target genes could be uncoupled. Ets-1 binds to the PF4 promoter in NIH 3T3 cells (Fig. 3B) where the gene is not active (Fig. 3C). Also, while Ets-1 binds to nonmegakaryocyte promoters, it does not activate them in megakaryocytes (Fig. 3D). The concept of uncoupled transcription and binding should be examined for other factors and promoters in mammalian cells.

The binding of Ets-1 to chromatin, independent of lineage-specific factors, prompted us to examine the binding affinity of Ets-1 to chromatin and to naked DNA. Some transcription factors, such as NFI and Max, have a greatly reduced (more than 20-fold) affinity for chromatin than for DNA (35, 56), while other factors, such as glucocorticoid receptor and HNF-3, are not strongly inhibited (9, 35). We found that Ets-1 could bind chromatin and naked DNA with similar affinity (threefold difference or less). No large difference in affinity was detected for Ets-1 binding to these templates, even when ATP was depleted or when the chromatin was treated with Sarkosyl. This excluded a requirement for ATP-dependent or Sarkosyl-sensitive chromatin-remodeling enzymes. At present, it is not fully understood why some DNA binding proteins recognize their sites in chromatin while others do so with reduced affinity. A recent crystal structure of the DNA binding domain of Ets-1 bound to DNA indicates that Ets-1 contacts one face of the major groove of DNA and projects out from that face (16), while in the nucleosome structure, histone-DNA interactions are through the minor groove (27). These features of the crystal structure are consistent with the ability of Ets-1 to bind nucleosomal DNA (Fig. 4) (43). The Ets-1 crystal structure also reveals DNA contacts that span at least 8 bp. A model of Ets interacting with the PF4 promoter in the context of chromatin is illustrated in Fig. 9A.

FIG. 9.

Models of Ets-1 binding to chromatin and its role in the activation of the PF4 promoter. (A) Ets-1 binds to chromatin and to naked DNA with similar affinity. The binding of Ets-1 to the PF4 promoter (DNA in black) in the context of chromatin is independent of lineage-specific factors, Sarkosyl-sensitive factors, or ATP-dependent chromatin remodeling. The similar affinity of Ets-1 to chromatin or to naked DNA suggests that the binding itself does not need any accessory factors. However, Ets-1 binding itself is not sufficient for full activation of transcription. The dashed arrow indicates the transcriptional start site and an inactive transcriptional state. (B) Ets-1 induces changes in chromatin. Binding of Ets-1 to the −379/−372 site recruits a Sarkosyl-sensitive chromatin-remodeling enzyme (Ch Rm), which increases positioning of an adjacent nucleosome (indicated by darker shading), resulting in increased accessibility of DNA at the DNase I HS site (indicated by two arrows). (C) Ets-1 activates transcription. Ets-1 may recruit basal transcriptional machinery and increase the activity of the PF4 promoter. Alternatively, Ets-1 may help recruit to the DNase I HS site other transcription factors (TF), which may now bind to chromatin and act as enhancers of transcription. The solid arrow indicates an active transcriptional state.

The binding of Ets-1 to chromatin and mononucleosomes has been examined using other in vitro systems. In one report, the authors utilized a Xenopus chromatin assembly system and demonstrated that when recombinant Ets-1 was added prior to chromatin assembly, it could bind to DNA; Ets-1 was not dissociated by the chromatin assembly machinery (31). This suggested that Ets-1 could bind DNA while chromatin is transiently disassembled during processes such as DNA replication or transcription. This report, however, did not examine the binding of Ets-1 after chromatin was assembled on the template. In another study, Steger et al (43) used a 151-bp fragment from the human immunodeficiency virus type 1 long terminal repeat (containing one Ets-1 site) reconstituted into nucleosome core particles by histone octamer transfer and analyzed the affinity of binding to Ets sites by EMSA. They showed that recombinant Ets-1 bound to mononucleosomes without the need for a remodeling enzyme, although at reduced affinity relative to naked DNA. While we do not fully understand why our results differ with respect to binding affinity, we do know of three experimental differences that may have determined the outcome. First, in our experiments, the nucleosome positioning is relatively random while the mononucleosome used was at a fixed position, which makes the Ets-1 binding site appear at a fixed position on the mononucleosome. It is possible that a certain region on the nucleosome is unfavorable for Ets-1 binding. Second, the templates we used naturally contain multiple Ets-1 sites, while the human immunodeficiency virus type 1 long terminal repeat naturally contains a single Ets-1 site; Ets proteins at different sites might interact. Third, we used arrays of nucleosomes, which are more dynamic than mononucleosomes; biophysical experiments from several laboratories have suggested that the structure of mononucleosomes differs from the structure of nucleosomes in arrays (2, 15, 50).

The second major aim of our study has been to better understand the role of Ets-1 in the regulation of PF4 gene expression. This is particularly interesting, since Ets-1 expression is significantly increased during megakaryocytopoiesis in either a primary cell culture or a cell line system (29, 45), and upregulation does not occur during erythropoiesis (45). Erythrocytes and megakaryocytes have common precursors during lineage differentiation and have similar spectra of transcription factors. Unfortunately, no specific transcription factor has been found in megakaryocytes to date. The answer for megakaryocyte-specific gene expression may therefore rely on a combination of factors acting in conjunction and/or on regulated levels or activation of a single transcription factor, which makes Ets-1 an intriguing candidate.

Although important Ets sites have been identified in many megakaryocytic promoters, only a few studies have examined Ets-1, one member of the Ets family of transcription factors, in megakaryocytic gene regulation (12, 22, 29). In these latter studies, Ets-1 was found (by EMSA) to bind to elements in the PF4 promoter. In the case of the GPIIb promoter, recombinant Ets-1 was reported to bind to a proximal Ets site (22). A recent study indicated that the Ets transcription factor Fli-1 binds in vitro and in vivo to a functional site adjacent to the transcriptional start, which mediates GATA-1/FOG-1 synergy on the GPIIb promoter; it was suggested that other Ets family members may also bind to this site (52). Our finding that Ets-1 bound to both the PF4 and GPIIb promoters in vivo (Fig. 1B) highlights Ets-1 as a physiological regulator for megakaryocyte-specific gene expression. We used the PF4 promoter as a model to address two questions: first, which of the Ets-1 binding sites are crucial for the activation of the PF4 promoter, and second, whether Ets-1 binding induces changes in chromatin structure (creating DNase I HS sites) and affects gene expression. We first identified in an in vitro system the binding sites of Ets-1 in the PF4 promoter by footprinting analyses. Some of these Ets-1 sites were previously reported, including the PF4 −75 and −379 sites (29), while others were newly identified in this study, such as the PF4 −50, −372, and −414 sites. Transient transfection of reporter constructs showed that mutation at PF4 −50 site alone yielded a very significant downregulation of promoter activity and, together with mutation at PF4 −75 site, rendered the promoter inactive. These results demonstrate that proximal Ets-1 binding sites in the PF4 promoter are essential for promoter activity.

While investigating whether Ets-1 binding might induce a change in chromatin structure, we observed that Ets-1 binding resulted in the formation of DNase I HS areas in the PF4 promoter (−388 to −398). Since Ets-1 binding to chromatin is independent of accessory factors and since HS site formation is Sarkosyl sensitive, one might deduce that Ets-1 functions not only as a transcriptional activator but also as a recruiter of chromatin-remodeling enzymes. We detected increased positioning signal for a nucleosome located right upstream of the −390 HS site. This suggests that the formation of HS site may be due to increased presence of internucleosomal areas at the region that allows increased accessibility to the HS site. Alternatively, a remodeling enzyme might create the DNase I HS sites by altering the path or conformation of the DNA in the HS regions without changing nucleosome positions. Mutations in the Ets-1 binding site that abolish HS site formation resulted in a mild decrease of reporter activity when transiently transfected and a stronger decrease when stably transfected. This suggests that the distal Ets sites are required for enhanced activation of the promoter. Since transiently transfected plasmids do not appear to form physiologic chromatin (42), activation through these sites is likely to occur through a chromatin-specific mechanism. Recruitment of chromatin-remodeling activities and subsequent chromatin changes may facilitate the binding of other transcription factors, thus allowing further activation of lineage-specific genes. Our observation that Ets-1 is not sufficient for megakaryocytic gene expression in NIH 3T3 cells (Fig. 3C) suggests that other transcription factors are required. A suggested model of Ets-1 role in the activation of the PF4 promoter is illustrated in Fig. 9.

In summary, our study is novel in demonstrating that Ets-1 binding to lineage-specific regulatory sequences does not require lineage-specific factors. We showed that Ets-1 binding to chromatin has similar affinity to binding to naked DNA and does not require ATP-dependent chromatin remodeling or Sarkosyl-sensitive factors. We found that Ets-1 regulates the PF4 promoter through novel Ets-1 binding sites and Ets-1-dependent chromatin changes. These include the sites at −75/−50, which are essential for transcription, and the sites at −379/−372 to which Ets-1 binds and consequently alters the chromatin and enhances promoter activity (all as illustrated in the scheme in Fig. 9). We also conclude that while Ets-1 is required for the expression of megakaryocyte-specific genes, it is not sufficient. Future studies should focus on the identity and properties of transcription factors that partner with Ets-1 to activate megakaryocyte gene transcription.